University of Southern California Dissertations and Theses

The roles of Tnni3k in heart regeneration, cardiac conduction system defects and cardiomyopathy

(USC Thesis Other)

The roles of Tnni3k in heart regeneration, cardiac conduction system defects and cardiomyopathy

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

The Roles of Tnni3k in Heart Regeneration, Cardiac
Conduction System Defects and Cardiomyopathy

By

Peiheng Gan

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Development, Stem Cells and Regenerative Medicine)

May 2020

Copyright 2020 Peiheng Gan
ii

ACKNOWLEDGEMENTS
This dissertation represents my individual Ph.D. work, but it would not be possible
without the support of my mentor, advisors, wife, parents and colleagues.
First of all, I sincerely thank my mentor Dr. Henry M. Sucov, who provided me the
superior research environment, strong support and wise guidance. He gave me every possible
opportunity to improve my science and myself without reservation. His intelligence, enthusiasm
and commitment to science made me realize how a successful scientist should be. I really
appreciate Dr. Sucov for not only being my scientific advisor but also my life navigator. Without
his constant encouragement and support, I could not have been a scientist that I am today. I
would like to thank my committee chair Dr. Ching-Ling Lien for her constructive advices to my
study and strong support to my career development. And I also like to thank my committee
members and advisors: Dr. Jian Xu, Dr. Qilong Ying and Dr. Ram Kumar Subramanyan for their
time and efforts devoted to help me shape my scientific thinking and broaden my view of
science.
I am really thankful for being with all current and former members of Sucov lab. I
especially thank Dr. Michaela Patterson for teaching me experimental techniques hand by hand
when I first joined the lab and encouraging me to think wisely. I also like to thank Dr. Hua Shen
and Dr. Susana Cavallero for their timely advices and helps. I would like to thank our current lab
technician Kristy Wang, who always helps me with my experiments and lab managements. And
I would also like to thank Dr. Hirofumi Watanabe for his assistance in my study. Next, I would
iii

like to thank my collaborators in the Medical University of South Carolina: Dr. Catalin Baicu and
Dr. Rupak D. Mukherjee for their helps with my experiments and advices to my study.
Last but not least, my heartfelt thanks go to my family. I especially thank my wife Lu Lu,
whose love, support and encouragement are my inexhaustible source of energy and courage to
overcome the tough situations and frustrations in my work and my life. In the end, I sincerely
thank my parents: Huajian Gan and Xiaoyang Li, who brought me up and gave me all their love.
I dedicate this dissertation to all of them.

iv

TABLE OF CONTENTS
ACKNOWLEDGEMENTS ii
LIST OF FIGURES AND TABLES vi
ABSTRACT viii
CHAPTER 1: General introduction
Heart regeneration 1
Mononuclear diploid cardiomyocyte 3
Tnni3k in heart physiology 6
Findings and significance 9
References 13
CHAPTER 2: Tnni3k alleles influence ventricular mononuclear diploid
cardiomyocyte frequency
Introduction 20
Results 23
Discussion 32
Materials and Methods 36
References 41
CHAPTER 3: Tnni3k loss-of-function causes concentric left ventricular remodeling
and beta-adrenergic-PKA signaling inhibition
Introduction 44
Results 47
Discussion 57
Materials and Methods 59
References 66
CHAPTER 4: The role of Tnni3k in cardiac conduction system defects
Introduction 78
v

Results 81
Discussion 87
Materials and Methods 89
References 91
CHAPTER 5: Allelic variants between mouse substrains BALB/cJ and
BALB/cByJ influence mononuclear cardiomyocyte composition
and cardiomyocyte nuclear ploidy
Introduction 97
Results 99
Discussion 106
Materials and Methods 112
References 117
CHAPTER 6: Conclusions and perspectives
MNDCM in heart 130
Tnni3k regulates MNDCM frequency 132
Tnni3k LOF causes heart diseases 133
References 136

vi

LIST OF FIGURES AND TABLES
CHAPTER 2: Tnni3k alleles influence ventricular mononuclear diploid
cardiomyocyte frequency
Figure 1. Tnni3k deletion mutants cause nonsense-mediated mRNA decay. 24
Figure 2. The Tnni3k gene in naked mole-rats (H. glaber). 26
Figure 3. Consequences of a kinase-dead Tnni3k allele in mice. 30
Figure 4. Consequences on in vitro kinase activity of common human
Tnni3k kinase domain variants. 31
Figure 5. Tnni3k converges with oxidative stress. 33
CHAPTER 3: Tnni3k loss-of-function causes concentric left ventricular remodeling and
beta-adrenergic-PKA signaling inhibition
Figure 1. Echocardiographic results of adult Tnni3k wildtype and ko mice. 69
Figure 2. Morphological analysis of Tnni3k wildtype and ko mouse hearts. 70
Figure 3. Cardiomyocyte architectural analysis. 71
Figure 4. Cardiomyocyte contractility analysis of Tnni3k wildtype, ko
and kinase-dead mutant (K489R) mice. 72
Figure 5. Cardiomyocyte calcium transient analysis of Tnni3k wildtype,
ko and kinase-dead mutant (K489R) mice. 73
Figure 6. PKA signaling is blunted in Tnni3k ko mice under iso
stimulation. 74
Figure 7. Tnni3k-I685T knock-in mouse heart and cardiomyocyte 76
physiology.
Figure 8. Echocardiographic results of adult Tnni3k-I685T/- mice. 77
CHAPTER 4: The role of Tnni3k in cardiac conduction system defects
Figure 1. Mouse lead II ECG and representative ECG profiles. 93
Table 1. ECG analysis in Tnni3k wildtype and mutant mice. 94
vii

Figure 2. ECG parameters of Tnni3k wildtype, heterozygous and ko mice. 95
CHAPTER 5: Allelic variants between the mouse substrains BALB/cJ and BALB/cByJ
influence mononuclear cardiomyocyte composition and cardiomyocyte
nuclear ploidy
Figure 1. CM nucleation and nuclear ploidy. 120
Figure 2. Identification and evaluation of two BALB/cByJ-specific
X-linked variants. 121
Supplementary Fig. S1. Quantitation of CM nuclear ploidy in parental
strains. 122
Supplementary Fig. S2. Polyploidy variation between BALB/cJ and
BALB/cByJ. 124
Supplementary Fig. S3. Evaluation of bone marrow cell polyploidy. 125
Supplementary Fig. S4. Quantitation of CM nuclear ploidy in F1 mice. 126
Supplementary Fig. S5. BALB/cJ variants found to not be polymorphic
with BALB/cByJ. 128
Supplementary Table 1. Sequences of primers used for gene amplification
and sequencing 129

viii

ABSTRACT
Adult mammalian heart is not generally thought to have regenerative capacity. Our
previous study demonstrated a positive correlation between the mononuclear diploid
cardiomyocyte (MNDCM) frequency and adult mammalian heart regeneration capacity. The
MNDCM frequency is a multigene trait, and very little is known about the underlying genetic
causes. We identified troponin I 3 interacting kinase (Tnni3k) to be a key regulator of the
MNDCM frequency in mouse heart. Tnni3k is a novel cardiac specific gene, while the function of
Tnni3k in heart remains largely unknown. There are evidences showing that Tnni3k is related to
heart diseases. However, no substrate or downstream signaling of Tnni3k has been reported
yet. In this dissertation, I address the role of Tnni3k in three aspects of heart physiology: heart
regeneration, cardiac conduction system defects and cardiomyopathy.
By creating a Tnni3k kinase-dead mutant mouse line, I confirmed that TNNI3K functions
as a kinase in regulating the MNDCM frequency. And I validated that TNNI3K is not functioning
through MAPK pathways in regulating the MNDCM frequency, even though TNNI3K is
characterized as a MAP3K based on sequence. To investigate the roles of Tnni3k in heart
diseases, I performed electrocardiography and echocardiography analysis on Tnni3k knockout
mice. I found that Tnni3k loss-of-function (LOF) directly causes cardiac conduction system
defects (AV-nodal reentrant tachycardia, bundle branch block, etc.) and cardiomyopathy
(concentric left ventricular remodeling). The conduction system defects in Tnni3k knockout
mice are variable and not fully penetrant, which are similar to the arrhythmic phenotypes seen
in human patients carrying TNNI3K hypomorphic mutations. I also demonstrated that the
ix

concentric left ventricular remodeling of Tnni3k knockout mouse hearts are caused by
cardiomyocyte architectural changes. I further demonstrated that Tnni3k LOF caused
cardiomyocyte contractility reduction through blunting beta-adrenergic-PKA-signaling, which is
a pivotal signaling regulating cardiomyocyte and heart physiology. My study revealed a novel
correlation between Tnni3k and PKA signaling. Furthermore, by surveying human TNNI3K
polymorphisms, I identified multiple common TNNI3K mutations in human population which
largely compromise the kinase activity. And I found that the most common human TNNI3K
polymorphism I685T leads to an increased MNDCM frequency, blunted PKA signaling, and
ventricular wall remodeling, which may predispose the heart to severe abnormalities. Besides
Tnni3k-related projects, my study also revealed that the allele difference between two closely-
related inbred mouse strains Balb/c and Balb/cBy underlies the MNCM frequency and ploidy
difference between the two strains.
Taken together, I demonstrated the important roles of Tnni3k in heart, especially in
cardiac conduction system defects and cardiomyopathy, which have strong human relevance
but were largely underestimated. I further identified PKA signaling as a downstream signaling of
Tnni3k, which provides a foundation for investigating more in-depth molecular mechanisms of
Tnni3k functions in heart and may benefit therapeutics to Tnni3k-related heart diseases.

1

CHAPTER 1:
General introduction
Heart regeneration
Heart disease is a leading cause of death in the United States and causes a huge public
health burden annually. Generally, adult mammalian heart is not thought to have regenerative
capacity, and heart injury results in irreversible loss of myocardium combined with fibrotic
scarring and declining cardiac function. Unlike hepatocytes, which are proliferative after liver
injury, most majorities of cardiomyocytes are permanently differentiated and cannot
proliferate. Whether there are cardiac stem cells in adult mammalian heart, which can
contribute to heart regeneration, is one of the most controversial topics in the last few
decades
1, 2
. Huge amount of endeavors have been taken to explore the possibility of promoting
adult mammalian heart regeneration, including both exogenous and endogenous approaches.
The exogenous approaches are based on the idea of adding back the lost contractile
units to injured myocardium. Direct engraftments of embryonic stem cell (ESC) or induced
pluripotent stem cell (iPSC)-derived human cardiomyocytes have been applied to restore
cardiac injuries in rodents
3
, pigs
4
, and non-human primates
5, 6
. Long-term survival of grafted
cardiomyocytes has been observed in host scar tissue, electromechanical coupling partially
occurs, and cardiac function of transplant models has improved
6
. However, ventricular
arrhythmias
6
and teratomas
7
are common side effects of stem cell-derived cardiomyocyte
transplants. Additionally, the stem cell-derived cardiomyocyte engraftment comes with several
2

other concerns, like the amount of engrafted cardiomyocytes should be large, considering the
low survival and coupling rate in the host heart; tissue scaffolds are also needed, which can
assist the loading of transplanted cardiomyocytes. The immunological rejection response is also
a potential problem when using allogeneic transplants
8
. Meanwhile, injection of adult stem
cells (bone marrow mononuclear cells and cardiac mesenchymal cells) to injured heart has been
demonstrated not to produce functional cardiomyocyte, but trigger an acute inflammatory-
based wound-healing response to rejuvenate infarcted region indirectly
9
.
On the other hand, exploring the endogenous regenerative capacity of adult mammalian
heart is another route to achieve heart regeneration, which can be subdivided into two major
categories: inducing cardiomyocyte proliferation directly and changing non-cardiomyocyte
conditions to promote cardiomyocyte proliferation. These approaches are all based on the fact
that adult mammalian cardiomyocytes can proliferate, but in a very low rate
10
, which opened
up possibilities of adult heart regeneration. Over the last few decades, several manipulations
have been shown to successfully trigger cardiomyocyte proliferation and initiate functional
restoration of injured adult rodent heart. For example, the inhibition of Hippo signaling, which
is a conserved pathway regulating cellular proliferation and organ size, can induce
cardiomyocyte cell-cycle reentry and proliferation after adult mouse heart injury
11
. Recently, by
manipulating cell cycle regulators (cyclin-dependent kinase 1 (CDK1), CDK4, cyclin B1, and cyclin
D1), researchers can efficiently induce cardiomyocyte division and functional recovery in adult
mouse heart after injury
12
. Meanwhile, researchers have also achieved adult rodent
cardiomyocyte proliferation by manipulating non-cardiomyocyte conditions, such as
extracellular matrix (ECM). They identified agrin, a component of neonatal ECM to be an
3

essential factor to induce adult cardiomyocyte proliferation in vivo, which functions through
Hippo-Yap signaling
13
.
Despite the various approaches shown to achieve adult mammalian cardiomyocyte
proliferation, the source of proliferative cardiomyocyte is still unclear. Recent studies have
revealed a small population of proliferative cardiomyocytes in adult mouse heart, which tend to
be hypoxic and mononuclear diploid. However, very little is known about this subpopulation of
cardiomyocyte
14
.
Mononuclear diploid cardiomyocyte
In mammals, almost all adult ventricular cardiomyocytes are polyploid (they have more
than two complete sets of chromosomes). Cardiomyocyte polyploidy arises by endoreplication,
in which the diploid genome is replicated in the S-phase of the cell cycle but not followed by
karyokinesis or cytokinesis, and then the cardiomyocytes become polyploid. Cardiomyocytes in
different mammalian species have different manners and timings of becoming polyploid
15
. In
human, cardiomyocytes are typically mononuclear tetraploid (1X4n), while most majorities of
adult mouse cardiomyocytes are binucleated (2X2n). In embryonic stage, all mouse
cardiomyocytes are mononuclear diploid (1X2n) and highly proliferative. By the time of birth,
cardiomyocyte proliferation rate drops down to minimal levels. Most majority of
cardiomyocytes remain to be 1X2n before postnatal day 4 (P4), and then there is a short period
of reactivation of cardiomyocyte DNA synthesis, which peaks around P4 and is accompanied by
cardiomyocyte binucleation. By P10, most majorities of cardiomyocytes have gone through at
least one cycle of endoreplication and become polyploid, while only a small percentage of
4

cardiomyocytes remain to be 1X2n after P10
16
. The timing of cardiomyocyte binucleation in
mouse is correlated with the neonatal heart regeneration time window
15
. In the first few days
after birth (by P7), neonatal mouse heart can fully regenerate after injury by reactivation of
cardiomyocyte proliferation
17
. However, the regeneration capacity is lost after the first
postnatal week. Hence, the neonatal mouse heart is regenerative when most cardiomyocytes
are still 1X2n, but the regeneration capacity is lost synchronously with cardiomyocyte
binucleation. Interestingly, newt and zebrafish cardiomyocytes remain to be 1X2n, and they
retain the capacity to fully regenerate heart after injury throughout their life spans
18
. However,
there has not been any direct evidence showing that the 1X2n cardiomyocytes (MNDCM)
proliferate and contribute to adult mammalian heart regeneration, which is mainly because of
the lacking of a MNDCM-specific marker to perform linage tracing in heart regeneration
studies
15
.
Our previous study demonstrated that the frequency of MNDCM in adult mouse heart is
highly variable across inbred mouse strains and ranges from 2% to 10%. We showed that the
mouse strains with high level of MNDCM had better cardiac functional recovery and higher
cardiomyocyte proliferation rate after adult heart injury. This result suggests that a high level of
MNDCM can predict a better outcome after adult heart injury in mouse
19
. Another study from a
different lab supported this finding from a different stand point. By overexpressing a dominant
negative version of cytokinesis component Ect2 in zebrafish heart, researchers had created a
transgenic zebrafish model with high degree of cardiomyocyte polyploidy, which then had been
found to lose cardiac regeneration capacity
20
. Their study demonstrated that the polyploidy
state of cardiomyocyte is a barrier to heart regeneration. Conceptually, these results broke the
5

dogma that the adult mammalian heart is nonregenerative and replaced it with a novel
paradigm that regeneration conditionally occurs based primarily on the degree of the MNDCM.
In all mammals, cardiomyocyte polyploidy is initiated after a period of cell cycle arrest,
which indicates that this is an induced process
15
. Besides genetic factors, several other inducers
of cardiomyocyte polyploidization have been demonstrated. Thyroid hormone is one important
endogenous trigger of CM polyploidization. In sheep
21, 22
and in mice
23
, thyroid hormone levels
rise coincident with the prenatal or neonatal onset of polyploidy, respectively. Thyroid
hormone treatment can prematurely activate polyploidization and interference in thyroid
hormone signaling delays and reduces the extent of polyploidization. One recent study has
surveyed the cardiomyocyte polyploidy status of more than 30 mammalian species, and they
found that the acquisition of endothermy causes the dramatical reduction of MNDCM level in
hearts, which is regulated by the change of thyroid hormone level
23
. Additionally, postnatal
oxidative stress has emerged as another important inducer of cardiomyocyte polyploidization.
A compelling case has been made that an increase in reactive oxygen species (ROS) after birth is
a cause of cell cycle arrest in mice
24
. After birth, with the establishment of breath cycle, blood
oxygen saturation is substantially increased. Exposure to high blood oxygen level and high
contractile workload, cardiomyocytes generate more ROS. Excessive amount of ROS can cause
DNA damage in cardiomyocytes, which in turn causes cell cycle arrest. Besides thyroid hormone
and oxidative stress changes, the events involved in the transition from fetal to postnatal life
might also contribute to the polyploidization, such as the loss of exposure to placental or
maternal hormones, increased neuronal activity of the sympathetic and parasympathetic
systems (on the heart and on peripheral vasculature), and a substantial anatomical
6

reorganization of the circulatory system (closure of the atrial septum, closure of the ductus
arteriosus, and increased pulmonary blood flow) that is associated with considerable
hemodynamic changes in blood pressure and heart wall strain. How these circumstances might
influence the process of polyploidization is unknown
15
.
Although the MNDCM is conceived to be a primary source of endogenous
cardiomyocyte proliferation in adult mammalian heart, the frequency of this group of
cardiomyocytes is extremely low and highly variable. By surveying 120 naturally occurring
inbred mouse strains for their MNDCM frequency, we found that the change of frequency is
continuous
19
. This observation implicated that multiple genes are involved in regulating this
trait. We performed genome-wide association study (GWAS) to search for genetic variants
influencing MNDCM frequency across those inbred mouse strains, and identified Tnni3k as one
of the regulatory genes
19
. However, the complete genetics of cardiomyocyte polyploidization
remains largely elusive.
Tnni3k in heart physiology
Tnni3k (troponin I 3 interacting kinase) is a cardiac-specific gene, and it was first
identified as a binding partner of cardiac troponin I by yeast two-hybrid screening. TNNI3K is
characterized as a kinase based on the protein structure: seven N-terminal ankyrin repeats,
kinase domain and a Ser-rich region on the C-terminal
25
. TNNI3K has been validated to have
serine and threonine dual kinase activity by autophosphorylation assay, and the C-terminal Ser-
rich domain can regulate the kinase activity
26
. The N-terminal ankyrin repeats function as a
protein-protein interacting domain, which implicated a potential protein binding role of TNNI3K
7

besides its kinase role. Researchers have performed a binding protein screening assay on
TNNI3K to identify its potential substrates or regulators, and found that antioxidant protein 1
(AOP-1) is a binding partner and regulator of TNNI3K
27
. However, no kinase substrate of TNNI3K
has been identified yet. It is controversial whether cardiac troponin I is the direct substrate of
TNNI3K or not, although they are binding partners to each other.
In the last two decades since Tnni3k was first identified, scientists have been
investigating its role in heart. So far there are three aspects of Tnni3k function have been
demonstrated: cardiomyopathy, cardiac conduction system defects and cardiomyocyte
maturation. Firstly, it has been reported that Tnni3k expression level was significantly elevated
in a hypertrophic cardiomyocyte model and overexpression of Tnni3k by adenovirus infection in
cardiomyocyte induced hypertrophy
28
. Another group of scientists have reported that the
overexpression of Tnni3k can accelerate the cardiomyopathy progression in a left-ventricular
pressure overload mouse model
29
. And they further demonstrated that Tnni3k can work
synergistically with Calsequestrin to modify cardiomyopathy progression and cause severe
heart function impairments
30
. A similar Tnni3k-overexpression experiment in a rat model of
cardiac hypertrophy generated by transverse aortic constriction reached a similar conclusion
that Tnni3k promoted cardiomyopathy
31
. Additionally, in a mouse heart ischemia/reperfusion
(I/R) injury model, Tnni3k has been shown to promote the injury by mediating oxidative stress
through MAPK-p38 pathway, while the inhibition of Tnni3k could provide a protective effect in
heart I/R injury
32
. Collectively, all these findings pointed to an adverse role of Tnni3k
overexpression in cardiomyopathy. However, the effects of Tnni3k loss-of-function in normal
8

conditions have not been clarified yet, and the direct targets or downstream signaling of Tnni3k
remain far from clear.
Secondly, Tnni3k has also been reported to involve in cardiac conduction system
defects. So far, four TNNI3K mutations have been reported in patients carrying familial
conduction system defects and dilated cardiomyopathy, in which T538A mutation causes a
severely reduced TNNI3K kinase activity
33
, and G526D mutation makes TNNI3K protein
insoluble or unstable
34
. c333+2T>C mutation results in a premature stop codon in TNNI3K gene
and totally eliminates the expression of protein
35
. All three TNNI3K mutations are loss-of-
function mutations. However, there is a fourth TNNI3K mutation (c.2302G>A) found in three
families carrying supraventricular tachycardias, conduction diseases, and cardiomyopathy,
which is characterized as a gain-of-function mutation with enhanced TNNI3K kinase activity
36
.
These results suggested that the normal activity of TNNI3K is essential to maintain the normal
physiological function of cardiac conduction system, while enhanced, reduced TNNI3K kinase
activity and completely loss of TNNI3K are all detrimental. However, none of these studies
provided a mechanistic explanation of why TNNI3K functional changes caused conduction
system defects. Another study has demonstrated that the expression level of Tnni3k is
correlated with PR interval duration by a genome-wide transcriptional profiling
37
, suggesting
that Tnni3k may regulate the atrio-ventricular conduction so as to influence cardiac conduction
system physiology.
Thirdly, Tnni3k is also correlated with cardiomyocyte maturation. Cardiomyocytes are
differentiated from mesodermal progenitors in early embryonic stages, which are orchestrated
9

by multiple genes and signaling. Tnni3k expression level is low in the embryonic stage, and
increases in an age-related manner
19
. Researchers have found that the overexpression of
Tnni3k in embryonic stem cells (ESCs) can promote the expression of mature cardiomyocyte
makers, which implicated that Tnni3k may regulate cardiomyocyte maturation in embryonic
stages
38
. After birth, cardiomyocytes undergo the polyploidization as introduced in the previous
topic. In our previous study, we have used natural genetic variations across inbred mouse
strains to seek genes that influence the polyploidization process and 1x2n level of
cardiomyocytes. We identified Tnni3k as one of the regulators, and the elimination of Tnni3k
caused a 2.5-fold increase in MNDCM frequency in adult mouse heart
19
. Tnni3k does not seem
to impact the cell cycle entry of cardiomyocyte, but interrupt the cytokinesis in mitosis, which
results in binucleated cardiomyocytes. The underlying mechanisms of how Tnni3k functions in
this role is not known yet, an interesting speculation is that it functions by mediating oxidative
stress (ROS level) or oxidative stress response, as Tnni3k is implicated in this pathway in the
adult heart I/R injury
32
.
Findings and significance
In this dissertation, I systematically investigated the functions of this novel cardiac-
specific gene Tnni3k in heart regeneration, cardiac conduction system defects, and
cardiomyopathy. In Chapter 2
39
, I clarified the kinase role of TNNI3K in regulating MNDCM
frequency by creating a Tnni3k kinase-dead mutant (K489R) knock-in mouse. I identified
multiple common TNNI3K polymorphisms in human which largely compromise its kinase
activity. Those hypomorphic mutations may not only influence MNDCM frequency but also
10

causes other heart diseases, such cardiac conduction system defects, cardiomyopathies.
Additionally, I demonstrated that Tnni3k is not functioning through the three arms of MPAK
pathways (p38, ERK1/2, JNK) in regulating MNDCM frequency, although Tnni3k is always
conceived as a member of MAPK family because of its gene sequence. Furthermore, I found
that the MNDCM frequency was also increased in two newly engineered mouse lines carrying
Tnni3k 4-base-pair (bp) or 8-bp deletion mutations. In another rodent species, naked mole rat
(NMR), we observed a high level of MNCM (17%). Interestingly, Tnni3k in two NMR species has
independently devolved into pseudogene by multiple mutations, suggesting that Tnni3k may
play a similar role in regulating MNCM frequency across species.
To understand the how Tnni3k influences cardiomyopathy, in Chapter 3, I found that
Tnni3k deletion causes concentric left ventricular remodeling of mouse heart. Although the
Tnni3k knock-out mouse hearts display reduced ejection fraction, which is an important index
reflecting cardiac systolic function, the overall heart function is still preserved. I further
demonstrated that the morphological change of heart is age-related and caused primarily by
the reduced aspect ratio of cardiomyocyte. To understand the cause of the reduced
cardiomyocyte aspect ratio, I measured the cardiomyocyte contractility and calcium fluctuation.
I found that Tnni3k deletion doesn’t affect cardiomyocyte basal contractility and calcium
fluctuation. When stimulated by isoproterenol (iso), which is a commonly used beta-adrenergic
receptor agonist, Tnni3k deletion mutant cardiomyocytes display reduced increase of
contractility and calcium amplitude compared with wildtype. This finding suggests that the
reduced aspect ratio of Tnni3k deletion mutant cardiomyocyte might be an adaptive change to
contractility loss under stress. I also demonstrated that the blunted PKA targets activation is the
11

cause of reduced contractility and calcium fluctuation under iso stimulation. Having confirmed
that Tnni3k functions as a kinase in all these scenarios by using Tnni3k-K489R cardiomyocyte, I
investigated the influence of Tnni3k-I685T mutation. TNNI3K-I685T is the most common human
TNNI3K mutation, which largely compromises kinase activity as demonstrated in chapter 2, but
has not been studied yet. We created a Tnni3k-I685T knock-in mouse line by CRISPR-Cas9
technology, and found that the knock-in mutant mice also have elevated MNCM frequency, and
blunted PKA signaling, and thickened left ventricular myocardium as seen in knock-out mice.
But the heart function of Tnni3k-I685T knock-in mouse is not obviously changed. In sum, in
chapter 3, I clarified that Tnni3k loss-of-function causes concentric left ventricular remodeling
and identified PKA signaling as a downstream signaling of TNNI3K.
In Chapter 4, I investigated the role of Tnni3k in cardiac conduction system defects,
since several TNNI3K mutations have been reported in families carrying arrhythmias
33, 34, 35, 36
. I
found that Tnni3k deletion caused variable manifestations of conduction system defects in
mice, including bundle branch block, atrioventricular nodal reentrant tachycardia, premature
ventricular contraction, premature atrial contraction, which resemble the arrhythmic
phenotypes seen in human patients. This result validated that Tnni3k loss-of-function is the
direct cause of conduction system defects. Moreover, by analyzing Tnni3k-K489R mice, I
confirmed that Tnni3k functions as a kinase in cardiac conduction system. Meanwhile, by using
Tnni3k deletion mutant mice as described in chapter 3, I found that the none-sense mediated
decay of Tnni3k mRNA can prevent the manifestations of arrhythmias even without Tnni3k
expression. This result explained why no arrhythmic phenotypes have been reported in mouse
strains carrying natural Tnni3k frameshift mutations and don’t express TNNI3K protein.
12

Moreover, I found that the human mutant Tnni3k-I685T mice don’t have obvious conduction
system defects.
Besides Tnni3k-related studies, I am also investigating other genetics factors influencing
MNDCM frequency. As demonstrated in our previous study, the MNDCM frequency is a
multigene trait
19
. Tnni3k is one of the regulatory genes, but not the only one. In chapter 5, I
found that the two closely-related mouse strains Balb/c and Balb/cBy have significant
differences in MNCM frequency and ploidy level. Balb/c mice have lower MNCM frequency but
high ploidy level, while Balb/cBy mice have higher MNCM frequency but low ploidy level, thus
Balb/c and Balb/cBy mice have relatively the same level of MNDCM frequency. Cross-mating
experiment revealed that the MNCM frequency difference is because of an X chromosome-
linked recessive allele, but the ploidy difference is an autosomal difference. We further
explored the X chromosomal polymorphisms between the two strains by sequencing, and
identified two candidate polymorphisms on Gdi1 and Irs4 gene. However, neither of the
polymorphisms was validated to be responsible for the MNCM difference between Balb/c and
Balb/cBy. In sum, this study demonstrated an interesting phenomenon that MNCM frequency
and ploidy level are independently regulated by different gene alleles.
In sum, my Ph.D. study covered a broad scope of heart biology: regeneration, cardiac
conduction system defects and cardiomyopathy. And I am particularly focused on Tnni3k in the
three aspects of heart biology. My study confirmed the significant role of Tnni3k in heart, which
was previously underestimated, and provided a foundation to explore more in-depth aspects of
Tnni3k function in heart.
13

References
1. Eschenhagen T, Bolli R, Braun T, et al. Cardiomyocyte Regeneration: A Consensus
Statement. Circulation. 2017;136(7):680–686. doi:10.1161/CIRCULATIONAHA.117.029343
2. Beltrami AP, Barlucchi L, Torella D, et al. Adult cardiac stem cells are multipotent and support
myocardial regeneration. Cell. 2003;114(6):763–776. doi:10.1016/s0092-8674(03)00687-1
3. Weinberger F, Breckwoldt K, Pecha S, et al. Cardiac repair in guinea pigs with human
engineered heart tissue from induced pluripotent stem cells. Sci Transl Med.
2016;8(363):363ra148. doi:10.1126/scitranslmed.aaf8781
4. Ye L, Chang YH, Xiong Q, et al. Cardiac repair in a porcine model of acute myocardial
infarction with human induced pluripotent stem cell-derived cardiovascular cells [published
correction appears in Cell Stem Cell. 2015 Jan 8;16(1):102]. Cell Stem Cell. 2014;15(6):750–761.
doi:10.1016/j.stem.2014.11.009
5. Liu YW, Chen B, Yang X, et al. Human embryonic stem cell-derived cardiomyocytes restore
function in infarcted hearts of non-human primates [published correction appears in Nat
Biotechnol. 2018 Sep 6;36(9):899]. Nat Biotechnol. 2018;36(7):597–605. doi:10.1038/nbt.4162
6. Chong JJ, Yang X, Don CW, et al. Human embryonic-stem-cell-derived cardiomyocytes
regenerate non-human primate hearts. Nature. 2014;510(7504):273–277.
doi:10.1038/nature13233
7. Lee AS, Tang C, Rao MS, Weissman IL, Wu JC. Tumorigenicity as a clinical hurdle for
pluripotent stem cell therapies. Nat Med. 2013;19(8):998–1004. doi:10.1038/nm.3267
14

8. Tzahor E, Poss KD. Cardiac regeneration strategies: Staying young at heart. Science.
2017;356(6342):1035–1039. doi:10.1126/science.aam5894
9. Vagnozzi RJ, Maillet M, Sargent MA, et al. An acute immune response underlies the benefit of
cardiac stem cell therapy. Nature. 2020;577(7790):405–409. doi:10.1038/s41586-019-1802-2
10. Senyo SE, Steinhauser ML, Pizzimenti CL, et al. Mammalian heart renewal by pre-existing
cardiomyocytes. Nature. 2013;493(7432):433–436. doi:10.1038/nature11682
11. Heallen T, Zhang M, Wang J, et al. Hippo pathway inhibits Wnt signaling to restrain
cardiomyocyte proliferation and heart size.Science. 2011;332(6028):458–461.
doi:10.1126/science.1199010
12. Mohamed TMA, Ang YS, Radzinsky E, et al. Regulation of Cell Cycle to Stimulate Adult
Cardiomyocyte Proliferation and Cardiac Regeneration. Cell. 2018;173(1):104–116.e12.
doi:10.1016/j.cell.2018.02.014
13. Morikawa Y, Heallen T, Leach J, Xiao Y, Martin JF. Dystrophin-glycoprotein complex
sequesters Yap to inhibit cardiomyocyte proliferation.Nature. 2017;547(7662):227–231.
doi:10.1038/nature22979
14. Kimura W, Xiao F, Canseco DC, et al. Hypoxia fate mapping identifies cycling cardiomyocytes
in the adult heart [published correction appears in Nature. 2016 Apr 14;532(7598):268]. Nature.
2015;523(7559):226–230. doi:10.1038/nature14582
15. Gan P, Patterson M, Sucov HM. Cardiomyocyte Polyploidy and Implications for Heart
Regeneration. Annu Rev Physiol. 2020;82:45–61. doi:10.1146/annurev-physiol-021119-034618
15

16. Soonpaa MH, Kim KK, Pajak L, Franklin M, Field LJ. Cardiomyocyte DNA synthesis and
binucleation during murine development. Am J Physiol. 1996;271(5 Pt 2):H2183–H2189.
doi:10.1152/ajpheart.1996.271.5.H2183
17. Porrello ER, Mahmoud AI, Simpson E, et al. Transient regenerative potential of the neonatal
mouse heart. Science. 2011;331(6020):1078–1080. doi:10.1126/science.1200708
18. Poss KD, Wilson LG, Keating MT. Heart regeneration in zebrafish. Science.
2002;298(5601):2188–2190. doi:10.1126/science.1077857
19. Patterson M, Barske L, Van Handel B, et al. Frequency of mononuclear diploid
cardiomyocytes underlies natural variation in heart regeneration. Nat Genet. 2017;49(9):1346–
1353. doi:10.1038/ng.3929
20. González-Rosa JM, Sharpe M, Field D, et al. Myocardial Polyploidization Creates a Barrier to
Heart Regeneration in Zebrafish. Dev Cell. 2018;44(4):433–446.e7.
doi:10.1016/j.devcel.2018.01.021
21. Chattergoon NN, Giraud GD, Louey S, Stork P, Fowden AL, Thornburg KL. Thyroid hormone
drives fetal cardiomyocyte maturation. FASEB J. 2012;26(1):397–408. doi:10.1096/fj.10-179895
22. Chattergoon NN, Louey S, Stork P, Giraud GD, Thornburg KL. Mid-gestation ovine
cardiomyocytes are vulnerable to mitotic suppression by thyroid hormone. Reprod Sci.
2012;19(6):642–649. doi:10.1177/1933719111432860
16

23. Hirose K, Payumo AY, Cutie S, et al. Evidence for hormonal control of heart regenerative
capacity during endothermy acquisition. Science. 2019;364(6436):184–188.
doi:10.1126/science.aar2038
24. Puente BN, Kimura W, Muralidhar SA, et al. The oxygen-rich postnatal environment induces
cardiomyocyte cell-cycle arrest through DNA damage response [published correction appears in
Cell. 2014 May 22;157(5):1243]. Cell. 2014;157(3):565–579. doi:10.1016/j.cell.2014.03.032
25. Zhao Y, Meng XM, Wei YJ, et al. Cloning and characterization of a novel cardiac-specific
kinase that interacts specifically with cardiac troponin I. J Mol Med (Berl). 2003;81(5):297–304.
doi:10.1007/s00109-003-0427-x
26. Zhao Y, Meng XM, Wei YJ, et al. Cloning and characterization of a novel cardiac-specific
kinase that interacts specifically with cardiac troponin I. J Mol Med (Berl). 2003;81(5):297–304.
doi:10.1007/s00109-003-0427-x
27. Feng Y, Liu DQ, Wang Z, et al. AOP-1 interacts with cardiac-specific protein kinase TNNI3K
and down-regulates its kinase activity. Biochemistry (Mosc). 2007;72(11):1199–1204.
doi:10.1134/s0006297907110053
28. Wang L, Wang H, Ye J, et al. Adenovirus-mediated overexpression of cardiac troponin I-
interacting kinase promotes cardiomyocyte hypertrophy. Clin Exp Pharmacol Physiol.
2011;38(4):278–284. doi:10.1111/j.1440-1681.2011.05499.x
17

29. Tang H, Xiao K, Mao L, Rockman HA, Marchuk DA. Overexpression of TNNI3K, a cardiac-
specific MAPKKK, promotes cardiac dysfunction. J Mol Cell Cardiol. 2013;54:101–111.
doi:10.1016/j.yjmcc.2012.10.004
30. Wheeler FC, Tang H, Marks OA, et al. Tnni3k modifies disease progression in murine models
of cardiomyopathy. PLoS Genet. 2009;5(9):e1000647. doi:10.1371/journal.pgen.1000647
31. Wang X, Wang J, Su M, et al. TNNI3K, a cardiac-specific kinase, promotes physiological
cardiac hypertrophy in transgenic mice. PLoS One. 2013;8(3):e58570.
doi:10.1371/journal.pone.0058570
32. Vagnozzi RJ, Gatto GJ Jr, Kallander LS, et al. Inhibition of the cardiomyocyte-specific kinase
TNNI3K limits oxidative stress, injury, and adverse remodeling in the ischemic heart. Sci Transl
Med. 2013;5(207):207ra141. doi:10.1126/scitranslmed.3006479
33. Xi Y, Honeywell C, Zhang D, et al. Whole exome sequencing identifies the TNNI3K gene as a
cause of familial conduction system disease and congenital junctional ectopic tachycardia. Int J
Cardiol. 2015;185:114–116. doi:10.1016/j.ijcard.2015.03.130
34. Theis JL, Zimmermann MT, Larsen BT, et al. TNNI3K mutation in familial syndrome of
conduction system disease, atrial tachyarrhythmia and dilated cardiomyopathy.Hum Mol
Genet. 2014;23(21):5793–5804. doi:10.1093/hmg/ddu297
35. Fan LL, Huang H, Jin JY, et al. Whole exome sequencing identifies a novel mutation
(c.333 + 2T > C) of TNNI3K in a Chinese family with dilated cardiomyopathy and cardiac
conduction disease. Gene. 2018;648:63–67. doi:10.1016/j.gene.2018.01.055
18

36. Podliesna S, Delanne J, Miller L, et al. Supraventricular tachycardias, conduction disease,
and cardiomyopathy in 3 families with the same rare variant in TNNI3K (p.Glu768Lys).Heart
Rhythm. 2019;16(1):98–105. doi:10.1016/j.hrthm.2018.07.015
37. Lodder EM, Scicluna BP, Milano A, et al. Dissection of a quantitative trait locus for PR
interval duration identifies Tnni3k as a novel modulator of cardiac conduction. PLoS Genet.
2012;8(12):e1003113. doi:10.1371/journal.pgen.1003113
38. Wang Y, Wang SQ, Wang LP, et al. Overexpression of Cardiac-Specific Kinase TNNI3K
Promotes Mouse Embryonic Stem Cells Differentiation into Cardiomyocytes. Cell Physiol
Biochem. 2017;41(1):381–398. doi:10.1159/000456400
39. Gan P, Patterson M, Velasquez A, et al. Tnni3k alleles influence ventricular mononuclear
diploid cardiomyocyte frequency. PLoS Genet. 2019;15(10):e1008354. Published 2019 Oct 7.
doi:10.1371/journal.pgen.1008354

19

CHAPTER 2
Tnni3k alleles influence ventricular mononuclear diploid
cardiomyocyte frequency

RESEARCH ARTICLE
Tnni3k alleles influence ventricular
mononuclear diploid cardiomyocyte
frequency
Peiheng Gan
ID
1,2,3
, Michaela Patterson
ID
4
, Alexa Velasquez
3
, Kristy Wang
ID
1,2
, Di Tian
5
,
Jolene J. Windle ID
6
, Ge Tao ID
1
, Daniel P. Judge
2
, Takako Makita ID
7
, Thomas J. Park
8
,
Henry M. Sucov ID
1,2
*
1 Department of Regenerative Medicine and Cell Biology, Medical University of South Carolina, Charleston,
South Carolina, United States of America, 2 Department of Medicine, Division of Cardiology, Medical
University of South Carolina, Charleston, South Carolina, United States of America, 3 Department of Stem
Cell Biology and Regenerative Medicine, University of Southern California Keck School of Medicine, Los
Angeles, California, United States of America, 4 Department of Cell Biology, Neurobiology and Anatomy, and
Cardiovascular Center, Medical College of Wisconsin, Milwaukee, Wisconsin, United States of America,
5 Department of Pathology and Laboratory Medicine, Tulane University School of Medicine, New Orleans,
Louisiana, United States of America, 6 Department of Human and Molecular Genetics, Virginia
Commonwealth University, Richmond, Virgina, United States of America, 7 Darby Children’s Research
Institute, Department of Pediatrics, Medical University of South Carolina, Charleston, South Carolina, United
States of America, 8 Laboratory of Integrative Neuroscience, Department of Biological Sciences, University
of Illinois at Chicago, Chicago, Illinois, United States of America
* sucov@musc.edu
Abstract
Recent evidence implicates mononuclear diploid cardiomyocytes as a proliferative and regen-
erative subpopulation of the postnatal heart. The number of these cardiomyocytes is a com-
plex trait showing substantial natural variation among inbred mouse strains based on the
combined influences of multiple polymorphic genes. One gene confirmed to influence this
parameter is the cardiomyocyte-specific kinase Tnni3k. Here, we have studied Tnni3k alleles
across a number of species. Using a newly-generated kinase-dead allele in mice, we show
that Tnni3k function is dependent on its kinase activity. In an in vitro kinase assay, we show
that several common human TNNI3K kinase domain variants substantially compromise
kinase activity, suggesting that TNNI3K may influence human heart regenerative capacity
and potentially also other aspects of human heart disease. We show that two kinase domain
frameshift mutations in mice cause loss-of-function consequences by nonsense-mediated
decay. We further show that the Tnni3k gene in two species of mole-rat has independently
devolved into a pseudogene, presumably associated with the transition of these species to a
low metabolism and hypoxic subterranean life. This may be explained by the observation that
Tnni3k function in mice converges with oxidative stress to regulate mononuclear diploid cardi-
omyocyte frequency. Unlike other studied rodents, naked mole-rats have a surprisingly high
(30%) mononuclear cardiomyocyte level but most of their mononuclear cardiomyocytes are
polyploid; their mononuclear diploid cardiomyocyte level (7%) is within the known range (2–
10%) of inbred mouse strains. Naked mole-rats provide further insight on a recent proposal
that cardiomyocyte polyploidy is associated with evolutionary acquisition of endothermy.
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 1 / 24
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Gan P, PattersonM, Velasquez A, Wang
K, Tian D, Windle JJ, et al. (2019) Tnni3k alleles
influence ventricular mononuclear diploid
cardiomyocyte frequency. PLoS Genet 15(10):
e1008354. https://doi.org/10.1371/journal.
pgen.1008354
Editor: Anthony B. Firulli, Indiana University Purdue
University at Indianapolis, UNITED STATES
Received: June 10, 2019
Accepted:August 7, 2019
Published: October 7, 2019
Copyright: © 2019 Gan et al. This is an open
access article distributed under the terms of the
Creative CommonsAttribution License, which
permits unrestricted use, distribution, and
reproduction in any medium,provided the original
author and source are credited.
Data Availability Statement:All relevant data are
within the manuscript and its Supporting
Information files.
Funding: This work was supported by grants
HL070123 and HL144938 from the National Inst.
of Health to HMS, and NS083265to TM, and by
grant 1655494 from the National Science
Foundationto TJP, and by grants to GT from the
American Heart Assn. (17SDG33400141) and the
National Science Foundation(EPSCoR RII Track-1:
MADE in SC OIA-1655740). PG was supportedby
Author summary
Embryonic cardiomyocytes have one diploid nucleus (like most cells of the body), but
most adult cardiomyocytes are polyploid. Most adult cardiomyocytes are also post-mitotic
and nonregenerative, and as a result, heart injury (such as from a heart attack) is followed
by scarring and impaired function rather than by regeneration. A subset of cardiomyo-
cytes in the adult heart remains mononuclear diploid, and recent evidence indicates that
this subpopulation has proliferative and regenerative capacity. Our previous work in mice
showed that the percentage of this cell population in the adult heart is a complex trait sub-
ject to the combined influence of a number of polymorphic genes. One gene that influ-
ences variation in this trait is a kinase gene known asTnni3k. This study addresses the
consequences of a number ofTnni3k alleles, both newly engineered in mice and naturally
occurring in a number of species, including human and mole-rat, and studied at the phe-
notypic and biochemical level. These results provide insight into inter- and intra-species
variation in the cardiomyocyte composition of the adult heart, and may be relevant to
understanding heart regenerative ability in humans and across other species.
Introduction
The ability of the heart to regenerate after injury is an indication of the proliferative capacity of
its cardiomyocytes (CMs). In all species with hearts, the embryonic heart grows during devel-
opment by CM proliferation, with a rate that peaks and then declines to minimal levels as the
proper number of CMs is reached. Long after the phase of rapid embryonic CM proliferation
is past, zebrafish and newts throughout life are able to regenerate their hearts after injury by
reactivating CM proliferation [1–3]. In the early mammalian neonate, heart injury also induces
robust reactivation of CM proliferation that leads to efficient regeneration, as observed in rat
[4], mouse [5], pig [6, 7], and human [8, 9]. However, this response is mostly lost in mammals
soon after birth (in mice, during the first postnatal week [5]). This has major clinical signifi-
cance. A common cause of human heart injury is atherosclerotic coronary artery occlusion
leading to myocardial ischemia or infarction followed by CM death in the impacted region.
For lack of adequate CM proliferation and regeneration, the aftermath of adult heart injury is
scar formation and permanent loss of myocardial function. Diminished heart function can
progress to heart failure, which is among the largest impacts on U.S. healthcare costs [10].
There is still grossly insufficient insight as to why and how the mammalian heart becomes
nonregenerative in the early postnatal period, and lack of understanding of this process is a
critical barrier to progress in improving human health and adult post-injury outcomes.
CMs become polyploid coincident with the loss of their proliferative ability. In all species,
all CMs are mononuclear and diploid (like most cells of the body) during embryonic develop-
ment. In mammals shortly after birth (during the first postnatal week in mice), there is a final
CM entry into the cell cycle (i.e., S phase DNA replication) but with arrest either before or
after karyokinesis but prior to completion of cytokinesis, such that most CMs at this time
become polyploid. In mice and rats and believed for all rodents, most CMs become binucle-
ated [11–14], whereas in humans there is a greater frequency of mononuclear CMs with tetra-
ploid nuclei [15–17]. CMs that are mononuclear tetraploid or CMs that are binucleated with
two diploid nuclei are both referred to as being polyploid; higher levels of CM polyploidy, both
in number of nuclei per CM and in the ploidy of each CM nucleus, also occur. All of these are
distinguished from mononuclear diploid CMs. CM polyploidization is thought to occur
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 2 / 24
a predoctoral fellowship from the Amer. Heart.
Assn. Services provided by the VCU Massey
Cancer Center Transgenic/Knockout Mouse Core
were supported in part with funding from NIH-NCI
Cancer Center Support Grant P30 CA016059.The
funders had no role in study design, data collection
and analysis, decision to publish, or preparationof
the manuscript.
Competinginterests: The authorshave declared
that no competing interests exist.
through an interruption in the cell cycle machinery that controls karyokinesis and/or cytoki-
nesis, and while there are initial insights [18, 19], relatively little is known of the mechanics of
this process. Only a limited fraction of neonatal CMs complete cytokinesis at this time to form
new mononuclear diploid CMs. Thereafter, cell cycle activity is mostly nonexistent, and the
CM composition (the ratio of mononuclear diploid to polyploid CMs) of the mammalian
heart remains constant through adulthood. In zebrafish and newts, polyploidization does not
occur, and these species retain the capacity for highly efficient CM proliferation and regenera-
tion throughout life. Thus, from a comparison of multiple species and life stages, there is a pre-
cise correlation between the abundance of mononuclear diploid CMs and competence to
support CM proliferation for embryonic heart growth or post-injury heart regeneration.
Two recent studies move the relationship of CM ploidy and regeneration from correlation
to causation. In mouse, the level of ventricular mononuclear diploid CMs in the normal adult
heart is typically measured as approx. 2%. In our work [20], rather than assuming this to be a
fixed feature of the adult heart, we showed that the percentage of these CMs in the adult heart
is surprisingly variable between inbred mouse strains, reaching as high as 10% in some. We
showed that strains with more of these had better regeneration at the functional and cellular
level after adult heart injury. By genome-wide association, we identified one gene, a CM-spe-
cific kinase namedTnni3k, as having a naturally-occurring loss-of-function variant that influ-
ences mononuclear CM percentage, and there are clearly other polymorphic genes that also
contribute to variation in this trait. TheTnni3k gene is functional in C57BL/6J mice and in
many other strains with lower mononuclear CM level, whereas many inbred strains with
higher mononuclear CM content are homozygous for the natural loss-of-function allele. These
strains all appear normal and healthy and have been maintained for decades, showing that the
Tnni3k gene is not required for life. Engineered knockout of this one gene in the C57BL/6J
strain background resulted in the predicted increase in mononuclear diploid CM percentage,
and a consequent improvement in CM cellular regeneration after adult heart injury. Recipro-
cally, transgenic overexpression ofTnni3k in zebrafish increased polyploidy and impaired
adult zebrafish heart regeneration.
A zebrafish study from a different lab [21] reached a similar conclusion. In this analysis,
transient transgenic expression of a dominant negative version of the cytokinesis component
Ect2 caused a high degree of CM polyploidy, and much later (after the dominant negative pro-
tein was no longer present, but CMs were still polyploid), adult fish were consequently not
able to regenerate effectively. The significance of the two studies is in their use of direct experi-
mental approaches, rather than observational correlations, to conclude that mononuclear dip-
loid CMs can be proliferative and regenerative, whereas the polyploid CM state is at least
mostly nonproliferative and nonregenerative. In most mammalian species and in most indi-
viduals within a species, there might be relatively few mononuclear diploid CMs, but in some
cases there could be more, and perhaps many more, of these potentially regenerative CMs.
TNNI3K in humans is 835 amino acids long, with an extended amino-terminal ankyrin
repeat domain comprising the first half of the protein, followed by a kinase domain, and a ser-
ine-rich carboxy-terminal domain. In several human pedigrees, mutations in theTNNI3K
gene have been associated with conduction system disease (of various manifestations) and
dilated cardiomyopathy. One of these is a mutation at the splice donor sequence following
exon 4 (of a total of 25 exons) that presumably terminates the open reading frame at a down-
stream intronic stop codon [22]. The others are point mutations in the kinase domain
(Thr539Ala and Gly526Asp) [23, 24] or in the C-terminal domain (Glu768Lys) [25]. These
pathogenic mutations are rare (i.e., are not found in the ExAC compilation [26] of human
whole exome sequence), and are disease-associated as heterozygous alleles.
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 3 / 24
Other thanTnni3k, no gene in any species has yet been identified to have natural variants
that influence CM polyploidy.Tnni3k therefore provides a unique opportunity to understand
the genetic basis of CM polyploidization in a way that has natural correlates and potentially
also influences human heart disease.
Results
NewTnni3k alleles in C57BL/6J mice subject to nonsense-mediated decay
The natural mouseTnni3k variant [27] that revealed this gene to be relevant to mononuclear
CM frequency in our prior genome-wide association is a splice site mutation at position +9 of
intron 19 (position 154875123 of NC_000069.6). This variant shifts the exon 19 splice donor
by 4nt, thereby creating a frameshift after Gly625 that terminates one amino acid later. By non-
sense-mediated decay, this frameshifted transcript is degraded. This variant allele is therefore a
loss-of-function mutation. The wild-type functional allele is present in C57BL/6J mice.
In the course of making other CRISPR-mediated alleles in C57BL/6J mice, we recovered
two lines in which small deletions, one of 4bp and one of 8bp, were introduced into exon 16
(S1 Fig). For simplicity, these alleles are named here as Δ4 and Δ8. Both result in frameshifts
after Ser532 and in termination downstream after 29 and 16 ectopic codons, respectively, both
well before the end of the wild-type open reading frame in exon 25. We therefore expected that
transcripts derived from these alleles, like the natural variant allele described above, would be
degraded by nonsense-mediated decay. When overexpressed in transfected 293 cells, these
transcripts resulted in only a very low level of truncated proteins (Fig 1A). To negate the possi-
bility that low protein level was a manifestation of proteasomal degradation, we treated trans-
fected 293 cells with the proteasomal inhibitor MG132. Beta-catenin, which is known to be
regulated by proteasomal degradation [28], served as the positive control, and was increased
after MG132 treatment. The level of the wild-type full length Tnni3k protein was not influ-
enced by MG132, indicating that the abundance of this protein is not regulated by proteasomal
degradation. Similarly, the low levels of the truncated Δ4 and Δ8 proteins were not impacted
by MG132. This implicates nonsense-mediated transcript decay or nonproteasomal degrada-
tion as the basis for the low protein levels observed in transfected 293 cells.
Founder mice heterozygous for the deletion alleles were crossed to wild-type C57BL/6J
partners to confirm germline establishment of the allele, and heterozygous F1 mice then
crossed to C57BL/6J-inbredTnni3k homozygous null mice (derived in our previous study
[20]) to generate littermate mice that were used for analysis. By this approach, any off-target
mutation co-introduced by CRISPR manipulation would at most be heterozygous but not
homozygous, and a uniform C57BL/6J background was maintained.Tnni3k Δ4/- and Δ8/-
mice, like null (-/-) mice, were externally normal. Ventricular lysates prepared from adult
hearts demonstrated a substantially reduced level of mRNA (Fig 1B) and no detectable Tnni3k
protein at the expected (62kDa) size range (Fig 1C; note that the antibody used here was tar-
geted to amino acids 187–403, so would be able to detect either truncated protein if present).
In comparison to transfected 293 cells, where a low level of truncated protein was observed
(Fig 1A), the absence of detectable protein in heart tissue may be explained by one or more of
several potential causes: nonsense-mediated decay may be more efficient in cardiomyocytes,
or the lower level of gene expression in vivo might not saturate the decay machinery as may
occur with high expression in transfected 293 cells, or the anti-Flag antibody may be more sen-
sitive to low protein levels in Western blots than the anti-Tnni3k antibody.
Following collagenase digestion and preparation of single cell suspensions, adult hearts
were evaluated for ventricular mononuclear CM content. In our previous analysis of a loss-of-
functionTnni3k null allele in C57BL/6J-inbred mice [20], we observed that the mononuclear
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 4 / 24
Fig 1. Tnni3k deletion mutants cause nonsense-mediated mRNA decay. A. Western blot of protein extracted 293 cells transfected
with constructs to express wild-type full-length mouse Tnni3k (WT), the 4bp or 8bp deletion mutant variants ( Δ4 and Δ8), and a
truncation mutation at position 492 corresponding to the premature stop codon observed inH.glaber. All constructs were based on
the mouse protein, and all carried a Flag epitope at the N-terminus. Transfected protein was detected by anti-Flag antibody. MG132
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 5 / 24
CM frequency of wild-type (+/+) and heterozygous (+/-) hearts was the same, and was
increased in homozygous null (-/-) mice. The same increase was observed in ventricles from
Δ4/- and Δ8/- mice as in -/- mice (Fig 1D). Although the nuclear ploidy of the mononuclear
CM subpopulation of these mice was not directly measured, as shown below, control and
Tnni3k null mice on an inbred C57BL/6J background have the same percentage of diploid
nuclei in their mononuclear CM subpopulation. Thus, mononuclear CM number but not
nuclear ploidy is affected byTnni3k gene status, such that mononuclear CM frequency is a
suitable surrogate for mononuclear diploid CM frequency in the C57BL/6J background. The
Δ4 and Δ8 alleles therefore behave as loss-of-function alleles that increase mononuclear CM
content, just as inferred for the natural intronic splice site mutant allele. The principle advance
afforded by this analysis lies in the derivation of the two frameshift alleles on an inbred
C57BL/6J background. Thus, rather than comparing mononuclear CM content between mice
of different strain backgrounds carrying differentTnni3k alleles, here we confirm on a single
strain background that a transcript encoding a prematurely truncatedTnni3k open reading
frame has the same phenotypic effect as a full null allele.
Analysis ofTnni3k and CM composition in naked mole-rats, a
poikilothermic subterranean rodent
Tnni3k expression in mammals is mostly restricted to the heart, likely in part through a Mef2
binding site in the promoter region [29]. ATnni3k gene is present in all mammals, and recog-
nizableTnni3k homologs exist in numerous invertebrate species among both the deutero-
stome and protostome subdivisions of the animal kingdom, including those not thought to
have a heart (e.g., sea urchins; S1 Table). The ancestralTnni3k gene was therefore present early
in animal evolution; it is not known when cardiac-restricted expression evolved.
From a survey of public databaseTnni3k gene sequences across mammalian species, we
found that the gene for the naked mole-rat (Heterocephalusglaber) contained a premature stop
codon at the position corresponding to mouse Arg492 in exon 16 early in the encoding portion
of the kinase domain. This sequence was observed in two independentH.glaber genomic
sequence assemblies (NW_004624742.1 and NW_004629930.1) and was confirmed by direct
genomic sequencing (Fig 2A) in both of twoH.glaber animals (albeit from the same colony
and so possibly related). Importantly, the stop codon was homozygous in both. Thus, this is
unlikely to be a minor variant and more likely represents the reference sequence of theH.gla-
ber species. We modeled a premature stop codon in the mouseTnni3k gene at position 492,
and in transfected 293 cells observed that this resulted in a low level of truncated protein,
exactly equivalent in behavior to what was observed with the Δ4 and Δ8 frameshift transcripts
(Fig 1A). Therefore, theH.glaber sequence at this position is predicted by itself to cause non-
sense-mediated decay and result in absence of protein expression in vivo, just as with the
mouse Δ4 and Δ8 alleles.
The available genomic sequence ofH.glaberTnni3k implied two additional premature stop
codons within the kinase domain coding region, one associated with a frameshift mutation in
the equivalent of mouse exon 20 and the other a nonsense codon in the equivalent of mouse
exon 21, and both were confirmed by direct genomic sequencing (panels A-B of S2 Fig). A full
was added at the concentrations indicated to test for proteasomal degradation. Gapdh was used as a loading control. B. Semi-
quantitative RT-PCR analysis ofTnni3k transcript abundance in ventricular tissue from adult mice (2 of each genotype) of the
indicated genotypes. C. Western blot to detect Tnni3k protein in ventricular lysate from adult mice of the indicated genotypes. The
filled arrow indicates full-length Tnni3k (93kDa), the open arrow indicates the expected size of the Δ4 and Δ8 truncated proteins
(62kDa) if they were stably expressed. D. Ventricular mononuclear CM% in adult hearts of the indicated genotypes; in both
evaluations, littermate +/- control mice were used as a reference for Δ4/- or for Δ8/- experimental mice.
https://doi.org/10.1371/journal.pgen.1008354.g001
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 6 / 24
comparison of database genomic sequence for theM.musculus andH.glaberTnni3k genes
demonstrated a very high degree of conservation, and notably in the first 15 exons encoding
the amino-terminal ankyrin repeat domains and the beginning of the kinase domain. With
allowance for one abnormal splice acceptor sequence (GG rather than AG preceding exon 13),
a continuous open reading frame with high homology to the mouse protein could be assem-
bled computationally from these 5’ exons (S2 Table, panels C-D of S2 Fig). Although non-
sense-mediated decay should degrade any transcript with a stop codon in any exon before the
terminal exon, exceptions are known, and it therefore remained possible that theH.glaber
gene might encode a protein with only ankyrin repeats that terminated at the stop codon
Fig 2. TheTnni3k gene in naked mole-rats (H.glaber). A. Genomic DNA sequence at the end of intron 15 and
beginning of exon 16 and deduced protein coding sequence (in 1 letter code above) in mouse (Mmu) and inH.glaber
(Hgla). The in-frame premature stop codon is indicated in red type, and lies at position 22288559–561 (reverse
complement) of NW_004624742.1 and at 1144837–839 (reverse complement) of NW_004629930.1. The sequence
trace confirms the accuracy of theH.glaber genomic sequence. B. Western blot to detect Tnni3k; the blot contains
ventricular protein from one wild-type (WT) C57BL/6 mouse, one homozygous null (-/-) mouse, and two individual
H.glaber animals. C. Ventricular mononuclear CM% in adult hearts of two inbred mouse strains (SJL/J and A/J),
which represent the lowest and highest known values of mononuclear CM% among common inbred strains, and in
adultH.glaber. Data for SJL/J and A/J mice are from our previous evaluation [20] and are shown here for reference. D.
Evaluation of nuclear ploidy in single cell preparations from one C57BL/6J mouse (at left) and four naked mole-rats, as
measured by DAPI fluorescence. Values shown represent fluorescence intensity per nucleus from mono- and bi-
nucleated cardiomyocytes and endothelial cells, normalized to the median intensity of endothelial cell nuclei; a
normalized value of 1.0 is assumed to be the median signal from a diploid nucleus. Each dot is one measured nucleus.
Numerical quantitations of 2N and�4N status as shown are only for the mononuclear CM group; compiled numerical
data for both CM subpopulations are shown in S5 Table.
https://doi.org/10.1371/journal.pgen.1008354.g002
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 7 / 24
corresponding to mouse Arg492 in the beginning of the kinase domain. Western blotting of
H.glaber ventricular lysates with Tnni3k antibody (recognizing the N-terminal domain and
the same as used above) did not reveal a band around the size predicted for a truncated protein
(56kDa) (Fig 2B), although the antibody might not have been able to recognize the naked
mole-rat protein sequence. Thus it remained uncertain if theH.glaber gene encodes a protein.
To resolve these questions, we amplified theH.glaber cDNA corresponding to mouse
Tnni3k exons 1–15. No band corresponding to the expected full-length sequence (i.e., 1479
bp) was detected, but a smaller band of 1257 bp was amplified on two separate occasions using
heart cDNA samples from two separate animals. These products were both sequenced and
both found to be identical. The deduced cDNA (panel E of S2 Fig) includes a number of
changes relative to the predicted transcript extracted from the genomic sequence assembly
based on homology to the mouse gene (S2 Table). Most significantly, exon 2 was spliced to
exon 6, bypassing the intervening potential exons 3–5. These three exons represent 310 bp of
sequence, and their exclusion results in a frameshift that terminates the open reading frame
after the 83rd encoded amino acid (the first 50 corresponding to the N-terminal mouse pro-
tein, the subsequent 33 being ectopic and resulting from the frameshift). Other observed
changes included a 51nt readthrough of the exon 12 predicted splice donor (based on the
mouse sequence) to employ a downstream splice donor sequence (intronic in mice) that was
spliced to exon 14 and bypassed exon 13, and complete readthrough of the small intron 14
(79nt inH.glaber; the corresponding intron is 93nt in mouse) directly into exon 15. As noted
above, the kinase domain premature stop codon that originally drew our attention to the
Tnni3k gene inH.glaber is in exon 16. We also observed one mismatch between the cDNA
sequence and the corresponding genomic sequence in the database, a G for A substitution in
exon 6 (position 22318710 of NW_004624742.1) that may be a natural variant or may repre-
sent a sequencing error in the genomic assembly. The altered splicing pattern of theH.glaber
Tnni3k gene as deduced from the cDNA sequence was predictable in one case: the database
genomic sequence at the splice acceptor site preceding exon 13 was GG rather than AG (as
noted above), explaining why this exon was skipped. However, all other intron junctions
appeared normal in the genomic sequence through the first 15 exons (these genomic
sequences were not directly confirmed by us). Despite the fairly high conservation between the
mouse and naked mole-rat genomic sequences corresponding to each mouse exon, we con-
clude that the naked mole-ratTnni3k gene does not encode any protein and may have
devolved into a pseudogene. Nonetheless, the gene is still transcriptionally active in the heart,
and generates a transcript that is predicted to mostly be degraded by nonsense-mediated
decay.
In the genome sequence database (S3 Table), we found that theTnni3k gene of another
mole-rat species, the Damaraland mole-rat (Fukomysdamarensis), also appeared to contain
multiple frame shifts and premature stop codons. We directly sequenced exons 5 and 15 from
one animal, and observed a frameshift and a premature stop codon in both exons (panel F of
S2 Fig). Importantly, theTnni3k gene frame-shifts, premature stop codons, and splice donor
or acceptor mutations inH.glaber were not found inF.damarensis, and vice versa. The two
mole-rat species diverged from each other approx. 25–30 million years ago, which is on the
order of the divergence time between mouse and rat [30–33]. We conclude that the ancestral
Tnni3k gene was present and functional in the common rodent ancestor, and then devolved
independently in bothH.glaber andF.damarensis. Possibly, this is an example of parallel evo-
lution associated with the independent transition of both species to subterranean life. Alterna-
tively, the last common ancestor of both species of mole-rat may have already acquired an
initial loss-of-function mutation in theTnni3k gene, perhaps associated with it becoming
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 8 / 24
subterranean, with the gene then acquiring further mutations independently in each species
after they became reproductively isolated.
Prior studies have reportedH.glaber heart morphology and physiology [34, 35], although
CM polyploidy had not been addressed. We conducted an analysis of ventricular CM compo-
sition in juvenile-adult stage naked mole-rats, using the same procedures as used for mouse
heart analysis. Unexpectedly, we found that that 30% of the CM population was mononuclear
(Fig 2C), which is substantially higher than any of our previously measured inbred mouse
strains. Although the great majority of CMs in mice and rats are binucleated (e.g., 95% in
C57BL/6J mice), human CMs are more commonly mononuclear tetraploid, such that a direct
measurement of nuclear ploidy in naked mole-rat mononuclear CMs was worthwhile. Because
fluorescence in situ hybridization probes for direct visualization of chromosome number (as
we used previously for mouse [20]) are not available forH.glaber, we instead estimated nuclear
ploidy in stained ventricular single cell suspensions by DAPI fluorescence intensity, normal-
ized to the intensity of nuclei in Pecam1+ endothelial cells (which are presumed to be diploid
unless in the process of cell division). This analysis (Fig 2D, S5 Table) indicated that the major-
ity (77%) of naked mole-rat mononuclear CMs are polyploid (tetraploid or higher nuclear
ploidy level), and that the prevalence of mononuclear diploid CMs in this species is therefore
on the order of 7%. We of course do not know the extent of genetic variation or the range of
mononuclear diploid CM frequency across theH.glaber species, so this assessment is a single
data point based on only one group of animals. One conclusion from this analysis is that
mononuclear tetraploid CMs can be a substantial fraction of ventricular CMs in rodents (e.g.,
forH.glaber CMs, 70% are binucleated, 23% are mononuclear tetraploid, 7% are mononuclear
diploid), and that it cannot be assumed for all rodent species that most mononuclear CMs are
diploid. While a 7% level of mononuclear diploid CMs is high compared to most mouse strains
and to most other mammalian species that have been studied, this is less than in mouse strain
A/J, which has the highest level of mononuclear diploid CMs (10%) of the inbred mouse
strains that we have surveyed [20]. While anH.glaber substrain with a functionalTnni3k gene
is not available for comparison, we suggest that absence ofTnni3k gene function in naked
mole-rats may contribute to the moderately high level of mononuclear diploid CMs seen in
this species, just as we have shown experimentally in mice.
Tnni3k kinase activity regulates mononuclear cardiomyocyte frequency
By sequence, Tnni3k is a member of the MLK family of kinases, which have both tyrosine and
serine/threonine phosphorylation activity, and is classified as a MAP3K. Although the protein
has previously been shown to have autophosphorylation activity in vitro [25, 36, 37], no natu-
ral proteins that serve as kinase substrates in vivo have yet been verified. Furthermore, kinase
activity in the context of establishment of the mononuclear diploid CM frequency cannot be
assumed, as the protein could act solely via interactions with other proteins through its N- or
C-terminal domains; this is not without precedent in other kinases [38].
To demonstrate the in vivo function of Tnni3k kinase activity, we created a kinase-dead
mutant allele in C57BL/6J mice by CRISPR-mediated homologous replacement, changing the
lysine codon at position 489 to arginine (AAA to AGA) in exon 15 (panel A of S3 Fig). This
lysine sits in the ATP binding pocket, and among all kinases, a lysine in this position is almost
ubiquitously conserved as it coordinates the terminal phosphate of ATP as it is transferred to a
substrate. In all other studied kinases, mutation of this lysine to any other amino acid (includ-
ing arginine) abolishes kinase activity. As previously shown [25, 37] and as reconfirmed here
(see below), Tnni3k with the K489R mutation was stable when expressed in 293 cells, and in
an in vitro kinase assay was devoid of autophosphorylation activity. A founder female mouse
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 9 / 24
heterozygous for the K489R allele was crossed to wild-type C57BL/6J males to confirm germ-
line establishment of the allele, and heterozygous F1 mice then crossed to C57BL/6J-inbred
Tnni3k homozygous null mice to generate littermate mice that were used for analysis. Just as
homozygosity of theTnni3k null allele has no externally obvious consequence, mice in which
the K489R allele and the null allele were combined were also not obviously impacted. Ventric-
ular lysates from these mice contained the expected level of full-length protein (Fig 3A), con-
firming that the K489R point mutation does not alter protein stability in vivo.
Adult mouse hearts were evaluated for their percentage of ventricular mononuclear CMs.
Mice with the kinase-dead K489R allele (K489R/-) had a significantly increased level of mono-
nuclear CMs relative to heterozygous (+/-) mice and equivalent to the level in null (-/-) mice
(Fig 3B), and equivalent also to Δ4/- and Δ8/- mice (Fig 1D; not compared in the same analy-
sis). Direct measurement of nuclear ploidy in mononuclear CMs (Fig 3C, S4 Fig, S5 Table)
showed no change across the +/-, K489R/-, and -/- genotypes. Thus, Tnni3k kinase activity
regulates mononuclear diploid cardiomyocyte frequency. Because absence of Tnni3k protein
has the same consequence as a normal level of the kinase-dead full-length protein, we also con-
clude thatTnni3k has no additional role through an independent function (e.g., as a scaffold
for interaction with other proteins). The amino-terminal ankyrin repeat domain and the car-
boxy-terminal serine-rich domain may also be required to modulate Tnni3k’s kinase role in
this context [36], but not in a manner that is independent of its kinase activity.
Common humanTNNI3K polymorphisms compromise TNNI3K kinase
activity
The ExAC database of aggregated human whole exome sequence [26] indicates thatTNNI3K
loss-of-function alleles are moderately rare in the human population (0.3% of sequenced
alleles). These include frameshifts, premature stops, and splice site mutations that presumably
result in frameshifts. These variants are distributed over the entire length of the gene, with
most therefore appearing in exons 5’ to the last large intron. Based on the rules of nonsense-
mediated decay [39] and by extrapolation from the variousTnni3k alleles described above, we
predict that these human variants would also be subjected to nonsense-mediated decay and
would not encode any protein. Supporting this prediction, one of the disease-associated rare
TNNI3K variants, a mutation in the splice donor sequence following exon 4, was measured to
cause nonsense-mediated decay [22]. For the two most common loss-of-function variants in
the ExAC database (a premature stop codon in exon 3 at position 65, and a 10bp insertion in
exon 8 causing a frameshift at position 241; both positions based on NP_057062.1), one indi-
vidual for each was identified in the ExAC database to be homozygous. No clinical or pheno-
typic information is available for these two individuals, but much as homozygousTnni3k loss-
of-function alleles in mice do not cause any externally obvious consequence, we predict these
individuals to also be overtly normal. If our mouse studies ofTnni3k can be extended to
human heart biology, we also predict these individuals to have a higher percentage of mononu-
clear diploid cardiomyocytes in their hearts.
Compared to the loss-of-function alleles, a much larger number of nonsynonymous amino
acid substitution variants of unknown significance occur in the humanTNNI3K gene. Many
of these lie in the kinase domain and therefore could influence kinase activity. We used the in
vitro kinase assay mentioned above to assess the activity of the five most common such vari-
ants (S4 Table), engineered into the mouse protein. For comparison, a human disease-associ-
ated rareTNNI3K allele, Thr539Ala [23], was also engineered into the mouse full-length
sequence, and the kinase-dead K489R mutant described above was also included in this analy-
sis. All proteins were stable when expressed in 293 cells, as evident by detection of the Flag
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 10 / 24
epitope at the N-terminus of each protein (Fig 4A). A summary compilation of the autopho-
sphorylation kinase activity of each protein is shown in Fig 4B (numbering based on the
mouse protein sequence, which is one fewer than the human sequence), and the primary data
used for quantitation of kinase activity are shown in S5 Fig. Two of the human kinase domain
variants (S591T and T637M, corresponding in Fig 4 to positions 590 and 636 in the mouse
sequence, respectively) had little if any impact on kinase activity and were not further exam-
ined. However, three others (V510L, A666T, I686T) prominently compromised kinase activity.
From quantitation (Fig 4B, S5 Fig), we estimate that each of these three variants reduced kinase
activity to approximately one-third of that of the wild-type protein. In principle, then, individ-
uals who are homozygous for one of these alleles may have a lower level of cellular TNNI3K
kinase activity than individuals who are heterozygous for a loss-of-function allele. Because the
I686T variant in particular is so common across the human population, many individuals are
homozygous for this allele (1.67% allele frequency and 0.06% homozygous frequency in
ExAC). In mice,Tnni3k heterozygosity does not have a consequence on mononuclear CM
content [20], so it is unlikely that heterozygosity of these hypomorphic variants would impact
this parameter in humans. We do not yet have hypomorphicTnni3k alleles in mice with which
to define the threshold level of gene activity below which a measurable change in mononuclear
CM frequency would result. Consequently, it is uncertain if homozygosity of the human hypo-
morphic alleles would impact CM composition.
The human variants may however be associated with heart disease. The rare human exon 4
splice donor site mutant that results in transcript nonsense-mediated decay, and the rare
Fig 3. Consequences of a kinase-deadTnni3k allele in mice. A. Western blot of Tnni3k protein from mice of the indicated
genotypes, demonstrating that the K489R protein is stable in vivo. B. Ventricular mononuclear CM% in mice of the indicated
Tnni3k genotypes; heterozygous (+/-) and null (-/-) mice were littermates of the K489R/- mice, but not of each other. C.
Evaluation of nuclear ploidy specifically in the mononuclear CM subpopulation of mice of the indicatedTnni3k genotypes. See
also S4 Fig.
https://doi.org/10.1371/journal.pgen.1008354.g003
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 11 / 24
human T539A variant that is significantly hypomorphic in kinase activity (13% of wild-type
activity; Fig 4), are both associated with adult conduction system disease and dilated cardiomy-
opathy as heterozygous alleles ([22, 23]; see also Introduction). Thus, the V510L, A666T, or
I686T variants may predispose to these adult disease phenotypes, or may combine with other
genetic or environmental conditions to result in these diseases.
Tnni3k converges with oxidative stress to influence mononuclear CM
frequency
The critical time when mouse CMs become polyploid or remain mononuclear and diploid is
midway through the first postnatal week [11]. There is little understanding of the mechanisms
that mediate this transition, and virtually nothing known of howTnni3k is involved in this
Fig 4. Consequences on in vitro kinase activity of common humanTnni3k kinase domain variants. A.
Representative Western blot showing autophosphorylation on Ser and Tyr residues of wild-type and variant Tnni3k
extracted from transfected 293 cells prior to (-) or following (+) incubation in an in vitro kinase reaction. Proteins were
expressed with an N-terminal Flag epitope, which was detected separately as a loading control. B. Quantitation of
kinase activity for the four indicated variants based on data shown in S5 Fig.
https://doi.org/10.1371/journal.pgen.1008354.g004
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 12 / 24
process. As a MAP3K, a plausible mechanism to explainTnni3k involvement is through phos-
phorylation of MAP2Ks (MEKs), which activate the MAP kinases Erk, Jnk, and p38. We tested
for the level of phosphorylation of all three MAPKs in postnatal day P3 hearts (Fig 5A). We
observed a prominent level of phosphorylated Erk and Jnk at this time point, but neither were
affected by the absence ofTnni3k. The level of phospho-p38 was so low as to be barely detect-
able, but was not obviously different between wild-type and null P3 hearts. This level, however,
was substantially lower than found in adult hearts (panel A of S6 Fig). Overall, this analysis
provides no evidence for the involvement of MAPK signaling pathways as downstream media-
tors of Tnni3k kinase activity in influencing the neonatal transition of CMs to become
polyploid.
In mice, a compelling case has been made that an increase in reactive oxygen species (ROS)
associated with birth and postnatal life is a cause of cell cycle arrest [40], which is in principle
equivalent to the onset of polyploidization. While excessive ROS (oxidative stress) is pathologi-
cal and causes cell death, physiological levels of ROS are not damaging but rather serve as nec-
essary intracellular signals. A primary ROS is superoxide, which is formed transiently in
mitochondria and converted into the more stable peroxide. To experimentally modulate ROS
level in mice, we used the mCAT transgenic line [41], which expresses catalase (which
degrades peroxide) in mitochondria and so diminishes the amount of cellular ROS. The
mCAT transgene was obtained on an inbred C57BL/6NJ background, which is a sister strain
to C57BL/6J. We first tested to determine if this subtle difference in strain background had an
impact on mononuclear CM frequency. Although there was a slight trend to higher mononu-
clear CM level in C57BL/6NJ mice, this difference was not statistically significant (panel B of
S6 Fig). Nonetheless, because the main difference between the two strains is in theNnt gene
(functional in C57BL/6NJ, not functional in C57BL/6J [42]), in the following analyses we con-
trolled for theNnt status of all animals. By crossing the mCAT transgenic line with C57BL/6J
mice to make an F1 generation (Nnt heterozygous), we first observed that mCAT expression
alone resulted in an increase in mononuclear CM level (Fig 5B). This supports the interpreta-
tion that higher oxidative stress is associated with more CMs becoming polyploid, and con-
versely that lower oxidative stress associates with persistence of more mononuclear CMs.
Crossing the mCAT transgene into theTnni3k null background (now with theNnt gene being
null), we observed that mCAT expression still had the same effect of increasing mononuclear
CM level, and that the effects of the two genetic manipulations were additive (Fig 5C). The
nuclear ploidy of mononuclear CMs was not influenced by mCAT expression (S4 Fig, S5
Table). Thus,Tnni3k does not act as an obligate downstream mediator of oxidative stress, but
rather converges with ROS level to influence the percentage of neonatal CMs that become
polyploid or remain mononuclear diploid.
Discussion
TheTnni3k gene is expressed almost exclusively in cardiomyocytes, a selectivity that is unique
among the kinase family. Presumably, this relates to a gene function that was subjected to posi-
tive selection in evolution. Nonetheless, theTnni3k gene can be mutated in mice without obvi-
ous deleterious consequence. As a result, a naturalTnni3k loss-of-function allele became
widely distributed among inbred mouse strains, and gave the strongest association signal in
our forward genetic screen for variants that affect the frequency of ventricular mononuclear
CMs. Relatively little is known of the biological roles of this gene, and prior to our work it was
unstudied in the context of the neonatal heart and the transition of neonatal cardiomyocytes
to become polyploid.
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 13 / 24
Fig 5. Tnni3k converges with oxidative stress. A. Western blot of ventricular protein from two wild-type and two
Tnni3k null P3 mice showing phosphorylated (activated) and total protein for the three MAPK members p38, Jnk, and
Erk. The very low level of phospho-p38 in P3 heart is compared against the normal level in adult heart in S6 Fig. B-C.
Ventricular mononuclear CM% in adult hearts, comparing absence vs. presence of the mCAT transgene in a
background that is wild-type forTnni3k and heterozygous forNnt (B), and comparing absence vs. presence of the
mCAT transgene in aTnni3k null andNnt null background (C).
https://doi.org/10.1371/journal.pgen.1008354.g005
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 14 / 24
By introducing a K489R point mutation into theTnni3k gene in mice, resulting in a nor-
mally-expressed full-length kinase-dead protein, we show that Tnni3k functions as a kinase in
the context of CM polyploidization. This was not a foregone conclusion, as the highly con-
served amino-terminal ankyrin repeat domain constitutes over half of the protein length, and
may have been the important functional domain in this process. The observation that the
kinase-dead K489R mutant shows the same increase in mononuclear diploid CM content as
alleles in which no Tnni3k protein is made demonstrates that the amino and carboxy terminal
domains of the protein do not independently influence this frequency outside of their connec-
tion to kinase activity. They may also be required, but not independent of kinase function. The
observation that mole-ratTnni3k genes became mutated not only by acquiring point muta-
tions in the kinase domain but also throughout the protein coding region is a further indica-
tion that the ankyrin repeat domain is unlikely to have a function independent of the protein’s
kinase activity.
Tnni3k variants in mice influence the frequency of mononuclear diploid CMs. If the same
pathways are active in the human heart, variants in theTNNI3K gene may similarly influence
this trait in humans. NullTNNI3K alleles are present in the ExAC database, mostly as hetero-
zygous alleles but with rare instances of humans that are homozygous null. If only limited to
rare homozygous null individuals, the contribution ofTNNI3K variants to variation in overall
human heart composition would be of only incidental significance. More intriguing is the rec-
ognition that several commonTNNI3K polymorphisms have approximately one-third the
kinase activity of full-length protein, and that many individuals are homozygous for these
hypomorphic alleles. These common humanTNNI3K hypomorphic variants may therefore
contribute to variation in mononuclear diploid CM level across a broad segment of the human
population. As pointed out above, this assumes that the same pathways and the same role of
Tnni3k in establishing mononuclear diploid CM level in the mouse heart are also similarly
active in the human heart. A further caveat is that we only defined kinase activity of these vari-
ants using an in vitro assay and only using autophosphorylation as an experimental endpoint.
The variants studied here may have more or less kinase activity in vivo on a true substrate pro-
tein compared to what was revealed by the in vitro assay. Furthermore, we cannot yet confirm
what level of reduction of kinase activity would be needed to change mononuclear diploid CM
level, as we do not have hypomorphicTnni3k alleles in mice with which to model the effects of
such variants.
In the neonatal heart, as CMs are induced to enter cell cycle,Tnni3k influences the percent-
age of these cells that successfully complete cell cycle with cytokinesis. This is consistent with
an impact on the machinery of karyokinesis or cytokinesis, or on regulators of mitotic progres-
sion, rather than G1- or S-phase components. Even with the understanding thatTnni3k func-
tions as a kinase in this context, a critical limitation is a lack of knowledge regarding its
primary substrates and downstream pathways of activity. One mechanistic insight comes from
an observation in adult heart ischemia-reperfusion injury where the presence ofTnni3k led to
ROS overproduction and thereby to increased CM death [43]. This was shown to be mediated
by p38, which is a known ROS sensor. Ischemia-reperfusion is a pathological situation of mas-
sive oxidative stress, and so observations made in this setting do not necessarily apply to the
uninjured heart. Indeed, we examined MAPK components, including p38, and did not see an
indication of relevance to the normal process by which neonatal CMs become polyploid. Even
if the pathway of action ofTnni3k in neonatal CM polyploidy does not involve p38, conver-
gence with ROS signaling is an appealing model. Under physiological conditions, lower ROS
is associated with higher mononuclear diploid CM level; this was implied in a prior study
showing the relation of oxidative stress to cell cycle interruption [40], and confirmed in our
study with expression of mCAT to reduce ROS and increase mononuclear diploid CM
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 15 / 24
content. Absence ofTnni3k has the same direction of effect as expression of mCAT. Conceptu-
ally, therefore, the normal function ofTnni3k may be to amplify ROS generation or signaling.
The onset of CM polyploidy, which in mice occurs during the first postnatal week, is one spe-
cific time when this function ofTnni3k is manifest.
Tnni3k is expressed in CMs throughout life, so we expect it to have a continuing role in CM
physiology, possibly also related to ROS signaling.Tnni3k function is maladaptive when the
heart is challenged with extreme ROS levels as in the context of ischemia-reperfusion men-
tioned above, but given its evolutionary conservation, the gene is presumed to have a beneficial
role under normal physiological circumstances, perhaps to augment responsiveness to meta-
bolic alterations via ROS signaling. The apparent evolution of theTnni3k gene in mole-rats to
become nonfunctional (at least for protein coding) may be interpretable in this light. Mole-
rats evolved to life in subterranean tunnels where oxygen levels are very low, but have a rela-
tively pro-oxidant intracellular milieu and withstand normoxia without apparent ill effect, pre-
sumably by the activity of a variety of cytoprotective mechanisms [44, 45].Tnni3k mediation
of some feature of ROS response as we posit in other mammals may not have been needed or
may even have been counterproductive for subterranean life. In human heart biology, though,
diminishment or absence of this beneficialTNNI3K function may predispose to conduction
system disease and dilated cardiomyopathy (as observed in the human pedigrees with these
conditions), and perhaps more so if combined with other challenges that increase reactive oxy-
gen generation. Clearly, it will be necessary to identify the kinase substrates and specific down-
stream pathways that might account for such effects to confirm this model.
One of the human disease-associatedTNNI3K variants is the T539A point mutation that
was included as a reference in our in vitro kinase analysis (Fig 4). This threonine lies in the
“gatekeeper” position in the ATP binding pocket where it coordinates the adenine base of
ATP. In all studied kinases, mutation of the gatekeeper residue to the smaller amino acids gly-
cine or alanine enlarges the ATP binding pocket and reduce ATP affinity and thereby reduce
kinase activity. As we measured, the T539A variant diminishes in vitro kinase activity to 13%
of the wild-type protein. A second disease-associatedTNNI3K variant is a splice donor muta-
tion following exon 4 that was previously measured to cause nonsense-mediated decay, similar
to the effect of several alleles explored in the present analysis. Importantly, both variants are
disease-associated as heterozygous alleles. If these have hypomorphic or loss-of-function con-
sequences, this implies that reduction of TNNI3K activity to below a threshold level even in
the presence of one wild-type allele could cause disease manifestation. If so, the common
hypomorphic variants described here (V510L, A666T, I686T), which based on the in vitro
assay are predicted to result in 30–40% of kinase activity, may predispose to a similar spectrum
of disease phenotypes. A recent hypothesis has suggested that disease-associatedTNNI3K vari-
ants may instead have either hyperactive or dominant negative (sequestering) activity, rather
than loss-of-function activity [25]. This is likely to be true for some disease-associated
TNNI3K variants (e.g., E768K), although seems difficult to reconcile with nonsense-mediated
decay of the exon 4 splice site mutation transcript and with the abundant biochemical infor-
mation regarding the functional consequences of Ala substitution of the gatekeeper threonine
as occurs with the T539A variant. If the common variants do contribute to disease as hypo-
morphic alleles, this may only be when combined with specific additional genetic or environ-
mental perturbations. Possibly, families with conduction system disease and dilated
cardiomyopathy share other genetic alterations or environmental conditions that combine
with the heterozygousTNNI3K variants to cause disease.
A recent evaluation [46] concluded that the acquisition of endothermy (generating body
heat by metabolism, commonly described as warm-blooded biology) in animal evolution was
associated with thyroid hormone-dependent metabolism, causing increased levels of CM
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 16 / 24
polyploidy and thereby loss of adult heart regenerative capacity. Naked mole-rats are an inter-
esting test of this model, in that they are one of the few mammalian species that are poikilo-
thermic (do not regulate their body temperature), and have a low basal metabolism and a low
level of thyroid hormone [47]. Our analysis indicates that the level of mononuclear diploid
CMs inH.glaber is on the higher side of the range seen in inbred mouse strains but is certainly
not remarkable. This species would appear to contradict the model that thyroid hormone and
endothermy are a primary explanation for CM polyploidy.H.glaber may be an exception to
this rule, or the model may not fully account for this process.H.glaber presumably evolved to
poikilothermia long after a high level of CM polyploidy evolved in endothermic vertebrates,
and may have retained additional mechanisms that drive CM polyploidy even as they lost
warm-blooded metabolism.
Methods
Animals
All mouse experiments were performed on age- and sex-matched animals (8–10 weeks of age),
with an equal ratio of male to female mice. Naked mole rats were provided by T.J.P. from his
colony at the University of Illinois at Chicago and sent as live animals to USC where hearts
were isolated for analysis.
Ethics Statement
Animal research was reviewed and approved by the IACUC committees of the Univ. of South-
ern California (#10173) and of the Medical Univ. of South Carolina (201800642). Animals were
euthanized by isoflurane anesthesia followed by cervical dislocation and removal of hearts.
Derivation of new mouseTnni3k alleles
The K489R allele was derived by the VCU Massey Cancer Center Transgenic/Knockout
Mouse Core. The protospacer/PAM target sequence 5’-AGATGAATTGCATCGTCAGC
TGG-3’, which is in the 5’ side ofTnni3k intron 15, was used to generate an Alt-R crRNA,
which was annealed to the Alt-R tracrRNA (both from Integrated DNA Technologies, IDT) to
generate the functional guide RNA. Homology-directed repair at the cleavage site utilized a
200-base single-stranded oligodeoxynucleotide (ssODN) identical to the WT locus non-coding
strand spanning from intron 14 to exon 16, with the exception of three altered bases: an A-to-
G substitution to create the lysine-to-arginine mutation at codon 489, and two G-to-T substi-
tutions to eliminate the Cas9 PAM sequence (to prevent retargeting) and to create a DraI site
for screening purposes. A mix containing 1.2μM Cas9 protein (IDT Alt-R S.p. HiFi Cas9
Nuclease 3NLS), 6μM annealed cr/tracrRNA (preincubated with the Cas9 protein at room
temperature for 10 min), and 300 ng/μl ssODN (IDT) in Opti-MEM medium (Thermo Fisher
Scientific) was electroporated into fertilized C57BL/6J mouse eggs using a NEPA 21 electro-
porator (NEPA GENE Co. Ltd., Chiba, Japan) and CUY501P1-1.5 electrode as follows: 5μl
containing ~30 embryos at a time received 4 “poring pulses” (40V, 3.0 msec, 50 msec intervals,
10% voltage decay rate, polarity +) followed by 5 “transfer pulses” (7V, 50 msec, 50 msec inter-
vals, 40% voltage decay rate, polarity +/- alternating.) The impedance of the solution on the
electrode was maintained at ~0.20–0.22. Following electroporation, the eggs were placed back
into KSOM medium (Specialty Media), and subsequently implanted into the oviducts of pseu-
dopregnant ICR females (~15 embryos per recipient). At weaning, DNA was purified from a
5 mm tail snip from each pup. The PCR genotyping protocol employed primers flanking the
targeted region, (forward: 5’- AAGCCCAGACCATGTGTGCTAAGG-3’ and reverse: 5’-
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 17 / 24
CAGTAGGGTCTCTGACTGGAAGTC-3’), yielding a 775-bp product for both the WT and
KI alleles. Subsequent digestion of the product with DraI yielded 428bp and 347bp fragments
for the KI allele, while the WT allele remained uncut. The Δ4 and Δ8 alleles were generated at
the Children’s Hospital of Los Angeles Saban Research Institute Transgenic/Knockout Mouse
Core, using the same IDT crRNA and tracRNA procedures described above with a protospa-
cer/PAM target sequence of 5’-GTGACAATGGCAAACTGGCTGGG-3’, which is located
within exon 16. The annealed crRNA-tracrRNA complex was diluted in injection buffer (1
mM Tris-HCl, pH 7.5, 0.1 mM EDTA) to 40 ng/μl and mixed with equal volume of Alt-R S.p.
HiFi Cas9 nuclease 3NLS (at a concentration of 40 ng/μl in injection buffer). The mixture of
crRNA-tracrRNA and Cas9 nuclease were incubated at room temperature for 15 min to form
crRNA-tracrRNA-Cas9 protein ribonucleoprotein complex used in pronuclear injection. The
final concentration of crRNA-tracrRNA and Cas9 ribonucleoprotein were both 20ng/μl. Pro-
nuclear injection of E0.5 embryos was performed according to standard procedures; after
injection, viable embryos were transferred to pseudopregnant CD1 female mice. For genotyp-
ing, primers flanking the targeted region (forward: 5’-AAAACACCCCGTGATGTTTTATT-3’
and reverse: 5’- CAGATGAATTGCATCGTCAGCT-3’) yielded a 942bp (wild-type allele;
slightly smaller for deletion alleles) product that was gel purified and then sequencing using
the primer 5’-TAGATACCGAGCCAACACC-3’.
Mouse allele and genotype designations
The wild-typeTnni3k allele in C57BL/6J mice is indicated as the + allele. An engineered loss-
of-functionTnni3k null allele described previously [20] is designated as the - allele. The alleles
newly created in this study are represented by Δ4, Δ8, and K489R. Unless otherwise noted,
alleles were combined such that all animals carried one null allele (e.g., +/-, Δ4/-, Δ8/-,
K489R/-, and -/-). The mCAT transgenic allele was obtained from The Jackson Laboratory
(JAX016197) and in all cases was kept as a hemizygous allele.
Single-cell ventricular cardiomyocyte suspensions
Hearts were digested with 1mg/mL collagenase type II via Langendorff retroaortic perfusion.
After digestion, atria and valves were removed and ventricular tissue alone was triturated in
Kruftbru ¨he (KB) solution (70mM potassium aspartate, 40mM KCl, 15mM KH
2
PO
4
, 10mM
glucose, 10mM taurine, 0.5mM EGTA, 10mM sodium pyruvate, 10mM HEPES, 5mM BDM,
0.5% BSA), filtered through a 250-μm nylon mesh, stained with LiveDead Fixable (Thermo-
Fisher, L10120) for 20 min at room temperature and then fixed in 2% PFA at room tempera-
ture for 15 min. Fixed ventricular cell suspensions were stained for cTnT (1:1,000, Abcam
ab8295) overnight at 4˚C followed by goat anti-mouse secondary (1:500, ThermoFisher
A11001) and DAPI. Cell suspensions were then pipetted across a slide and coverslipped. Car-
diomyocyte nucleation was quantified on an Olympus BX41 fluorescence microscope with a
20x objective. Only live cardiomyocytes were counted for mono-, bi-, tri- and tetranucleation;
at least 600 cells were counted per heart. An unpaired, two-tailed Student t-test was used to
assess statistical significance when only two groups were compared.
Cardiomyocyte ploidy analysis
Cardiomyocyte suspensions were spread on slides and stained for mouse anti-Myl2 (1:250)
and anti-CD31 (1:250) with an Alexa Fluor secondary (1:250, ThermoFisher A11001) and
DAPI using standard procedures. Cells were coverslipped with ProLong Gold antifade reagent
(Invitrogen). Slides were scanned by fluorescence microscopy. IMARIS software was used to
quantify cardiomyocyte ploidy. Briefly, nuclei were identified and outlined with a standard
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 18 / 24
threshold requirement for all samples. A histogram of each nucleus was generated and total
fluorescence was calculated as area under the curve, thereby accounting for both size of the
nucleus and fluorescence intensity. The median value of DAPI fluorescence intensity of
CD31+ endothelial cell nuclei was used as a diploid nucleus standard.
Reverse transcription polymerase chain reaction (RT-PCR)
RNA was extracted from heart ventricle using Quick-RNA Mini Prep Kit (Zymo Research).
Equal amounts of RNA between samples were used to synthesize cDNA using M-MLV reverse
transcriptase (Invitrogen). Primer sequences used to amplifyTnni3k cDNA to detect its rela-
tive abundance across samples were: 5
0
-CAGCACAGGAGGAAAGCAGA-3
0
and 5
0
-GCAGG
TCATCTTCCAGCCTT-3
0
. Gapdh was used as an internal control.
Plasmids and molecular cloning
Mouse cDNA was synthesized using SuperScript III First-Strand Synthesis System (Invitro-
gen). The mouseTnni3k gene was amplified from C57BL/6J mouse cDNA by PCR with Q5
High-Fidelity DNA Polymerase (New England Biolabs) and the primers forward: 5’-GAGAG
AAGATCTGCCACCATGGGGAATTACAAATCCAGACCG-3’ and reverse: 5’-GAGAGAC
TCGAGTTACTTGTACAGCTCGTCCATGCCG-3’. The naked mole ratTnni3k gene was
amplified fromH.glaber cDNA by PCR with primers forward: 5’-GAGAGAGGATCCGCCA
CCATGGGAAATTATAAATCTAGACCAACACA-3’ and reverse: 5’- GAGAGACTCGAGT
CAACAGTGTTTTATAGCCACTATTTTATTTCT-3’. Amplified fragments were cloned into
the pcDNA3 vector (Addgene) with an N-terminal Flag tag. AllTnni3k point mutant expres-
sion plasmids (K489R, T538A, V509L, S590T, T636M, A665T, I685T), and all truncation
mutants ( Δ4, Δ8, 492
�
) were constructed from the wild-type complete open reading frame
pcDNA3-Flag-Tnni3k plasmid.
Western blotting
Heart ventricular tissue was snap frozen in liquid nitrogen then homogenized with an OMNI TH
homogenizer in lysis buffer (50mM Tris-HCl pH7.5, 150mM NaCl, 0.1% NP-40) on ice. Heart
lysates were centrifuged at 14,000 RPM for 20 min at 4˚C to remove insoluble material. Lysis
buffer contained 1x Complete protease inhibitor mixture (Roche), and 1x PhosStop phosphatase
inhibitor (Roche). Immunoblotting was performed by standard protocols with 50μg heart lysate
using anti-Tnni3k (1:1,000, Invitrogen PA5-21989), anti-GAPDH (1:1,000, GeneTex GT239),
anti-p38 (1:1,000, Cell Signaling Technology, CST9212), anti-phospho-p38 (1:1,000, Cell Signal-
ing Technology, CST4511), anti-ERK1/2 (1:1,000, Cell Signaling Technology, CST4695), anti-
phospho-ERK1/2 (1:1,000, Cell Signaling Technology, CST9101), anti-SAPK/JNK (1:1,000, Cell
Signaling Technology, CST9252), anti-phospho-SAPK/JNK (1:1,000, Cell Signaling Technology,
CST4668) and HRP-coupled secondary (1:10,000, Jackson ImmunoResearch sc2040) antibodies.
20μg HEK-293T cell lysate or 1μl of beads with immunoprecipitated TNNI3K were loaded and
blotted in the same manner, then using anti-phospho-tyrosine (1:1,000, Cell Signaling Technol-
ogy, CST9411), anti-phospho-serine (1:1,000, Millpore Sigma, AB1603), anti-Flag (1:1,000, Invi-
trogen, MA1-91878), or anti- β-catenin (Ser33/37/Thr41) (1:1,000, Cell Signaling Technology,
CST8814). Quantification of Western blotting signal was done using ImageJ.
Tissue culture, transfection and immunoprecipitation
HEK293T cells were grown using standard conditions in DMEM (Gibco) containing 10%
(vol/vol) FBS (Equitech-Bio, Inc.), 1x sodium pyruvate (Gibco), 1x MEM NEAA (Gibco), and
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 19 / 24
1x penicillin-streptomycin (Gibco). Cells were transfected with Lipofectamine 3000 (Invitro-
gen) at 60% confluence for 24h, and harvested at 90–100% confluence. Cell lysis was per-
formed in lysis buffer (50mM Tris pH7.5, 150mM NaCl, 0.1% NP-40, 1x Roche Complete
protease inhibitor mixture, and 1x PhosStop phosphatase inhibitor (Roche)). MG132 (Sigma-
Aldrich) was applied 24h after transfection for 8h before harvest.
In vitro kinase assays
Immunoprecipitation of protein was performed from 1mg HEK293T cell lysate per sample,
using 10μL of anti-Flag magnetic beads (Sigma) at 4˚C overnight. For use in in vitro kinase
assays and Western blotting, beads were washed five times in 500μL lysis buffer, and resus-
pended in 10μL lysis buffer. 1μL of immunoprecipitated flag-tagged TNNI3K protein on beads
was used for in vitro kinase assay, conducted in 50mM Tris pH7.5, 150mM NaCl, 10mM
MgCl
2
, 20μM ATP, 2mM DTT. The reactions were incubated at 30˚C with gentle agitation
for 30 min. Reactions were quenched by adding gel loading buffer and heating to 95˚C before
SDS/PAGE separation.
Supporting information
S1 Fig. Sequence of the Δ4 and Δ8 alleles. Sequence traces of genomic DNA from heterozy-
gous Δ4/+ (top) and Δ8/+ (bottom) mice, illustrating the location where the sequences of the
wild-type and deletion mutant alleles diverge. The deleted bases correspond to positions
154939648–651 ( Δ4) and 154939641–648 ( Δ8) of NC_000069.6. Both alleles result in frame-
shifts that cause downstream sequences to become termination codons (in red).
(PDF)
S2 Fig. TheTnni3k gene inH.glaber andF.damarensis mole-rats. A. Alignment of portions
of mouse exons 16, 20, and 21 with genomic DNA from the two mole-rat species (see S2 Table
and S3 Table for coordinates of the mole-rat exon sequences), emphasizing the premature stop
codons in exons 16 and 21 and the gap in exon 20 inH.glaber (in the center line), whereas F.
damaraensis conserves the open reading frame in these regions. B. Sequence traces fromH.
glaber genomic DNA of the portions of exons 20 and 21 shown in panel A. C. Putative mRNA
and translated reading frame as deduced fromH.glaber genomic DNA, beginning with the
ATG codon in exon 1 and ending with the TGA codon in exon 25. The premature stop codon
in exon 16 is underlined and in red type. The putative translated product begins with the ATG
codon in exon 1 and ends at the premature stop codon in exon 16. D. Alignment of the puta-
tive translatedH.glaber (Hgla) Tnni3k protein with that of mouse (Mmu). E. Sequence of the
Tnni3k cDNA as obtained experimentally fromH.glaber ventricular mRNA, beginning with
the ATG codon in exon 1 and ending with the premature stop codon in exon 16. The first in-
frame stop codon is underlined and in red type. The translated product is shown below. F.
Alignment of portions of mouse exons 5 and 15 with genomic DNA from the two mole-rat
species, here withF.damarensis in the center line and organized to show the gaps and prema-
ture stop codons in both exons, whereasH.glaber conserves the open reading frame in these
regions. Sequence traces ofF.damarensis for these exon regions are shown below.
(PDF)
S3 Fig. Construction of the K489R kinase deadTnni3k allele in mice. A. Diagram of the
wild-type allele and the conversion of the AAA codon encoding K489 to AGA (Arg). An intro-
nic TGGAAA sequence was converted at the same time to the DraI restriction site TTTAAA.
B. Sequence trace of a K489R/+ mouse, illustrating the changes introduced into the gene as in
panel A. C. DraI restriction digest of PCR-amplified genomic DNA from mice of the indicated
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 20 / 24
genotypes.
(PDF)
S4 Fig. Measurement of nuclear ploidy. A. Graphical representation of nuclear ploidy in
mononuclear and binuclear CMs from mice of the indicated genotypes. The first three col-
umns of the Mononuclear panel are duplicated from Fig 3C for easier comparison to the
remainder of the figure. B. Primary data evaluating mononucleated and binucleated CM pop-
ulations for nuclear DAPI fluorescence in ventricle cell preparations from mice of the indi-
catedTnni3k genotypes. Each dot represents one nucleus. Numbers above plots indicate
animal identifiers, and compiled numerical data are shown in S5 Table.
(PDF)
S5 Fig. Quantitation of kinase activity of Tnni3k variants in an in vitro kinase reaction.
293 cells were transfected with plasmids to express full-length wild-type mouse Tnni3k or
human variants introduced into the mouse sequence (numbering based on the mouse protein,
which is one less than the human protein). Within a group of three plasmids, each was trans-
fected in the same experiment and cell lysates prepared and used in an in vitro kinase reaction
at the indicated time points; all samples were run on gels and blotted at the same time and
then probed with antibody and visualized together. Each variant was assayed twice, and all
were assayed with the wild-type construct as an internal reference. B. Quantitation of the nor-
malized signals from the blots shown in panel A. Each experiment was quantitated individually
because blotting or antibody conditions and exposure times may have varied between individ-
ual experiments.
(PDF)
S6 Fig. Tnni3k signaling pathways. A. Comparison of phospho-p38 signal in the same four
P3 samples shown in Fig 5A, now also with two wild-type adult heart samples included on the
same blot. B. Ventricular mononuclear CM% in adult hearts, comparing C57BL/6J vs. C57BL/
6NJ mice.
(PDF)
S1 Table.Tnni3k homologs in other species. Accession numbers forTnni3k or close homo-
logs in human, mouse, sea urchin (S.purpuratus), golden apple snail (P.canaliculate) and Cai-
lifornia two spot octopus (O.bimaculoides) are provided.
(DOCX)
S2 Table. Candidate assembly ofTnni3k gene inHeterocephalusglaber from
NW_004624742.1 by homology to mouse gene.
(DOCX)
S3 Table. Candidate assembly ofTnni3k gene inF.damarensis from NW_011046421.1 by
homology to mouse gene.
(DOCX)
S4 Table. Relatively common humanTNNI3K kinase-domain variants (from ExAC).
(DOCX)
S5 Table. Compilation of diploid nuclei percentages of mononucleated (Mono) and binu-
cleated (Bi) cardiomyocytes in this study. These data correspond to Figs. 2D, 3B, 5B and 5C.
(DOCX)
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 21 / 24
Author Contributions
Conceptualization: Peiheng Gan, Michaela Patterson, Takako Makita, Thomas J. Park, Henry
M. Sucov.
Data curation: Michaela Patterson, Alexa Velasquez, Henry M. Sucov.
Formal analysis: Peiheng Gan, Michaela Patterson, Alexa Velasquez, Ge Tao, Daniel P. Judge,
Henry M. Sucov.
Funding acquisition: Takako Makita, Henry M. Sucov.
Investigation: Peiheng Gan, Michaela Patterson, Alexa Velasquez, Kristy Wang.
Methodology: Peiheng Gan, Michaela Patterson, Alexa Velasquez, Kristy Wang.
Project administration: Henry M. Sucov.
Resources: Di Tian, Jolene J. Windle, Ge Tao, Takako Makita, Thomas J. Park, Henry M.
Sucov.
Software: Alexa Velasquez.
Supervision: Michaela Patterson, Henry M. Sucov.
Writing – original draft: Peiheng Gan, Michaela Patterson, Di Tian, Jolene J. Windle, Ge
Tao, Daniel P. Judge, Takako Makita, Thomas J. Park, Henry M. Sucov.
Writing – review & editing: Henry M. Sucov.
References
1. Jopling C, Sleep E, Raya M, Marti M, Raya A, Izpisua Belmonte JC. Zebrafish heart regeneration occurs
by cardiomyocyte dedifferentiation and proliferation. Nature. 2010; 464: 606–9. https://doi.org/10.1038/
nature08899 PMID: 20336145
2. Kikuchi K, Holdway JE, Werdich AA, Anderson RM, Fang Y, Egnaczyk GF, et al. Primary contribution to
zebrafish heart regeneration by gata4(+) cardiomyocytes. Nature. 2010; 464: 601–5. https://doi.org/10.
1038/nature08804 PMID: 20336144
3. Oberpriller JO, Oberpriller JC. Response of the adult newt ventricle to injury. J Exp Zool. 1974; 187:
249–53. https://doi.org/10.1002/jez.1401870208 PMID: 4813417
4. Robledo M. Myocardial regeneration in young rats. Am J Pathol. 1956; 32: 1215–39. PMID: 13372753
5. Porrello ER, Mahmoud AI, Simpson E, Hill JA, Richardson JA, Olson EN, et al. Transient regenerative
potential of the neonatal mouse heart. Science. 2011; 331: 1078–80. https://doi.org/10.1126/science.
1200708 PMID: 21350179
6. Zhu W, Zhang E, Zhao M, Chong Z, Fan C, Tang Y, et al. Regenerative Potential of Neonatal Porcine
Hearts. Circulation. 2018; 138: 2809–16. https://doi.org/10.1161/CIRCULATIONAHA.118.034886
PMID: 30030418
7. Ye L, D’Agostino G, Loo SJ, Wang CX, Su LP, Tan SH, et al. Early Regenerative Capacity in the Por-
cine Heart. Circulation. 2018; 138: 2798–808. https://doi.org/10.1161/CIRCULATIONAHA.117.031542
PMID: 30030417
8. Westaby S, Archer N, Myerson SG. Cardiac development after salvage partial left ventriculectomy in an
infant with anomalous left coronary artery from the pulmonary artery. J Thorac Cardiovasc Surg. 2008;
136: 784–5. https://doi.org/10.1016/j.jtcvs.2007.10.085 PMID: 18805288
9. Haubner BJ, Schneider J, Schweigmann U, Schuetz T, Dichtl W, Velik-Salchner C, et al. Functional
Recovery of a Human Neonatal Heart After Severe Myocardial Infarction. Circ Res. 2016; 118: 216–21.
https://doi.org/10.1161/CIRCRESAHA.115.307017 PMID: 26659640
10. Mozaffarian D, Benjamin EJ, Go AS, Arnett DK, Blaha MJ, Cushman M, et al. Executive Summary:
Heart Disease and Stroke Statistics-2016 Update, A Report From the American Heart Association. Cir-
culation. 2016; 133: 447–54. https://doi.org/10.1161/CIR.0000000000000366 PMID: 26811276
11. Soonpaa MH, Kim KK, Pajak L, Franklin M, Field LJ. Cardiomyocyte DNA synthesis and binucleation
during murine development. Am J Physiol. 1996; 271: H2183–9. https://doi.org/10.1152/ajpheart.1996.
271.5.H2183 PMID: 8945939
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 22 / 24
12. Li F, Wang X, Capasso JM, Gerdes AM. Rapid transition of cardiac myocytes from hyperplasia to hyper-
trophy during postnatal development. J Mol Cell Cardiol. 1996; 28: 1737–46. https://doi.org/10.1006/
jmcc.1996.0163 PMID: 8877783
13. Kellerman S, Moore JA, Zierhut W, Zimmer HG, Campbell J, Gerdes AM. Nuclear DNA content and
nucleation patterns in rat cardiac myocytes from different models of cardiac hypertrophy. J Mol Cell Car-
diol. 1992; 24: 497–505. https://doi.org/10.1016/0022-2828(92)91839-w PMID: 1386113
14. Brodsky WY, Arefyeva AM, Uryvaeva IV. Mitotic polyploidization of mouse heart myocytes during the
first postnatal week. Cell Tissue Res. 1980; 210: 133–44. https://doi.org/10.1007/bf00232149 PMID:
7407859
15. Xavier-Vidal R, Mandarim-de-Lacerda CA. Cardiomyocyte proliferation and hypertrophy in the human
fetus: quantitative study of the myocyte nuclei. Bull Assoc Anat (Nancy). 1995; 79: 27–31.
16. Mollova M, Bersell K, Walsh S, Savla J, Das LT, Park SY, et al. Cardiomyocyte proliferation contributes
to heart growth in young humans. Proc Natl Acad Sci U S A. 2013; 110: 1446–51. https://doi.org/10.
1073/pnas.1214608110 PMID: 23302686
17. Laflamme MA, Murry CE. Heart regeneration. Nature. 2011; 473: 326–35. https://doi.org/10.1038/
nature10147 PMID: 21593865
18. Hesse M, Doengi M, Becker A, Kimura K, Voeltz N, Stein V, et al. Midbody Positioning and Distance
Between Daughter Nuclei Enable Unequivocal Identification of Cardiomyocyte Cell Division in Mice.
Circ Res. 2018; 123: 1039–52. https://doi.org/10.1161/CIRCRESAHA.118.312792 PMID: 30355161
19. Leone M, Musa G, Engel FB. Cardiomyocyte binucleation is associated with aberrant mitotic microtu-
bule distribution, mislocalization of RhoA and IQGAP3, as well as defective actomyosin ring anchorage
and cleavage furrow ingression. Cardiovasc Res. 2018; 114: 1115–31. https://doi.org/10.1093/cvr/
cvy056 PMID: 29522098
20. Patterson M, Barske L, Van Handel B, Rau CD, Gan PH, Sharma A, et al. Frequency of mononuclear
diploid cardiomyocytes underlies natural variation in heart regeneration. Nat Genet. 2017; 49: 1346–53.
https://doi.org/10.1038/ng.3929 PMID: 28783163
21. Gonzalez-Rosa JM, Sharpe M, Field D, Soonpa MH, Field LJ, Burns CE, et al. Myocardial Polyploidiza-
tion creates a barrier to heart regeneration in zebrafish. Dev Cell. 2018; 44: 433–46. https://doi.org/10.
1016/j.devcel.2018.01.021 PMID: 29486195
22. Fan LL, Huang H, Jin JY, Li JJ, Chen YQ, Zhao SP, et al. Whole exome sequencing identifies a novel
mutation (c.333+2T >C) of TNNI3K in a Chinese family with dilated cardiomyopathy and cardiac conduc-
tion disease. Gene. 2018; 648: 63–7. https://doi.org/10.1016/j.gene.2018.01.055 PMID: 29355681
23. Xi Y, Honeywell C, Zhang D, Schwartzentruber J, Beaulieu CL, Tetreault M, et al. Whole exome
sequencing identifies the TNNI3K gene as a cause of familial conduction system disease and congenital
junctional ectopic tachycardia. Int J Cardiol. 2015; 185: 114–6. https://doi.org/10.1016/j.ijcard.2015.03.
130 PMID: 25791106
24. Theis JL, Zimmermann MT, Larsen BT, Rybakova IN, Long PA, Evans JM, et al. TNNI3K mutation in
familial syndrome of conduction system disease, atrial tachyarrhythmia and dilated cardiomyopathy.
Hum Mol Genet. 2014; 23: 5793–804. https://doi.org/10.1093/hmg/ddu297 PMID: 24925317
25. Podliesna S, Delanne J, Miller L, Tester DJ, Uzunyan M, Yano S, et al. Supraventricular tachycardias,
conduction disease, and cardiomyopathy in 3 families with the same rare variant in TNNI3K (p.
Glu768Lys). Heart Rhythm. 2019; 16: 98–105. https://doi.org/10.1016/j.hrthm.2018.07.015 PMID:
30010057
26. Lek M, Exome Aggregation Consortium. Analysis of protein-coding genetic variation in 60,706 humans.
Nature. 2016; 536: 285–91. ExAC.broadinstitute.org. https://doi.org/10.1038/nature19057 PMID:
27535533
27. Wheeler FC, Tang H, Marks OA, Hadnott TN, Chu PL, Mao L, et al. Tnni3k modifies disease progres-
sion in murine models of cardiomyopathy. PLoS Genet. 2009; 5: e1000647. https://doi.org/10.1371/
journal.pgen.1000647 PMID: 19763165
28. Aberle H, Bauer A, Stappert J, Kispert A, Kemler R. beta-catenin is a target for the ubiquitin-proteasome
pathway. EMBO J. 1997; 16: 3797–804. https://doi.org/10.1093/emboj/16.13.3797 PMID: 9233789
29. Wang H, Chen C, Song X, Chen J, Zhen Y, Sun K, et al. Mef2c is an essential regulatory element
required for unique expression of the cardiac-specific CARK gene. J Cell Mol Med. 2008; 12: 304–15.
https://doi.org/10.1111/j.1582-4934.2007.00155.x PMID: 18021318
30. Patterson BD, Upham NS. A newly recognized family from the Horn of Africa, the Heterocephalidae
(Rodentia: Ctenohystrica). Zool J Linnean Soc. 2014; 172: 942–63.
31. Nei M, Xu P, Glazko G. Estimation of divergence times from multiprotein sequences for a few mamma-
lian species and several distantly related organisms. Proc Natl Acad Sci U S A. 2001; 98: 2497–502.
https://doi.org/10.1073/pnas.051611498 PMID: 11226267
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 23 / 24
32. Fabre PH, Hautier L, Dimitrov D, Douzery EJ. A glimpse on the pattern of rodent diversification: a phylo-
genetic approach. BMC Evol Biol. 2012; 12: 88. https://doi.org/10.1186/1471-2148-12-88 PMID:
22697210
33. Eigenbrod O, Debus KY, Reznick J, Bennett NC, Sanchez-Carranza O, Omerbasic D, et al. Rapid
molecular evolution of pain insensitivity in multiple African rodents. Science. 2019; 364: 852–9. https://
doi.org/10.1126/science.aau0236 PMID: 31147513
34. Grimes KM, Voorhees A, Chiao YA, Han HC, Lindsey ML, Buffenstein R. Cardiac function of the naked
mole-rat: ecophysiological responses to working underground. Am J Physiol Heart Circ Physiol. 2014;
306: H730–7. https://doi.org/10.1152/ajpheart.00831.2013 PMID: 24363308
35. Grimes KM, Barefield DY, Kumar M, McNamara JW, Weintraub ST, de Tombe PP, et al. The naked
mole-rat exhibits an unusual cardiac myofilament protein profile providing new insights into heart func-
tion of this naturally subterranean rodent. Pflugers Arch. 2017; 469: 1603–13. https://doi.org/10.1007/
s00424-017-2046-3 PMID: 28780592
36. Feng Y, Cao HQ, Liu Z, Ding JF, Meng XM. Identification of the dual specificity and the functional
domains of the cardiac-specific protein kinase TNNI3K. Gen Physiol Biophys. 2007; 26: 104–9. PMID:
17660584
37. Tang H, Xiao KH, Mao L, Rockman HA, Marchuk DA. Overexpression of TNNI3K, a cardiac-specific
MAPKKK, promotes cardiac dysfunction. Journal of Molecular and Cellular Cardiology. 2013; 54: 101–
11. https://doi.org/10.1016/j.yjmcc.2012.10.004 PMID: 23085512
38. Rauch J, Volinsky N, Romano D, Kolch W. The secret life of kinases: functions beyond catalysis. Cell
Commun Signal. 2011; 9: 23. https://doi.org/10.1186/1478-811X-9-23 PMID: 22035226
39. Kurosaki T, Popp MW, Maquat LE. Quality and quantity control of gene expression by nonsense-medi-
ated mRNA decay. Nat Rev Mol Cell Biol. 2019.
40. Puente BN, Kimura W, Muralidhar SA, Moon J, Amatruda JF, Phelps KL, et al. The oxygen-rich postna-
tal environment induces cardiomyocyte cell-cycle arrest through DNA damage response. Cell. 2014;
157: 565–79. https://doi.org/10.1016/j.cell.2014.03.032 PMID: 24766806
41. Schriner SE, Linford NJ, Martin GM, Treuting P, Ogburn CE, Emond M, et al. Extension of murine life
span by overexpression of catalase targeted to mitochondria. Science. 2005; 308: 1909–11. https://doi.
org/10.1126/science.1106653 PMID: 15879174
42. Huang TT, Naeemuddin M, Elchuri S, Yamaguchi M, Kozy HM, Carlson EJ, et al. Genetic modifiers of
the phenotype of mice deficient in mitochondrial superoxide dismutase. Hum Mol Genet. 2006; 15:
1187–94. https://doi.org/10.1093/hmg/ddl034 PMID: 16497723
43. Vagnozzi RJ, Gatto GJ Jr., Kallander LS, Hoffman NE, Mallilankaraman K, Ballard VL, et al. Inhibition of
the cardiomyocyte-specific kinase TNNI3K limits oxidative stress, injury, and adverse remodeling in the
ischemic heart. Sci Transl Med. 2013; 5: 207ra141.
44. Lewis KN, Andziak B, Yang T, Buffenstein R. The naked mole-rat response to oxidative stress: just deal
with it. Antioxid Redox Signal. 2013; 19: 1388–99. https://doi.org/10.1089/ars.2012.4911 PMID:
23025341
45. Munro D, Baldy C, Pamenter ME, Treberg JR. The exceptional longevity of the naked mole-rat may be
explained by mitochondrial antioxidant defenses. Aging Cell. 2019; 18: e12916. https://doi.org/10.1111/
acel.12916 PMID: 30768748
46. Hirose K, Payumo AY, Cutie S, Hoang A, Zhang H, Guyot R, et al. Evidence for hormonal control of
heart regenerative capacity during endothermy acquisition. Science. 2019; 364: 184–8. https://doi.org/
10.1126/science.aar2038 PMID: 30846611
47. Buffenstein R, Woodley R, Thomadakis C, Daly TJ, Gray DA. Cold-induced changes in thyroid function
in a poikilothermic mammal, the naked mole-rat. Am J Physiol Regul Integr Comp Physiol. 2001; 280:
R149–55. https://doi.org/10.1152/ajpregu.2001.280.1.R149 PMID: 11124146
Tnni3k and cardiomyocyte composition
PLOS Genetics | https://doi.org/10.1371/journal.pgen.1008354 October 7, 2019 24 / 24

44

CHAPTER 3
Tnni3k loss-of-function causes concentric left ventricular remodeling
and beta-adrenergic-PKA signaling inhibition
Introduction
TNNI3K was first identified as cardiac troponin I (cTnI) interacting kinase, and it is
characterized as a kinase based on protein structure
1
. cTnI is a pivotal regulatory component in
cardiomyocyte sarcomere. It is the inhibitory subunit of the cardiac troponin complex, acting to
inhibit actin-myosin interaction. When Ca
2+
binds to troponin C, conformational changes occur
in cTnI that release the inhibition of troponin T and a-tropomyosin, allowing actomyosin cross-
bridge formation and sarcomere shortening
2
. Although TNNI3K was identified as a novel
binding partner of cTnI, whether TNNI3K regulates the function of cTnI remains unknown, and
whether cTnI is a direct kinase substrate of TNNI3K is still controversial.
A small number of families with rare mutations in the gene TNNI3K have been
characterized for familial cardiac conduction system defects. In all of the studied pedigrees,
progression to dilated cardiomyopathy (DCM, a ventricular myocardial phenotype with
numerous causes) was common
3, 4, 5, 6
, suggesting that TNNI3K functional changes may cause
cardiomyopathy. DCM is the most common cardiomyopathy, and is defined by the dilation of
left ventricle and cardiac contractile dysfunctions. DCM is a severe and decompensatory
cardiomyopathy, which may lead to congestive heart failure or sudden death
7
. Before the
progression into DCM, adaptive morphological changes may take place in response to cardiac

45

dysfunctions. Left ventricular (LV) concentric remodeling is one of the adaptive morphological
changes of heart in response to increased cardiac afterload. To maintain the cardiac output
under increased afterload, the heart has to generate more contractile force than usual. At the
cellular level, more contractile force can be generated by adding cardiomyocyte sarcomeres in
parallel, which results in increased cardiomyocyte width. Thus, the thickening of
cardiomyocytes causes the thickened ventricular wall. Besides the extrinsic conditions
increasing cardiac afterload, like hypertension and aortic stenosis, LV concentric remodeling
can also be caused by cardiomyocyte intrinsic changes either genetic or non-genetic, which is
still far from clear
8
.
Cardiomyocyte contraction is tightly regulated by cytosolic calcium fluctuation, during
which the sarcoplasmic reticulum (SR) plays a pivotal role in calcium release and reuptake.
When cardiomyocyte membrane is depolarized following the impulse, the membranal voltage-
dependent L-type calcium channels open and augment the calcium influx into cardiomyocytes.
Increasing of cytosolic calcium concentration activates the ryanodine receptor (RyR) on the SR
membrane, which is an essential mediator regulating calcium release from SR into cytoplasma
and results in the calcium concentration peak. The increase of cytosolic calcium level triggers
the contraction of cardiomyocyte by activating cTnC and allowing actin-myosin cross-bridging
to proceed. This is a process called calcium-induced calcium release (CICR). The decay of
calcium transient involves outflux of calcium and the reuptake of calcium into SR. On the SR
membrane sarco/endoplasmic reticulum Ca
2+
-ATPase (SERCA) is the major regulator of calcium
reuptake from the cytosol into the SR. The decrease of cytosolic calcium concentration leads to
the relaxation of cardiomyocyte. The activity of SERCA is mainly controlled by the inhibitor

46

phospholamban (PLN)
9
. Thus the activities of RyR and PLN regulate the speed of calcium
release and reuptake in SR, which in turn determine the shortening fraction and shortening,
relaxation velocities of cardiomyocyte.
RyR and PLN are both downstream effectors of beta-adrenergic receptor (β-AR)-PKA
signaling, which is a crucial signaling regulated by sympathetic nervous system (SNS) to provide
a coordinated control of contractility, metabolism and gene regulation in heart all the time. β-
ARs are members of the G protein-coupled receptors (GPCRs) on cardiomyocyte membrane,
which are regulated by SNS through the secretion of cateolamines (CAs) like norepinephrine
and epinephrine to blood. Activation of β-ARs activates adenylyl cyclase (AC), which in turn
catalyzes the adenosine triphosphate (ATP) into cyclic adenosine monophosphate (cAMP). The
rise of cAMP then activates protein kinase A (PKA) by binding to its regulatory subunits and
releasing the catalytic subunits to function. PKA then activates series of downstream targets by
phosphorylation, such as RyR2, PLN, cTnI etc, to initiate subsequent functional responses
10
.
By using the several Tnni3k genetic mouse models, we investigated the genuine role of
Tnni3k in uninjured heart. Our results revealed that Tnni3k loss-of-function (LOF) causes left
ventricular concentric remodeling, which may predispose the heart to DCM or other severe
cardiomyopathies. With our Tnni3k kinase-dead mutant (K489R) mice line, I confirmed the
kinase role of TNNI3K in regulating cardiomyocyte contractility and calcium transient.
Additionally, the molecular mechanism study demonstrated a novel perspective of TNNI3K
function that Tnni3k LOF reduces cardiomyocyte contractility and calcium transient through
blunting the β-AR-PKA pathway.

47

Results
Mice with Tnni3k LOF mutation have left ventricular concentric remodeling
In the previous studies, TNNI3K mutations (G526D, T539A, c333+2T>C) have been
reported in familial cardiac conduction system defects and DCM, which are all severely
hypomorphic LOF mutations
3, 4, 5
. If human TNNI3K alleles associated with DCM are LOF alleles,
it is reasonable to infer that mice with Tnni3k LOF mutations might manifest similar
cardiomyopathy. We analyzed heart functions of 8-month old mice under isoflurane anesthesia
for echocardiography. We have now evaluated 10 Tnni3k homozygous null mutant (KO) mice
and 8 littermate wildtype (WT) mice, all inbred on a C57BL/6 background. We observed a
significant reduction of cardiac ejection fraction (EF%) in Tnni3k ko mice (39.4%) compared with
the wildtype control mice (59.3%) (Figure 1A, 1B). However, no obvious DCM is observed in
Tnni3k ko mice. Cardiac EF% reflects the systolic function of LV, and the reduction of EF% seen
in Tnni3k ko mice implicates that the blood pumping capacity of ko heart is compromised.
Interestingly, the Tnni3k ko mouse heart has a thicker LV wall and ventricular septum, which is
an opposite morphological change to DCM (Figure 1D, 1E). However, the volume and dimension
of LV chamber are not significantly different between wildtype and ko mice (Figure 1C, 1F, 1G).
To clarify whether the thickened myocardium is because of LV hypertrophy, I measured the
heart weight to body weight ratio, LV mass to body weight ratio (Figure 2A, 2B). I found no
change in heart weight between wildtype and ko mice, which suggests that Tnni3k ko mice
don't have LV hypertrophy, and the morphological change of Tnni3k ko mice heart resembles
concentric LV remodeling.

48

LV concentric remodeling is an adaptive change in cardiac geometry caused by
increased cardiac afterload, as introduced before. LV concentric remodeling is a compensated
status under pressure overload before proceeding to concentric or eccentric hypertrophy or
congestive heart failure
8
. To evaluate whether Tnni3k ko mice have pressure overload
conditions, such as hypertension. I evaluated mouse blood pressure by a noninvasive tail-cuff
blood pressure system, but I didn't see any difference in systolic or diastolic blood pressure
between Tnni3k wildtype and ko mice (Figure 2C, 2D). Although Tnni3k ko mouse hearts
displayed reduced ejection fraction and LV concentric remodeling, the cardiac output is
maintained and overall heart function is still preserved, as shown by the unchanged blood
pressure. Moreover, as we've previously reported, Tnni3k ko mice are externally normal, and
have a normal life span
11
.
To confirm the morphological changes of Tnni3k ko mouse heart, I performed
histological analysis with 6-month-old mouse hearts by trichrome staining, and found that
Tnni3k ko mouse has a thicker LV wall than wildtype (Figure 2E, 2F). However, the overall
volume of Tnni3k ko mouse heart is virtually smaller than wildtype (Figure 2G). With a decrease
of heart volume and increase of myocardium thickness, but no change in heart weight, this
morphological change of ko mouse heart can be characterized as LV concentric remodeling. To
investigate whether this morphological change of heart is a congenital or an acquired
phenotype, I analyzed mice at early stages of life, postnatal day 14 (P14), 4 weeks of age, but no
obvious difference of LV wall and ventricular septum thickness were seen between wildtype
and ko mouse hearts, suggesting that the morphological change of ko mouse heart is an
acquired phenotype after birth (Figure 2H, 2I).

49

Cardiomyocyte architectural changes caused LV concentric remodeling
Thickened ventricular wall can be a result of abnormal enlargement or thickening of
cardiomyocytes, or changes in other cardiac components, such as excessive fibrosis. Our
trichrome staining of Tnni3k ko heart sections didn't reveal any obvious collagen deposition,
compared with aortic valve regions which is collagen-rich (Figure 2E), suggesting that the
morphological changes of Tnni3k ko hearts might be because of the architectural changes of
cardiomyocyte. Tnni3k is cardiomyocyte specific in expression, it is unlikely to affect the blood
vessel physiology to change cardiac afterload. Therefore, the concentric remodeling of LV is
most likely because of the intrinsic changes in cardiomyocyte. Hence, I analyzed the size and
shape of isolated adult cardiomyocytes (6-month-old). Considering the difference in
mononuclear cardiomyocyte frequency between wildtype and ko mice, I only measured the
binucleated cardiomyocytes, which make up of 90% of all cardiomyocytes and could represent
the overall cardiomyocyte morphology. I found that the cardiomyocyte area is not different
between wildtype and ko, while the ko cardiomyocyte tends to be wider and shorter (Figure 3A,
3B, 3C). To understand whether the morphological changes of ko cardiomyocyte are because of
sarcomere disorganization, I performed immunofluorescence staining on isolated
cardiomyocyte for alpha-actinin 2, which is localized to Z-disc of sarcomere and commonly used
as a marker to visualize the sarcomere structure. Confocal microscopy of cardiomyocyte with
alpha-actinin 2 immunofluorescence staining revealed a detailed sarcomere structure, but no
obvious disorganization was observed in Tnni3k ko cardiomyocyte (Figure 3D). Additionally, I
didn't find any difference in the sarcomere length between wildtype and ko cardiomyocyte
(Figure 3E).

50

I've demonstrated that the LV concentric remodeling of Tnni3k ko mouse heart shows
an age-related pattern (Figure 2E, 2F, 2H, 2I). To investigate whether cardiomyocyte
architectural changes also display an age-related pattern, I analyzed the binucleated
cardiomyocyte from P14 pup hearts, which haven't shown LV concentric remodeling yet (Figure
2H). However, no obvious thickening of cardiomyocyte was seen in ko hearts at P14, which
implicated that cardiomyocyte architectural changes are also age-related and similar to heart
morphological changes (Figure 3F, 3G, 3H). Although it is possible that the thickened
myocardium in Tnni3k ko mouse is because of some other extracellular matrix components
accumulation other than collagen, considering the cardiomyocyte-specific expression pattern of
Tnni3k, the primary effect of Tnni3k LOF most likely causes cardiomyocyte intrinsic changes.
The thickening of ko cardiomyocyte explained the thickened myocardium of LV in Tnni3k ko
mice. Meanwhile, the area of ko cardiomyocyte does not differ from wildtype, given that the
heart weight of Tnni3k ko mouse doesn't differ from wildtype mice (Figure 2A, 2B), I infer that
the number of cardiomyocyte is not changed, even though Tnni3k LOF changes the ratio of
mononuclear cardiomyocyte as we've previously demonstrated
11
. With the same number of
cardiomyocyte and heart weight, the shorter and wider cardiomyocyte results in the thicker
myocardium but smaller overall heart size
Although I didn't distinguish left and right ventricular cardiomyocyte, given that the left
ventricle makes up of 66% ventricle weight, my cardiomyocyte size analysis should
predominantly be LV cardiomyocyte, which can explain the morphological change of LV. The
biological significance of reducing cardiomyocyte aspect ratio (length: width) is to compensate
the contractile force loss in physiological or pathological circumstances
12, 13
. I hypothesized that

51

the reduced aspect ratio in ko cardiomyocyte was a compensatory effect to the contractility
loss after Tnni3k LOF. In other words, Tnni3k LOF compromises cardiomyocyte contractility.
Tnni3k LOF cardiomyocyte has reduced contractility under stress
To evaluate the contractility of Tnni3k ko cardiomyocyte, I used freshly isolated viable
cardiomyocytes and performed the contractility measurement by IonOptix instruments.
However, I didn't observe any basal contractility difference between wildtype and ko
cardiomyocytes, as indicated by the sarcomere shortening fraction (Figure 4A, 4B). It is possible
that there is too subtle contractility difference between wildtype and ko cardiomyocyte to be
detected. Therefore, I applied a mild concentration of isoproterenol (iso, 10 nM), which is
known to induce inotropic effects on cardiomyocyte, to stress cardiomyocytes. Under Iso
stimulation, Tnni3k ko cardiomyocyte showed a significantly lower contractility (11%) compared
with wildtype cardiomyocyte (14%) (Figure 4A, 4B). Interestingly, Tnni3k kinase-dead mutant
(K489R) cardiomyocyte displayed the same level of contractility reduction as ko cardiomyocyte
under stress (Figure 4A, 4B), which implicates that TNNI3K functions as a kinase in affecting
cardiomyocyte contractility. Besides shortening fraction, the shortening and relaxation
velocities of cardiomyocyte are also blunted in ko and K489R cardiomyocytes under iso
stimulation (Figure 4C, 4D). As discussed in the introduction, cardiomyocyte contraction is
regulated by calcium fluctuation. Tnni3k ko cardiomyocyte displayed not only a reduced
contractility under stress, but also the shortening and relaxation velocity (Figure 4C, 4D),
suggesting that the reduced contractility in ko cardiomyocyte may be because of the inefficient
calcium release and reuptake of SR. TNNI3K has been reported to work synergistically with

52

Calsequestrin to modify the progression of dilated cardiomyopathy in mouse
14
. Calsequestrin is
a calcium-binding protein, which helps to regulate calcium storage in SR. The interaction
between TNNI3K and Calsequestrin implicates a potential involvement of TNNI3K in
cardiomyocyte calcium handling.
Tnni3k LOF cardiomyocyte has reduced calcium fluctuation under stress
I next tested the calcium transient of cardiomyocyte with or without isoproterenol
treatment by IonOptix instruments with fluorescent calcium indicator Fura-2. At basal level, no
difference was observed in calcium amplitude between wildtype and ko cardiomyocyte (Figure
5A, 5B). After iso treatment, the ko cardiomyocyte displayed a lower level of calcium amplitude
compared with wildtype. This result explained the reduced contractility seen in ko
cardiomyocyte under iso stimulation was because of the blunted calcium transient (Figure 4A,
4B).
To understand why ko cardiomyocyte has reduced calcium amplitude compared with
wildtype under iso stimulation. I analyzed the calcium upstroke and decay velocity, and found
that ko cardiomyocyte also has decreased speeds of calcium release and reuptake (Figure 5C,
5D). The K489R mutant cardiomyocyte also displayed reduced calcium transient, calcium
upstroke and decay velocity, which substantiates that TNNI3K functions as kinase in regulating
the calcium fluctuation (Figure 5A, 5B, 5C, 5D). The reduced contractility and calcium transient
of Tnni3k ko cardiomyocyte may lead to the reduced aspect ratio of cardiomyocyte to
compensate the contractility loss under stress, which in turn causes the concentric LV
remodeling.

53

Tnni3k LOF causes blunted beta-adrenergic-PKA signaling
Although the calcium upstroke and decay are partially because of ion channels activities
on cell membrane, the SR predominantly regulates the calcium fluctuation in cardiomyocyte.
Mouse cardiomyocyte cycles 90% of cytosolic calcium through the SR but only 10% through
extracellular spaces
9
. The activities of RyR and PLN determine the calcium release and reuptake
speeds in SR, which further determine the shortening and relaxation velocities of
cardiomyocyte. Therefore, I hypothesized that the reduced contractility and calcium transient
in ko cardiomyocyte was because of the reduced activities of calcium regulators on SR, such as
RyR, SERCA, and PLN.
The activities of RyR and PLN are positively correlated with their phosphorylation levels.
Thus I analyzed the basal phosphorylation level of RyR and PLN by using fresh heart ventricular
lysate of adult Tnni3k wildtype and ko mice, but there is no obvious difference between them
(Figure 6A, 6B). This result may explain why we didn't see any difference in cardiomyocyte
contractility and calcium transient between wildtype and ko at basal level. Then I tested the
phosphorylation status of RyR and PLN after iso treatment. Instead of treating mice with iso
and analyzing the heart lysate, I used isolated ventricular cardiomyocyte in culture, which
turned out to be a clean in vitro system to study the phosphorylation of PKA targets in a well-
controlled and consistent manner. The isolated cardiomyocytes were placed in culture dish, and
treated with gradient concentration of iso at 0, 10, 30, 100 nM for 8 min before lysing. The ko
cardiomyocyte displayed compromised phosphorylation of RyR and PLN under iso treatment
(Figure 6C, 6D, 6G). So far, I have demonstrated that the activation of RyR and PLN under iso

54

treatment was reduced in ko cardiomyocyte, which is consistent with the previous
cardiomyocyte contractility and calcium fluctuation findings.
RyR and PLN are both PKA substrates and localized to the SR membrane. It is likely that
the activation of PKA targets were preferentially blunted on the SR membrane in ko mice heart,
given that the subcellular localization of PKA is precisely regulated by a group of scaffold
proteins A-kinase anchoring proteins (AKAPs) to allow specific targeting of substrates
15
. To
understand whether RyR and PLN were preferentially-affected PKA targets in Tnni3k ko mouse
heart, I analyzed other PKA targets located in different subcellular spaces, like cTnI on the
sarcomere, CREB1 in cytoplasma and nuclear. Again, I found no basal phosphorylation level
difference of cTnI and CREB1 between wildtype and ko heart lysate (Figure 6A, 6B), while the
phosphorylation was reduced in ko cardiomyocyte after iso treatment (Figure 6C, 6E, 6F). This
result strongly implicated that the whole PKA signaling was blunted in Tnni3k ko mouse heart.
Since isoproterenol is a commonly used β-AR receptor agonist, to understand how Tnni3k
affects the activation of PKA targets, I investigated the β-AR-PKA signaling.
To elucidate whether the blunted PKA signaling in ko moue heart is because of
inefficient β-AR activation or cAMP production, I treated cardiomyocyte in vitro with gradient
level of 8-Br-cAMP, which is a membrane-permeable, non-degradable derivative of cAMP, to
directly activate PKA. Interestingly, I observed a similar reduction of PKA targets
phosphorylation in Tnni3k ko cardiomyocyte to the iso treatment (Figure 6H, 6I, 6J, 6K, 6L).
These results rejected the possibility that the blunted PKA signaling in Tnni3k ko mouse heart
was because of inefficient β-AR activation or cAMP production. On the other hand, it was also

55

possible that the increased phosphatase activity caused the reduction of PKA substrates
phosphorylation in Tnni3k ko mouse heart. To resolve this puzzle, I performed an experiment to
evaluate the phosphatase activity in cardiomyocyte. I fully activated β-AR-PKA signaling in
isolated cardiomyocyte by a maximal level of iso (300 nM) treatment, then replaced the
medium with fresh medium supplemented with propranolol, which is a β-AR antagonist, to
diminish the residual activity of iso. I lysed cardiomyocyte at different time points and analyzed
PLN phosphorylation level changes. As expected, PLN phosphorylation level is decreasing in a
time-course manner, which reflects the natural dephosphorylation by phosphatase (Figure 6M).
However, there is no difference in the dephosphorylation rate between wildtype and ko, as
shown in the curve (Figure 6N). PLN is mainly dephosphorylated by protein phosphatase (PP1),
PP2A, which are the two most common protein phosphatases in heart and dephosphorylating
other PKA substrates, such as RyR2, cTnI etc. Although PLN is the only PKA target I tested for
the dephosphorylation rate, it indicated that the overall protein phosphatase activity is not
changed. This result rejected the possibility that the reduced PKA targets phosphorylation was
because of the increased phosphatase activity.
So far, I can draw a conclusion that Tnni3k is necessary to maintain the regular activity
of PKA signaling in cardiomyocyte, while Tnni3k LOF blunts β-AR-PKA signaling and reduces the
contractility and calcium transient of cardiomyocyte under iso stimulation. However, the
underlying mechanisms of how Tnni3k LOF blunts PKA signaling remain unknown.
Human polymorphism Tnni3k-I685T mice have blunted beta-adrenergic-PKA signaling

56

Having confirmed that Tnni3k LOF function causes the LV concentric remodeling,
cardiomyocyte contractility and calcium transient reduction and β-AR-PKA signaling blunting, I
wanted to investigate if human hypomorphic TNNI3K mutations also cause the same
abnormalities. Three human hypomorphic TNNI3K mutations (G526D, T539A, c333+2T>C) have
been identified in families carrying cardiac conduction defects and DCM, while all three are rare
mutations in human population
3, 4, 5
. In our previous study
16
, I have identified multiple common
TNNI3K hypomorphic polymorphisms in human, in which TNNI3K-I685T is the most common
one. TNNI3K-I685T mutant is found in approximately 2% human population world-wide, which
preserves 38% of regular kinase activity based on our previous study
16
. I have also validated
that TNNI3K functions as a kinase in regulating cardiomyocyte contractility, calcium transient
and β-AR-PKA signaling. Thus, I hypothesized that the hypomorphic human polymorphism
TNNI3K-I685T may also affect heart physiology to some extent.
To demonstrate the in vivo function of Tnni3k-I685T mutation, we created a Tnni3k-
I685T knock-in allele in C57BL/6J mice by CRISPR-mediated homologous replacement, changing
the isoleucine codon at position 685 to threonine (ATC to ACA) in exon 21 (Figure 7A, 7B).
Founder mice heterozygous for the knock-in alleles were crossed to wild-type C57BL/6J
partners to confirm germline establishment of the allele, and heterozygous F1 mice then
crossed to C57BL/6J-inbred Tnni3k ko mice to generate littermate mice that were used for
analysis. Tnni3k-I685T mice, like ko (-/-) mice, were externally normal. Ventricular lysates
prepared from adult hearts demonstrated a stable expression of TNNI3K-I685T protein (Figure
7C). I performed the mononuclear cardiomyocyte frequency analysis first, and found that
Tnni3k-I685T allele (I685T/-) had a significantly increased level of mononuclear cardiomyocyte

57

relative to heterozygous (+/-) mice and equivalent to the level in ko (-/-) mice (Figure 7D). Thus,
this hypomorphic Tnni3k-I685T mutant allele is elevating mononuclear cardiomyocyte
frequency as Tnni3k ko allele, which implicates that the reduced kinase activity of TNNI3K might
be penetrant to affect heart physiology.
To investigate whether β-AR-PKA signaling is also affected in Tnni3k-I685T mutant
cardiomyocyte, I performed the iso treatment experiment on isolated cardiomyocytes. I
observed that Tnni3k-I685T cardiomyocyte also showed a reduced phosphorylation of PKA
targets, RyR2, PLN, cTnI as Tnni3k ko cardiomyocyte (Figure 7E). This result suggests that the
hypomorphic Tnni3k-I685T mutation also blunts the β-AR-PKA signaling in cardiomyocyte. I
hypothesized that the blunted β-AR-PKA signaling in Tnni3k-I685T mutant may also change the
heart physiology. Thus, I performed echocardiography analysis on Tnni3k-I685T mice. However,
we didn't observe any obvious functional changes of Tnni3k-I685T mouse hearts, as
demonstrated by EF%, EDV, ESV (Figure 7A, 7B, 7C). I found that Tnni3k-I685T mouse has
thicker LV posterior wall (Figure 7D, 7E, 7F), suggesting that the Tnni3k-I685T mouse hearts also
display mild LV concentric remodeling as Tnni3k ko mice. Although Tnni3k-I685T mouse didn't
show obvious functional changes, the mutant allele affected MNCM frequency level, blunted
PKA signaling, and caused LV remodeling as Tnni3k ko mice. Thus, it is reasonable to infer that
the Tnni3k-I685T mutation may predispose heart to more severe abnormalities.
Discussion
Tnni3k is an ancient gene present in all mammals, and recognizable Tnni3k homologs
exist in numerous invertebrate species, which suggests the indispensable role of Tnni3k in

58

animal evolution. Tnni3k is cardiac specific in mammals, but also expressed in animals don't
have hearts, like sea urchin
16
. During evolution, animals developed sophisticated organ systems
to adapt to environmental changes and increasing physiological requirements. Tnni3k might be
a derivative from an ancestral gene to function specifically in hearts. As we've shown, the
deletion of Tnni3k causes LV concentric remodeling and cardiomyocyte architectural changes,
but the overall heart function is still preserved and life-spans of deletion mutant mice are
virtually normal. These results suggest that Tnni3k may not function independently as a pivotal
protein in heart. The seven ankyrin repeats on N terminus of TNNI3K protein, which are the
most common protein-protein interaction motifs, strongly implicate a protein-protein binding
role of TNNI3K
1
. Thus, it is likely that TNNI3K works synergistically with other binding partners
in a regulatory complex, in which the missing of TNNI3K dampens the regulatory machinery but
does not cause lethal effect to animals.
As we've demonstrated, Tnni3k LOF affects cardiomyocyte contractility or calcium
fluctuation under iso stimulation, but not at basal level. This is an evidence suggesting the
biological significance of Tnni3k is in stressed instead of normal conditions. Therefore, Tnni3k
LOF may not lead to, but predispose the heart to cardiomyopathy or other heart diseases, like
arrhythmias. Fight-or-flight response is a common physiological reaction when animals are
facing harmful events or threats to life, and it is a process orchestrated by multiple organs,
systems, and hormones. In heart, increased level of epinephrine and norepinephrine activates
β-AR-PKA signaling to increase cardiomyocyte contractility and cardiac output in fight-or-flight
response, while the blunting of β-AR-PKA signaling after Tnni3k LOF can adversely change the
cardiac function. However, different species and different individuals are facing different

59

environmental conditions, which result in different frequency of fight-or-flight response in their
lives. Additionally, Tnni3k LOF can converge with other cardiac risk factors either genetic or
non-genetic to influence heart functions. Therefore, the outcomes of Tnni3k LOF might be
dramatically different between individuals. That's may explain why TNNI3K hypomorphic and
LOF mutations are common in human populations, but very few of them have been linked to
heart diseases. Although the direct influence of Tnni3k LOF is subtle in heart, this gene is still an
important regulatory unit of cardiomyocyte physiology, which may explain why it is conserved
across species.
There are naturally occurring mouse strains don't express Tnni3k because of a mutation
at rs49812611 in Tnni3k gene causing alternative splicing and NMD of Tnni3k mRNA, such as A
strain, DBA/2 strain. But why those mouse strains haven't displayed the same phenotypes as
our Tnni3k ko mice in C57BL/6 background. One explanation is that those mice strains may also
have cardiac abnormalities similar to Tnni3k ko mice, but too subtle to be found out. Only when
coupled with other genetic factors in strain background, Tnni3k LOF can manifest obvious
abnormalities. For example, the DBA/2 strain has been reported as a natural hypertrophic
cardiomyopathy (HCM) model
17
. Another explanation is that, there are genes compensating
Tnni3k loss in other mouse strains.
Materials and Methods
Animals
Tnni3k knock-out (Patterson et al., 2017), Tnni3-K489R (Gan et al., 2019) alleles used in
this study were also used in our prior work, and maintained on a C57BL/6 background.

60

Derivation of Tnni3k-I685T allele followed the procedure as creating Tnni3-K489R (Gann et al.,
2019). Animals were euthanized by isoflurane anesthesia followed by cervical dislocation and
removal of hearts. Animal research was reviewed and approved by the IACUC committees of
the Univ. of Southern California (#10173) and of the Medical Univ. of South Carolina (2018-
00642), and all experiments were performed in accordance with relevant guidelines and
regulations.
Echocardiography
Mice were anesthetized with 3–5% isoflurane vapor in an anesthesia chamber, weighed,
and then placed on a biofeedback warming station with nose cone anesthesia of 1.5–2.5%
isoflurane, which was regulated to maintain a heart rate between 400 to 500 beats/min, while
providing anesthesia (abolition of the toe pinch reflex). The hair over the chest was removed
using a commercially available depilatory cream (Nair®). Electrical coupling gel was applied to
the electrical contact pads on the heated stage and proper contact of the animal’s paws to the
gel was ensured. The electrocardiogram signal was isochronously recorded and used for timing
and synchronization of the recorded echocardiographic images to the cardiac cycle. Ultrasound
gel was placed on the chest, and echocardiography measurements were performed using a 40-
MHz probe with a spatial resolution of 30 μm (Vevo3100; Visualsonics). Two-dimensional and
M-mode echo images were obtained in the parasternal short- and long-axis views. LV volumes
and ejection fractions were computed from the parasternal long-axis recordings and LV mass
was computed from the short axis measurements.1-3 The entire echocardiography procedure
took 30-45 minutes per mouse.

61

Tail-cuff mouse blood pressure measurements
The blood pressure of male, age-matched animals (2–4 months) was measured by tail
cuff on awake, conscious animals during daytime using a two-channel CODA system (Kent
Scientific, Torrington, CT), following the manufacturer's instructions and using 20
acclimatization and 10 measurement cycles.
Cardiomyocyte isolation and calcium reintroduction
Hearts were digested with 2.4 mg/mL collagenase type II in perfusion buffer (NaCl 120.4
mM, KCl 14.7 mM, KH
2
PO
4
0.6 mM, Na
2
HPO
4
0.6 mM, MgSO
4
.7H
2
O 1.2 mM, Na-HEPES 10 mM,
NaHCO3 4.6 mM, Taurine 30 mM, BDM 10 mM, Glucose 5.5 mM) via Langendorff retroaortic
perfusion. After digestion, atria and valves were removed and ventricular tissue alone was
triturated in stop buffer (perfusion buffer with 10% Fetal Bovine Serum, CaCl
2
12.5 µM), filtered
through a 250-μm nylon mesh. Spin down cardiomyocytes at 20 g for 3 min at room
temperature, aspirate supernatant and wash cardiomyocytes with 10 mL stop buffer once. Prior
to the calcium reintroduction, add ATP to the tube to reach the 2 mM final concentration of
ATP. Prepare three 15-mL tubes containing 10 mL myocyte stopping buffer and 100 µM, 400
µM, 900 µM CaCl
2
. Resuspend and spin down (20 g for 3 min) cardiomyocytes in three tubes of
calcium reintroduciton buffer sequentially, and let the cardiomyocytes stand for 2 min in each
buffer. After finishing the last reintroduction buffer, cardiomyocytes can be used for culture
and contractility and calcium transient measurements.
Cardiomyocyte contractility and Ca2+ transient measurements

62

Sarcomere shortening and Ca
2+
transient measurements were performed using IonOptix
calcium and contractility following the manufacturers’ standard operating instructions. Briefly:
freshly isolated cardiomyocytes were loaded with fura2 Ca
2+
indicator by incubation with
2ng/µL fura2-AM (Life Technologies) in ‘stop buffer’ containing 1.2 mM CaCl
2
for 5mins in dark
at room temperature, followed by a 10 minute wash in stop buffer containing 1.2 mM CaCl
2
to
remove any excess label. The loaded cells were then allowed to settle to the bottom of a
perfusion chamber with a 1.5 thickness cover slip base, which was mounted on an inverted
fluorescence microscope. Cells were electrically paced at 20 volts and pacing frequency was set
at 1Hz. Sarcomere shortening was captured by Fourier transform of the cardiomyocyte
striations under phase contrast microscopy using a switching rate of 100Hz. fura2
Ca
2+
transients were captured simultaneously, using the ratio of fura2 fluorescence emission at
365/380nm at a switching rate of 1000Hz. All contracting cardiomyocytes were measured for
contractility and fura2 Ca
2+
, any cells displaying asynchronous contractility, excessive
blebbing/dysmorphology, and abnormally high or low shortening fraction or calcium amplitude
were ignored for acquisition. To compare sarcomere shortening and fura2 Ca
2+
transients in the
presence of isoproterenol (10 nM), 3 µL of 1 µM iso were added into 300 µL fura2-AM-loaded
cells right before recording. No preparation of cells was left for more than 10 mins before being
replaced for a fresh batch of pre-incubated cells. At least 10 cells were analyzed for each
sample with/without iso treatment. At least 6 groups of cardiomyocytes from different animals
were analyzed per genotype.
Cardiomyocyte culture

63

Coat 12-well plate with laminin (Thermo Fisher Scientific 23017015) at 10 µg/mL in PBS.
Incubate the plate at 4°C with gentle shaking on a rocker platform overnight. Prepare
cardiomyocye culture medium (Medium 199, Earle's salts 42.5 mL/50 mL, Fetal bovine serum
10%, BDM 10 mM, penicillin 100 U/mL, gultamine 2 mM) and plating medium (Medium 199,
Earle's salts 48.5 mL/50 mL, BSA 0.1%, penicillin 100 U/mL, gultamine 2 mM ). Equilibrate both
at 37 °C in a 2% CO
2
incubator for at least 2 h to adjust temperature and pH. Remove the
laminin coating solution just prior to plating the cardiomyocyte. Resuspend freshly isolated
cardiomyocyte in plating medium, approximately 2 mL/heart. Calculate the total number of
rod-shaped myocytes and determine the volume of myocyte suspension need for each well,
and plate 50,000 rod-shaped myocytes/well. Immediately place finished plate in a 2% CO
2

incubator at 37 °C, and incubate for 1 h to allow cardiomyocyte attachment. After 1 h, gently
aspirate the plating medium and wash each well with 1 mL pre-equilibrated cardiomyocyte
culture medium to remove dead cardiomyocyte. Add 1 mL culture medium into each well and
put the plate back to the 2% CO
2
incubator at 37 °C. After an additional hour, treat
cardiomyocytes in different wells with gradient levels of isoproterenol (Millipore Sigma I5627)
at 0 nM, 10 nM, 30 nM, 100 nM for 8 min, or with 8-Br-cAMP at 0 µM, 40 µM, 80 µM, and 120
µM for 10 min. For PKA substrate dephosphorylation assay, cardiomyocytes were treated with
300 nM iso for 8 min, then replaced with fresh culture medium supplemented with propranolol
(Fisher scientific 06-241-00) 100nM.
Histology

64

Heart samples were dehydrated through increasing ethanol concentrations and then
embedded in paraffin. 10 µm sections were used for histological studies. To visualize heart
morphology, Masson trichrome stain kit (Richard-Allen Scientific) was used. Evaluation of
ventricular wall thickness was determined by using ImageJ software, and series of transverse
section from at least 5 hearts per genotype.
Western Blot
Heart ventricular tissue was snap frozen in liquid nitrogen then homogenized with an
OMNI TH homogenizer in lysis buffer (NaCl 150 mM, Tri-HCl pH=7.5, 50 mM, NP-40 0.1%, 1x
Complete protease inhibitor mixture (Roche), and 1x PhosStop phosphatase inhibitor (Roche))
on ice. Heart lysates were centrifuged at 14,000 RPM for 20 min at 4°C to remove insoluble
material. For cardiomyocyte lysate, aspirate medium after treatment, and wash cardiomyocytes
with 1 mL/well ice-cold PBS twice. Then add 150 µL ice-cold lysis buffer into each well to lyse
cardiomyocyte by pipetting. Centrifuge the lysate at 14,000 rpm for 20 min at 4 °C to remove
insoluble material. Immunoblotting was performed by standard protocols with 50μg heart
lysate using anti-Tnni3k (1:1,000, Invitrogen PA5-21989), anti-GAPDH (1:1,000, GeneTex
GT239), and HRP-coupled secondary (1:10,000, Jackson ImmunoResearch sc2040) antibodies. 5
μg cardiomyocyte lysate was loaded and blotted in the same manner, then using anti-phospho-
Ryanodine receptor 2(RYR-2)-S2808 (1:1,000, abcam ab59225), Anti-RyR2 (1:1,000, Invitrogen
PA5-38329), anti-phospho-CREB1 -S133 (1:1,000, Cell Signaling Technology, 9196), anti-CREB1
(1:1,000, Cell Signaling Technology, 9104), anti-phospho-cardiac troponin I (cTnI)-S23/24
(1:1,000, Cell Signaling Technology, 4004), anti-cTnI (1:1,000, abcam ab47003), anti-phospho-

65

phospholamban (PLN)-S16 (1:1,000, Millipore Sigma, 07-052), anti-PLN (1:1,000, abcam
ab2865), anti-PKA-c-alpha (1:1,000, Cell Signaling Technology, 5842), anti-GAPDH (1:1,000,
GeneTex GT239). Quantification of Western blotting signal was done using ImageJ.
Cardiomyocyte architectural analysis
Freshly isolated cardiomyocytes were fixed in 4% PFA for 15 min. Blocking and
permeabilization of cardiomyocyte with 10% Normal Goat Serum (NGS) (Thermo Fisher
Scientific, 50062Z) at room temperature with gentle agitation. Cardiomyocytes were stained for
anti-cTnT (1:500, abcam, ab8295) and Alexa Fluor secondary (1:250, ThermoFisher A11001) and
DAPI using standard procedures. Cells were coverslipped with ProLong Gold antifade reagent
(Invitrogen). Fluorescent images were captured with 20X objective and the same settings for all
slides. The area, length and width of binucleated cardiomyocytes were analyzed with ImageJ.
For cardiomyocyte sarcomere structure imaging, cells were stained for ant-alpha-actinin2
(1:1,000, Invitrogen 701914) and Alexa Fluor secondary (1:250, ThermoFisher A11001).
Confocal microscopy was performed using Leica TCS SP5 microsystems with 60X objective.
Images were analyzed with ImageJ.

66

References
1. Zhao Y, Meng XM, Wei YJ, et al. Cloning and characterization of a novel cardiac-specific
kinase that interacts specifically with cardiac troponin I. J Mol Med (Berl). 2003;81(5):297–304.
doi:10.1007/s00109-003-0427-x
2. Polakit Teekakirikul, Carolyn Y.Ho, Christine E.Seidman. Chapter 47 - Inherited
Cardiomyopathies. Emery and Rimoin's Principles and Practice of Medical Genetics (Sixth
Edition) 2013, Pages 1-38
3. Xi Y, Honeywell C, Zhang D, et al. Whole exome sequencing identifies the TNNI3K gene as a
cause of familial conduction system disease and congenital junctional ectopic tachycardia. Int J
Cardiol. 2015;185:114–116. doi:10.1016/j.ijcard.2015.03.130
4. Theis JL, Zimmermann MT, Larsen BT, et al. TNNI3K mutation in familial syndrome of
conduction system disease, atrial tachyarrhythmia and dilated cardiomyopathy.Hum Mol
Genet. 2014;23(21):5793–5804. doi:10.1093/hmg/ddu297
5. Fan LL, Huang H, Jin JY, et al. Whole exome sequencing identifies a novel mutation
(c.333 + 2T > C) of TNNI3K in a Chinese family with dilated cardiomyopathy and cardiac
conduction disease. Gene. 2018;648:63–67. doi:10.1016/j.gene.2018.01.055
6. Podliesna S, Delanne J, Miller L, et al. Supraventricular tachycardias, conduction disease, and
cardiomyopathy in 3 families with the same rare variant in TNNI3K (p.Glu768Lys).Heart Rhythm.
2019;16(1):98–105. doi:10.1016/j.hrthm.2018.07.015

67

7. Robert G Weintraub, Christopher Semsarian, Peter Macdonald. Dilated cardiomyopathy. The
Lancet. VOLUME 390, ISSUE 10092, P400-414, JULY 22, 2017
DOI:https://doi.org/10.1016/S0140-6736(16)31713-5
8. Beverly H. Lorell, Blase A. Carabello. Left Ventricular Hypertrophy Pathogenesis, Detection,
and Prognosis. Circulation. 2000;102:470–479.https://doi.org/10.1161/01.CIR.102.4.470
9. MacLennan, D., Kranias, E. Phospholamban: a crucial regulator of cardiac contractility. Nat
Rev Mol Cell Biol 4, 566–577 (2003). https://doi.org/10.1038/nrm1151
10. Madamanchi A. (2007). Beta-adrenergic receptor signaling in cardiac function and heart
failure. McGill journal of medicine : MJM : an international forum for the advancement of
medical sciences by students, 10(2), 99–104.
11. Patterson M, Barske L, Van Handel B, et al. Frequency of mononuclear diploid
cardiomyocytes underlies natural variation in heart regeneration. Nat Genet. 2017;49(9):1346–
1353. doi:10.1038/ng.3929
12. Kuo PL, Lee H, Bray MA, et al. Myocyte shape regulates lateral registry of sarcomeres and
contractility. Am J Pathol. 2012;181(6):2030–2037. doi:10.1016/j.ajpath.2012.08.045
13. Rosen BD, Edvardsen T, Lai S, et al. Left ventricular concentric remodeling is associated with
decreased global and regional systolic function: the Multi-Ethnic Study of
Atherosclerosis. Circulation. 2005;112(7):984–991. doi:10.1161/CIRCULATIONAHA104.500488
14. Wheeler FC, Tang H, Marks OA, et al. Tnni3k modifies disease progression in murine models
of cardiomyopathy. PLoS Genet. 2009;5(9):e1000647. doi:10.1371/journal.pgen.1000647

68

15. Skroblin P, Grossmann S, Schäfer G, Rosenthal W, Klussmann E. Mechanisms of protein
kinase A anchoring. Int Rev Cell Mol Biol. 2010;283:235–330. doi:10.1016/S1937-
6448(10)83005-9
16. Gan P, Patterson M, Velasquez A, et al. Tnni3k alleles influence ventricular mononuclear
diploid cardiomyocyte frequency. PLoS Genet. 2019;15(10):e1008354. Published 2019 Oct 7.
doi:10.1371/journal.pgen.1008354
17. Zhao, W., Zhao, T., Chen, Y., Zhao, F., Gu, Q., Williams, R. W., Bhattacharya, S. K., Lu, L., &
Sun, Y. (2015). A Murine Hypertrophic Cardiomyopathy Model: The DBA/2J Strain. PloS
one, 10(8), e0133132. https://doi.org/10.1371/journal.pone.0133132

WT
LV
LA
Ao
KO
LV
LA
Ao
A
Figure 1. Echocardiographic results of adult Tnni3k wildtype and komice. A. Representative long-axis view B-
Mode images of WT and KO hearts. LV: left ventricle, LA: left atria, Ao: aortic opening. Echocardiographic
measurements of B. ejection fraction (EF%), C. left ventricular internal dimension at end-diastole (LVIDd), D.
interventricularseptum thickness at end-diastole (IVSd), E. left ventricular posterior wall thickness at end-diastole
(LVPWd), F. end-systolic volume (ESV), G. end-diastolic volume (EDV). (WT: n=8, KO: n=10, 8 months old)
B
WT
KO
0
20
40
60
80
EF%
P=0.0002
WT KO
WT
KO
3.0
3.5
4.0
4.5
5.0
5.5
LVIDd
C
LVIDd (mm)
P=0.15
WT KO
D
WT
KO
0.0
0.5
1.0
1.5
IVSd
P=0.0102
IVSd(mm)
WT KO
WT
KO
0.0
0.5
1.0
1.5
LVPWd
E
LVPWd (mm)
P=0.014
WT KO
WT KO
WT
KO
0
10
20
30
40
ESV
F
ESV (µL)
P=0.039
WT KO
WT
KO
0
20
40
60
80
EDV
EDV (µL)
P=0.19
G
WT KO
69
E
RV
LV
LV
RV
WT KO
F
G
WT KO
WT KO
I H
P14 1 Month
6 Months
Figure 2. Morphological analysis of Tnni3k wildtype and ko mouse hearts. A. Measurements of heart weight to
body weight ratio (HW/BW) (WT: n=9; KO: n=15). B. Measurements of left ventricular weight to body weight
ratio (LV mass/BW) (WT: n=5; KO: n=6). Tail-cuff measurements of C. systolic, D.diastolic blood pressure of WT
(n=6) and KO (n=5) mice. Histology showing heart morphology of WT and KO mice (6 months old). E. Trichrome
staining, F. H&E staining,G. Images of representative whole hearts. Histology showing heart morphology of WT
and KO mice at H. postnatal day 14, I. 1 month old.
A B
wt
ko
0
20
40
60
80
100
120
140
160
180
Systolic BP (mmHg)
wt
ko
0
20
40
60
80
100
120
140
160
180
Diastolic BP (mmHg)
C D
P=0.13
+
-
0
1
2
3
4
5
6
HW/BW(mg/g)
WT Tnni3k ko WT KO WT
KO
0
1
2
3
4
5
LV mass/BW (mg/g)
P=0.64
WT KO
P=0.92
P=0.98
WT KO WT KO
WT KO WT KO
70
A B
6 Months
WT
KO
0
2000
4000
6000
8000
P=0.85
BinucleatedCMarem(µm
2
)
WT
KO
0
50
100
150
200
250
300
P<0.0001
BinucleatedCM length(µm)
C
WT
KO
0
10
20
30
40
50
60
P<0.0001
BinucleatedCM width(µm)
P14
alpha-actinin2, DAPI
D
WT KO WT KO WT KO
KO
WT
WT
KO
0
2000
4000
6000
8000
P=0.27
F
BinucleatedCM arem(µm
2
)
WT
KO
0
50
100
150
200
250
300
P=0.75
G
BinucleatedCM length(µm)
WT
KO
0
10
20
30
40
50
60
P=0.55
H
BinucleatedCM width(µm)
WT KO WT KO WT KO
Figure 3. Cardiomyocyte architectural analysis. A. Measurements of cellular area, B. length, C. width of
binucleated cardiomyocytes from 6 months old mice (WT: n=267 from 5 mice, KO: n=249 from 6 mice); D.
Representative confocal images of WT and KO binucleated cardiomyocyte (6 months old), stained for alpha-
actinin2 (green), DAPI (blue). E. Quantification of sarcomere length. F. Measurements of cellular area, G. length,
H. width of binucleated cardiomyocytes from P14 mice. (WT: n=208 from 4 mice, KO: n=236 from 5 mice)
WT
KO
0.0
1.5
1.5
2.0
2.5
Sarcomere Length (µm)
P=0.45
E
WT KO
71
B
Figure 4. Cardiomyocyte contractility analysis of Tnni3k wildtype, ko and kinase-dead mutant (K489R) mice. A.
Quantification of the sarcomere shortening fraction without or with isoproterenol (10 nM) stimulation. B. A
representative twitch of shortening fraction of cardiomyocyte with or without iso stimulation. Quantification of
the sarcomere C. shortening velocity, and D. relaxation velocity. WT: n=102 from 11 mice, KO: n=98 from 9 mice,
K489R: n=92 from 10 mice, 2-4 month old.
WT
WT ISO
KO
KO ISO
K489R
K489R ISO
0
5
10
15
20
Sarcomere Shortening
(%)
P=0.32
P=0.28
P=0.0001
P<0.0001
P=0.35
A
Sarcomere Shortening
WT
WT ISO
KO
KO ISO
K489R
K489R ISO
0
2
4
6
8
10
Shortening velocity
(µm/s)
P=0.78
P=0.58
P=0.0045
P=0.0012
P=0.93
C
Shortening velocity (µm/s)
WT
WT ISO
KO
KO ISO
K489R
K489R ISO
0
2
4
6
8
10
Relaxation velocity
(µm/s)
P=0.73
P=0.84
P=0.0096
P=0.0028
P=0.35
D
Relaxation velocity (µm/s)
72
WT
WT ISO
KO
KO ISO
K489R
K489R ISO
0
10
20
30
40
50
Calcium
(%)
P=0.17
P=0.53
P<0.0001
P=0.0005
P=0.22
A
WT
WT ISO
KO
KO ISO
K489R
K489R ISO
0
2
4
6
8
10
Calcium Upstroke Vmax
P=0.34
P=0.085
P<0.0001
P=0.018
P=0.039
WT
WT ISO
KO
KO ISO
K489R
K489R ISO
0
1
2
3
Calcium Decay Vmax
P=0.99
P=0.61
P=0.0075
P=0.0044
P=0.52
C
B
WT WT ISO
KO
KO ISO
K489R
K489R ISO
D
Figure 5. Cardiomyocyte calcium transient analysis of Tnni3k wildtype, ko and kinase-dead mutant (K489R)
mice. A. Quantification of the calcium amplitude without or with isoproterenol (10 nM) stimulation. B.
Representative calciumtransients in WT, KO and K489R cardiomyocytes before and after iso stimulation.
Quantification of the calcium C. upstroke Vmax, and D. decay Vmax. WT: n=64 from 8 mice, KO: n=75 from 8
mice, K489R: n=69 from 7 mice 2-4 month old.
Calcium (%)
73
TNNI3K
Phospho-RYR2-S2808
RYR2
cTnI
Phospho-cTnI-S23/24
GAPDH
Phospho-CREB1-S133
PLN
CREB1
Phospho-PLN-S16
WT1 WT2 WT3 KO1 KO2 KO3
Adult heart lysate
A
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
p-RYR2-S2808 p-CREB1-S133 p-cTnI-S23/24 p-PLN-S16
WT
KO
P=0.58 P=0.87 P=0.81 P=0.068
B
Normalized band density
Iso (nM) 0 10 30 100 0 10 30 100 0 10 30 100
WT KO K489R
Phospho-cTnI-S23/24
Phospho-PLN-S16
Phospho-RYR2-S2808
Phospho-CREB1-S133
GAPDH
PLN
cTnI
CREB1
RYR2
PKAc-alpha
Iso(nM)
Phospho-RYR2-S2808
Band density after normalization
D
C
Band density after normalization
Phospho-CREB1-S133
Iso(nM)
E G
Band density after normalization
Phospho-cTnI-S23/24
Iso(nM)
F
Phospho-PLN-S16
Band density after normalization
Iso(nM)
#
# #
#
74
Phospho-RYR2-S2808
8-Br-cAMP (µM) 0 40 80 120 0 40 80 120 0 40 80 120
WT KO K489R
GAPDH
Phospho-cTnI-S23/24
cTnI
Phospho-CREB1-S133
CREB1
PKAc-alpha
PLN
Phospho-PLN-S16
RYR2
H
Band density after normalization
Phospho-RYR2-S2808
8-Br-cAMP (µM)
Phospho-CREB1-S133
Band density after normalization
8-Br-cAMP (µM)
I
J
K
Phospho-PLN-S16
Band density after normalization
8-Br-cAMP (uM)
L
Time (min) 0 2 4 6 8 10 0 2 4 6 8 10
WT KO
Phospho-PLN-S16
GAPDH
PLN
M
N
Phospho-cTnI-S23/24
Band density after normalization
8-Br-cAMP (uM)
Figure 6. PKA signaling is blunted in Tnni3k ko mice under iso stimulation. A.Western blot using ventricular lysates
from adult WT and KO mice hearts, comparing the phosphorylation of PKA substrates RyR2, CREB1, cTnI, PLN. B.
Quantification of band densities (Fig. 6A) of WT (n=3) and KO (n=3) samples. C. Western blot using cardiomyocyte
lysates from adult WT, KO, K489R mice in iso treatment experiment, comparing the phosphorylation of PKA
substrates RyR2, CREB1, cTnI, PLN. Graphs show quantification of the relative levels of band density of D. phospho-
RYR2-S2808, E. phospho-CREB1-S133 , F. phospho-cTnI-S23/24, G. phospho-PLN-S16 from biological replicates (WT:
n=3; KO: n=3; WT: n=3; K489R: n=3). #: P<0.05. H. Western blot using cardiomyocytelysates from adult WT, KO,
K489R mice in 8-Br-cAMP treatment experiment, comparing the phosphorylation of PKA substrates RyR2, CREB1,
cTnI, PLN. Graphs show quantification of the relative levels of band density of I. phospho-RYR2-S2808, J. phospho-
CREB1-S133 , K. phospho-cTnI-S23/24, L. phospho-PLN-S16 from biological replicates (WT: n=3; KO: n=3; WT: n=3;
K489R: n=3). #: P<0.05. M. Western blot using cardiomyocyte lysates from adult WT, KO mice in iso and propranolol
treatment experiment. N. Quantification of band densities (Fig. 6M) of WT and KO samples.
#
#
#
#
# # #
#
75
A
C
WT WT I685T I685T KO KO
TNNI3K
GAPDH
B
Mononuclear CM%
D
p=0.52
P=0.0002
+/-
-/-
K489R/-
I685T/-
0
2
4
6
8
10
12
14
WT KO K489R/- I685T/-
Iso(nM) 0 10 30 100 0 10 30 100 0 10 30 100
WT I685T KO
Phospho-cTnI-S23/24
Phospho-PLN-S16
Phospho-RYR2-S2808
GAPDH
PLN
cTnI
RYR2
E
Figure 7. Tnni3k-I685Tknock-in mouse heart and cardiomyocyte physiology. A. Diagram of the wildtype allele
and the conversion of the ATC codon encoding I685 to ACA (Thr), and a synonymous mutation GGC to GGA was
created simultaneously to generate EcoNI restriction site. B. Sequence trace of a I685T homozygous mouse,
illustrating the changes introduced into the gene as in Fig.7A. C. Western blot of TNNI3K protein from the
indicated genotypes, demonstrating that the I685T protein is stable in vivo. D. Ventricular mononuclear CM% in
mice of the indicated Tnni3kgenotypes. E. Western blot using cardiomyocyte lysates from adult WT, KO, I685T
mice in iso treatment experiment, comparing the phosphorylation of PKA substrates RyR2, cTnI, PLN.
76
Figure 8. Echocardiographic results of adult Tnni3k-I685T/- mice. Echocardiographic measurements of A.
ejection fraction (EF%), B. end-systolic volume (ESV), C. end-diastolic volume (ESV), D. interventricularseptum
thickness at end-diastole (IVSd), E. left ventricular posterior wall thickness at end-diastole (LVPWd). F. left
ventricular internal dimension at end-diastole(LVIDd). (I685T/-, n=8, 4 months old)
WT
I685T/-
KO
0
20
40
60
80
EF%
WT
I685T/-
KO
0.0
0.5
1.0
1.5
IVSd
WT
I685T/-
KO
0.0
0.5
1.0
1.5
LVPWd
WT
I685T/-
KO
0
10
20
30
40
50
ESV
WT
I685T/-
KO
0
20
40
60
80
EDV
WT
I685T/-
KO
3.0
3.5
4.0
4.5
5.0
5.5
LVIDd
B A C
D F E
ESV (µL)
EDV (µL)
IVSd(mm)
LVPWd(mm)
LVIDd(mm)
p=0.076
p=0.022
p=0.37
p=0.41
p=0.69
p=0.45
p=0.64
p=0.45
p=0.028
p=0.84
p=0.17
p=0.95
77
78

CHAPTER 4
The role of Tnni3k in cardiac conduction system defects
Introduction
In normal heart rhythm, atrial contraction is initiated by depolarization of the sinoatrial
(SA) node, and because cardiomyocytes (CMs) are electrically coupled by gap junctions,
depolarization rapidly spreads throughout the atrial chambers. Because the SA node is located
in the upper quadrant of the right atrium, the vector of atrial depolarization spreads from
upper right to lower left, which in an electrocardiograph (ECG) using a lead II configuration is
manifest as a positive deflection of a P wave. Propagation of conduction is delayed at the
atrioventricular (AV) node, and electrical impulse passes into the bundle of His in the
ventricular septum then divides into the left and right bundle branches in the ventricular
myocardium. Bundle branches further divide into millions of Purkinje fibres, which interdigitate
with cardiomyocytes to induce coordinated, rapid, and synchronous physiologic depolarization
of heart ventricle. The depolarization of ventricle is seen as the QRS component of a normal
ECG. Repolarization happens soon after ventricle depolarization, which causes the relaxation of
the cardiac muscle of the ventricles. In ECG, ventricle repolarization is signified by T wave.

Supraventricular tachycardia (SVT) refers to a group of arrhythmias in which the
initiation or propagation of atrial conduction is abnormal. A common subtype of SVT is AV node
reentrant tachycardia (AVNRT), in which atrial depolarization originates from the AV node and
propagates retrograde into the atria; this is evident in an ECG by either an apparent absence of
79

a P wave (by virtue of overlap with and being obscured by the much stronger QRS complex) or
by an inverted P wave that trails the preceding QRS complex, depending on the subtype of
AVNRT. Another type of SVT is ectopic atrial tachycardia, in which initial depolarization occurs
in one or more portions of the atrial myocardium that do not include the SA node; depending
on the location of the ectopic foci, this can be evident also by an inverted P wave. In humans,
these conditions are associated with rapid heartbeat, and while this itself is not life threatening,
the association of SVT with various types of cardiomyopathy suggest that long-term
dysregulation of atrial heart rhythm can progress to complications that are more serious
1
. The
etiology of SVT is complex and several mechanisms may independently cause these conditions.
Inflammation, scarring, infectious agents, or infarction is logical primary causes if damage
occurs along atrial conduction pathways
2
. Recent genetic studies have also implicated variants
in ion channel genes that might affect the initiation or propagation of atrial depolarization
3
.
Bundle branch block (BBB) refers to the blockage of electrical conduction in His-Purkinje
system. BBB is commonly caused by conditions damaging bundle branches, like myocardium
infarction, cardiomyopathy, myocarditis and hypertension. It has also been reported that
genetic variants cause hereditary familial BBBs
4
. Without normal bundle branch propagation,
electrical impulses move through cardiac muscle fibers in a slow and altered direction of
propagation, which in turn cause ventricular asynchrony. In ECG, a widened QRS waves can be
observed. BBB causes loss of ventricular synchrony, and may result in drop in cardiac output.

80

A small number of families with rare heterozygous mutations in the gene Tnni3k have
been characterized for familial SVTs and BBBs, which appear to be haploinsufficient
5, 6, 7, 8
.
Tnni3k encodes a cardiomyocyte-specific kinase, and while named as troponin-I 3 kinase
(troponin i3 is cardiac troponin I, an obligate component of the contractile apparatus), troponin
i3 and TNNI3K physically interact but it is uncertain if the former is an actual substrate for
phosphorylation by the kinase. Pedigrees exhibited a variable arrhythmic phenotype that
included AV node reentrant tachycardia, junctional (AV nodal) and ectopic atrial tachycardia,
and various subtypes of conduction block. How TNNI3K variants might lead to conduction
system diseases has been unclear
5, 6, 7, 8
. Tnni3k has been implicated in oxidative stress response
in the extreme pathological context of ventricular ischemia-reperfusion injury
9
, although the
relevance of this insight to the uninjured heart has not been evident. In all of the studied
pedigrees, progression to dilated cardiomyopathy (DCM, a ventricular myocardial phenotype
with numerous causes) was common
5, 6, 7, 8
, suggesting either a causal relation between
arrhythmias and DCM, or independent roles for Tnni3k in supraventricular conduction and in
ventricular myocardium.

In this study, I have taken advantage of several Tnni3k genetic mouse models and used
molecular approaches to address the etiology of conduction system dysfunction associated
with Tnni3k mutations. My results reveal a surprising propensity for conduction system
dysfunction in the human population, explain why cardiac conduction system defect has a
variable frequency in humans even with this propensity to disease, and provide a potential
81

means of testing for and suppressing the onset of both conduction dysfunction and its
progression to severe outcomes.

Results
Mice with Tnni3k LOF mutation have conduction system defects
Heterozygous point mutations (G526D, T539A, c333+2T>C) in the TNNI3K gene were
implicated in the first three reports of TNNI3K-associated familial conduction system diseases
5, 6,
7
. None of those alleles is found in the ExAC collection or in other human genome sequences,
indicating that these are private mutations limited to these families. G526D and T539A
mutations are in the ATP binding pocket of the TNNI3K kinase domain. In chapter 2 of this
dissertation, I modeled these and other Tnni3k variants in an in vitro kinase reaction; this
revealed that the T538A mutation in the mouse protein (equivalent to human T539A) resulted
in substantially diminished kinase activity
5, 10
. Protein with the G526D equivalent mutation was
insoluble or unstable in 293 cells or when expressed in bacteria, and the original study also
reported an inability to prepare protein in insect Sf9 cells
6
. Moreover, the novel splice site
mutation (c.333 + 2 T > C) results in a premature stop codon in exon 4 of the TNNI3K gene and
subjects the transcript to nonsense-mediated mRNA decay
7
. Thus, all three human TNNI3K
point mutations are severely hypomorphic loss-of-function (LOF) mutations, and the human
heterozygotes are therefore haploinsufficient.

82

If the human TNNI3K alleles associated with conduction system diseases are LOF alleles,
mice with Tnni3k LOF gene mutation might have similar conduction system dysfunction. I
examined this in adult mice under isoflurane anesthesia using subcutaneous needle electrodes
for lead II ECG recording (Fig. 1A). I have now evaluated 60 mice, all inbred on a C57BL/6
background. 26 of 30 Tnni3k homozygous null mutant (ko) mice were observed with a
conduction phenotype (a crucial subdivision of this cohort that accounts for the incomplete
penetrance of this phenotype is explained below) (Table 1). The frequency of manifestations
ranged from virtually constant to occasional. As in the human families
5, 6, 7
, mouse Tnni3k
mutation was associated with a spectrum of conduction abnormalities, indicating a variable and
perhaps partly stochastic or nongenetic component to their etiology. Many mice showed
multiple types of patterns during recordings. Observed phenotypes in Tnni3k ko mice includes
premature atrial contraction (PAC), premature atrial contraction (PVC), BBB and AVNRT, which
are consistent with human arrhythmic phenotypes (Fig. 1B, 1C, 1D, 1E). One of 15 wildtype
controls was scored as abnormal with some PACs. Interestingly, 6 of 15 heterozygotes (het)
were abnormal, and one has AVNRT, which evokes the haploinsufficiency of the human
pedigrees (Table 1). Tnni3k ko mice live at least to 1-year-old without obvious incidence of
premature death, so their conduction abnormalities do not cause early fatality. Human
pedigrees also do not exhibit death until well into adulthood
5, 6, 7
. We conclude that arrhythmic
phenotypes in Tnni3k ko mice are similar to phenotypes observed in the human families.

Mice with Tnni3k LOF mutation have reduced PR intervals
83

After atrial depolarization, the electrical impulse passes through AV node to initiate
ventricular depolarization. On ECG profile, the time between the beginning of atrial
depolarization (the onset of P wave) until the beginning of ventricular depolarization (the onset
of QRS complex) is termed as PR interval. PR interval reflects the conductivity of AV node. The
duration of PR interval is an important parameter to evaluate the function of AV node. PR
interval prolongation is normally caused by AV block, while short PR interval is correlated with
preexitations (such as AVNRT) and AV node junctional rhythm. It has been reported that the
expression level of Tnni3k in mouse is positively correlated with PR interval duration.
Overexpression of Tnni3k in mice significantly prolonged the PR interval
11
, which indicates that
TNNI3K directly affects AV node conductivity.

After analyzing the ECG parameters, I found that Tnni3k ko mice (31.1 ms) have a
significantly shorter PR interval than wildtype mice (35.1 ms). In addition, Tnni3k het mice (32.6
ms) also have a significantly shorter PR interval duration than wildtype, which confirmed the
haploinsufficiency of Tnni3k in conduction system physiology (Fig. 2A). While all other
parameters, such as P wave duration, P wave amplitude, heart rate, QRS interval, QTc were not
significantly different between wildtype and ko mice (Fig. 2B, 2C, 2D, 2E). This result is
consistent with the previous report that Tnni3k expression level regulated PR interval
duration
11
. Since Tnni3k LOF causes AVNRT in both humans and mice, it is reasonable to infer
that Tnni3k regulates the AV node function and Tnni3k LOF causes PR interval shortening and
series of relevant arrhythmia.
84

Tnni3k functions as kinase in cardiac conduction system defects
Since the human TNNI3K mutant (T539A) has been validated to compromise kinase
activity, implicating that TNNI3K may function as a kinase in cardiac conduction system defects.
And in chapter 2 and 3 of this dissertation, I have demonstrated that TNNI3K plays a kinase role
in regulating MNDCM frequency
10
and cardiomyocyte physiology. To clarify whether TNNI3K
truly functions as a kinase in this scenario, I utilized our Tnni3k kinase-dead mutant (K489R)
mice
10
for ECG analysis. Of the 20 Tnni3k-K489R mice I have analyzed, I found that one mouse
had BBB, while 10 had mild ECG phenotypes, such as PACs/PVCs, and 9 mice were virtually
normal (Table. 1). This result suggests that Tnni3k kinase-dead mutant mice also have cardiac
conduction system defects, which implicates that TNNI3K also plays a kinase role in cardiac
conduction system defects.

Human TNNI3K polymorphism I685T does not cause obvious cardiac conduction system
defects
The ExAC database contains a large number of TNNI3K nonsynonymous amino acid
substitution variants of unknown significance. In chapter 2 of this dissertation, I modeled
several for their possible impacts on kinase activity, using in vitro kinase assay. Some had no
effect, but others severely compromised kinase activity
10
. In particular, an Ile>Thr transversion
(equivalent to pos. 685 in mouse) is the most common no synonymous human polymorphism in
all of TNNI3K, has an allele frequency ranging from 0.5-4% in various subpopulations, and
reduces kinase activity to a level similar to the T539A mutation that is associated with human
85

conduction system disease. Several other human TNNI3K nonsynonymous variants of
moderate allele frequency were also found to compromise kinase activity in this in vitro assay.
Thus, hypomorphic or complete LOF TNNI3K mutations are common in the human population.

To demonstrate the in vivo function of the most common human polymorphism
TNNI3K-I685T, I took advantages of the Tnni3k-I685T knock-in mice we created, which is
demonstrated in chapter 3. ECG analysis of Tnni3k-I685T (I685T/-) mice did not reveal any
severe ECG abnormalities based on the 15 mice I have scored so far (Table 1). 4 out 15 of mice
displayed mild conduction system defects, like PACs/PVCs, but most majorities of them were
normal. Although the heterozygous Tnni3k-I685T (I685T/+) should be the most common type of
I685T polymorphism carrier in human population, I don't expect to see any severer ECG
abnormalities in those carriers given that I685T/- didn't show any severe phenotypes. In
conclusion, Tnni3k-I685T polymorphism will not cause obvious conduction system defects by
itself. Since I have already demonstrated that Tnni3k-I685T mutation also blunted beta-
adrenergic-PKA signaling, caused ventricular wall remodeling in heart in the previous chapter, it
is highly likely that Tnni3k-I685T may predispose heart to some severe abnormalities, especially
in people having cardiotoxic conditions, like alcohol abuse, cardiotoxic drugs use, hypertension
etc.

Tnni3k-related cardiac conduction defect is background-dependent
86

If engineered absence of Tnni3k in C57BL/6 mice causes a conduction phenotype, the
same might be expected in mice, which naturally do not express TNNI3K protein. I examined
adult A/J and SWR/J mice (n=3 each), and none had any abnormality. To our awareness,
conduction defects of the types described above were also never seen in past studies of any
strains with the natural LOF allele. One possibility is that the C57BL/6J background is permissive
for developing conduction system disease when Tnni3k is mutated. Another possibility is that
the natural and engineered Tnni3k alleles are not equivalent in terms of conduction system
impact. The strategy I used to create Tnni3k ko mice results in a near complete elimination of
TNNI3K protein. However, the natural Tnni3k LOF allele is because of an alternative splicing
mutation, which results in a premature stop codon and unstable mRNA subjected to none-
sense mediated mRNA decay (NMD). Recently, it has been reported that the mRNA NMD is a
novel mechanism to compensate original gene loss by triggering homologous gene expression
12,
13
. Thus, it might also be true that Tnni3k mRNA NMD triggers homologous gene expression in
mouse strains carrying the natural LOF allele, which in turn prevents the manifestation of
conduction system defects.

To resolve this puzzle, I took advantage of our engineered mouse lines carrying the small
deletions in Tnni3k gene (Δ4 and Δ8). As I've demonstrated in chapter 2, both deletion
mutations caused framshfit after Ser532 and premature stop codon on Tnni3k gene, which
result in NMD of Tnni3k mRNA and no stable TNNI3K protein expression. The Tnni3k deletion
mutant mice turned out to be a good model for us to study the Tnni3k frameshift mutation and
87

mRNA NMD in a C57BL/6J background. ECG analysis on 12 Tnni3k mutant mice (Δ4 or Δ8)
didn't reveal any conduction system defects (Table 1). This observation supports the hypothesis
that Tnni3k mRNA NMD prevents the manifestation of conduction system defects in a C57BL/6J
background, and excludes the possibility that the C57BL/6J background is permissive to
conduction system defects. However whether there is any compensatory gene expression is still
an open question.

Discussion
In this study, I reconcile several disparate observations that link Tnni3k to arrhythmic
phenotypes in mice and humans. I show in mice that Tnni3k LOF causes conduction system
defects. I further confirmed that TNNI3K plays a kinase role in conduction system defects. By
using our engineered Tnni3k mutant mice (Δ4 and Δ8), I found that the NMD of Tnni3k mRNA
could prevent the manifestation of conduction system defects in a C57BL/6 background. This
result helps to explain why the many inbred mouse strains that naturally lack Tnni3k have (or
are presumed to have) normal heart rhythm, which is not because of the permissive
background of specific strains.

In Tnni3k ko mice, the manifestation of conduction system defects is variable and not
completely penetrant, and one possibility is that it might be a multigene trait. In the human
populations, TNNI3K-related arrhythmic phenotypes are only identified in very few families
with rare TNNI3K LOF mutations. Whereas TNNI3K LOF or hypomorphic mutations are relatively
88

common in the human populations, based on the ExAC database and our previous study
10
.
Additionally, TNNI3K gene mutations associated with conduction system disease were first
identified in pedigrees, which are most likely because the predisposing second gene variant are
inherited within families. These observations strongly implicate that there might be other genes
orchestrating with Tnni3k mutations to manifest arrhythmic phenotypes.

Another possibility is that conduction systems defect in Tnni3k LOF carrier is an acquired
phenotype. In chapter 3, I have demonstrated that Tnni3k LOF unequivocally causes LV
concentric remodeling and cardiomyocyte physiological abnormalities. In the human pedigrees
with TNNI3K mutations and conduction disease, it is relevant that dilated cardiomyopathy is
also common, either as an isolated condition or as a progression from earlier-onset arrhythmia.
The mice in our study did not obviously exhibit DCM, but we may not have allowed them to
reach an age when this condition would emerge, or mice may be more resistant to developing
this condition than humans. DCM is a common terminal phenotype associated with a large
number of challenges to the heart. The common TNNI3K LOF alleles in the human population
might therefore underlie not only idiopathic arrhythmias but also idiopathic DCM, the latter
that is a much more prevalent and more serious disease. At the cellular level, it is likely that the
direct effect of Tnni3k LOF is cardiomyocyte physiological changes, which predispose the heart
to cardiomyopathy. Given that different individuals are exposed to different cardiac risk factors,
like alcohol abuse, cardiotoxic drug use, hypertension, TNNI3K polymorphisms carriers may
have different propensity to cardiomyopathies. Conduction system defects are known to be
89

caused by cardiomyopathy, such as DCM, HCM. It may explain why conduction system defects
and DCM were observed in families carrying Tnni3k LOF mutations but not in every member of
those families. Actually, the multigene hypothesis and the acquired arrhythmic phenotype
hypothesis are not necessarily mutually exclusive. It is possible that the primary effect of Tnni3k
LOF is causing cardiomyocyte physiological abnormalities, which may not lead to but predispose
the heart to severe outcomes (conduction system defects and DCM) when combining with
other genetic or environmental risk factors.

I also observe that the most common human TNNI3K polymorphism I685T didn't cause
obvious conduction system defects. One explanation is that the Tnni3k-I685T mutation reduces
the kinase activity to 38%, while the disease-associated T539A mutation reduces the kinase
activity to 13%, as I've previously demonstrated
10
. There might be a threshold of TNNI3K kinase
activity, below which the cardiac conduction system defects show up. While in other aspects of
heart biology, Tnni3k-I685T mutation increases MNCM frequency to the level of ko mice, and
beta-adrenergic-PKA signaling is also blunted in Tnni3k-I685T cardiomyocyte. These
observations implicate that Tnni3k-I685T can alter cardiomyocyte physiology. Although Tnni3k-
I685T mutation causes relatively mild cardiac abnormalities, it might predispose the heart to
severe consequences when coupled with other risk factors, so do other human TNNI3K LOF and
hypomorphic mutations.

Materials and Methods
90

Animals
Tnni3k knockout, Tnni3-K489R, Tnni3-4del and 8del alleles used in this study were also
used in our prior work
10, 14
, and maintained on a C57BL/6 background. Animals were euthanized
by isoflurane anesthesia followed by cervical dislocation and removal of hearts. Animal
research was reviewed and approved by the IACUC committees of the Univ. of Southern
California (#10173) and of the Medical Univ. of South Carolina (2018-00642), and all
experiments were performed in accordance with relevant guidelines and regulations.

Electrocardiography (ECG)
4-8 month-old mice were anesthetized by isoflurane inhalation via a SomnoSuite small
animal anesthesia system (Kent Scientific), and needle electrodes were inserted subcutaneously
in a standard lead II arrangement (at the right thoracic limb and the left and right pelvic limb) to
obtain ECG readings with a Powerlab 35 data acquisition system (AD Instruments). Start to
record when stable baseline heart rate was confirmed. The ECG for each subject was recorded
continuously for 15 min. LabChart eight software (AD Instruments) was used for cardiac
conduction system defects analyses with the last 10-min recording. And ECG parameters
analyses were based on 10 stable heart beats.

91

References
1. Ellis, E. R., & Josephson, M. E. (2013). What About Tachycardia-induced Cardiomyopathy?.
Arrhythmia & electrophysiology review, 2(2), 82–90. https://doi.org/10.15420/aer.2013.2.2.82
2. Salerno JC, Seslar SP. Supraventricular Tachycardia. Arch Pediatr Adolesc Med.
2009;163(3):268–274. doi:10.1001/archpediatrics.2008.547
3. Dehghani-Samani, A., Madreseh-Ghahfarokhi, S., & Dehghani-Samani, A. (2019). Mutations of
Voltage-Gated Ionic Channels and Risk of Severe Cardiac Arrhythmias. Acta Cardiologica Sinica,
35(2), 99–110. https://doi.org/10.6515/ACS.201903_35(2).20181028A
4. Pérez-Riera, Andrés & Barros, Raimundo & Barbosa, Marianne & Raimundo, Rodrigo & Abreu,
Luiz & Nikus, Kjell. (2018). Left bundle branch block: Epidemiology, etiology, anatomic features,
electrovectorcardiography, and classification proposal. Annals of Noninvasive Electrocardiology.
24. 10.1111/anec.12572.
5. Xi Y, Honeywell C, Zhang D, et al. Whole exome sequencing identifies the TNNI3K gene as a
cause of familial conduction system disease and congenital junctional ectopic tachycardia. Int J
Cardiol. 2015;185:114–116. doi:10.1016/j.ijcard.2015.03.130
6. Theis JL, Zimmermann MT, Larsen BT, et al. TNNI3K mutation in familial syndrome of
conduction system disease, atrial tachyarrhythmia and dilated cardiomyopathy.Hum Mol Genet.
2014;23(21):5793–5804. doi:10.1093/hmg/ddu297
7. Fan LL, Huang H, Jin JY, et al. Whole exome sequencing identifies a novel mutation
(c.333 + 2T > C) of TNNI3K in a Chinese family with dilated cardiomyopathy and cardiac
conduction disease. Gene. 2018;648:63–67. doi:10.1016/j.gene.2018.01.055
92

8. Podliesna S, Delanne J, Miller L, et al. Supraventricular tachycardias, conduction disease, and
cardiomyopathy in 3 families with the same rare variant in TNNI3K (p.Glu768Lys).Heart Rhythm.
2019;16(1):98–105. doi:10.1016/j.hrthm.2018.07.015
9. Vagnozzi RJ, Gatto GJ Jr, Kallander LS, et al. Inhibition of the cardiomyocyte-specific kinase
TNNI3K limits oxidative stress, injury, and adverse remodeling in the ischemic heart. Sci Transl
Med. 2013;5(207):207ra141. doi:10.1126/scitranslmed.3006479
10. Gan P, Patterson M, Velasquez A, et al. Tnni3k alleles influence ventricular mononuclear
diploid cardiomyocyte frequency. PLoS Genet. 2019;15(10):e1008354. Published 2019 Oct 7.
doi:10.1371/journal.pgen.1008354
11. Lodder EM, Scicluna BP, Milano A, et al. Dissection of a quantitative trait locus for PR
interval duration identifies Tnni3k as a novel modulator of cardiac conduction. PLoS Genet.
2012;8(12):e1003113. doi:10.1371/journal.pgen.1003113
12. El-Brolosy, M.A., Kontarakis, Z., Rossi, A. et al. Genetic compensation triggered by mutant
mRNA degradation. Nature 568, 193–197 (2019). https://doi.org/10.1038/s41586-019-1064-z
13. Ma, Z., Zhu, P., Shi, H. et al. PTC-bearing mRNA elicits a genetic compensation response via
Upf3a and COMPASS components. Nature 568, 259–263 (2019).
https://doi.org/10.1038/s41586-019-1057-y
14. Patterson M, Barske L, Van Handel B, et al. Frequency of mononuclear diploid
cardiomyocytes underlies natural variation in heart regeneration. Nat Genet. 2017;49(9):1346–
1353. doi:10.1038/ng.3929

A
AV nodal reentrant tachycardia (AVNRT)
Bundle Branch Block (BBB)
Premature ventricular contraction (PVC)
B
Premature atrial contraction (PAC)
C
D
E
F
Normal ECG
P
Q
R
S
T
Figure 1. Mouse lead II ECG and representative ECG profiles. A. Schematic of lead II mouse ECG acquisition with
injected electrodes. The various ECG parameters are indicated.P, atrial depolarization (P-wave); QRS, ventricular
depolarization (QRS complex); T, ventricular repolarization (T-wave). B. The representative profile of normal ECG. C.
The representative ECG profile of AV nodal reentrant tachycardia (AVNRT). D. The representative ECG profile of
bundle branch block (BBB). E. The representative ECG profile of premature ventricular contraction (PVC). F. The
representative ECG profile of premature atrial contraction (PAC).
93
BBB AVNRT PVC/PAC Normal
Tnni3k wt 0/15 0/15 1/15 14/15
Tnni3k het 0/15 1/15 5/15 9/15
Tnni3k ko 5/30 1/30 20/30 4/30
Tnni3k-K489R 1/20 0/20 10/20 9/20
Tnni3k-I685T/- 0/15 0/15 4/15 11/15
Tnni3k-4delor8del 0/12 0/12 0/12 12/12
Table 1. ECG analysis in Tnni3k wildtype and mutant mice. AVNRT, AV nodal reentrant tachycardia; BBB, bundle
branch block; PVC, premature ventricular contraction; PAC, premature atrial contraction.
94
WT
HET
KO
400
450
500
550
600
650
700
750
Heart rate (bpm)
p=0.19
p=0.24 p=0.041
WT
HET
KO
20
25
30
35
40
45
PR interval (ms)
p=0.00017
p=0.06312
A
WT
HET
KO
0
5
10
15
P duration (ms)
WT
HET
KO
0
5
10
15
20
QRS interval (ms)
WT
HET
KO
0
50
100
150
QTc (ms)
B
E
F
p=0.055
p=0.12
p=0.29
p=0.074
p=0.49
p=0.98
p=0.0069
p=0.58
WT
HET
KO
0.00000
0.00005
0.00010
0.00015
0.00020
P Amplitude (mV)
C
p=0.94
p=0.49 p=0.72
p=0.25 p=0.51
Figure 2. ECG parameters of Tnni3k wildtype, heterozygous and ko mice. A. PR interval (WT: n=14; HET: n=17; KO:
n=17). B. P duration (WT: n=13; HET: n=17; KO: n=17). C. P amplitude (WT: n=14; HET: n=17; KO: n=17). D. Heart
rate (WT: n=14; HET: n=17; KO: n=17). E. QRS interval (WT: n=14; HET: n=17; KO: n=17). F. Corrected QT interval
(QTc) (WT: n=8; HET: n=14; KO: n=12) (All animals were 4-6 months old).
D
95

96

Chapter 5
Allelic variants between mouse substrains BALB/cJ and BALB/cByJ
influence mononuclear cardiomyocyte composition and
cardiomyocyte nuclear ploidy
Abstract
Most mouse cardiomyocytes (CMs) become multinucleated shortly after birth via
endoreplication and interrupted mitosis, which persists through adulthood. The very closely
related inbred mouse strains BALB/cJ and BALB/cByJ differ substantially (6.6% vs. 14.3%) in
adult mononuclear CM level. This difference is the likely outcome of a single X-linked
polymorphic gene that functions in a CM-nonautonomous manner, and for which the
BALB/cByJ allele is recessive to that of BALB/cJ. From whole exome sequence we identified two
new X-linked protein coding variants that arose de novo in BALB/cByJ, in the genes Gdi1
(R276C) and Irs4 (L683F), but show that neither affects mononuclear CM level individually. No
BALB/cJ-specific X-linked protein coding variants were found, implicating instead a variant that
influences gene expression rather than encoded protein function. A substantially higher
percentage of mononuclear CMs in BALB/cByJ are tetraploid (66.7% vs. 37.6% in BALB/cJ), such
that the overall level of mononuclear diploid CMs between the two strains is similar. The
difference in nuclear ploidy is the likely result of an autosomal polymorphism, for which the
BALB/cByJ allele is recessive to that of BALB/cJ. The X-linked and autosomal genes
independently influence mitosis such that their phenotypic consequences can be combined or
segregated by appropriate breeding, implying distinct functions in karyokinesis and cytokinesis.

97

Introduction
In mammals, only a small percentage of adult ventricular cardiomyocytes (CMs) are
diploid (i.e., have a single 2n nucleus (n is the haploid chromosomal content), also called
mononuclear diploid or abbreviated as 1x2n) (reviewed in [1,2]). Through a process known as
endoreplication, CMs enter cell cycle and progress through S-phase DNA replication, but then
fail to complete mitosis. The first iteration of endoreplication results in CMs having either a
single tetraploid nucleus (1x4n) or two diploid nuclei (2x2n), depending on whether mitosis was
interrupted before completion of karyokinesis or of cytokinesis, respectively. Both states are
considered to be polyploid, as both have four chromosome sets per cell. In rats and mice [3-5]
and several large animals and likely also humans [6], the peak of CM endoreplication occurs
shortly after birth, whereas in lambs it likely occurs in late gestation [7]. Subsequent
reiterations of this process, as occur naturally in postnatal life and in the aftermath of injury or
disease, can result in CMs with higher numbers of nuclei, higher numbers of genomes per
nucleus, or both. Polyploidy is not unique to CMs, as certain other cell types also have polyploid
subpopulations. The liver has been extensively studied in this regard; in adult mice,
approximately 90% of hepatocytes are polyploid [8].

The relevance of polyploidy to cell biology and organ physiology has been the subject of
speculation for decades. Polyploid CMs are larger than diploid CMs; candidate roles other than
size that may distinguish diploid and polyploid CMs include regenerative capacity, sensitivity to
oxidative stress, contractility, gene expression, metabolism, and others [1,2]. Our approach has
been to identify genetically-encoded variation in the extent of polyploidy as a first step towards

98

defining how polyploidy occurs and its biological relevance. We proposed and then showed in
mice [9] that the frequency of diploid CMs in the normal adult mammalian ventricle is not a
fixed trait, but rather exhibits substantial natural variation based on the combined effects of a
number of alleles that are polymorphic between individuals. In mice, most polyploid CMs are
binucleated with two diploid nuclei (2x2n). Therefore, in this prior analysis, we first surveyed a
large number of inbred mouse strains for the percentage of mononuclear CMs, and then
measured nuclear ploidy specifically within the mononuclear CM subset only for selected
strains. We found 7-fold variation in the percentage of mononuclear CMs (range 2.3-17.0%),
but less than 2-fold variation in the percentage of diploid nuclei in mononuclear CMs (range 40-
70%). Thus, we used the simple measurement of mononuclear CM level as a surrogate for
diploid CM level. Because other variables (housing, age, sex, etc.) were controlled, genetic
variation is the most likely explanation for the observed phenotypic variation between strains.
As a demonstration of this principle, by genome-wide association we identified a natural loss-
of-function variant in the gene Tnni3k in many inbred strains that have a high level of
mononuclear CMs, and confirmed in a controlled C57BL/6J strain background (which normally
carries the functional wild-type Tnni3k allele) that mutation of this gene resulted in a 2-3-fold
increase in the percentage of adult mononuclear CMs and in the percentage of diploid CMs
[9,10]. Clearly, many genes in mice in addition to Tnni3k have natural variants that also
influence the frequency of this CM population. The identification of these genes and their
natural variants, and how their products function in cell cycle control, karyokinesis, and
cytokinesis, is of significance for reaching a better understanding of the causes and
consequences of CM polyploidy and how this influences adult heart biology.

99

In our survey of mononuclear CM content across inbred mouse strains [9], we observed
a surprising discrepancy between the sister strains BALB/cJ (5.9% mononuclear CM) and
BALB/cByJ (14.0%) (both values are slightly higher in the reevaluation reported in the present
study). The BALB/cJ level was very close to the median (6.1%) among the 120 inbred mouse
strains evaluated, whereas BALB/cByJ was quite high (rank 116 of 120). The difference in
mononuclear CM content between BALB/cJ and BALB/cByJ is therefore substantial. These two
strains originated from an already-inbred Balb stock, were separated in 1935, and have since
been kept in reproductive isolation [11]. Thus, phenotypic differences between these substrains
today arose by spontaneous mutation during the past 85 years and then became fixed by
inbreeding. Because they are so closely related, it is possible that one or a small number of
mutations unique to one or the other strain might account for this difference. Because of the
prediction that the relevant variant(s) arose uniquely and recently within only one of the two
sublines and is therefore not widely distributed over many inbred strains, genome-wide
association as in our previous analysis [9] would not be expected to detect its presence
regardless of the magnitude of its effect.

The goal of this study was to explore the genetic basis of the divergence in CM
composition between BALB/cJ and BALB/cByJ, and the implications of this divergence for the
general subject of diploid and polyploid CMs in mice.

Results
Analysis of BALB/cJ and BALB/cByJ parental strains

100

In order to control sex as a variable, our original survey of mononuclear CM content
across inbred mouse strains [9] evaluated only female mice. We first addressed whether the
striking difference in mononuclear CM content between BALB/cJ and BALB/cByJ adult females
was also true in males. Indeed, for both substrains, males demonstrated the same mononuclear
CM level as females (Fig. 1A). Our past work with C57BL/6J mice also showed no sex difference
in this parameter [9,10]. Combining male and females together, our revised calculation of
ventricular mononuclear CM frequency is 6.6% for BALB/cJ, and 14.3% for BALB/cByJ, which is
more than a 2-fold difference. Although the ventricular CM populations of BALB/cJ and
BALB/cByJ are both overwhelmingly polyploid (93.4% and 85.7%, respectively), the >2-fold
elevation in diploid CMs in BALB/cByJ is a substantial difference and is of a magnitude similar to
Tnni3k gene mutation [9].

In general, because most adult mouse CMs are binucleated, many studies in the
literature have evaluated CM nuclear number but not also nuclear ploidy, under the
assumption that a relatively consistent percentage of mononuclear CMs are mononuclear
diploid and mononuclear tetraploid. As noted above, our initial large-scale survey of inbred
mouse lines also scored first for the frequency of mono- vs. multi-nucleated CMs, and
conducted direct measurement of nuclear ploidy only for specific selected strains. To address
this explicitly for BALB/cJ and BALB/cByJ, we measured nuclear ploidy by quantification of
nuclear DAPI signal intensity (normalized to the signal intensity from endothelial cells
(Supplementary Fig. S1), which are assumed to be diploid unless in the process of mitosis), as
we [10] and others have used in the past. Surprisingly, we found a striking distinction in

101

mononuclear CM nuclear ploidy between the two strains (Fig. 1B, Supplementary Fig. S1): in
BALB/cJ mice, 66.7% of the mononuclear CM nuclei were diploid, compared to only 37.6% in
BALB/cByJ. While these numbers are consistent with the 40-70% range previously observed
when comparing other inbred mouse strains [9], it was unexpected to find this magnitude of
difference in such closely related substrains. The combination of mononuclear CM percentage
(Fig. 1A) and mononuclear CM nuclear ploidy (Fig. 1B) allows calculation of the level of diploid
ventricular CMs in the two substrains: 4.4% for BALB/cJ, and 5.4% for BALB/cByJ. Thus, the two
BALB strains differ in mononuclear CM percentage in one direction, and differ in mononuclear
CM nuclear ploidy in the opposite direction, such that their levels of diploid CMs are similar.

We also evaluated hepatocyte polyploidy in BALB/cJ and BALB/cByJ mice. In both
strains, slightly less than half of hepatocytes were mononuclear (43.7% for BALB/cJ; 42.6% for
BALB/cByJ; Fig. 1C), and in both strains approximately 30% of the mononuclear hepatocytes
had diploid nuclei (31.9% for BALB/cJ; 26.8% for BALB/cByJ; Fig. 1D). Thus, the prominent
differences between these strains in mononuclear CM level and mononuclear CM nuclear
ploidy level are not also manifest in mononuclear hepatocytes. Combining these values yielded
a calculated level of diploid hepatocytes of 13.6% and 11.4% in BALB/cJ and BALB/cByJ,
respectively, which is consistent with prior observations in mice [8]. Interestingly, a closer
evaluation revealed that BALB/cByJ had a higher degree of polyploidy in both polyploid CMs
and polyploid hepatocytes (Supplementary Fig. S2). That is, in both cell types, BALB/cJ had a
higher percentage of tetraploid (1x4n and 2x2n) cells, whereas BALB/cByJ had a higher level of
octaploid (1x8n and 2x4n) cells for both cell types. The spectrum of polyploid CMs in BALB/cJ

102

was similar to C57BL/6J and other mouse strains we have studied in the past [9,10], whereas
that of BALB/cByJ was novel. Endothelial cells in both strains were uniformly mononucleated
and their nuclear DAPI fluorescence intensity clustered narrowly around a median value (i.e.,
were diploid) (Supplementary Fig. S1A). Similarly, analysis of bone marrow cells (stromal and
hematopoietic combined) showed no binucleated cells and only a modest percentage (identical
in the two strains) of polyploid nuclei (Supplementary Fig. S3) that are presumed to be of
diploid cells in mitosis and polyploid megakaryocytes.

Analysis of F1 mice
To gain insight into the inheritance pattern of the variants that distinguish BALB/cJ and
BALB/cByJ, we crossed the two parental lines in both directions and evaluated male and female
F1 mice in the same manner as above. Variation in mononuclear CM percentage (Fig. 1E)
segregated in a binary manner consistent with an X-linked allele, and with the variant in
BALB/cByJ being recessive to that in BALB/cJ. That is, the higher mononuclear CM level was
only observed in F1 male mice that carried a single X chromosome inherited from their
BALB/cByJ mother, and not in F1 female mice regardless of direction of cross. Other
explanations, including maternal biology and epigenetic/imprinting inheritance, are
inconsistent with these results. It is likely that the relevant gene functions in a CM-
nonautonomous manner, as F1 females showed only the lower (BALB/cJ-parental) level and not
an intermediate level as would be predicted for a CM autonomous trait subject to X-
inactivation.

103

Assessment of nuclear ploidy within the mononuclear CM subpopulation of F1 mice
revealed an equivalent percentage of diploid and polyploid nuclei (Fig. 1F, Supplementary Fig.
S4). This level (64-70% diploid nuclei) is equivalent to that of the parental BALB/cJ strain (66.7%;
Fig. 1B). This suggests that difference in nuclear diploid/polyploid ratio between these two
substrains is inherited in an autosomal manner, with the BALB/cJ allele being dominant to that
from BALB/cByJ. If the relevant gene is autosomal, no conclusion can be made regarding
whether gene function is CM-autonomous or nonautonomous. The analysis also reveals that
the two traits of mononuclear CM percentage and mononuclear CM nuclear diploidy appear to
segregate independently in these crosses. As a result, the level of diploid CMs in F1 male mice
derived from BALB/cByJ mothers (13.1% x 69.9% = 9.2%) is substantially higher than in either
parental line (4.4% and 5.4%; see above). A 9.2% level of diploid CMs is among the highest
levels we have measured among natural inbred strains (e.g., 10.0% in A/J, 9.3% in SWR/J)[9],
and much higher than in C57BL/6J mice (2.5%). The tendency of polyploid CMs to reach higher
ploidy levels in BALB/cByJ mice was not seen in the F1 mice (Supplementary Fig. S2), which is
consistent with this feature also being a manifestation of a recessive allele in BALB/cBy. We
cannot yet say if this trait is independent of, or related to, either the X-linked allele that
influences mononuclear CM percentage or the autosomal allele that influences the
mononuclear CM nuclear diploid/polyploid ratio.

We assume in the above that a single polymorphic X-linked gene is responsible for
variation in mononuclear CM level, and similarly that a single polymorphic autosomal gene is
responsible for variation in the diploid level of mononuclear CM nuclei. It is formally possible

104

that two (or more) genes are involved in each case, although this seems less likely given the
relatively recent divergence between the two substrains. Furthermore, because karyokinesis
and cytokinesis are both linked to cell cycle, we also note the alternative possibility that a more
complex explanation of inheritance that involves interaction between the autosomal and X-
linked alleles is possible.

Identification and evaluation of Gdi1 and Irs4 as candidate genes
We took an informatics approach to derive candidate X-linked genes that might be
responsible for mononuclear CM variation between BALB/cJ and BALB/cByJ mice. Because both
current strains originated from the same already-inbred stock, we based our approach on the
assumption that the relevant variant must have arisen de novo in one or the other line after
their segregation in 1935, and therefore would be unlikely to be present in any other inbred
strain unless derived from either parental source. Complete genome sequence is available for
36 inbred mouse strains [12], including BALB/cJ but not BALB/cByJ. Among these fully
sequenced strains, we surveyed for all X-linked gene variants in BALB/cJ that have
nonsynonymous or premature stop protein coding changes or splice donor/acceptor mutations,
but none were unique to BALB/cJ or shared only with SEA/GnJ (which was derived from a cross
of BALB/cJ to P/J in the mid-1940s [13]). The BALB/cJ coding variant shared with the fewest
other sequenced inbred strains was rs31755951 in the Vgll1 gene (Supplementary Fig. S5).
Direct sequencing showed that this variant is also present in BALB/cByJ (Supplementary Fig. S5),
indicating that it predates the 1935 segregation of the two. Indeed, strains A, C3H, and BALB/c,
which all share the Vgll1 variant, have some common early (pre-1920) ancestry, likely including

105

this portion of the X chromosome. The X chromosome of BALB/cJ in the 159.6-164.04Mb region
is derived from wild mice and is highly divergent from common inbred mouse strains although
shared by SEA/GnJ. Within this region, we sequenced four coding region variants in BALB/cJ
(and SEA/GnJ) but found all were shared by BALB/cByJ (Supplementary Fig. S5). Thus, genome
sequence failed to reveal an X-linked protein coding change in BALB/cJ mice that could account
for the divergence in mononuclear CM content with BALB/cByJ mice.

Alternatively, the relevant mutation might have arisen uniquely in the BALB/cByJ
lineage. Whole genome sequence is not available for BALB/cByJ, but we obtained whole exome
sequence and filtered for nonsynonymous or functional variants on the X chromosome. We
identified two previously unreported polymorphisms (Gdi1, X:74309969 C/T, R276C; Irs4,
X:141723152 G/A, L683F) uniquely in BALB/cByJ relative to all other sequenced mouse
genomes, including that of BALB/cJ, and confirmed these as being different between BALB/cJ
and BALB/cByJ by direct genome sequencing (Fig. 2A,B). We conducted a parallel assessment of
whole exome sequence from BALB/cJ mice, although this failed to reveal any new
polymorphism not noted above.

To address the candidate role of the Gdi1 and Irs4 alleles, we crossed the two parental
lines to generate F1 females, then backcrossed these to BALB/cByJ males for 3 further
generations, selecting females at each generation that by meiotic crossover retained only one
of the BALB/cJ-specific X-linked alleles (heterozygous to the BALB/cBy allele). F4 females were
again crossed to BALB/cByJ males, and F5 males were evaluated for mononuclear CM level. In

106

principle, these F5 males are >90% homozygous for BALB/cByJ alleles over the entire genome
except around the individual BALB/cJ variants that were subject to selection. We determined
that the strain origin of both variant genes did not alter the mononuclear CM content of the
mice (Fig. 2C,D), which was equivalent to the parental BALB/cByJ level. These two variants are
thus unlikely to individually explain divergence in this trait between the two strains. Because
there are many autosomal variants between the two strains, we did not attempt in this study to
evaluate these for their influence on the nuclear ploidy phenotype.

Discussion
The polyploid nature of almost all adult cardiomyocytes in human, mouse, rat, and
several other species has been known for decades, and yet the mechanisms that account for
this outcome have remained obscure. Analysis of the natural genetic variation present among
inbred mouse strains is one strategy to discover genes that are functionally relevant to any
variable mouse trait and therefore to its underlying processes. In this way, by surveying a large
number of inbred mouse strains, we previously identified one polymorphic gene (Tnni3k) that
influences how many diploid CMs are present [9]. The analysis also uncovered the substantial
difference in mononuclear CM level between BALB/cJ and BALB/cByJ that served as the starting
point for the present study.

Our analysis of BALB/cJ and BALB/cByJ mice revealed two significant principles. First, the
simplifying assumption that the easily measured level of mononuclear CMs in the adult mouse
heart is a suitable surrogate for the level of diploid CMs is incorrect, at least in some cases if the

107

mononuclear tetraploid CM population is unusually high or low. As we found here, inclusion of
nuclear ploidy measurement essentially erased the substantial difference in mononuclear CMs
between the very closely related BALB/cJ and BALB/cByJ substrains. Because differences in
nuclear ploidy are based on genetic polymorphisms, this observation is a particularly relevant
caution for studies that are conducted on an outbred or mixed strain background, where
littermate animals or animals from the same colony would not necessarily serve as appropriate
genetic controls.

The second general conclusion of this study is that genes can selectively influence
mononuclear CM or CM nuclear ploidy levels in an independent and separable manner. Here, of
the two presumptive genes that differ between BALB/cJ and BALB/cByJ, because one gene is X-
linked and the other is autosomal, we could easily discern the separable nature of both in the
F1 crosses. We interpret differences in nuclear ploidy to indicate effects at the time of
karyokinesis, and similarly, genes that influence mononuclear CM level to indicate effects on
cytokinesis. Because the machinery and regulation of karyokinesis and cytokinesis are
overlapping but also distinct, that there are genes that selectively influence one or the other
may not be surprising. Indeed, among mammalian species, although the vast majority of CMs
are polyploid, the nature of CM polyploidy varies from being primarily binucleated with two
diploid nuclei (2x2n) as in mice and rats to primarily mononuclear tetraploid (1x4n) in humans
[14]. Different inbred mouse strains also vary in mononuclear CM nuclear ploidy [9]. Thus,
between species and within a species, and now shown even between very closely related sister
strains of mice, there is a genetic basis that distinguishes interruption of endoreplication prior

108

to karyokinesis or prior to completion of cytokinesis as CMs become polyploid.

In retrospect, because our initial survey of 120 mouse strains [9] only addressed
mononuclear CM level, genome-wide association based on strain variation in this parameter
might be predicted to yield genes that are selectively relevant to cytokinesis. Indeed, our
identification and evaluation of Tnni3k from that study confirmed that this gene primarily
impacts mononuclear CM level and not nuclear ploidy, at least when addressed on a controlled
C57BL/6J background [9,10]. Statistical significance in genome-wide association is dependent
on allele effect size and allele frequency; the latter favors discovery of genes with polymorphic
alleles that are widely distributed among the sampled population. For example, the natural
Tnni3k variant allele was present in 60 of the 120 strains that we surveyed for mononuclear CM
content [9]. For being recent de novo mutations likely present in only a single parental strain,
the two (or potentially more) still-unknown BALB variants that influence CM ploidy would not
likely have been evidenced in our genome-wide association analysis. The 120 surveyed strains
included 8 CXB recombinant inbred lines, which were derived from crosses started in 1959 of
BALB/cByJ with C57BL/6ByJ [15], and also the Sea/GnJ strain, which was derived from a cross
begun in the mid-1940s between BALB/cJ and P/J [13], i.e., both occurred after the 1935
separation of BALB/cJ and BALB/cByJ. However, even with these, the population would still
have been underpowered to reveal a statistically significant association. Because BALB/cJ and
BALB/cByJ are so closely related, we were able to use a different approach (F1 analysis) to
demonstrate that variants in two distinct genes are present, one of which influences
mononuclear CM level and one that influences mononuclear CM nuclear ploidy. Such an

109

approach would be not be possible with more distantly related strains because of the
complicating influence of numerous additional gene variants that impact either or both
features.
While the existence of two variant alleles (one X-linked, one autosomal) is the most
parsimonious explanation for the differences between BALB/cJ and BALB/cByJ, we note the
possibility that additional genes or a more complex pattern of inheritance might be involved. A
definitive conclusion will only be possible once the relevant variants are identified. In the
present study, we did not attempt to identify autosomal variants for their role in CM nuclear
ploidy, and we were not able to identify the X-linked gene that distinguishes BALB/cJ and
BALB/cByJ in mononuclear CM level. Both BALB/cByJ alleles appear to be recessive to their
counterparts in BALB/cJ, and thus in principle could be loss-of-function protein coding variants
that arose selectively in the former. We identified two new X-linked protein coding variants in
BALB/cByJ (in Gdi1 and Irs4) but our data exclude both from being individually relevant to this
trait, and analysis of whole exome and full genome sequence of BALB/cJ did not reveal any X-
linked protein coding or functional variant that isn’t also shared with BALB/cByJ or among at
least several other fully sequenced inbred strains. There are several possible explanations. First,
the relevant allele might not be a coding region variant but rather a regulatory variant that
influences gene expression. There are too many noncoding variants in BALB/cJ to evaluate
these as individual candidates, and we do not have whole genome sequence of BALB/cByJ with
which to compare nontranslated regions of the genome. Second, more than one X-linked gene
might be involved. In an attempt to specifically address the candidacy of Gdi1 and Irs4 in the
phenotype, we segregated the two BALB/cJ alleles by meiotic recombination and backcrossed

110

these individually to BALB/cByJ. However, the two might work together, or either might require
an autosomal BALB/cJ variant in order to manifest its effect. Other explanations are possible as
well.
Our data reveal that two (or perhaps more) recessive alleles distinguish BALB/cByJ from
BALB/cJ mice, one influences mononuclear CM percentage and the other influences CM nuclear
ploidy. Curiously, the two influence CM polyploidy in opposite directions: relative to BALB/cJ,
BALB/cByJ has a higher number of mononuclear CMs but a higher percentage of these nuclei
are polyploid. Thus, the level of diploid CMs between the two substrains is similar (4.4% and
5.4%). This could suggest that an excess of diploid CMs is detrimental such that the two alleles
arose together out of necessity. However, because other inbred mouse strains [9], and F1 male
mice derived from BALB/cByJ mothers and BALB/cJ fathers (this study) all have a high diploid
CM level without apparent ill effect, we think the occurrence of two variant alleles in BALB mice
that both influence CM ploidy is more likely to be a coincidence rather than the outcome of
selection.

All cell types, whether diploid or polyploid in the adult, have gone through numerous
rounds of mitosis earlier in their lineages. Gene variants that influence either karyokinesis or
cytokinesis to result in polyploid cells are clearly not doing so in all circumstances of mitosis,
implying their participation in a unique program that influences the outcome specifically in
endoreplication. An unexpected observation in this study was that phenotypic variation
between BALB/cJ and BALB/cByJ in mononuclear CM level (the X-linked variant) and of
mononuclear CM nuclear diploidy (the autosomal variant) were both manifest in CMs but not in

111

hepatocytes (Fig. 1). This suggests that the mechanisms governing cytokinesis and karyokinesis
in these two types of endoreplicating cells might be sufficiently distinct from each other to be
selectively influenced by variants in these genes. Alternatively, the genes may be selectively
expressed in CMs, just as Tnni3k is only expressed in CMs.

A feature that was shared between CMs and hepatocytes was the tendency of both to
reach higher levels of polyploidy in the BALB/cByJ background (Supplementary Fig. S2). This
could reflect a tendency to initiate additional rounds of endoreplication, but other explanations
are also possible. It is unknown if this is the manifestation of an additional genetic locus in
BALB/cByJ, or is in some manner the result of the influence of the X-linked and autosomal loci
that are presumptively involved in cytokinesis and karyokinesis, respectively. This behavior was
inherited in autosomal manner, with the BALB/cByJ allele(s) being recessive to BALB/cJ.

Although the new BALB/cByJ variants in Gdi1 and Irs4 discovered in this study do not
seem to be relevant to CM ploidy, it is possible that these variants may contribute to other
phenotypic effects. The PolyPhen-2 prediction for the BALB/cByJ Irs4 Leu683Phe variant is
“probably damaging” (score 0.998). The protein encoded by this gene is named as an insulin
receptor substrate, and while it not clear that this protein is actually a substrate for the insulin
receptor [16], it is thought to participate in signaling processes that control metabolism [17,18].
In humans, mutations in IRS4 are associated with central hypothyroidism, although mice
carrying an Irs4 null allele had unchanged serum thyroid hormone concentrations [19]. It
remains unknown if the BALB/cByJ allele has related or other effects. Gdi1 encodes a Rab GDP

112

dissociation inhibitor, which is involved in vesicular trafficking. In humans, GDI1 loss-of-function
gene mutations are associated with X-linked intellectual disability (mental retardation) [20], and
in mice, deletion of Gdi1 results in memory and behavioral alterations that resemble the human
condition [21]. The BALB/cByJ variant (Arg276Cys) has the potential to compromise protein
function (PolyPhen-2 prediction of “probably damaging”, score of 0.994). Interestingly, Gdi1
deletion in mice is associated with less aggressive behavior [21], and BALB/cByJ mice are
reported to be less aggressive than BALB/cJ mice [22], which would be consistent with the
natural BALB/cByJ variant identified here having functional consequences.

Materials and Methods
Animals
All mice in this study were obtained from The Jackson Laboratories (BALB/cJ JAX
#000651; BALB/cByJ JAX #001026) or were bred in-house from these stocks. All mouse analyses
were performed on mice 8–10 weeks of age. Animals were euthanized by isoflurane anesthesia
followed by cervical dislocation and removal of hearts. Animal research was reviewed and
approved by the IACUC committees of the Univ. of Southern California (#10173) and of the
Medical Univ. of South Carolina (2018-00642), and all experiments were performed in
accordance with relevant guidelines and regulations.

Single-cell ventricular cardiomyocyte suspensions and nuclear ploidy analysis
Following a methodology we have used previously [10], hearts were digested ex vivo
with 1mg/ml collagenase type II in calcium-free Tyrode’s solution (120mM NaCl, 4mM KCl,

113

0.33mM NaH
2
PO
4
, 1mM MgCl
2
, 10mM HEPES, 11mM glucose, 20mM taurine, 20mM BDM) via
Langendorff retroaortic perfusion. After digestion, atria and valves were removed and
ventricular tissue alone was triturated in Kruftbrühe (KB) solution (70mM potassium aspartate,
40mM KCl, 15mM KH
2
PO
4
, 10mM glucose, 10mM taurine, 0.5mM EGTA, 10mM sodium
pyruvate, 10mM HEPES, 5mM BDM, 0.5% BSA), filtered by gravity through a 250μ nylon mesh,
stained with LiveDead Fixable (ThermoFisher, L10120) in PBS for 20 min at room temperature
and then fixed in 2% paraformaldehyde (PFA) in PBS at room temperature for 15 min. Fixed
ventricular cell suspensions were stained for cTnT (1:1,000, Abcam ab8295) overnight at 4°C
followed by goat anti-mouse secondary (1:500, ThermoFisher A11001), washed with PBS, and
resuspended in PBS containing 5µg/ml DAPI for 5min with rocking. Cell suspensions were
washed once in PBS then pipetted across a slide and coverslipped. Numbers of nuclei per
cardiomyocyte were quantified using photographs taken at a uniform setting for all cell
preparations with a Leica DFC3000G camera in full frame mode (1296 x 966 pixels; 3.75µ
2
pixel
size) through an Olympus BX41 fluorescence microscope (20x objective). Only live
cardiomyocytes were counted; at least 300 cells were counted per heart. An unpaired, two-
tailed Student t-test was used to assess statistical significance when only two groups were
compared. To evaluate the ploidy of CM nuclei, using ImageJ software, nuclei in photographs
were identified and outlined with a standard threshold requirement for all samples, and DAPI
fluorescence intensity of each nucleus automatically quantified by ImageJ. The median value of
DAPI fluorescence intensity of CD31+ endothelial cell nuclei was used as a diploid nucleus
standard and given a value of 1, all other nuclear fluorescence signals were normalized to this
value. Nuclei were assigned as being diploid if their intensity value was within the 0.5-1.5 range

114

(indicated by a red box in some figures), tetraploid for values 1.5-2.5, and octaploid for values
>3. The latter were confirmed individually to not be the result of microscopy artifact.

Hepatocyte isolation and ploidy analysis
After severing the portal vein, mouse livers were perfused via a 24 gauge needle placed
in the inferior vena cava with prewarmed 37°C perfusion buffer (0.14M NaCl, 6.7mM KCl,
10mM HEPES pH=7.4, 0.1mM EGTA) for 5-10 min at 7 ml/min until the liver was pale. The
buffer was then changed to digestion solution (66.7mM NaCl, 6.7mM KCl, 100mM HEPES
pH=7.4, 4.7mM CaCl
2
, 1 mg/ml collagenase type II) and perfusion was continued for
approximately 15 min at 3 ml/min. The liver was transferred to a petri dish containing ice-cold
DMEM, and after the gallbladder was removed, was minced with forceps. The cell suspension
was pipetted several times then filtered by gravity through 70µ nylon mesh into a 50 ml tube.
Cells were centrifuged at 210 x g for 3 min, resuspended in 5ml 0.05% trypsin in PBS with 1mM
EDTA and incubated at 37°C for 10 min with rocking, then centrifuged at 210 x g for 3 min. Cells
were washed with PBS three times then fixed in 70% ethanol for 15 min. 10µl of the fixed cell
suspension was pipetted onto a glass microscope slide and air-dried. Slides were blocked with
10% normal goat serum (Thermo Fisher Scientific 50062Z) with 0.1% Triton-X100 for 1 h, then
incubated with primary antibodies anti-CD31 (1:250, BD Pharmingen 553370 ) and anti-albumin
(1:250,GeneTex GTX102419) at 4°C overnight, followed by secondary antibodies Alexa Fluor
488 (Invitrogen A11001) and Alexa Fluor 546 (Invitrogen A10040) and with DAPI using standard
procedures. Slides were coverslipped with ProLong Gold antifade reagent (Invitrogen) and
photographed under fluorescence microscopy. Hepatocyte nuclei were identified and their

115

fluorescence intensity quantified as performed for cardiomyocytes; roughly 200 hepatocytes
were analyzed for each sample.

Bone marrow cells isolation and ploidy analysis
Trimmed femurs from euthanized mice were flushed three times with 0.5 ml PBS using a
1-ml insulin syringe with a 29 gauge needle. The collected cells were transferred into a 1.5ml
eppendorf tube and fixed by adding 0.5ml of 4% PFA in PBS (final PFA concentration 2%) and
incubated at room temperature for 10min. Cells were centrifuged at 100 x g rpm for 3 min,
washed three times with PBS, and stained with DAPI using standard procedures. Photography
and nuclear fluorescence intensity were as for cardiomyocytes; roughly 400 bone marrow cells
were analyzed for each sample.

Exome analysis and validation.
BALB/cJ (6 independent samples) and BALB/cByJ (5 samples) exome sequence data
were obtained from a previous analysis [23] and from the Mouse Mutant Resource variant
database (MMRdb: https://mmrdb.jax.org/mmr/). Using tools available in MMRdb and
SAMTools (http://samtools.github.io/bcftools/bcftools.html), annotated variant calls in the
variant caller format (VCF) from these samples were filtered to remove variants with rsID
numbers, low quality variants (QUAL<70), heterozygous variants (GT=0/1), and autosomal
variants. Calls that were shared between BALB/cJ and BALB/cByJ or shared with at least several
other inbred strains were likewise removed. The remaining variants were novel, homozygous
variant calls from the X chromosome. Gene fragments were amplified from mouse genomic

116

DNA by PCR with GoTaq Green Master Mix (Promega); primers used are listed in
Supplementary Table S1. Sanger sequencing of amplified fragments was performed by GeneWiz
(genewiz.com) using one of the amplification primers.
PolyPhen-2 SNP assessment
An on-line tool at http://genetics.bwh.harvard.edu/pph2/ was used with the human
GDI1 (P31150) or IRS4 (O14654) Uniprot entry. For mouse Gdi1, the R276 position is the same
in the human sequence. For mouse Irs4, the L883 position corresponds to position 711 of
Uniprot O14654.

117

References
1. Gan P, Patterson M, and Sucov HM. Cardiomyocyte Polyploidy and Implications for
Heart Regeneration. Annu Rev Physiol, 2019.
2. Patterson M and Swift SK. Residual Diploidy in Polyploid Tissues: A Cellular State with
Enhanced Proliferative Capacity for Tissue Regeneration? Stem Cells Dev, 2019, 28(23):
1527-1539.
3. Brodsky WY, Arefyeva AM, and Uryvaeva IV. Mitotic polyploidization of mouse heart
myocytes during the first postnatal week. Cell Tissue Res, 1980, 210(1): 133-44.
4. Li F, Wang X, Capasso JM, and Gerdes AM. Rapid transition of cardiac myocytes from
hyperplasia to hypertrophy during postnatal development. J Mol Cell Cardiol, 1996,
28(8): 1737-46.
5. Soonpaa MH, Kim KK, Pajak L, Franklin M, and Field LJ. Cardiomyocyte DNA synthesis
and binucleation during murine development. Am J Physiol, 1996, 271(5 Pt 2): H2183-9.
6. Velayutham N, Agnew EJ, and Yutzey KE. Postnatal Cardiac Development and
Regenerative Potential in Large Mammals. Pediatr Cardiol, 2019, 40(7): 1345-1358.
7. Jonker SS, Zhang L, Louey S, Giraud GD, Thornburg KL, and Faber JJ. Myocyte
enlargement, differentiation, and proliferation kinetics in the fetal sheep heart. J Appl
Physiol, 2007, 102(3): 1130-42.
8. Duncan AW. Aneuploidy, polyploidy and ploidy reversal in the liver. Semin Cell Dev Biol,
2013, 24(4): 347-56.
9. Patterson M, Barske L, Van Handel B, Rau CD, Gan PH, Sharma A, Parikh S, Denholtz M,
Huang Y, Yamaguchi Y, Shen H, Allayee H, Crump JG, Force TI, Lien CL, Makita T, Lusis AJ,

118

Kumar SR, and Sucov HM. Frequency of mononuclear diploid cardiomyocytes underlies
natural variation in heart regeneration. Nature Genetics, 2017, 49(9): 1346-1353.
10. Gan P, Patterson M, Velasquez A, Wang K, Tian D, Windle JJ, Tao G, Judge DP, Makita T,
Park TJ, and Sucov HM. Tnni3k alleles influence ventricular mononuclear diploid
cardiomyocyte frequency. PLoS Genet, 2019, 15(10): e1008354.
11. Les EP A brief history of the two substrains of BALB/c, BALB/cJ and BALB/cByJ. JAX
eNews, 1990.
12. Keane TM, et al. Mouse genomic variation and its effect on phenotypes and gene
regulation. Nature, 2011, 477(7364): 289-94.
13. Green MC. Further morphological effects of the short ear gene in the house mouse. J.
Morphol., 1951, 88(1): 1-21.
14. Korecky B, Sweet S, and Rakusan K. Number of nuclei in mammalian cardiac myocytes.
Can J Physiol Pharmacol, 1979, 57(10): 1122-9.
15. Bailey DW. Recombinant-inbred strains. An aid to finding identity, linkage, and function
of histocompatibility and other genes. Transplantation, 1971, 11(3): 325-7.
16. Schreyer S, Ledwig D, Rakatzi I, Kloting I, and Eckel J. Insulin receptor substrate-4 is
expressed in muscle tissue without acting as a substrate for the insulin receptor.
Endocrinology, 2003, 144(4): 1211-8.
17. Fantin VR, Wang Q, Lienhard GE, and Keller SR. Mice lacking insulin receptor substrate 4
exhibit mild defects in growth, reproduction, and glucose homeostasis. Am J Physiol
Endocrinol Metab, 2000, 278(1): E127-33.

119

18. Wauman J, De Smet AS, Catteeuw D, Belsham D, and Tavernier J. Insulin receptor
substrate 4 couples the leptin receptor to multiple signaling pathways. Mol Endocrinol,
2008, 22(4): 965-77.
19. Heinen CA, de Vries EM, Alders M, Bikker H, Zwaveling-Soonawala N, van den Akker ELT,
Bakker B, Hoorweg-Nijman G, Roelfsema F, Hennekam RC, Boelen A, van Trotsenburg
ASP, and Fliers E. Mutations in IRS4 are associated with central hypothyroidism. J Med
Genet, 2018, 55(10): 693-700.
20. D'Adamo P, Menegon A, Lo Nigro C, Grasso M, Gulisano M, Tamanini F, Bienvenu T,
Gedeon AK, Oostra B, Wu SK, Tandon A, Valtorta F, Balch WE, Chelly J, and Toniolo D.
Mutations in GDI1 are responsible for X-linked non-specific mental retardation. Nat
Genet, 1998, 19(2): 134-9.
21. D'Adamo P, Welzl H, Papadimitriou S, Raffaele di Barletta M, Tiveron C, Tatangelo L,
Pozzi L, Chapman PF, Knevett SG, Ramsay MF, Valtorta F, Leoni C, Menegon A, Wolfer
DP, Lipp HP, and Toniolo D. Deletion of the mental retardation gene Gdi1 impairs
associative memory and alters social behavior in mice. Hum Mol Genet, 2002, 11(21):
2567-80.
22. Velez L, Sokoloff G, Miczek KA, Palmer AA, and Dulawa SC. Differences in aggressive
behavior and DNA copy number variants between BALB/cJ and BALB/cByJ substrains.
Behav Genet, 2010, 40(2): 201-10.
23. Fairfield H, et al. Exome sequencing reveals pathogenic mutations in 91 strains of mice
with Mendelian disorders. Genome Res, 2015, 25(7): 948-57.

Figure 1.CM nucleation and nuclear ploidy. A.Evaluation of CM mononuclear percentage of male and female
mice of the BALB/cJ (abbreviated as cJ) and BALB/cByJ (abbreviated as cByJ) parental strains. 3 data points for
BALB/cJ females and 3 data points for BALB/cByJ females are from a prior analysis [9], all others were newly
generated in this analysis. B. Evaluation of the nuclear ploidy specifically of the mononuclear CM subpopulation
of the parental strains. The data points graphed are measurements of the percentages of diploid nuclei in
individual mice. C.Evaluation of hepatocytemononuclear percentage in the two substrains. D. Evaluation of the
nuclear ploidy specifically of the mononuclear hepatocyte subpopulation. E. Evaluation of CM mononuclear
percentage of male and female F1 mice bred from crosses of the two parental strains. F. Evaluation of nuclear
ploidy of the mononuclear CM subpopulation of F1 mice. Primary data for panels B and F are in Supplementary
Fig. S1 and S2. Error bars in all panels are standard deviation.
120
Figure 2. Identification and evaluation of two BALB/cByJ-specific X-linked variants. A, B. Sequence traces of the
Gdi1 (A) and Irs4 (B) genes in BALB/cJ and BALB/cByJ mice. C, D. Mononuclear CM percentage in backcrossed F5
male mice grouped by the parental strain origin of their Gdi1 (C) or Irs4 (D) genes.
121
DAPI
cTnT
CD31
A
Fibroblast
Bi CM
0
1
2
3
4
Balb/cBy
Relative DAPI density
Endo Bi CM
RelativeDAPI intensity
Fibroblast
Bi CM
0
1
2
3
4
Relative DAPI density
Balb/c
Endo Bi CM
RelativeDAPI intensity
B
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
Balb/c 2
62%
mono
bi
0
1
2
3
4
5
Relative DAPI density
Balb/c 1
68%
Mono Bi Mono Bi
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
Balb/c 3
65%
Mono Bi
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
Balb/c 4
72%
Mono Bi
BALB/cJ 1 BALB/cJ 2 BALB/cJ 3 BALB/cJ 4
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
mono
bi
0
1
2
3
4
5
Balb/cBy 1
Relative DAPI density
25%
mono
bi
0
1
2
3
4
5
Balb/cBy 2
Relative DAPI density
45%
Mono Bi Mono Bi
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
Balb/cBy 3
50%
Mono Bi
Mono
Bi
0
1
2
3
4
5
Balb/cBy 4
Relative DAPI density
31%
Mono Bi
BALB/cByJ 1 BALB/cByJ 2 BALB/cByJ 3 BALB/cByJ 4
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
122
Supplementary Fig. S1. Quantitation of CM nuclear ploidy in parental strains. A. An example of using DAPI
nuclear fluorescence of CD31+ endothelial cells as a reference for diploid nuclei, compared to nuclei from
binucleated CMs. Cardiac troponin T (cTnT) is a CM-specific marker. All fluorescence intensity values are
normalized to the median value of the endothelial cell nuclei population. The red box indicates the 0.5-1.5
threshold for inclusion of a nucleus as diploid. B.Primary data of four BALB/cJ and four BALB/cByJhearts; each
graph represents ventricular cardiomyocytes of a different animal. None of these were from the same mice
shown in panel A. The numerical values shown indicate the percentage of diploid nuclei specifically within the
mononuclear CM population; these data points are graphed in Fig. 1B.
123
124
Supplementary Fig. S2.Polyploidy variation between BALB/cJand BALB/cByJ. A. CM analysis. Within the
mononuclear CM and binuclear CM subgroups, the percentages of nuclei of the indicated ploidy (2n, diploid; 4n,
tetraploid; 8n, octoploid) are shown. The 2n (diploid) percentage of mononuclear CMs is shown in Fig. 1B and
not repeated here. Data were calculated in this manner because a disproportionately greater number of
mononuclear CM nuclei were evaluated relative to their frequency in the heart. B. Hepatocyte analysis. Same
evaluation as for CMs; the 2n (diploid) percentage of mononuclear hepatocytes is shown in Fig. 1D and not
repeated here. C.Distribution of different polyploid subtypes for both CMs and hepatocytes, multiplying the data
from panels A and B with the percentage of mononuclear and binuclear cells shown in Fig. 1A and C. 1x4n and
2x2n cells are both tetraploid; 1x8n and 2x4n cells are octoploid. A small number of 2x8n cells were observed;
there were too few cells with other types of polyploidy to show on this chart. D. Analysis of CM polyploidy levels
in F1 mice derived from crosses of BALB/cJ and BALB/cByJ parents in both directions; same color scheme as in
panel C.
Supplementary Fig. S3. Evaluation of bone marrow cell polyploidy. All cell nuclei were from mononuclear
cells; no binucleated cells were observed.
125
A cJ x cByJ
Mono
Bi
0
1
2
3
4
5
McBy Fc M 1
Relative DAPI density
53%
Mono Bi
♂1
Mono
Bi
0
1
2
3
4
5
McBy Fc M 2
Relative DAPI density
61%
Mono Bi
♂2
Mono
Bi
0
1
2
3
4
5
McBy Fc M 3
Relative DAPI density
66%
Mono
Bi
♂3
Mono
Bi
0
1
2
3
4
5
McBy Fc M 4
Relative DAPI density
59%
Mono Bi
♂4
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
McBy Fc M 5
79%
Mono Bi
♂1
F1 1 F1 2 F1 3 F1 4 F1 5
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
McBy Fc F1
c
56%
Mono Bi
♀1
Mono
Bi
0
1
2
3
4
5
McBy Fc F2
Relative DAPI density
71%
Mono Bi
♀2
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
McBy Fc F3
80%
Mono Bi
♀3
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
McBy Fc F4
66%
Mono Bi
♀4
F1 1 F1 2 F1 3 F1 4
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Mono
Bi
0
1
2
3
4
5
Mc FcBy M1
Relative DAPI density
67%
Mono Bi
♂1
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
Mc FcBy M2
64%
Mono Bi
♂2
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
Mc FcBy M3
75%
Mono Bi
♂3
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
Mc FcBy M4
71%
Mono Bi
♂4
F1 1 F1 2 F1 3 F1 4
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Mono
Bi
0
1
2
3
4
5
Mc FcBy F1
Relative DAPI density
73%
Mono Bi
♀1
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
Mc FcBy F2
71%
Mono Bi
♀2
Mono
Bi
0
1
2
3
4
5
Mc FcBy F3
Relative DAPI density
63%
Mono Bi
♀3
Mono
Bi
0
1
2
3
4
5
Relative DAPI density
Mc FcBy F4
72%
Mono Bi
♀4
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
Relative DAPI intensity
F1 1 F1 2 F1 3 F1 4
B cByJ x cJ
126
Supplementary Fig. S4. Quantitation of CM nuclear ploidy in F1 mice. A.Primary data for five male and four
female F1 mice derived from crosses of BALB/cJ females to BALB/cByJ males. B. Primary data for four male and
four female F1 mice derived from crosses of BALB/cByJ females to BALB/cJ males. The numerical quantitation
values shown indicate the percentage of diploid nuclei specifically within the mononuclear CM population, and
are graphed as data points in Fig. 1F.
127
Vgll1
rs31755951
Scml2
rs255759965
Nhs
rs29303490
Zrsr2
rs213006026
Car5b
rs31413241
128
Gene Forward primer Reverse primer
Vgll1 ACCTGAAAGCAGTCAAAAACCG AGAGCAGCAAACTCAGCCTT
Scml2 CCATGACACCTGGCCTACAAA ACATCTCCTGTGAGGCGACA
Nhs TGTTGTTGGGGTCATCCAGC TCCAAGCAGCCCAAGTACAC
Zrsr2 GCCACGACTTCTGGACCTTG GGACTAGCTCCTCCTTCGGTA
Car5b TTGTCTCCCGGTCTTTGCTTT GTTGGAAGCCAACCTGGTCTA
Gdi1 TGTGATCATTGTGTTTTAGCAGTAGGG CCCCAACCTCCCCATAAAGTTT
Irs4 CCTTTATTGGTGTCCGGTGCT AGACCTGGAGATGGTCATGGC
Supplementary Fig. S5. BALB/cJvariants found to not be polymorphic with BALB/cByJ.The table shows a
screenshot from the phenome.jax.org web tool of selected SNPs identified in BALB/cJbased on comparison to all
available mouse whole genome sequence. SEA/GnJ was derived in part from BALB/cJ and so is grouped with
BALB/cJ in this table. The sequence traces show genomic sequencing of these candidate SNPs in BALB/cByJ; all
SNPs are identical in BALB/cJ and BALB/cByJ.
Supplementary Table 1. Sequences of primers used for gene amplification and sequencing
129

130

CHAPTER 6
Conclusions and perspectives
Tnni3k is a novel cardiac specific gene, and very little is known about its role in heart.
Our previous study identified Tnni3k to be a key regulator of mononuclear diploid
cardiomyocyte (MNDCM) frequency to influence adult mammalian heart regeneration
1
.
Additionally, Tnni3k was also reported in heart diseases, including cardiac conduction system
defects, dilated cardiomyopathy
2, 3, 4, 5
and heart ischemia/reperfusion (I/R) injury
6
. However,
the substrates and downstream signalings of Tnni3k remain largely unknown. In this
dissertation, I addressed the role of Tnni3k in regulating the MNDCM frequency with more
details
7
. And I confirmed that Tnni3k loss-of-function (LOF) is a direct cause of cardiac
conduction system defects and left ventricular remodeling. My study also revealed PKA
signaling as a downstream signaling of Tnni3k.
MNDCM in heart
The presence of MNDCM in adult mammalian heart is rare, and the biological
significance of this group of cardiomyocytes remains far from clear. Our previous study
revealed a positive correlation between the MNDCM frequency and adult mammalian heart
regeneration capacity
1
. However, there are still many unresolved puzzles about the presence of
the MNDCM in adult mammalian hearts. For example, if the MNDCM is an important
subpopulation of cardiomyocyte contributing to heart regeneration, why there are so few of
them in adult mammalian hearts. In embryonic stage, all cardiomyocytes are MNDCMs. Most

131

majorities of mammalian cardiomyocytes become polyploid and lose proliferative capacity soon
after birth, while only few of MNDCMs remain in adult heart. It has been reported that the
postnatal oxygen-rich environment triggers the polyploidization of MNDCM
8
. Researchers have
also reported that a subpopulation of proliferative cardiomyocytes in adult mouse heart are
mononuclear, diploid and hypoxic, residing in a microenvironment with low capillary density
9
.
Thus, it is likely that these hypoxic microenvironments in adult mammalian heart happen to
facilitate the maintenance of the MNDCM after birth, while most majorities of cardiomyocytes
exposed to the high level of oxidative stress become polyploid. As I've discussed before, the
acquisition of polyploidy is important for the efficient contractile and metabolic functions of
adult cardiomyocytes, and it also contributes to the hypertrophic growth of heart
9
. One other
speculation of the presence of MNDCM is that the MNDCM may play a specific role rather than
just regular contractile unit in heart. So far, there has been no demonstration of the
electrophysiological characteristics or the genetic, molecular signatures of the MNDCM in adult
mammalian heart. Thus, to unveil of the genuine function of the MNDCM, it might be
worthwhile to investigate its physiological and molecular properties.
Although there are accumulating evidences showing the correlation between the
MNDCM and heart regeneration
1, 10, 11, 12
, how to boost the proliferation of the MNDCM to
efficiently restore heart function after adult cardiac injuries is still an insurmountable obstacle.
Meanwhile, solely increasing the MNDCM frequency in adult mammalian hearts may cause
unexpected troubles to cardiac function. Therefore, much more efforts have to be taken to
investigate the MNDCM physiology in adult mammalian hearts.

132

Tnni3k regulates MNDCM frequency
In our previous study, we found that the MNDCM frequency is a multigene trait, and
Tnni3k was identified to be one of the genetic regulators
1
. In chapter 2 of this dissertation, I
addressed that Tnni3k functions as a kinase to regulate the MNDCM frequency, while it is not
functioning through the MAPK pathways
7
. However, the detailed molecular mechanisms of how
Tnni3k regulates the MNDCM frequency are unknown. One hypothesis is that Tnni3k may
mediate the oxidative stress response in cardiomyocyte to influence cardiomyocyte
polyploidization. As discussed previously, postnatal oxidative stress is an important trigger of
cardiomyocyte polyploidization
8
, but the oxidative stress experienced by each individual
cardiomyocyte might be different because of both cardiomyocyte-extrinsic and intrinsic
differences. The extrinsic conditions include coronary microvascular density, proximity to
ventricular lumen and presence of modulatory cells or molecules etc. The intrinsic conditions
can be the expression of genes that mediate oxidative stress production or response. I infer
that Tnni3k is a gene mediating oxidative stress response in cardiomyocyte, because there is
evidence showing that Tnni3k is mediating oxidative stress to modulate disease progression in
heart I/R injury
6
. Additionally, the subcellular localization of TNNI3K protein is partially in
mitochondria
13
, which implicated a potential involvement of Tnni3k in mitochondrial function
and reactive oxygen species (ROS) production.
It is also possible that Tnni3k doesn't influence cardiomyocyte cell cycle directly, in other
words, the MNDCM frequency increase is an indirect effect of Tnni3k LOF. In chapter 3 of this
dissertation, I've demonstrated that the Tnni3k LOF blunts beta-adrenergic-PKA signaling and

133

reduces cardiomyocyte contractility. In a very recent study, researchers found that the
inactivation of beta-adrenergic receptor (β-AR) improved the expression of Ect2, which is an
important cytokinesis component, and resulted in the elevation of MNDCM frequency
14
. This
result suggests that the blockage of β-AR-PKA signaling can increase the MNDCM frequency.
Hence, I infer that the increase of the MNDCM seen in Tnni3k LOF mutant mice might be
caused by inhibition of β-AR-PKA signaling. Additionally, Tnni3k LOF caused reduced
cardiomyocyte contractility under stress, which may affect mechanical properties of
cardiomyocyte and influence cardiomyocyte polyploidization. It has been reported that the
decreased mechanical loading on adult cardiomyocyte can promote cardiomyocyte
proliferation
15
. To my knowledge, there has not yet been any study on how the intrinsic
mechanical property change of cardiomyocyte can alter the cell cycle progression.
Tnni3k LOF causes heart diseases
In chapter 3 of this dissertation, I demonstrated that Tnni3k LOF caused a concentric left
ventricular remodeling of mouse heart. Concentric left ventricular remodeling is characterized
by the thickening of ventricular wall, which is an adaptive morphological change of heart in
response to increased cardiac afterload. In our Tnni3k LOF mutant mice, I reasoned that the
remodeling of ventricular wall is because of the cardiomyocyte architectural change (reduced
length to width ratio). By testing the contractility and calcium transient of cardiomyocyte, I
found that the Tnni3k LOF mutant cardiomyocytes have compromised cardiomyocyte
contractility and calcium transient under stress, which provided an explanation to the adaptive
morphological changes of cardiomyocyte and heart. I further identified PKA signaling to

134

function downstream of Tnni3k in regulating cardiomyocyte contractility and calcium
fluctuation. However, as I've shown, the heart function of Tnni3k LOF mutant mice is preserved
and the mice are externally normal, suggesting that the loss of Tnni3k might not cause serious
consequences independently. Additionally, I found that Tnni3k LOF mice have various
conduction system defects, as seen in human TNNI3K mutation carriers.
There are several TNNI3K mutations have been reported in families with familial cardiac
conduction system defects and dilated cardiomyopathy
2, 3, 4, 5
. To understand why those TNNI3K
mutations caused severe heart diseases in humans, a comprehensive analysis of the genetic
background in each family is essential. It is possible that TNNI3K mutations happened to couple
with other unrecognized genetic abnormalities in their genomes and manifest those severe
cardiac dysfunctions synergistically. Besides genetic variants, TNNI3K LOF is also likely to couple
with other cardiac risk factors like alcohol abuse, cardiotoxic drugs (for example, anti-tumor
drugs) use to manifest severe outcomes, which is worthwhile to investigate in lab settings.
According to the ExAC database, Tnni3k LOF and hypomorphic mutations are common in
human population. I created the Tnni3k-I685T knock-in mouse, which recapitulates the most
common TNNI3K mutation in human population. I demonstrated that Tnni3k-I685T mutation
caused increased MNDCM frequency, blunted PKA targets phosphorylation and myocardium
wall thickening, but not obvious conduction system defects and cardiomyopathy. The I685T
allele might contribute to the manifestations of conduction system diseases and
cardiomyopathies when the appropriate constellation genetic and environmental risk factors
are present. By this perspective, individuals with TNNI3K LOF alleles might be considered at-risk

135

for heart diseases, and were such individuals identified prior to disease onset, could adopt
dietary or lifestyle approaches to prevent the development of disease regardless of their
genetic background.
When considered with earlier studies, TNNI3K has several functions that at a first glance
might appear to be opposing. Here, we show that Tnni3k LOF mutations predispose to
conduction system dysfunction and (based on the human families) to onset of DCM, implying a
beneficial function for the normal protein that would presumably have been subjected to
positive selection during evolution. In contrast, when Tnni3k is mutated, hearts are more
regenerative, and in adult ischemia-reperfusion, absence of Tnni3k is cardioprotective; these
properties would seem to favor gene loss during evolution. These phenotypes are not opposite
or mutually exclusive, as they pertain to different times, time scales, levels of ROS, and
biological processes. By mediating oxidative stress response, we propose that Tnni3k has a
cardioprotective role under normal conditions (i.e., without experimental challenge), perhaps
to modulate normal fluctuations in oxidative stress that cardiomyocytes experience as part of
normal heart biology. Paradoxically, this function is maladaptive in the context of extreme
oxidative stress (in adult ischemia-reperfusion), and is antiregenerative in causing
cardiomyocyte cell cycle arrest. To the extent that all of these processes are inextricably
associated with oxidative stress and oxidative stress response, this may provide at least one
clue for why the mammalian heart evolved to become mostly nonregenerative after birth: it
may have been more beneficial to have Tnni3k and its associated pathways to prevent adult
conduction system problems and DCM, even if this involved cardiomyocytes becoming
hypersensitive to oxidative stress-inducing injury and mostly unable to regenerate.

136

References
1. Patterson M, Barske L, Van Handel B, et al. Frequency of mononuclear diploid
cardiomyocytes underlies natural variation in heart regeneration. Nat Genet. 2017;49(9):1346–
1353. doi:10.1038/ng.3929
2. Xi Y, Honeywell C, Zhang D, et al. Whole exome sequencing identifies the TNNI3K gene as a
cause of familial conduction system disease and congenital junctional ectopic tachycardia. Int J
Cardiol. 2015;185:114–116. doi:10.1016/j.ijcard.2015.03.130
3. Theis JL, Zimmermann MT, Larsen BT, et al. TNNI3K mutation in familial syndrome of
conduction system disease, atrial tachyarrhythmia and dilated cardiomyopathy.Hum Mol
Genet. 2014;23(21):5793–5804. doi:10.1093/hmg/ddu297
4. Fan LL, Huang H, Jin JY, et al. Whole exome sequencing identifies a novel mutation
(c.333 + 2T > C) of TNNI3K in a Chinese family with dilated cardiomyopathy and cardiac
conduction disease. Gene. 2018;648:63–67. doi:10.1016/j.gene.2018.01.055
5. Podliesna S, Delanne J, Miller L, et al. Supraventricular tachycardias, conduction disease, and
cardiomyopathy in 3 families with the same rare variant in TNNI3K (p.Glu768Lys).Heart Rhythm.
2019;16(1):98–105. doi:10.1016/j.hrthm.2018.07.015
6. Vagnozzi RJ, Gatto GJ Jr, Kallander LS, et al. Inhibition of the cardiomyocyte-specific kinase
TNNI3K limits oxidative stress, injury, and adverse remodeling in the ischemic heart. Sci Transl
Med. 2013;5(207):207ra141. doi:10.1126/scitranslmed.3006479

137

7. Gan P, Patterson M, Velasquez A, et al. Tnni3k alleles influence ventricular mononuclear
diploid cardiomyocyte frequency. PLoS Genet. 2019;15(10):e1008354. Published 2019 Oct 7.
doi:10.1371/journal.pgen.1008354
8. Puente BN, Kimura W, Muralidhar SA, et al. The oxygen-rich postnatal environment induces
cardiomyocyte cell-cycle arrest through DNA damage response [published correction appears in
Cell. 2014 May 22;157(5):1243]. Cell. 2014;157(3):565–579. doi:10.1016/j.cell.2014.03.032
9. Gan P, Patterson M, Sucov HM. Cardiomyocyte Polyploidy and Implications for Heart
Regeneration. Annu Rev Physiol. 2020;82:45–61. doi:10.1146/annurev-physiol-021119-034618
9. Kimura W, Xiao F, Canseco DC, et al. Hypoxia fate mapping identifies cycling cardiomyocytes
in the adult heart [published correction appears in Nature. 2016 Apr 14;532(7598):268]. Nature.
2015;523(7559):226–230. doi:10.1038/nature14582
10. Han L, Choudhury S, Mich-Basso JD, et al. Lamin B2 Levels Regulate Polyploidization of
Cardiomyocyte Nuclei and Myocardial Regeneration [published online ahead of print, 2020 Feb
26]. Dev Cell. 2020;S1534-5807(20)30062-9. doi:10.1016/j.devcel.2020.01.030
11. González-Rosa JM, Sharpe M, Field D, et al. Myocardial Polyploidization Creates a Barrier to
Heart Regeneration in Zebrafish. Dev Cell. 2018;44(4):433–446.e7.
doi:10.1016/j.devcel.2018.01.021
12. Hirose K, Payumo AY, Cutie S, et al. Evidence for hormonal control of heart regenerative
capacity during endothermy acquisition. Science. 2019;364(6436):184–188.
doi:10.1126/science.aar2038

138

13. Tang, H., Xiao, K., Mao, L., Rockman, H. A., & Marchuk, D. A. (2013). Overexpression of
TNNI3K, a cardiac-specific MAPKKK, promotes cardiac dysfunction. Journal of molecular and
cellular cardiology, 54, 101–111. https://doi.org/10.1016/j.yjmcc.2012.10.004
14. Liu H, Zhang CH, Ammanamanchi N, et al. Control of cytokinesis by β-adrenergic receptors
indicates an approach for regulating cardiomyocyte endowment. Sci Transl Med.
2019;11(513):eaaw6419. doi:10.1126/scitranslmed.aaw6419
15. Canseco DC, Kimura W, Garg S, et al. Human ventricular unloading induces cardiomyocyte
proliferation. J Am Coll Cardiol. 2015;65(9):892–900. doi:10.1016/j.jacc.2014.12.027

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

TNNI3K expression affects nucleation of cardiomyocytes and skeletal myoblasts

PDF

A novel role for post-migratory neural crest cells in cardiac outflow tract alignment: insights into the molecular basis of concomitant congenital cardiac and cleft malformations

PDF

Proliferation and maturation events in second heart field cells during cardiovascular development activated by the Delta like ligand-4 mediated notch signaling

PDF

Investigating the role of cardiac fibroblasts in modulating angiogenesis during heart injury

PDF

Zebrafish as a blueprint for cardiac regeneration

PDF

Developmental biology of epicardial adipose tissue

PDF

The role of ERK1/2 in mouse embryonic stem cell fate control

PDF

Retinoic acid and TGFβ signaling regulate cardiovascular development

PDF

Characterization of Cxcr7 in zebrafish cardiac lymphatic vessel development

PDF

Investigation of butyrate’s effects on colonic stem cell development

PDF

β-catenin couples self-renewal, induction and epithelial morphogenesis and patterning at the outset of mammalian nephrogenesis

PDF

Cross-species comparison of non-canonical Wnt signaling in the developing retina

PDF

Fibroblastic connective tissue cells: the blastema stem cells and source of large-scale chondrogenesis in the regenerating lizard tail

PDF

Role of beta-catenin in mouse epiblast stem cell, embryonic stem cell self-renewal and differentiation

PDF

Characterization of transgenic zebrafish lines for studying role of platelet derived growth factor (PDGF) signaling on coronary vessel formation in regenerating zebrafish heart after cryoinjury

PDF

Identification of therapeutic targets in human cerebral brain organoid models of neurodegeneration

PDF

DNA methylation in the mouse cochlea promotes maturation of supporting cells and contributes to the failure of hair cell regeneration

PDF

Investigating the role of STAT3 in mouse and rat embryonic stem cell self-renewal and differentiation

PDF

Characterization of metabolically distinct muscle resident fibro-adipogenic subpopulations reveals a potentially exploitable mechanism of skeletal muscle aging and disease

PDF

The role of Runx2 in the development of the tendon-bone attachment unit

Asset Metadata

Creator Gan, Peiheng (author)

Core Title The roles of Tnni3k in heart regeneration, cardiac conduction system defects and cardiomyopathy

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Development, Stem Cells and Regenerative Medicine

Publication Date 04/30/2020

Defense Date 03/06/2020

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag cardiac conduction system defect,cardiomyopathy,heart regeneration,mononuclear diploid cardiomyocyte,OAI-PMH Harvest,TNNI3K

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Lien, Ching-Ling (committee chair), Sucov, Henry M. (committee member), Xu, Jian (committee member), Ying, Qilong (committee member)

Creator Email ganp@musc.edu,peihengg@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-293376

Unique identifier UC11664315

Identifier etd-GanPeiheng-8380.pdf (filename),usctheses-c89-293376 (legacy record id)

Legacy Identifier etd-GanPeiheng-8380.pdf

Dmrecord 293376

Document Type Dissertation

Rights Gan, Peiheng

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA