University of Southern California Dissertations and Theses

Discovery of mitochondria-DNA-encoded microproteins in neurodegeneration

(USC Thesis Other)

Discovery of mitochondria-DNA-encoded microproteins in neurodegeneration

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

DISCOVERY OF MITOCHONDRIA-DNA-ENCODED MICROPROTEINS IN
NEURODEGENERATION

by

Brendan Miller

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
(NEUROSCIENCE)

August 2022

Copyright 2022 Brendan Miller
ii
ACKNOWLEDGEMENTS
Completing doctoral research is not a one-person effort. My work presented here is a
collective effort from family, friends, and my lab.
To my parents David and Jolene, thank you for your constant support, encouragement, and
willingness to listen. The Lou Malnati’s deep dish pizza you shipped on dry ice were the true
support, though. To my brother Justin and sister-in-law Amanda, your inclusiveness and friendship
are invaluable. I can’t imagine what it’s like to have unfortunately seen me so often over these six
years. To my nephew West, thank you for only throwing up on me once and sharing your cards
for consecutive hours at a time. To my friends who likely won’t read this enormous document
(understandably so), I appreciate your constant backing. To my friends within the neuroscience
program – especially Artemis, Clio, and Yuni – I’m grateful to have shared this experience with
you.
I spent most of my time over the last six years with the people in my lab. They are
colleagues but more importantly friends and mentors. Dr. Pinchas Cohen, my personal and
professional mentor, instilled within me confidence to approach intimidating problems. However,
the wine we uncorked and golf balls we lost (mostly I lost) were our most impressive work. Dr.
Su-Jeong Kim was the most influential experimental mentor in lab. From her I learned how to
comprehensively design experiments and creatively consider solutions to difficult problems.
Hemal Mehta, our lab manager, is the glue that holds the lab together. I spent the most time with
Hemal of anyone in lab over the last six years. She always encouraged me, purchased reagents for
all my stupid ideas, and helped me with animal work. Professor Dr. Kelvin Yen’s unique positive
attitude mixed in with realism made lab meetings enjoyable. But Dr. Yen always provided honest
feedback and pointed me towards scientific directions with greater probabilities of success.
iii
Professor Dr. Junxiang Wan never said no to my asks, whether it be making my antibodies or
measuring any analyte for which I asked. Dr. Hiroshi Kumugai, a post-doc and one of the hardest
workers whom I’ve ever met, always offered help even when I didn’t ask. Dr. Joyce Xiao, a
graduate student in lab when I first joined, helped form my scientific thinking process. Senior
research technician Ricardo Ramirez had a positive attitude even with the Dodgers always lost in
the playoffs. Kevin Cao, my right-hand man and research technician for multiple years, provided
exceptional technical work. Regina Gonzalez-Braniff, a skilled research technician, always gave
her best effort and was extremely trustworthy. Graduate students Mill, Zef, Ana, and Melanie,
were extremely patient with me when offering support and ideas while my experience was limited.
I also received strong support from people outside my lab but within the Leonard Davis
School of Gerontology. Dr. Eileen Crimmins provided me with succinct, genuine feedback that
molded my science communication. Dr. Thalida “Em” Arpawong formed my bioinformatic
thinking processes, provided thorough feedback, and consistently made time for me. Dr. Jennifer
Ailshire first connected me with people and resources to grow my bioinformatic skills when I
joined USC. My committee (Drs. Pike, Braskie, Lee, and Tower) provided invaluable feedback
and made time to discuss project goals or even process bulk loads of data for me. The Business
Office always gave me batteries and helped fix urgent lab problems. David Hong, the best FedEx
employee in its history, made sure I received packages and always offered to personally ensure
my time-sensitive packages were shipped. Dr. Sean Curran trusted me with giving presentations
and talking with outside scientists and students. To the rest of the faculty, thank you for letting me
steal your monthly meeting lunches and welcoming me to dinners.
iv
TABLE OF CONTENTS

Acknowledgements ............................................................................................................. ii
List of Figures .................................................................................................................... vi
Abbreviations ................................................................................................................... viii
Abstract ................................................................................................................................x
Chapter 1: Introduction ........................................................................................................1
1.1 Discovery of Microproteins ...............................................................................1
1.2 Mitochondria-sORF-Encoded Microproteins ....................................................4
1.3 Human Mitochondrial Genomics .......................................................................7
1.4 Mitochondrial-Derived Peptides in Age-Related Diseases ..............................10
Chapter 2: The Utility of Human Mitochondrial DNAVariation Data for Genome Wide
Association Studies ............................................................................................................12
2.1 Abstract ............................................................................................................12
2.2 Introduction ......................................................................................................14
2.3 Results ..............................................................................................................17
2.4 Discussion ........................................................................................................20
2.5 Methods............................................................................................................22
Chapter 3: Mitochondrial DNA Variation in Alzheimer’s Disease Reveals a Unique
Microprotein Called SHMOOSE .......................................................................................30
3.1 Abstract ............................................................................................................30
3.2 Introduction ......................................................................................................31
3.3 Results ..............................................................................................................33
3.4 Discussion ........................................................................................................41
v
3.5 Methods............................................................................................................45
Chapter 4: Mitochondrial DNA Variation in Extreme Longevity Reveals Rare Humanin
Microprotein Variant that Promotes APOE4 Resilience ...................................................73
4.1 Abstract ............................................................................................................73
4.2 Introduction ......................................................................................................74
4.3 Results ..............................................................................................................76
4.4 Discussion ........................................................................................................82
4.5 Methods............................................................................................................87
Chapter 5: Conclusion ......................................................................................................104
References ........................................................................................................................110

vi
LIST OF FIGURES

Figure 2.1 Multi-ethnic principal component analysis on nuclear and mitochondrial
single nucleotide polymorphisms.. ........................................................................25
Figure 2.2 Intra-ethnic principal component analysis on mitochondrial SNPs .................26
Figure 2.3 Comparing amount of variance captured by nuclear and mitochondrial
principal components .............................................................................................27
Figure 2.4 Comparison of nuclear, mitochondrial, and combined nuclear/mitochondrial
PCA for statistically classifying individuals into broader ethnic sub-groups.. ......28
Figure 2.5 Effects of 20 mitochondrial principal components on height (centimeters) in
inter-ethnic and intra-ethnic samples .....................................................................29
Figure 3.1 Mitochondrial rs2853499 changes the amino acid sequence of SHMOOSE
and associated with AD and neuroimaging modalities ..........................................65
Figure 3.2 SHMOOSE levels in CSF correlate to age, tau, and brain white matter
microstructure ........................................................................................................67
Figure 3.3 SHMOOSE associates with differential mitochondrial and ribosomal gene
expression in humans, cells, and mice ...................................................................68
Figure 3.4 SHMOOSE is a biologically active microprotein that localizes to
mitochondria and boosts metabolic activity and oxidative consumption rate .......70
Figure 3.5 SHMOOSE binds the inner mitochondrial membrane protein mitofilin ..........72
Figure 4.1 Humanin CSF by APOE4 and dementia ..........................................................96
Figure 4.2 The humanin variant P3S is enriched in APOE4 centenarians .........................97
Figure 4.3 Humanin binds APOE, but humanin P3S binds APOE4 with much higher
affinity ....................................................................................................................98

vii
Figure 4.4 Molecular dynamics simulations (umbrella sampling) of humanin and
APOE ....................................................................................................................99
Figure 4.5 Humanin co-expression analysis by APOE genotype in the human brain
temporal cortex ....................................................................................................100
Figure 4.6 Humanin P3S increases Aβ phagocytosis in APOE4 glia ..............................101
Figure 4.7 Humanin P3S reduces brain Aβ burden in APP/PS1/APOE4 mice ...............102
Figure 4.8 Humanin and humanin P3S differentiates the hippocampal transcriptome
in APP/PS1/APOE4 mice ....................................................................................102
viii
ABBREVIATIONS

AD, Alzheimer's disease
APOE, Apolipoprotein E
APP, amyloid precursor protein
Aβ, amyloid beta
Co-IP, co-immunoprecipitation
CSF, cerebrospinal fluid
DNA, deoxyribonucleic acid
FAD, familial Alzheimer's disease
GWAS, genome wide association study
iPSCs , induced pluripotent stem cells
MDP, mitochondrial-derived peptide
MICOS, mitochondrial contact site and cristae organizing system
MiWAS, mitochondrial wide association study
mRNA, messenger RNA
MS, mass spectrometry
mtDNA, mitochondrial DNA
mtSNP, mitochondrial SNP
nucDNA, nuclear DNA
PCA, principal component analysis
PheWAS, phenome wide association study
PS1, presenilin-1
RNA-Seq, RNA sequencing
ix
RNA, ribonucleic acid
rRNA, ribosomal RNA
SHMOOSE, small human mitochondrial open reading frame over serine tRNA
SNP, single nucleotide polymorphism
sORF, small open reading frame
tRNA, tranfser RNA
UTR, untranslated region
x
ABSTRACT
Microproteins are a new class of bioactive peptides, but the genes that encode these
microproteins have escaped biologists for decades. Now, due to enhanced technology and high
volume of ‘omics’ data, microprotein discoveries are growing in number. This thesis highlights an
integrative microprotein discovery approach that consists of population genetics, transcriptomics,
proteomics, and in vitro and in vivo experiments. The goal of this thesis was to identify and
functionalize new microproteins encoded by the mitochondrial genome. In chapter 1, we introduce
the biological significance of microproteins with an emphasis on human genetic variation. In
chapter 2, we present analyses that illustrate the significant degree of human mitochondria DNA
variation in population cohorts – these analyses are the foundation upon which we developed our
microprotein-specific population genetics methods. In chapter 3, we carry out a mitochondrial
genome wide association study to identify microprotein-encoding DNA regions that associate with
Alzheimer’s disease (AD). Indeed, we show that an AD-risk mitochondria single nucleotide
polymorphism changes the amino acid sequence of a previously unannotated microprotein called
SHMOOSE, a peptide that we ultimately detected in neuronal mitochondria and cerebrospinal
fluid. SHMOOSE is the first reported microprotein encoded by mitochondria DNA that has been
detected by mass spectrometry. In chapter 4, we reveal a rare mitochondria DNA variant that
changes the third amino acid of the microprotein humanin and protects against APOE4-related
decline. Simultaneously, we show that humanin binds APOE and – notably – the APOE4-
associative variant binds APOE 15 times greater and dramatically attenuates AD pathology in the
APP/PS1/APOE4 mouse model. Altogether, our microprotein results could impact gene
annotation, AD therapeutic development, AD biomarker development, and AD experimental
models.
1
CHAPTER 1: INTRODUCTION
1.1 Discovery of Microproteins
Nearly twenty years ago, the Human Genome Project estimated that 20,000-25,000 genes
encode functional proteins (International Human Genome Sequencing, 2004). Today, over 18,000
of these proteins have been validated by the Human Proteome Project (Omenn, 2021). However,
one element has been missed: microproteins.
The term microprotein refers to biologically active peptides shorter than 100 amino acids
(Saghatelian & Couso, 2015). Bioinformatic analyses suggest millions of putative small open
reading frames (sORFs) could encode for functional microproteins. Of these millions of sORFs,
molecular ribosome profiling sequencing techniques have pinpointed tens of thousands of potential
microproteins (Martinez et al., 2020). These data have been added to genome-wide information on
protein synthesis visualization (GWIPs-viz; https://gwips.ucc.ie/), allowing researchers to explore
sORFs across species, models, and experimental conditions.
Bona fide microproteins challenge traditional gene annotation. Most human genes have
been described as monocistronic, but nearly three quarters of microprotein sORFs detected by
ribosome profiling are encoded within 5’ untranslated regions (UTR) (Martinez et al., 2020;
Mouilleron, Delcourt, & Roucou, 2016). As a result, many transcriptomes might not actually be
monocistronic and in fact encode for multiple unique proteins ranging from dozens of amino acids
to hundreds of amino acids (large downstream coding region). One example of this phenomenon
is a microprotein encoded by a sORF in the 5’ UTR of the gene mitochondrial elongation factor
1(MIEF1) (Rathore et al., 2018). Both the MIEF1 microprotein (MIEF-1MP) and the MIEF
annotated large protein act together. MIEF1-MP localizes to mitochondria – as does the larger
2
MIEF protein – and modifies mitochondrial translation rates. Other examples include the
microproteins ASDURF, BiP ORF, HJV uORF, MP31, PRL-1 and PRL-2 uORF, and SEHBP –
all of which have diverse functionality related to protein chaperones, ion homeostasis, and
metabolic regulation (Cloutier et al., 2020; Eltermaa et al., 2019; Hardy et al., 2019; Huang et al.,
2021; Koh et al., 2021; Onofre, Tome, Barbosa, Silva, & Romao, 2015). Without advances in
proteomic and genomic technologies, the transcripts on which these sORFs reside would still be
considered monocistronic.
Many other transcript types in both prokaryotic and eukaryotic genomes contain sORFs.
Ironically, several “long noncoding RNAs” encode for biologically active microproteins. The
microprotein ASAP – encoded by LINC00467 – induces age-associated colorectal cancer
proliferation, while the microprotein CIP2A-BP– encoded by LINC00665 – inhibits triple-negative
breast cell invasion (Ge et al., 2021; B. Guo et al., 2020). Additionally, in a comparative genomics
study on almost 2,000 metagenomes, ~4,500 candidate microproteins were categorized into cell-
cell communication, antimicrobial, antiphage, and adaption activities (Sberro et al., 2019). Since
the gut microbiome has been connected to age-related disease progression – including AD and
metabolic dysfunction (Jiang, Li, Huang, Liu, & Zhao, 2017) – the repercussions of both
eukaryotic and prokaryotic microproteins are significant to human biology.
Several limitations and areas of improvement must be considered when studying
microproteins. For instance, most ribosome profiling-identified microproteins haven’t been
detected by mass spectrometry because they are small, low abundant, or hydrophobic (Martinez et
al., 2020). Furthermore, it is computationally challenging to detect the exact sORF undergoing
translation because codon periodicity often overlaps several sORFs. As a result, calling algorithms
make two choices: throw away the reads as low confident, leading to false negatives; or infer active
3
sORF translation, leading to false positives (Ji, 2018). Due to these restrictions, hundreds of
thousands of exclusive sORF might be called across experiments, leading to reproducibility
problems. To overcome these problems, antibodies have been made against select bona fide
microproteins such as the mitochondrial-modifying peptides BRAWNIN, humanin, and MOTS-c
(S.-J. Kim et al., 2020; S. Zhang et al., 2020). If the field is to continue to grow, solving these
problems is necessary.

4
1.2 Mitochondria-sORF-Encoded Microproteins (i.e., Mitochondrial-derived Peptides;
MDPs)
Human mitochondrial DNA contain hundreds of small open reading frames that encode
putative microproteins called “mitochondrial-derived peptides” (MDPs). Some of these act
intracellularly and other others are found in the systemic circulation that target various tissues
(Lee, Yen, & Cohen, 2013). The first MDP discovered is humanin, a 24 amino acid peptide
encoded from the 16S rRNA region of mtDNA (Hashimoto, Niikura, et al., 2001). Hashimoto et
al. initially cloned humanin from the resilient occipital lobe of an AD patient brain and found that
the peptide protected against amyloid-beta toxicity (Hashimoto, Ito, et al., 2001). Around the same
time, two additional labs discovered humanin as a cytoprotective peptide that bind the pro-
apoptotic molecules IGFBP3 and BAX (B. Guo et al., 2003; Ikonen et al., 2003). Since then,
humanin has been described as a cytoprotective factor in cardiovascular, metabolic, and
neurological contexts. These effects have been in part mediated by the interaction between
humanin and the tripartite receptor gp130, WSX1, and CNTR as well as with a second interacting
receptor formyl peptide receptor 2 (Hashimoto, Kurita, Aiso, Nishimoto, & Matsuoka, 2009).
Downstream effects of this humanin cascade include activation of the AKT/ERK1/2, and STAT3
pathways (S. J. Kim et al., 2016).
Seven additional mitochondrial microproteins have been identified since the discovery of
humanin. Of these seven, six were named Small Humanin Like Peptides 1 – 6 (SHLPs 1-6), which
are encoded from the 16S rRNA region and share some biological features with humanin (Cobb
et al., 2016). For example, SHLP2 protects cells from amyloid beta-induced toxicity and age-
related macular degeneration (Nashine, Cohen, Nesburn, Kuppermann, & Kenney, 2018). SHLP2
also has been characterized as a chaperone because it bound IAPP species and blocked amyloid
5
seeding (Okada et al., 2017). This chaperone-like activity might link its cytoprotective roles,
suggesting that SHLP2 has potential as a metabolic therapeutic. Moreover, administration of
SHLP2 and SHLP3 promote mitochondrial biogenesis, reduce reactive oxygen species, and
decrease mtDNA oxidation (Cobb et al., 2016). Unlike these cytoprotective SHLPs, SHLP6 was
shown to induce apoptosis in multiple cell lines (Cobb et al., 2016). Much remains to be learned
about the mechanisms of these SHLPs with future experimentation.
Another MDP that has been studied deeply over the past several years is MOTS-c, a 16-
amino-acid peptide encoded by a mitochondrial sORF within the 12S rRNA (Lee et al., 2015).
MOTS-c was first described as an exercise mimetic-peptide because it prevented weight gain in
high-fat-diet-induced obese mice, improved insulin sensitivity, and increased exercise capacity in
both obese and old mice (Lee, Kim, & Cohen, 2016; Reynolds et al., 2021). In addition, MOTS-c
acts as retrograde signaling molecule by translocating from mitochondria to the nucleus and
binding to metabolism-regulating transcription factors (e.g., NRF1) (K. H. Kim, Son, Benayoun,
& Lee, 2018). Separate reports showed that MOTS-c increased glucose uptake and stimulated
glycolysis (S.-J. Kim, Miller, Kumagai, Yen, & Cohen, 2019). These glycolysis-stimulated effects
of MOTS-c were notably muted when knocking down AMPK and SIRT1, suggesting that AMPK
and SIRT1 might be part of MOTS-c action (Lee et al., 2015) and involved in longevity.
Currently, mitochondrial DNA is annotated with 13 large mRNAs, 22 tRNAs, and 2
rRNAs. Yet mitochondrial genomic regulation appears much more complex since the emergence
of MDPs, long non-coding RNAs, and small RNAs. Indeed, in their hallmark paper, Mercer et al.
found dozens of previously uncharacterized cleavage sites and small RNAs derived from tRNAs
with unknown function (Mercer et al., 2011). In another report, nearly 400 putative MDPs between
9 and 40 amino acids in silico were annotated and considered putative (B. Miller et al., 2020). To
6
characterize these putative orphan MDPs, existing technology needs to be enhanced, especially
ribosome profiling technology. Specifically, a mitochondrial ribosome inhibitor that stalls
ribosomes at the start codon could yield additional MDPs. Overall, technical advancement in
mitochondrial ribosome profiling and small peptide enrichment mass spectrometry has potential
to discovery of new MDPs.

7
1.3 Human Mitochondrial Genomics
Given that there are approximately 100-1000 copies of mtDNA per cell and 37 trillion cells
in the human body, one human might contain nearly 2 x 10
15
copies of mtDNA (Bianconi et al.,
2013; Rooney et al., 2015). However, existing genomic tools are primarily designed to study
nuclear DNA, as mtDNA does not undergo recombination or follow Hardy Weinberg equilibrium.
As a result, bioinformatic pipelines and genetic editing techniques for mtDNA are limited. For
instance, during genome wide association studies (GWAS), mtDNA variants are usually filtered
from analytic plans. Just a few years ago, a commonly used GWAS tool called PLINK was updated
to accurately estimate the effect of mtDNA variants during mtDNA-exclusive analysis (B. Miller
et al., 2019). Yet unlike traditional GWAS, there is no gold standard method for mitochondrial
wide association studies (MiWAS). In GWAS, genetic population structures (genetic ancestry)
are controlled by data reduction techniques such as principal component analysis. Perhaps no
better illustration of this is the 2008 report by Novembre et al. in which principal component
analysis on half a million DNA variants in Europeans mirrored the geography of Europe
(Novembre et al., 2008). In previous MiWAS reports, though, many analytic methods did not
considered controls for mtDNA-specific genetic ancestry. Some of these analytic methods instead
considered mitochondrial haplogroups based on mitochondrial SNPs, but these haplogroup
assignments are largely based on genome arrays that might lack depth (Biffi et al., 2010; McRae,
Byrne, Zhao, Montgomery, & Visscher, 2008; van Oven & Kayser, 2009). Further complicating
MiWAS is that population cohorts often have extremely variable mtSNP frequencies.
Nevertheless, recent reports highlighted significant effects of frequent mtSNPs on human
phenotypes in large population cohorts. For example, a report by Yonova-Doing et al. included a
phenome wide mtDNA-phenotype association analysis on 260 candidates in over 300,000
8
individuals (Yonova-Doing et al., 2021). They found significant associations between mtDNA
variants and type 2 diabetes, multiple sclerosis, height, and liver and renal function. Likewise,
Kraja et al. found several mtDNA variants that associated with multiple metabolic traits in 45
combined cohorts (Kraja et al., 2019). Independent reports on smaller cohorts noted associations
between mtDNA variation and neurodegeneration including Parkinson’s disease, Alzheimer’s
disease, and eye disease (Hudson et al., 2013; Lakatos et al., 2010; Brendan Miller, Mina Torres,
et al., 2020). While MiWAS can reveal meaningful mitochondrial genomic regions, its statistical
limitations necessitate experimental validation.
Validating MiWAS associations experimentally is incredibly challenging due to the
fundamental problem that mitochondrial DNA editing lacks fidelity. Whereas labs can edit single
nuclear nucleotides with CRISPR, mitochondrial DNA cannot be edited with similar precision
(Hussain, Yalvac, Khoo, Eckardt, & McLaughlin, 2021). Thus, MiWAS is rarely followed up with
comprehensive functional experimentation, although in vitro models called cybrids – whereby a
cell line is depleted of mitochondrial DNA and then replaced by donor mitochondrial DNA – have
been used (Fang et al., 2018; Khan et al., 2000). Cybrid approaches have revealed functional
effects of certain mtDNA variants, but they are limited by the fact that other mtDNA variants are
transferred to the parent cell line. To bypass problems with cybrids and mtDNA gene editing,
overexpression or recombinant administration of mitochondrial-encoded proteins have been
considered. In cells that harbored mtATP6 mutations, over expression of mtATP6 restored
homeostasis (Chin, Panavas, Brown, & Johnson, 2018). Similarly, a SNP in MOTS-c leading to a
variant of MOTS-c called K14Q raises the risk of T2D in Japanese men and unlike WT MOTS-c
failed to protect from metabolic dysfunction in vivo proving it to be a bioinactive form of the
hormone (Zempo et al., 2021). Moreover, a separate mitochondrial SNP within the humanin sORF
9
associated with lower circulating humanin peptide and with more severe cognitive decline,
suggesting that the variant affects translation of the humanin transcript (Kelvin Yen et al., 2018)
leading to decreased neuroprotection. In forthcoming years, precise mtDNA editing, whole
genome sequencing of large population cohorts, and functional mitochondrial gene annotation can
all help validate MiWAS associations.

10
1.4 Mitochondrial-Derived Peptides in Age-Related Diseases
MDPs have been extensively studied in the context of age-related diseases. For example,
since humanin was originally detected in the occipital lobe of an AD patient, it has been tested as
a therapeutic agent in several models of neurodegeneration. In pre-clinical in vivo experiments,
humanin prevented synaptic loss in hippocampal neurons and reduced astrocytic inflammation
(Zarate, Traetta, Codagnone, Seilicovich, & Reines, 2019). In double and triple transgenic mice
models of Alzheimer’s disease (AD), an analogue of humanin called S14G (HNG) improved
cognition (Niikura, Sidahmed, Hirata-Fukae, Aisen, & Matsuoka, 2011; W. Zhang et al., 2012). In
humans, patients with AD had lower humanin levels in CSF than controls (Kelvin Yen et al.,
2020), and humanin genetic variation was linked to cognition, as the naturally occurring
m.2706A>G polymorphism (rs2854128) within the humanin sORF associated with accelerated
cognitive aging in African Americans (Kelvin Yen et al., 2018). Altogether, these observations
suggest that humanin could be a potential biomarker and therapeutic target for cognitive decline
and neurological disorders such as AD.
Moreover, several studies suggest MOTS-c and humanin are possible biomarkers for CVD.
People with endothelial dysfunction – one of the strong risk factors for cardiovascular events
(Widlansky, Gokce, Keaney, & Vita, 2003) – displayed low MOTS-c and humanin levels.
Likewise, circulating MOTS-c and humanin levels positively correlated with coronary endothelial
function (Qin et al., 2018; Widmer et al., 2013). A follow-up study by Ikonomidis et al.
demonstrated that T2D patients with low circulating MOTS-c levels (< 167 ng/ml) exhibited over
3-fold higher risk of cardiac events than those with high MOTS-c levels (Ikonomidis et al., 2020).
Similarly, Cai et al. demonstrated that circulating humanin levels at baseline were an independent
risk factor for major adverse cardiac events in patients with angina (Cai et al., 2022). In vivo and
11
in vitro experiments support these observations. Wei et al. reported that MOTS-c prevented
vascular calcification by activating the AMPK signaling pathway and suppressing of AT-1
(angiotensin II type I) and ET-B (endothelin B) expression in rats treated with Vitamin D3 and
nicotine (Wei et al., 2020). Moreover, humanin has increased KLF2 expression –an essential
transcriptional regulator of endothelial function – and regulated eNOS (endothelial nitric oxide
synthase) and ET-1 (endothelin-1) in vitro. Comparably, humanin has suppressed endothelial
dysfunction and atherosclerosis progression in vivo (Wang et al., 2018). These findings suggest
MOTS-c and humanin are associated with CVDs via endothelin and vasoactive regulation.
Studies have demonstrated associations between MDPs and cancer. Xiao et al. showed that
prostate cancer (PCa) patients had low circulating SHLP2 levels (Xiao et al., 2017). They
suggested that circulating SHLP2 levels may be useful for predicting the risk of PCa in patients
undergoing biopsy (Xiao et al., 2017). Separately, several studies suggest humanin ameliorates
negative side effects of chemotherapy (Cohen, 2014; Eriksson et al., 2014; Jia et al., 2015). In
addition, Lue et al. demonstrated that HNG-treatment in mouse models not only decreased
negative side effects of chemotherapy but also decreased metastasis of cancer cells (Lue et al.,
2015).
The relationship between MDPs and age-related diseases is far from complete, though. As
technology continues to evolve and disease-specific cohorts continue to grow, it is possible
additional MDPs will be discovered within the context of age-related disease.

12
CHAPTER 2: THE UTILITY OF HUMAN MITOCHONDRIAL DNA VARIATION
DATA FOR GENOME WIDE ASSOCIATION STUDIES
2.1 Abstract
Mitochondrial genome-wide association studies identify mitochondrial single nucleotide
polymorphisms (mtSNPs) that associate with disease or disease-related phenotypes. Most
mitochondrial and nuclear genome-wide association studies adjust for genetic ancestry by
including principal components derived from nuclear DNA, but not from mitochondrial DNA, as
covariates in statistical regression analyses. Furthermore, there is no standard when controlling for
genetic ancestry during mitochondrial and nuclear genetic interaction association scans, especially
across ethnicities with substantial mitochondrial genetic heterogeneity. The purpose of this study
is to (1) compare the degree of ethnic variation captured by principal components calculated from
microarray-defined nuclear and mitochondrial DNA and (2) assess the utility of mitochondrial
principal components for association studies. Analytic techniques used in this study include a
principal component analysis for genetic ancestry, decision-tree classification for self-reported
ethnicity, and linear regression for association tests. Data from the Health and Retirement Study,
which includes self-reported White, Black, and Hispanic Americans, was used for all analyses. We
report that (1) mitochondrial principal component analysis (PCA) captures ethnic variation to a
similar or slightly greater degree than nuclear PCA in Blacks and Hispanics, (2) nuclear and
mitochondrial DNA classify self-reported ethnicity to a high degree but with a similar level of
error, and 3) mitochondrial principal components can be used as covariates to adjust for population
stratification in association studies with complex traits, as demonstrated by our analysis of
height—a phenotype with a high heritability. Overall, genetic association studies might reveal true
13
and robust mtSNP associations when including mitochondrial principal components as regression
covariates.

14
2.2 Introduction
Mitochondrial genome wide association studies (MiWAS) are used to identify
mitochondrial single nucleotide polymorphisms (mtSNPs) that associate with disease or disease-
related phenotypes. A human mitochondrion has several copies of a condensed circular genome
that encodes 13 large proteins, 22 tRNAs, 2 rRNAs, and many peptides (e.g., humanin, MOTS-c,
and SHLPs) (Hashimoto, Ito, et al., 2001; Ikonen et al., 2003; S. J. Kim et al., 2016; Lee et al.,
2015; Mercer et al., 2011; Yen, Lee, Mehta, & Cohen, 2013). Genetic variation in these genes
could alter mitochondrial function, cell biology, and increase risk for certain diseases in specific
ethnicities, as mitochondria DNA (mtDNA) reflects historical human migration patterns.
Approximately 100,000 years ago, mitochondria genetic variation began to rapidly disseminate
across the globe. For instance, the estimated age of American-defined haplogroups of A, B, C, and
D are just 25,000-50,000 years old, whereas the African-originating haplogroups of L2 and L3 are
approximately 70,000-80,000 years old (Malhi et al., 2002). Given the rapid pace of mitochondrial
genetic variation, it is plausible that mtSNPs help explain health disparities in ethnicities and are
therapeutic targets for disease care and prevention.
Several mtSNPs have been associated with metabolic disease and neurodegenerative
disease risk. For example, Kraja et al. reported two mtSNPs significantly associated with metabolic
outcomes in one of the largest published MiWAS, consisting of ~170,000 individuals in 45 cohorts
(Kraja et al., 2019). Our group identified a mtSNP in the humanin-coding region that associates
with cognitive impairment and lower circulating humanin levels predominantly in Black
Americans. We also showed that administering humanin in vivo improved cognition and attenuated
neuroinflammation in aging mice (K. Yen et al., 2018). Additional associations between mtSNPs
and Parkinson’s disease, Alzheimer’s disease, diabetes, and other chronic diseases have also been
15
noted but with and without ample validation (Fang et al., 2018; Hudson et al., 2013; Lakatos et al.,
2010; Xiao et al., 2017). Hence, investing in MiWAS and related association techniques could be
an invaluable tool to identify ethnic-specific mtSNPs that increase risk for disease.
MiWAS and GWAS, however, are prone to confounding associations in part because of
population substructures embedded within the study samples used, particularly if drawn from
multi-ethnic or admixed groups (Price, Zaitlen, Reich, & Patterson, 2010). To statistically adjust
for differences in population substructure, principal components are calculated as eigenvalues
from nuclear DNA (nucDNA) principal component analysis (PCA) and included in regression
analyses as covariates. PCA is a statistical method whereby the number of variables that
characterize variation in the data, in this case the number of SNPs representing variation in
ancestral subgroups, can be reduced to a smaller number of values, or principal components that
similarly represent the variation. But for MiWAS, groups have statistically adjusted for genetic
ancestry in the regression models by using varying numbers of these principal components
calculated from either nucDNA or mtDNA. There is also no standard for controlling genetic
ancestry when addressing mitochondrial and nuclear genetic interactions, especially across
ethnicities with heterogeneous mitochondrial genetic ancestry. A recent analysis showed that
mtDNA principal components recapitulated mitochondrial haplogroups and outperformed
haplogroup and nucDNA PCA when adjusting for genetic ancestry in simulated phenotype/mtSNP
analyses (Biffi et al., 2010). These observations provide rationale to evaluate how well mtPCA
recapitulates self-reported ethnicity compared to nucPCA, which has yet to be done in such a large,
nationally representative, longitudinal, and multi-ethnic cohort. Comparing the utility of nucPCA
and mtPCA will inform researchers who use PCA to adjust for genetic ancestry in genetic
association studies.
16
The purpose of analyses presented here is three-fold: use a large cohort with a high-level
of admixture to 1) assess whether White, Black, and Hispanic individuals can be grouped into
ancestral clusters using principal components derived from array-based mtSNPs, 2) compare how
nucDNA and mtDNA represent ethnic variation, and 3) examine effects of mtPCA on height – a
phenotype with a high heritability – in White, Black, and Hispanic Americans (Weedon &
Frayling, 2008).

17
2.3 Results
Inter-ethnic analysis of mitochondrial genetic reduction techniques
The self-reported race of HRS is as followed: Self-reported Whites made up most of the
sample (70.2%), followed by Blacks (15.9%), Hispanics (11.2%), and Other (2.7%). Nuclear PCA
revealed that the first three principal components captured the highest amount of variance and,
when these three were plotted, reflected an expected pattern of ancestry. As shown in Figure 2.1
in the left-hand plot, self-reported Whites, Blacks, and Hispanics clustered in the lower left corner,
lower right, and along the left plane, respectively, with those in the Other group being dispersed
across clusters. Variance explained by the first five nuclear principal components totaled 94.0
percent.
Mitochondrial PCA revealed an expected pattern of ancestry when plotting the first three
principal components. Also shown in Figure 2.1, in the right-hand plot, self-reported Whites were
grouped in separate clusters on the top right and lower right; Blacks grouped into clusters on the
left midline; Hispanics grouped adjacent to the top right cluster of Whites; and those in the Other
group were again dispersed across clusters. Variance explained by the first five mitochondrial
principal components totaled 58.7 percent.

Intra-ethnic analysis of mitochondrial genetic reduction techniques
The analysis of intra-ethnic mtPCA revealed population substructures. As seen in Figure
2.2, several sparse substructures were identified by conducting mtPCA within just Whites, Blacks,
and Hispanics (Figure 2.2). These data suggest that mtPCA captures genetic variation even within
White, Black, and Hispanic subgroups, which is informative for researchers attempting to examine
18
the effect of mtSNPs during mitochondrial gene association studies and nuclear/mitochondrial
interaction genetic association studies.
The amount of variation captured by nucPCA and mtPCA was examined by self-reported
ethnicity. While inter-ethnic nucPCA captured much more variance in fewer components than
mtPCA, intra-ethnic mtPCA captured similar to slightly more variance compared to nucPCA
(Figure 2.3). Slightly greater variation was captured within the first 10 mitochondrial principal
components for Hispanics (nucPC1:10 = 71.6%; mtPC1:10 = 74.8%) and Blacks (nucPC1:10 =
67.8%; mtPC1:10 = 72.7%), but not in Whites (nucPC1:10 = 80.0%; mtPC1:10 = 71.1%).

Machine learning-based classification of self-reported ethnicity using mitochondrial and nuclear
genetic derived principal components to assess self-reported error rate
We assessed how well nuclear and mitochondrial principal components classified broader
ethnic sub-groups when defined by self-report. Using optimal decision tree algorithms derived
from a 30% training sample and then separately implemented on the remaining 70% of the data,
we found that nuclear and mitochondrial principal components comparably classified individuals
into ethnic sub-groups to a high degree: at a 94.9 percent and 92.0 percent rate, respectively.
Combining both nuclear and mitochondrial principal components increased statistical
classification accuracy into self-reported ethnic subgroups to 96.8 percent. This analysis is notable
because the nuclear and mitochondrial misclassification error suggests controlling for genetic
ancestry within self-reported ethnic analyses or assigning individuals into genetically homogenous
groups for analyses is necessary. Cutoffs and nodes are illustrated in Figure 2.4.

Heritable phenotype as a function of mitochondrial and nuclear genetic derived principal
19
To evaluate the utility of mitochondrial principal components as covariates to adjust for
population stratification in association studies, we tested its association with height (Figure 2.5),
which is highly heritable and strongly linked to ancestry. In these analyses, 18 of 20 mitochondrial
principal components derived from complete ethnic PCA significantly predicted height. This
supports the strong association between height and ancestry markers. In intra-ethnic analysis, in
which we used mitochondrial principal components derived from separately conducted sub-ethnic
PCA, three components were significant among Hispanics, one component was significant among
Blacks, and two components were significant among Whites. These analyses suggest that
mitochondrial genetic dimension reduction strategies could be useful for identifying mtSNPs that
associate with phenotypes in mitochondrial-specific analyses such as MiWAS.

20
2.4 Discussion
Our analyses demonstrated the utility of mtPCA for mitochondrial and nuclear genetic
association studies. First, we showed genetically admixed substructures from mtDNA in all
ethnicities in HRS. Second, we illustrated that the amount of variance captured by mitochondrial
principal components in Hispanics and Blacks is similar to slightly greater than that captured by
nuclear principal components, whereas nuclear principal components captured substantially more
variance in combined ethnic analysis and in Whites. Third, using mitochondrial and nuclear
principal components to train a decision tree for self-reported ethnicity classification showed high
statistical accuracy yet similar misclassification error between mitochondrial and nuclear analyses.
This misclassification rate suggests that conducting MiWAS by ethnic-specific stratification
without adjusting for genetic ancestry might not be a sufficient way to control for genetic
admixture. Hence, we showed that factoring in principal components during stratified analysis can
provide an analytic approach to further address the more complex admixture. Our analysis shows
that mitochondrial principal components associated with a high heritability phenotype, height,
when evaluated across ethnicities and intra-ethnically.
One novel aspect of our analyses is that mtSNPS derived from an array capture within
ethnic variation, which could be critical when designing analytic strategies to minimize
confounding due to admixture. In the absence of nuclear DNA data during mitochondrial gene
association studies (e.g., targeted whole mitochondria DNA sequencing), controlling for genetic
ancestry using mitochondrial principal components could reduce type one error and provides a
solution for analyses lacking nuclear DNA.
As nationally representative cohorts continue to grow larger, it is likely that research
groups will attempt to identify the effects of mtSNPs on a variety of phenotypes. Based on previous
21
publications, groups might design their analytic strategies by assigning terms to mitochondrial
haplogroups or single mtSNPs while controlling for genetic ancestry. The former is limited by
reference group classification and the latter is limited by no standard method to control for genetic
ancestry. Notably, Biffi et al. examined mtSNPs derived from commercially based arrays—similar
to that used by HRS— and showed that mitochondrial haplogroup analysis was inferior to mtPCA
for discovery of true associations and nucPCA had little effect on mitochondrial association testing
(Biffi et al., 2010). Since prior groups who conducted MiWAS have controlled for genetic ancestry
using nucPCA, it is possible that the loss of degrees of freedom from the addition of unnecessary
nuclear principal components suppressed mitochondrial genetic associations and/or limited
estimates of mitochondrial genotypes.
Future mitochondrial gene studies might reveal true and robust mtSNP associations by
controlling for mitochondrial genetic principal components. Moreover, it is plausible that there is
significant interaction between nuclear and mitochondrial SNPs, and that controlling for
mitochondrial genetic ancestry to identify such nuclear and mitochondrial genetic associations
might be an important consideration when defining the analytic strategy. However, subpopulation
genetic architecture will vary from cohort-to-cohort due to SNP-based array and sample variation.
Therefore, before conducting genetic association studies, comparing nuclear and mitochondrial
genetic substructures by ethnicity could guide analytic plans.

22
2.5 Methods
Data
Data from the Health and Retirement Study (HRS), an on-going nationally representative
study of United States adults over 50 years-old, was used for all analyses. The goal of HRS is to
track changes in aging-related outcomes over time. HRS has genotyped nearly 16,000 individuals
using either the Illumina HumanOmni2.5-4v1 and HumanOmni2.5-8v1 arrays for samples
collected in 2006, 2008, and 2010. Genotyping was performed by the NIH Center for Inherited
Disease Research (CIDR, X01HG005770-01) with standard quality control procedures
implemented by the University of Washington Genetic Coordinating Center (details available here:
http://hrsonline.isr.umich.edu/sitedocs/genetics/HRS2_qc_report_SEPT2013.pdf?_ga=2.901113
23.1994933777.1551124734-1821727996.1513803556) (Laurie et al., 2010). The total number of
SNPs retained were those that overlapped across arrays and passed quality control standards,
yielding 2,315,518 nucSNPs and 90 mtSNPs

Principal component analysis on mtSNPs and nucSNPs
Principal component analysis was conducted separately using mtSNPs and nucSNPs in
ethnic-stratified and ethnic-combined analysis (i.e., PCA for combined ethnicities and PCA
exclusively for White, Black, and Hispanic individuals). For nucSNPs (coded 0, 1, 2), the PLINK
2.0 --pca command was used to extract principal components. In this process, nucSNPs are used
to calculate eigenvectors using a variance-standardized genetic relationship matrix between
individuals, which is similar to that implemented in EIGENSTRAT (Purcell et al., 2007). No
clumping by linkage disequilibrium was conducted prior to nucPCA because there is no way to
ensure compatibility in pruning processes across both nucSNPs and mtSNPs. For binary coded
23
mtSNPs (coded 0, 1), the prcomp function in R was used to generate principal components. The
prcomp function uses singular value-decomposition of the data matrix to provide eigenvectors that
are the closest approximation of the matrix using a minimum number of values (Abraham &
Inouye, 2014). Visualizations of PCA plots were generated by standardizing components to a mean
of zero with a standard deviation of 1 and by using the scatterplot3D and ggplot2 package in R
(version 3.5.1).

Machine learning decision-tree classification of self-reported ethnicity using nucPCA and mtPCA
The caret R package contains a group of functions to create predictive models, and was
used to generate a cross-validated (kfold = 10; repeat = 5) decision-tree training algorithm on 30
percent of the data (n=4,584). The optimal model was selected using the largest accuracy value
derived from the train function (rpart method and a tune length of 10). The optimal model
subsequently predicted self-reported ethnicity on the remaining 70 percent of the data using the
the predict function. Plots were generated using the prp function of the rpart.plot R package.

Effect of mitochondrial principal components on the heritable phenotype height
Effects of mtSNP principal components on height was estimated by constructing
multivariable linear regression models separately for each ethnic group (White, Black, and
Hispanic Americans) and in a combined ethnicity model using the lm function in R. The dependent
variable was height (in centimeters) and the predictors included a total of 20 principal components,
biological sex, and centered age. The purpose of these analyses was to 1) understand whether
reducing mitochondrial genetic variation with principal components could explain the variation in
24
height within and across ethnic groups and 2) serve as proof-of-concept for using mtSNPs to
characterize genetic ancestry in association studies.
25

Figure 2.1. Multi-ethnic principal component analysis on nuclear and mitochondrial single
nucleotide polymorphisms. Each data point represents one individual. Colors are coded by
self-reported ethnicity.

26

Figure 2.2. Intra-ethnic principal component analysis on mitochondrial SNPs. Red indicates
Whites; Blue indicates Blacks; and Green indicates Hispanics.

27

Figure 2.3. Comparing amount of variance captured by nuclear and mitochondrial principal
components
a Across all ethnicities; b Hispanic comparison; c White comparison; d Black comparison.

28

Figure 2.4. Comparison of nuclear, mitochondrial, and combined nuclear/mitochondrial
PCA for statistically classifying individuals into broader ethnic sub-groups.
a nucPCA shows a statistical classification accuracy rate of 94.9%; b mtPCA shows a statistical
classification accuracy rate of 92.0%; c Combined nuclear and mtPCA shows a statistical
classification accuracy rate of 96.8%.

29

Figure 2.5. Effects of 20 mitochondrial principal components on height (centimeters) in inter-
ethnic and intra-ethnic samples.
a Combined analysis; b Hispanic analysis; c Black analysis; d White analysis. X axes are the
coefficient estimate for mitochondrial principal component number (Y axes).

30
CHAPTER 3: MITOCHONDRIAL DNA VARIATION IN ALZHEIMER’S DISEASE
REVEALS A UNIQUE MICROPROTEIN CALLED SHMOOSE
3.1 Abstract
Mitochondrial DNA variants have previously associated with disease, but the underlying
mechanisms have been largely elusive. Here, we report that mitochondrial SNP rs2853499
associated with a variety of neurobiological phenotypes including Alzheimer’s disease (AD),
neuroimaging modalities, and transcriptomic signatures. We followed up on these significant
rs2853499 associations by mapping the variant to a novel microprotein-encoding gene called
SHMOOSE. Indeed, we detected SHMOOSE in mitochondria by using mass spectrometry – the
first unique mass spectrometry-based detection of a mitochondrial-encoded microprotein.
Furthermore, cerebrospinal fluid (CSF) SHMOOSE levels in humans correlated with age, CSF
tau, and brain white matter volume. Functional experiments revealed SHMOOSE localizes to
mitochondria and binds the inner mitochondrial membrane protein mitofilin. SHMOOSE
administration to cells boosted mitochondrial oxygen consumption and enriched mitochondrial
gene expression in neuronal cells and mice. Altogether, this multi-omics discovery of SHMOOSE
has vast implications for neurobiology, Alzheimer’s disease, gene annotation, and biomarker
development.
31
3.2 Introduction
Recent omics have revealed novel functional genomic elements in neurobiology and
Alzheimer’s disease (AD), but two components have yet to be rigorously examined: microproteins
and mitochondrial DNA variation. Microproteins are biologically active peptides encoded by small
open reading frames (sORFs). These peptides have been missed for decades due to computational
power and biochemical limitations. Yet today, high resolution genomics and proteomics have
revealed thousands of uncharacterized microproteins (Martinez et al., 2020; Saghatelian & Couso,
2015).
Microproteins represent an enormous opportunity to understand neurobiology. Several
mitochondrial-encoded microproteins have been studied for the past twenty years. One such
microprotein is humanin, a 24 amino-acid peptide that was cloned out of the occipital lobe of an
Alzheimer’s disease (AD) patient. Since its discovery, humanin has been found to attenuate AD
pathology in part through its trimeric receptor signaling and amyloid beta toxicity protection (F.
Guo et al., 2010; Hashimoto, Niikura, et al., 2001; Ikonen et al., 2003; S. J. Kim et al., 2016;
Tajima et al., 2005; Tsukamoto et al., 2003). Recently, Yen et al. reported that cognitive age and
circulating human levels associated with a single nucleotide polymorphism (SNP) within the
humanin sORF (Kelvin Yen et al., 2018), suggesting other mitochondrial SNPs might influence
uncharacterized microproteins.
However, mitochondrial SNP association studies have yielded mixed results. One
explanation is genetic cohort-specific mitochondrial SNP frequency, as was the case in the
Alzheimer’s Disease Neuroimaging Initiative Cohort (ADNI) and Health, Aging, and Body
Composition Cohort. In ADNI, mitochondrial haplogroup U associated with AD and occurred in
over 20% of the cohort, whereas haplogroup U frequency in the Health, Aging and Body
32
Composition cohort occurred in 13% of the cohort and did not associate with dementia (Lakatos
et al., 2010; Tranah et al., 2012). Further complicating mitochondrial SNP association studies is
no gold standard analytic method. Various statistical methods (i.e., reducing mitochondrial SNPs
to haplogroups or principal components in addition to testing individual SNPs) have been used to
estimate effects of mitochondrial SNPs (Brendan Miller et al., 2019; B. Miller et al., 2019; Brendan
Miller, Mina Torres, et al., 2020; Zempo et al., 2021). In addition, most of these association studies
were tested on case-control outcomes rather than continuous outcomes that permit greater
statistical power. Overall, the estimated effects of mitochondrial SNPs vary by analytic methods,
cohort composition, statistical power, and outcome of interest. Regardless, mitochondrial SNP
association studies are typically not interpreted in the context of mitochondrial-encoded
microproteins.
Here, we proposed mitochondrial DNA variants within microprotein-encoding-sORFs will
associate with neurobiological phenotypes. We hypothesized that this genetic variation would
reveal biological relevance of a novel microprotein.
33
3.3 Results
Mitochondrial rs2853499 occurs within a previously unannotated microprotein and associates
with AD and neuroimaging phenotypes
First, we tested the hypothesis that mitochondrial SNPs within sORFs will associate with
AD. Previously, Lakatos et al. reported that mitochondrial haplogroup U associated with AD in
the ADNI1 cohort (n = ~300)(Lakatos et al., 2010). Since then, ADNI has expanded its cohort size
with multiple new phases, which we included in our mitochondrial-wide association study
(MiWAS). By assessing ADNI 1, GO, 2, and 3 (n = ~800), we confirmed haplogroup U SNPs (i.e.,
base pair positions 11467, 12308, and 12372) associated with AD (Figure 3.1A). These three
haplogroup SNPs do not change the amino acid sequences of the mt-ND4 and mt-ND5 proteins.
However, mtSNP at base pair position 12372 (rs2853499) does change the amino acid sequence
of a microprotein encoded by a sORF that we call SHMOOSE (Small Human Mitochondrial ORF
Over SErine tRNA). Specifically, rs2853499 (henceforth referred to as SHMOOSE.D47N)
changes the 47
th
amino acid from glutamine to aspartic acid (Figure 3.1D). In ADNI,
SHMOOSE.D47N carriers presented an odds ratio of 1.56 (case frequency: 22.9%; control
frequency: 15.7%; 95% CI: 1.06-2.30; permutation empirical p value < 0.03).
Furthermore, we examined the effects of SHMOOSE.D47N in three additional cohorts:
Religious Orders Study (ROS) and Memory and Aging Project (MAP), Late-Onset Alzheimer's
Disease (LOAD), and NIA Alzheimer Disease Centers (ADC1 and ADC2). SHMOOSE.D47N
carriers in ROSMAP, LOAD, and ADC1/2 presented odds ratios, respectively, of 1.55 (case
frequency: 25.3%; control frequency: 17.6), 1.04 (case frequency: 24.7%; control frequency:
23.0%) and 1.13 (case frequency: 24.0%; control frequency: 23.3%). We analyzed these cohorts
with consideration to cohort-specific allele frequency differences. Whereas we did not observe
34
significant mitochondrial genetic heterogeneity for SHMOOSE.D47N carriers in ADNI and
ROSMAP, we did observe mitochondrial genetic heterogeneity for SHMOOSE.D47N in LOAD
and ADC1/2, which we corrected in our statistical models. Altogether, random effects meta-
analysis estimated an odds ratio of 1.30 for SHMOOSE.D47N (95% CI: 1.06-1.59; p value < 0.005;
Figure 3.1B). In addition, we estimated the effect of SHMOOSE.D47N on cognitive decline in the
Health and Retirement Study (HRS), a population-based study of US adults aged 50 years or older
(n = ~15,000) (Crimmins, Kim, Langa, & Weir, 2011), noting that alternative allele carriers had
faster cognitive decline over time.
Since GWAS/MiWAS are prone to spurious associations, we carried out a phenome wide
association study (PheWAS) that included approximately 4,000 neuroimaging modalities on a
large sample of ~18,300 European-ancestral individuals. A significant advantage of PheWAS is
the replication of SHMOOSE.D47N across related neurobiological phenotypes, which is both
targeted and statistically robust due to abundant power. In our PheWAS, SHMOOSE.D47N
(frequency: 25.5%) significantly associated with cortical thickness, volume, pial surface area, WM
surface Jacobian, and GM/WM contrast in several paralimbic regions, including the
parahippocampal gyri, the entorhinal cortex (EC), the anterior cingulate cortex (ACC), the
posterior cingulate cortex (PCC), and the temporal pole (TPO) (clusterwise, RFT-corrected p value
< 0.05). Over age-span, SHMOOSE.D47N carriers exhibited accelerated thinning of the
parahippocampus (Figure 3.1C), EC, and PCC. That is, the crossed age trajectories of
SHMOOSE.D47N showed inversed effects at younger and older ages. For instance, the
SHMOOSE reference allele associated with smaller brain structural measures at middle (45-65
years) and/or young-old (65-75 years) ages, whereas the alternative allele (i.e., SHMOOSE.D47N)
associated with structural loss at old ages (>75 years). In addition to the predominant results in the
35
paralimbic areas, at a lenient threshold of uncorrected p value < 0.05, we observed distributed
trends of SHMOOSE.D47N effects in the language centers (superior temporal and inferior frontal
gyri), dorsolateral and medial prefrontal cortex, central motor, and occipital visual cortices.
Separately, we observed similar trends in the ADNI PheWAS, which was conducted on the same
sample used for our ADNI permutation MiWAS. As we observed in UK Biobank,
SHMOOSE.D47N significantly associated with limbic regions such as the medial temporal cortex
and posterior cingulate cortex at a lenient threshold of uncorrected p value < 0.05.
Since rs2853499 associated with neuroimaging outcomes and AD – and that the variant
mutates the SHMOOSE amino acid sequence (Figure 1D) – we sought to detect endogenous
SHMOOSE. Therefore, we started by assessing the correlation between SHMOOSE RNA counts
and nuclear-encoded gene counts on 69 non-demented post-mortem brains temporal cortices.
These analyses revealed the mitochondria cellular compartment co-expressed with SHMOOSE
(Figure 3.1E). Indeed, using a custom antibody against the c-terminus of SHMOOSE, we detected
SHMOOSE via western blot from neuronal mitochondria and nuclei fractions at the predicted
~6kDa molecular weight (Figure 3.F). Yet in rho zero cells (i.e., cells without mtDNA), we did
not detect SHMOOSE, further confirming SHMOOSE is derived from mitochondrial DNA.
Notably, by precipitating SHMOOSE from neural cell mitochondria, we detected two unique
SHMOOSE in mass spectrometry, one towards the N-terminus and the other at the C-terminus
(Figure 3.1G).
Altogether, we targeted and detected the novel microprotein SHMOOSE after identifying
a genetic variant within its sORF that associated with AD and brain structure. To the best of our
knowledge, SHMOOSE is the first mitochondrial-derived microprotein detected using mass
spectrometry.
36

SHMOOSE levels in cerebrospinal fluid correlate with age, tau, and brain white matter
microstructure
In the temporal cortex of AD brains, SHMOOSE RNA was ~15% greater compared to
controls (AD n = 82; control n = 78; Figure 3.2A). But to assess microprotein levels directly, we
developed a SHMOOSE enzyme-linked immunosorbent assay (ELISA) with sensitivity ranging
from 100-250,000 pg/ml. In 79 cerebrospinal samples (CSF) of non-demented individuals from
the USC AD Research Center, we analyzed the effects of age, CSF amyloid beta, CSF total tau,
CSF phosphorylated tau at residue number 181 (p tau 181), and white matter microstructure
(measured by diffusion tensor imaging fractional anisotropy) on SHMOOSE levels. CSF
SHMOOSE levels positively and significantly correlated with age, CSF total tau, and CSF p tau
181 (Figure 3.2B). To determine if the effect of p tau 181 on SHMOOSE levels was mediated
through total tau, we performed a mediation analysis in which total tau was considered the
mediator. The indirect effect of this mediation was not considered statistically significant, and the
direct effect of p tau 181 on SHMOOSE tau was slightly attenuated when controlling for total tau
(p value = 0.060). Together, the direct and indirect effect of p tau 181 on SHMOOSE was
significant (p value < 0.01). Therefore, p tau 181 appears to be specifically driving the relationship
with SHMOOSE and is not mediated by total tau. Separately, we did not observe significant
correlations between SHMOOSE and CSF levels of amyloid beta 42. Finally, we found that
individuals with higher CSF SHMOOSE levels also had lower DTI fractional anisotropy FA in the
body of the corpus callosum and bilateral superior corona radiata (Figure 3.2C), even after
controlling for age, biological sex, and cognitive status (CDR test score).

37
SHMOOSE associates with mitochondrial and ribosomal gene expression changes
Given that we observed neuroimaging modality differences by SHMOOSE genotype, we
further hypothesized brain gene expression will differ by SHMOOSE genotype (i.e.,
SHMOOSE.D47N). Therefore, we analyzed RNA-Seq data derived from 69 post-mortem brain
temporal cortexes (Mayo Clinic; n = 14 alternative allele and n = 55 reference allele). We observed
that SHMOOSE.D47N associated with 2,122 differentially expressed genes in the brain temporal
cortex under a p adjusted value of 0.05. Remarkably, principal components derived from the gene
count matrix revealed clustering SHMOOSE reference allele gene expression, while the signature
for SHMOOSE.D47N carriers drifted further from the reference allele cluster. By categorizing
samples based on the median value of the second principal component, SHMOOSE.D47N carrier
exhibited significant global gene expression deviation (p value < 0.05; Figure 3.3A). Considering
that post-mortem brain gene expression is influenced by several factors (i.e., environment, etc.)(Ng
et al., 2019), we consider the significant effects of SHMOOSE.D47N extremely noteworthy. As a
contrast, we noticed a similar trend for APOE4 carriers, in which APOE4 carriers trended away
from the population norm, although the trend was not statistically significant. These statistically
differentially expressed genes by SHMOOSE.D47N carriers enriched GO cellular compartment
terms for mitochondria and ribosomes.
Next, to determine whether the effects for SHMOOSE.D47N on gene expression can be
recapitulated in vitro, we conducted differential transcriptomics after administering SHMOOSE
or SHMOOSE.D47N to neural cells for 24 hours. SHMOOSE.D47N-treated cells induced
differential expression of 1,400 genes under a p adjusted value of 0.2. Indeed, these significant
genes enriched mitochondrial and ribosome cellular compartments, like what was observed by
38
SHMOOSE genotype in the 69 post-mortem human brains. “Mitochondrial inner membrane” was
the top enriched GO cellular compartment term (Figure 3.3B).
We also injected SHMOOSE intraperitoneally (IP) to 12-week-old C57BL/6J mice fed a
high fat diet (i.e., a mild metabolic perturbation). At the conclusion of the two-week study, we
harvested brain and liver for RNA-Seq. As we observed for SHMOOSE.D47N carriers and in in
vitro experiments, 367 differentially expressed genes under a p adjusted value of 0.05 enriched
mitochondrial and ribosomal terms in the liver (Figure 3.3C). However, while we observed global
gene expression changes as illustrated by PCA in the mouse cortex and hypothalamus, most
enriched terms included central nervous system-related terms, not ribosomes nor mitochondrial
specifics. We also did not observe robust gene expression differences in the hippocampus. Mice
treated with SHMOOSE did not display toxic effects nor behavior changes and showed attenuated
weight gain and rise in the liver enzymes AST and ALT (without change in food intake), as
compared to control treated mice.

SHMOOSE modifies cellular metabolic activity and protects against amyloid beta toxicity
Since SHMOOSE associated with differential mitochondrial gene expression (Figure
3.4A-C), we hypothesized SHMOOSE would induce metabolic changes in vitro. Indeed, when
SHMOOSE was exogenously administered to cells, the peptide localized prominently to
mitochondria (Figure 3.4A). In a dose-dependent response from 1uM to 10uM, both SHMOOSE
and SHMOOSE.D47N increased neural cell metabolic activity by 10% and 20%, respectively
(Figure 3.4B). During mitochondrial stress, in which maximum proton flow was permitted through
the inner mitochondrial membrane following oligomycin and FCCP administration, both
SHMOOSE and SHMOOSE.D47N boosted mitochondrial spare capacity (Figure 3.4C),
39
suggesting general effects for both forms of SHMOOSE on mitochondrial biology. However,
compared to SHMOOSE.D47N, wild type SHMOOSE significantly increased basal oxygen
consumption rate by approximately 20% (Figure 3.4D).
Additionally, in neurons derived from iPSCs with both APP and PSEN1 mutations – two
familial AD mutations – SHMOOSE RNA expression was three-fold higher than neurons derived
with just one mutation, suggesting a role for SHMOOSE in amyloid beta biology (Figure 3.4E).
Likewise, in neuronal cells stressed with oligomerized amyloid beta, SHMOOSE administration
protected against cell death, but SHMOOSE.D47N did not similarly protect cells (Figure 3.4F).

SHMOOSE interacts with the inner mitochondrial membrane protein mitofilin
We searched for SHMOOSE protein interacting partners by conducting co-
immunoprecipitation assays. Ultimately, we identified mitofilin as a protein interacting partner
with SHMOOSE. Mitofilin was targeted based on proteomics analysis of neuronal lysates that
were spiked with SHMOOSE followed by SHMOOSE antibody-based immunoprecipitation. Mass
spectrometry-based analysis of these lysates suggested SHMOOSE bound 98 proteins; however,
during protein quantification filtering and indexing by p value and fold change, we considered
mitofilin a top SHMOOSE binding protein candidate (Figure 5A). Indeed, after administering
SHMOOSE to neural cells, we validated the SHMOOSE-mitofilin interaction by reciprocal
immunoprecipitation followed by western blot (Figure 5B). In addition, we conducted reciprocal
dot-based immunoblots between recombinant SHMOOSE and mitofilin, confirming binding
between mitofilin and SHMOOSE as well as SHMOOSE.D47N (Figure 5C). Moreover, after
knocking down mitofilin using siRNA, we observed no effect of SHMOOSE on neural cell
metabolic activity, as measured by MTT assay (Figure 5D). We modeled the biophysical
40
interaction between SHMOOSE and mitofilin using HDOCK, a hybrid algorithm of template-
based and template-free docking (Figure 5E), noting that the predicted interacting between
SHMOOSE and mitofilin center on the c-terminus of mitofilin (residues 332-413) (Yan, Tao, He,
& Huang, 2020).

41
3.4 Discussion
We initially targeted SHMOOSE because a mitochondrial SNP within the SHMOOSE
sORF associated with AD, neuroimaging modalities, and brain gene expression in large
epidemiological cohorts. In four cohorts, individuals with SHMOOSE.D47N exhibited increased
risk for AD (OR: 1.30; Figure 1B). In neuroimaging based PheWAS, SHMOOSE.D47N carriers
had greater atrophy over age in medial temporal areas such as the parahippocampus, entorhinal
cortex, and anterior and parietal cingulate cortex. The medial temporal cortex and parietal
cingulate cortex are known to be vulnerable in Alzheimer’s disease (AD), and the pathological
atrophy likely appears years before clinical symptoms (Davatzikos, Xu, An, Fan, & Resnick, 2009;
Jack et al., 2010). We also examined brain neuroimaging in ADNI and observed similar effects in
the limbic area, but no results survived multiple correction, and we lacked the power that UK
Biobank (n = ~18,300) provided. Moreover, the crossed age trajectories of SHMOOSE.D47N
showed inversed effects at younger and older ages. The SHMOOSE reference allele associated
with smaller brain structural measures at middle (45-65 years) and young-old (65-75 years) ages,
whereas the alternative allele was associated with structural loss at old ages (>75 years). Such
contrary genetic effects in younger and older samples have been observed for several other genes
that were frequently associated with age-sensitive cognitive functions and neurodegeneration (e.g.,
brain-derived neurotrophic factor) (Egan et al., 2003; Harris et al., 2006; Pezawas et al., 2004;
Ventriglia et al., 2002; Zhao et al., 2019).
Given the genetic association between this SHMOOSE SNP and neurobiological
phenotypes (i.e., AD and neuroimaging modalities), we targeted SHMOOSE biochemically by
developing a polyclonal antibody against amino acid residues 32-58 of SHMOOSE. By
immunoprecipitating SHMOOSE from neuronal mitochondrial lysates, we detected two unique
42
fragments of SHMOOSE in mass spectrometry. Likewise, we detected SHMOOSE at the predicted
~6kDa via western blot in neuronal mitochondrial as well as nuclei, while we did not observe
SHMOOSE detection in cells void of mitochondrial DNA. Furthermore, we found CSF
SHMOOSE positively correlated with age, tau, and brain white matter. Since higher levels of CSF
tau have previously predicted AD (Grangeon et al., 2016), correlations between SHMOOSE and
tau suggest SHMOOSE could be involved in the progressive etiology of AD and might be poised
as a biomarker. Moreover, we observed that higher CSF SHMOOSE levels associated with DTI
FA in non-demented older adults. Various factors can contribute to lower DTI FA, but it may
reflect lower levels of myelination and is often associated with a disease state. Myelin maintenance
and repair is metabolically demanding and is particularly vulnerable to damage when energy
deficits exist, (Bartzokis, 2011), providing a possible link between a mitochondrial peptide and
white matter microstructure. Possibly the relationship between higher CSF SHMOOSE and lower
regional DTI FA indicates an incomplete compensation for metabolic or other stressors in the
brain.
We further assessed the SHMOOSE SNP (i.e., SHMOOSE.D47N) in human population
cohorts to infer biological mechanism. Remarkably, the SHMOOSE.D47N SNP alone
differentiated the human brain transcriptome, as the gene expression signature via PCA of post-
mortem brains with SHMOOSE.D47N drifted from SHMOOSE reference allele cluster. This was
surprising given the reported effects of environment, lifespan, etc. on human brain transcriptomics
(Ng et al., 2019). Likewise, in vitro, gene expression differences by SHMOOSE and
SHMOOSE.D47N enriched inner mitochondrial membrane and ribosomes compartments.
Furthermore, in our in vivo studies involving IP injections of SHMOOSE, we also observed
ribosomal and mitochondrial inner membrane gene expression changes beyond the brain in the
43
liver, suggesting that SHMOOSE could act on non-neural systems. These mice treated with
SHMOOSE experienced attenuated weight gain during high-fat diet ad lib with mild reductions in
liver enzymes ALT and AST.
In all our transcriptomics studies, we observed a common theme for mitochondrial inner
membrane enrichment, which we consider noteworthy because SHMOOSE bound the inner
mitochondrial membrane mitofilin in multiple models. Mitofilin is a component of the MICOS
complex that regulates mitochondrial crista junctions and inner membrane organization (Feng,
Madungwe, & Bopassa, 2019; Gieffers, Korioth, Heimann, Ungermann, & Frey, 1997).
Separately, we observed transcription enrichment for ribosomal terms, which might be explained
through SHMOOSE-mitofilin interaction, as the Pathway Commons Protein-Protein Interactions
data set shows nearly 1800 interacting proteins to mitofilin, 137 of which are ribosomal proteins.
Although past MiWAS studies have examined the effects of mtSNPs on
neurodegeneration, few have followed up experimentally (Hudson et al., 2013; Lakatos et al.,
2010). One functional limitation of MiWAS is isolating the effects of individuals mtSNPs because
these SNPs define broader haplogroups (Malhi et al., 2002; McRae et al., 2008). As a result, we
cannot rule out that other SNPs within the SHMOOSE.D47N haplogroup (i.e., Haplogroup U) has
effects independent from SHMOOSE (e.g., effects on tRNA). Nevertheless, the only missense
effect of this SNP is to the SHMOOSE microprotein, and we used this multi-phenotype strategy
to identify a microprotein candidate for experimental validation (i.e., SHMOOSE).
Our data has several implications. First, separate from the discovery of SHMOOSE, we
showed mitochondrial DNA variants can associate with several neurobiological phenotypes that
can aid functional interpretation (i.e., disease classification, structural anatomy, and gene
expression). Second, we revealed mitochondrial DNA variants can be mapped to sORFs that
44
encode biologically functional microproteins. We identified SHMOOSE as the first biologically
active mitochondrial-encoded microprotein detected by mass spectrometry, immunoblot, and
ELISA. As large human cohorts with genetic data continue to add whole genome sequencing data,
it is foreseeable that this refined mtDNA resolution will yield additional microproteins. Moreover,
as proteomics technology improves, it is also conceivable more mitochondrial-encoded
microproteins will be detected (Brendan Miller, Su-Jeong Kim, et al., 2020). Third, the correlation
among CSF levels of SHMOOSE, CSF AD-related biomarkers (e.g., tau), and brain white matter
suggests SHMOOSE has potential as a biomarker. Finally, SHMOOSE appears to be another
microprotein that affects mitochondrial biology, as recent microprotein discoveries (e.g.,
mitoregulin, BRAWNIN, MIEF-MP1) have also noted profound effects on mitochondrial biology
(Rathore et al., 2018; Stein et al., 2018; S. Zhang et al., 2020).

45
3.5 Methods
Mitochondrial-wide association study (MiWAS) and mtSNP association analyses
Effects of mitochondrial genetic variants on probable AD in ADNI1, ADNI GO, ADNI2,
and ADNI3 were tested. We followed up on MiWAS results that were previously reported by
Lakatos et al on ADNI1 (Lakatos et al., 2010). In our analyses, mitochondrial genotypes and
diagnosis from ADNI1, ADNI GO, ADNI2, and ADNI3 were all merged for analysis. ADNI1
samples were genotyped using the Illumina 610-Quad BeadChip, and ADNI GO/2 samples were
genotyped using the Illumina HumanOmniExpress BeadChip, which does not contain mtSNPs.
Nevertheless, ADNI1/GO/2 samples were whole genome sequenced and include mitochondrial
genotypes. These whole mitochondrial genotypes were processed with stringent quality controls
and made available by Ridge et al. in variant call format (Ridge et al., 2018). ADNI3 samples were
genotyped using the Illumina Infinium Global Screening Array v2 (GSA2). Mitochondrial whole
genome sequencing data was converted to suitable format using PLINK (v1.9) and merged with
ADNI1 and ADNI3 in PLINK bed/bam/bim format. After merging genetic data, 138 mtSNPs
remained for a total of 448 clinical probable cases and 290 controls during MiWAS. The MiWAS
permutation model included a minor allele frequency threshold of 5% on individuals of European
descent noted by ADNI, leaving 29 mtSNPs qualified for permutation. We further assessed the
degree of mitochondrial genetic admixture by conducting a principal component analysis on
mtSNPs. These principal components were generated via singular value-decomposition of the
mtSNP matrix, outputting eigenvectors that approximates the matrix with a minimal number of
values (prcomp function in R), as portrayed elsewhere (Brendan Miller et al., 2019; B. Miller,
Haghani, Ailshire, & Arpawong, 2020; Brendan Miller, Mina Torres, et al., 2020; Z. Zhang &
Castello, 2017). The degree of mitochondrial genetic admixture was low and ideal for a
46
permutation approach, the same approach that Lakatos et al previously conducted (Lakatos et al.,
2010). Any mtSNP with an empirical p value under 0.05 was considered statistically significant.
A total of 957 permutations were conducted for the most significant mtSNP, which occurred in the
SHMOOSE sORF and became our microprotein candidate. Separately, we estimated the effect of
the SHMOOSE mtSNP in the Rush Alzheimer's Disease Center (RADC) comprised of Religious
Orders Study (ROS) and Memory and Aging Project (MAP), Late-Onset Alzheimer's Disease
(NIA-LOAD), and NIA Alzheimer Disease Centers (ADC1 and ADC2) cohorts using logistic
regression. ROSMAP samples were genotyped using whole genome sequencing, and
mitochondrial genetic variants were made available to qualified users in VCF format. LOAD
samples were genotyped using the Illumina 610-Quad BeadChip, and the ADC1 and ADC2
samples were genotyped using the Ilumina Human660W-Quad BeadChip. Mitochondrial genetic
admixture in ROSMAP (n = 281 cases and n = 233 controls), LOAD (n = 993 cases and n = 883
controls) and ADC1/2 (n = 2261 cases and n = 654 controls) were also assessed by implementing
mitochondrial principal component analysis (Supplemental Figure 4); we observed mitochondrial
genetic heterogeneity for SHMOOSE alternative allele carriers in LOAD and ADC1/2 and
therefore included the first three mitochondrial principal components in the logistic regression
model, with age and biological sex as additional covariates. For meta-analysis, age and biological
sex were included in the ADNI model. After the effects of the SHMOOSE mtSNP were estimated
in ROSMAP, LOAD, and ADC1/2, meta-analysis was conducted using a random effects model
approach from the metafor R package (Balduzzi, Rucker, & Schwarzer, 2019).

Neuroimaging phenome-wide association study (PheWAS)
This work was conducted using ADNI and the UK Biobank Resource
47
(https://www.ukbiobank.ac.uk) under approved project 25641. We used brain MRI imaging (UK
Biobank data-field: 110) from the 2018 August release of 22,392 participants
(http://biobank.ctsu.ox.ac.uk/crystal/label.cgi?id=110). Details of the MRI acquisition is described
in the UK Biobank Brain Imaging Documentation
(http://biobank.ctsu.ox.ac.uk/crystal/refer.cgi?id=1977) and in a protocol form
(http://biobank.ctsu.ox.ac.uk/crystal/refer.cgi?id=2367) (Alfaro-Almagro et al., 2018; K. L. Miller
et al., 2016). This study discarded 1,002 participants whose MRI scans did not pass manual quality
assessment, 45 participants due to data withdrawal or failed image processing, and 3,055
participants who did not have white British ancestry and/or did not pass the sample quality control
for the genetic data (https://biobank.ctsu.ox.ac.uk/crystal/label.cgi?id=100313), resulting a sample
of 18,330 individuals with age range from 45 to 81 years (mean age = 63.27 + 7.45 years), 8,729
males (47.62%), and 4,680 SHMOOSE mtSNP carriers (25.53%). All MR images were processed
using the FreeSurfer software package v6.0 (https://surfer.nmr.mgh.harvard.edu) to extract brain-
wide morphological measures. The FreeSurfer workflow includes motion correction and averaging
of volumetric T1-weighted images (Reuter, Rosas, & Fischl, 2010), removal of non-brain
tissue,(Segonne et al., 2004) automated Talairach transformation, brain volume segmentation
(Fischl et al., 2002; Fischl et al., 2004), intensity normalization (Sled, Zijdenbos, & Evans, 1998),
tessellation of the boundary between gray matter (GM) and white matter (WM), automated
topology correction (Segonne, Pacheco, & Fischl, 2007), and surface deformation following
intensity gradients to optimally place the GM/WM and GM/cerebrospinal fluid borders at the
location where the greatest shift in intensity defines the transition to the other tissue class (Fischl
& Dale, 2000). Each hemispheric GM and WM surface is composed of 163,842 vertices arranged
as 327,680 triangles. Once the surface models are complete, a number of deformable procedures
48
were performed for further data processing and analysis, including surface inflation, registration
to a spherical atlas using individual cortical folding patterns to match cortical geometry across
subjects (Fischl, Sereno, Tootell, & Dale, 1999), and finally creation of a variety of surface-based
brain morphological metrics. All the procedure for MRI processing were implemented on the
LONI pipeline system (http://pipeline.loni.usc.edu) for high-performance parallel computing (I.
Dinov et al., 2010; I. D. Dinov et al., 2009). This study included 9 vertex-wise brain morphological
measures: cortical thickness, volume, WM surface area, pial surface area, sulcal depth, WM
surface Jacobian, GM/WM contrast, mean curvature and Gaussian curvature. Detailed information
about these surface-based metrics is available at https://surfer.nmr.mgh.harvard.edu/fswiki/.
Briefly, cortical thickness values were calculated as the shortest distance between the gray and
white matter surfaces at each vertex. Vertex-wise volume is calculated by dividing each obliquely
truncated trilateral pyramid between the GM and WM surfaces into three tetrahedra. Vertex-wise
surface area measures on the pial and WM surfaces are estimated by assigning one third of the
area of each triangle to each vertex. Sulcal depth conveys information on how far removed a
particular vertex point on a surface is from a hypothetical mid-surface that exists between the gyri
and sulci. It gives an indication of linear distance and displacements: how deep and high are brain
folds. Surface Jacobian measures how much the surface is distorted to register to the spherical
atlas. GM/WM contrast presents the vertex-by-vertex percent contrast between white and gray
matter, where WM is sampled 1mm below the white surface, and GM is sampled 30% the thickness
into the cortex. Mean curvature is the average of the two principal curvatures at a vertex. The
Gaussian curvature is the product of the two principal curvatures at a vertex. Prior to statistical
analysis, these surface-based data were smoothed on the tessellated surfaces using a Gaussian
kernel with the full width half maximum of 20 mm to increase the signal-to-noise ratio and to
49
reduce the impact of mis-registration. All UK Biobank participants were genotyped using the
Affymetrix UK BiLEVE Axiom array (on an initial ~50,000 participants) and the Affymetrix UK
Biobank Axiom array (on the remaining ~450,000 participants) were genotyped using the
Affymetrix UK Biobank Axiom array. SHMOOSE genotype was extracted from the genotyping
array using the PLINK2.0 software. To capture population structure, the UKB team computed the
top 40 principal components (PCs) from the high-quality genotyping dataset (Bycroft et al., 2018).
Furthermore, to capture the population structure hidden in the mitochondrial genome, we also
computed mitochondrial PCs using a mitochondrial principal component analysis (B. Miller et al.,
2019). To test effects of the SHMOOSE mtSNP on age-related brain structural differences, we
assessed the interaction between the SHMOOSE mtSNP genotype and age by implementing linear
mixed-effects regression at each cortical surface vertex ! for a given morphological measure "
!

with the model: "
!
= !$%&'(&)%+ ,
"
-+ ,
#
./&+ ,
$
./&×-+&
!
, where - is the SHMOOSE
mtSNP genotype, ./& is individuals’ age in the scanner, & is the residual error, and the intercept
and , terms are the fixed effects. Sex, intracranial volume (ICV), nuclear and mitochondrial PCs
were added to the model as confounding variables. Statistical results at all vertices were corrected
for the family-wise error rate (FWER) across the brain surface using the random field theory (RFT)
method that adapts to spatial smoothness of the neuroimaging data (Worsley, Evans, Marrett, &
Neelin, 1992). All surface-based analyses were conducted using our Neuroimaging PheWAS
system, which is a cloud-computing platform for big-data, brainwide imaging association studies
(Zhao et al., 2020).

Cognitive decline analysis
50
In HRS, we assessed the effect of SHMOOSE genotype on longitudinal cognitive decline
over the aging process by implementing a mixed effects regression approach. HRS used the
HumanOmni2.5 array to directly genotype 256 mitochondrial SNPs. SHMOOSE genotype was
extracted using PLINK2.0. The validated HRS cognitive score represents episodic memory
learning, episodic memory retrieval, semantic fluency, and orientation (Crimmins et al., 2011).
The mixed effects model included fixed effect terms for biological sex, linear and quadratic age,
mitochondrial genetic ancestry, and SHMOOSE genotype for European-ancestral individuals.
Subject-specific random effects contained between-individual variation at the age of 65 in addition
to inter-individual variation in the rate of cognitive score change during aging (i.e., follow-up visits
every two years). The lme4 package in R was used to carry out the analysis. A total of 8,072
individuals were individually assessed with 45,465 total data points.

SHMOOSE protein structure prediction
RoseTTAFold was used to predict the microprotein structure of SHMOOSE and was
developed by Baker lab at the University of Washington. Full algorithm details have been
comprehensively detailed elsewhere (Baek et al., 2021). In comparison to Alphafold2,
RoseTTAFold achieved similar degree of accuracy for complex proteins. The wild-type version
of SHMOOSE and SHMOOSE.D47N were modeled, and output files were downloaded into PDB
format.

Co-expression analysis
We utilized transcriptome data generated by Mayo (Synapse ID: syn5550404). SHMOOSE
transcript count matrices were created from made-available bam files. This was done by
51
constructing a sORF database in GTF format and implementing the summarizeOverlaps function
of the GenomicAlignments package in R. Thereafter, normalized counts were used to conduct
correlation between SHMOOSE counts and all nuclear-encoded gene counts, corrected for
multiple hypotheses using a false discovery rate (FDR) of 0.05. Genes that statistically correlated
with SHMOOSE expression were tested for enrichment using the enrichGo function from the
clusterProfiler package, which returns enrichment of Gene Ontology (GO) categories after FDR
control. Data output from the enrichGo function were used to generate plots using ggplot2 in R.

Cells
SH-SY5Y cells used in the study were purchased from ATCC (CRL-2266). Cells were
grown in DMEM/F12 with 10%FBS at 37°C with 5% CO2 and split every 4-7 days depending on
confluency. In addition, for rho zero cells, SH-SY5Y cells were depleted of mitochondrial DNA
by adding 5 ug/ml ethidium bromide, 50 ug/ml uridine, and 1mM pyruvate for approximately two
months, as previously described [39]. For all experiments, cells were differentiated by addition of
10uM retinoic acid in DMEM/F12 with 1%FBS, and the media was changed once every 48 hours
for a total of two changes, as described previously [40]. When indicated, cells were treated with
chemically synthesized SHMOOSE, which was made by GenScript by solid-phase peptide
synthesis methods. Triflouracetic acid (TFA) was used to cleave synthesized peptide from resin.
After peptide synthesis, residual TFA was removed and the pH of reconstituted SHMOOSE was
neutral.

Subcellular fractionation
52
Cytosolic, nuclear, and mitochondrial fractions were prepared from cultured SH-SY5Y
cells. To extract nuclei, cells were washed in ice-cold DPBS and resuspended in fractionation
buffer containing 10 mM HEPES pH 7.6, 3 mM MgCl2, 10 mM KCl, 5% (v/v) glycerol, 1%
Triton-X100, and protease/phosphatase inhibitors for 15 minutes, followed by centrifugation for 5
minutes at 250 x g and 4°C. The resulting supernatant was further centrifuged once more at 18,000
x g for 10 minutes at 4°C to obtain a relatively pure cytoplasmic fraction. The original pellet prior
to the 18,000 x g centrifugation was washed in 10mM HEPES pH 7.6, 1.5mM MgCl2, 10 mM
KCl, and protease/phosphatase inhibitors and centrifuged at 250 x g and 4°C. The washed pellet
was then resuspended in nuclear extraction buffer containing 20 mM HEPES pH 7.6, 1.5mM
MgCl2, 420 mM NaCl, 25% (v/v) glycerol, 0.2 mM EDTA, and protease/phosphatase inhibitors,
followed by three sonication periods of 5 seconds (separated by 10 seconds) with 30% amplitude
on ice. The sonicated pellet was centrifuged at 18,000 x g for 10 minutes at 4°C to obtain a
relatively pure nuclear lysate. To extract mitochondria, cells were washed in ice-cold DPBS and
resuspended in 2 ml hypotonic buffer containing 10mM NaCl, 1.5mM MgCl2, and 10mM Tris-
HCl pH 7.5 for 7.5 minutes. After the hypotonic incubation, cells were transferred to a glass
homogenizer and homogenized by pressing straight down with the pestle 20 times, bursting the
cells open while maintaining the integrity of mitochondria. Mitochondrial homogenization buffer
(MHB) was then added to the 2ml homogenized sample to achieve a 1X concertation (210 mM
mannitol, 70 mM sucrose, 20mM HEPES, and 2mM EGTA). The homogenate was then
transferred to a clean 5 ml tube and centrifuged at 17,000 x g for 15 minutes at 4°C. The resulting
pellet was washed in MHB buffer and centrifuged two more times, followed by a resuspension of
the mitochondrial pellet in RIPA lysis buffer, and final centrifugation step of 14,000 x g for 10
minutes at 4°C to obtain a relatively pure mitochondrial lysate. For exogenous SHMOOSE
53
administration, 1 uM of SHMOOSE was administered to cells for 30 minutes, washed twice in
cold PBS, and fractionated. 5-15 mg of protein were reduced in NuPAGE sample buffer and run
on NuPAGE 4-12% Bis-Tris gels. Proteins were transferred to PVDF membranes, blocked with
5% BSA in TBS 0.1% tween, and incubated with respective antibodies at 1:1000 dilutions
overnight at 4C. The next day, membranes were washed with TBST0.1% and incubated with
1:30,000 secondary antibody conjugated to HRP against the respective primary antibody species
of origin, then excited using ECl reagent for 5 minutes.

Endogenous SHMOOSE detection using mass spectrometry
Eluents following SHMOOSE immunoprecipitation from mitochondrial fractions were
precipitated by methanol/ chloroform and redissolved in 8 M urea/100 mM TEAB, pH 8.5.
Proteins were reduced with 5 mM tris(2-carboxyethyl)phosphine hydrochloride (TCEP, Sigma-
Aldrich) and alkylated with 10 mM chloroacetamide (Sigma-Aldrich). SHMOOSE was
immunoprecipitating coupling 10 ug of polyclonal SHMOOSE antibody to Dynabeads Protein A
(Thermo). Approximately 5 x 10
7
differentiated SH-SY5Y cells were used for
immunoprecipitation. Eluted immunoprecipitation sample was digested overnight at 37 oC in 2 M
urea/100 mM TEAB, pH 8.5, with trypsin (Promega). Digestion was quenched with formic acid,
5 % final concentration.
The digest was analyzed on a Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo),
and the digest was injected directly onto a 25 cm, 100 um ID column packed with BEH 1.7um C18
resin (Waters). Samples were separated at a flow rate of 400 nl/min on an Easy nLC 1000
(Thermo). Buffer A and B were 0.1% formic acid in water and 0.1% formic acid in 90%
acetonitrile, respectively. A gradient of 1-25% B over 100 min, an increase to 40% B over 20 min,
54
an increase to 90% B over 10 min and held at 90%B for a final 10 min was used for 140 min total
run time. Column was re-equilibrated with 15 ul of buffer A prior to the injection of sample.
Peptides were eluted directly from the tip of the column and nanosprayed directly into the mass
spectrometer by application of 2.5 kV voltage at the back of the column. The Lumos was operated
in a data dependent mode. Full MS scans were collected in the Orbitrap at 120K resolution with
a mass range of 400 to 1500 m/z and an AGC target of 4e5. The cycle time was set to 3 sec, and
within this 3 sec the most abundant ions per scan were selected for CID MS/MS in the ion trap
with an AGC target of 2e4 and minimum intensity of 5000. Maximum fill times were set to 50 ms
and 35 ms for MS and MS/MS scans respectively. Quadrupole isolation at 1.6 m/z was used,
monoisotopic precursor selection was enabled and dynamic exclusion was used with exclusion
duration of 5 sec.
Protein and peptide identification were done with Integrated Proteomics Pipeline – IP2
(Integrated Proteomics Applications). Tandem mass spectra were extracted from raw files using
RawConverter (He, Diedrich, Chu, & Yates, 2015) and searched with ProLuCID (T. Xu et al.,
2015) against Uniprot human database with the SHMOOSE sequence. The search space included
all full-tryptic and half-tryptic peptide candidates. Carbamidomethylation on cysteine was
considered as a static modification. Data was searched with 50 ppm precursor ion tolerance and
600 ppm fragment ion tolerance. Identified proteins were filtered to using DTASelect (Tabb,
McDonald, & Yates, 2002) and utilizing a target-decoy database search strategy to control the
false discovery rate to 1% at the protein level (Peng, Elias, Thoreen, Licklider, & Gygi, 2003)

SHMOOSE antibody production and ELISA development
55
Rabbit ant-SHMOOSE sera were produced by Yenzyme Antibodies (San Francisco, CA).
SHOOSE affinity antibody was purified from rabbit anti-SHMOOSE sera using CarboxyLink
Immobilization kit with UltraLink Support (Thermo Scientific) according to manufacturer’s
protocol. Briefly, anti-sera were applied onto the synthetic SHMOOSE peptide immobilized
column and the eluted fractions were quantitated by UV absorbance at 280nM. Circulating levels
of SHOOSE were measured by in-house ELISA. Prior to assay, CSF was extracted with 90%
acetonitrile and 10% 1N HCl. To measure endogenous SHMOOSE levels, synthetic SHMOOSE
peptide was used as standard within range 100 pg/ml to 20,000 pg/ml. Briefly, 96-well microtiter
plate was coated with anti-SHMOOSE polyclonal antibody for 3 hours followed by blocking the
plate with SuperBlock buffer (Thermo Scientific). Next, standards, controls or extracted samples
and pre-tittered detection antibody were added to the appropriate wells and incubated overnight.
Followed by 3 washes, wells were added streptavidin-HRP conjugate and incubated for 30
minutes. After four washes, ultra-sensitive TMB (Thermo Scientific) were added and incubated
for 10-20 minutes. The reaction was stopped by the addition of 2N sulfuric acid and absorbance
was measured on a plate spectrophotometer at 450 nm. The intra- and inter-assay coefficient
variations (CV) of SHOOSE ELISA were less than 10%, respectively.

Correlation between CSF SHMOOSE levels and CSF tau, CSF p tau 181, and brain DTI
We considered data for 79 subjects recruited through the University of Southern California
Alzheimer Disease Research Center (ADRC) who had available diffusion MRI (dMRI) scans and
CSF measures of SHMOOSE. Of those, one did not have a usable dMRI scan and 6 were excluded
for preprocessing failures identified through quality assessment (see Diffusion MRI
Preprocessing). Our final sample included 72 non-demented older (mean 65.7; 47-82 years old)
56
adults who had diffusion MRI scans that passed all quality checks and available CSF measures of
SHMOOSE. Subjects had a clinical dementia rating (CDR) score of 0 (56 subjects) or 0.5 (16
subjects). Subject race/ethnicities were self-reported as: White (53), Asian (12), American Indian
or Alaska Native (3), more than one race (4), race not reported (1); Hispanic/Latino (any race)
(10), non-Hispanic/Latino (any race) 62. MR images were acquired on a 3 Tesla Siemens Prisma
scanner at the University of Southern California Alzheimer’s Disease Research Center (ADRC).
Anatomical sagittal T1-weighted magnetization prepared rapid acquisition gradient-echo
(MPRAGE) scan parameters were acquired (TR 2300 ms; TE 2.95 ms; 1.2 × 1.0 × 1.0 mm
3
voxel
size). We also acquired a 64-direction (b=1000 s/mm
2
) diffusion MRI scan (TR 7100 ms; TE 71.0
ms; 2.5 × 2.5 × 2.5 mm
3
voxel size). All scans were visually assessed for quality. For each subject,
the diffusion images were denoised with MATLAB version R2014b software (MathWorks,
Natick, MA) using a local primary components analysis (LPCA) tool with the Rician filter, with
intensity bias correction (Manjón et al., 2013). Distortion correction of DWI included correction
for Gibbs ringing using MRtrix3 (Tournier et al., 2019) and eddy current correction using the
eddy_correct tool in FSL utilities (FSL 5.0.9; (www.fmrib.ox.ac.uk/fsl; (Jenkinson, Beckmann,
Behrens, Woolrich, & Smith, 2012)). We performed bias field correction using MRtrix3 (Tournier
et al., 2019). Echo planar imaging (EPI) susceptibility artifacts were corrected using FSL and
ANTS software (Avants, Epstein, Grossman, & Gee, 2008; Avants et al., 2011) to align the average
b0 maps to subject-specific T1-weighted MPRAGE structural scan. Each step was visually quality
checked. Fractional anisotropy (FA) maps – indicating diffusion restriction within a voxel – were
created using FSL software. FA is a metric of microstructural integrity shown to augment the
power to detect AD-specific deficits (Nir et al., 2017) with lower FA values typically representing
poor white matter microstructural integrity in AD (as reviewed in (Lo Buono et al., 2020).
57
Voxelwise statistical analysis of DWI FA data was performed using the FSL-based tool(Smith et
al., 2004), tract based-spatial statistics (TBSS) (Smith et al., 2006). TBSS applies nonlinear
registration to bring all FA maps into standard template space. A mean FA skeleton was created
and then thresholded at 0.2, resulting in a 4D skeletonised FA image used in voxelwise statistical
analyses detailed below. We used the general linear model (GLM) with FSL’s Threshold-Free
Cluster Enhancement (TFCE) option to evaluate the relationship between SHMOOSE and
voxelwise white-matter FA within the mean FA skeleton, covarying for CDR score, age, and
reported sex (Smith & Nichols, 2009). SHMOOSE values <100 were coded as 50 for this analysis
(5 subjects). For CSF tau and p tau 181, linear regression analysis was conducted with biological
sex and age as a covariate, and SHMOOSE CSF levels as the dependent variable. Separately, we
considered the effects for p tau 181 were mediated through total tau levels. Hence, we used the
mediation package in R and modeled the effect of p tau 181, age, and biological sex on the mediator
(i.e., tau); modeled the effect of tau, p tau 181, biological sex, and age on SHMOOSE; and used
these models to determine the indirect (ACME) and direct (ADE) effects using the mediate
function.

Human brain SHMOOSE mtSNP differential expression analysis
We utilized genotype and transcriptome data generated by Mayo (Synapse ID:
syn5550404). “RNAseq TCX” data were analyzed by SHMOOSE genotype. Subject SHMOOSE
genotype was extracted from Mayo LOAD GWAS data that was generated from the
HumanHap300-Duo Genotyping BeadChips. A complete description of the processing and
individual sub cohorts has been described previously (Allen et al., 2016). Briefly, gene expression
fastq files from human brain temporal cortex were aligned using the Mayo MAP-RNAseq pipeline
58
(Kalari et al., 2014). Normalized read counts were then examined for differential expression by
SHMOOSE genotype using multi-variable linear regression to adjust for age at death, biological
sex, and RNA integrity. Source code in R provided by Mayo was modified to conduct the
differential expression analysis by SHMOOSE genotype (Allen et al., 2018). Results contain all
genes that have non-zero raw counts in at least 1 subject, and each gene contains a beta value
representing the effect size by SHMOOSE mtSNP. Multiple hypothesis correction was performed
using Benjamin Hochberg. Significant genes were included in gene enrichment analyses using the
clusterProfiler package in R. By using the enrichGo and enrichWP functions, significantly
enriched pathways were extracted according to a hypergeometric model (Yu, Wang, Han, & He,
2012). A total of 14 SHMOOSE mtSNP carriers were assessed against 55 reference allele
SHMOOSE individuals. These samples were selected by extracting non-demented individuals at
time of death that also contained SHMOOSE genotype data.

SHMOOSE-treated cell transcriptomics
Differentiated neural cells were incubated with 10uM SHMOOSE or SHMOOSE.D47N
for 24 hours followed by rapid RNA extraction. Cells were washed once with ice-cold DPBS and
immediately lysed with TRIzol (Thermo Scientific), and RNA was extracted using the Quick-RNA
Miniprep Kit (Zymo Research). High quality RNA used for library preparation (mRNA-Seq Nu
Quant), which captures poly-adenylated RNA. From there, prepped samples were sequenced on
an Illumina NextSeq 550 platform for 75 single end cycles. Each sample achieved a read depth of
nearly 25 million. High quality fastq files were ensured using FastQC and mapped to the human
reference genome (GRCh38.p13) using kallisto. Normalized fold changes were then used to
estimate differential gene expression between SHMOOSE and SHMOOSE.D47N treated cells
59
using the DESeq2 package in R. Gene enrichment was carried out on significantly different gene
(FDR < 0.2) using the clusterProfiler package in R.

SHMOOSE in vivo
To examine the transcriptomes of mice treated with SHMOOSE, 12-week-old male
C57Bl/6N mice were obtained from The Jackson Laboratory. Mice were fed a high fat diet for 10
days (60% total calories) prior to initiation of SHMOOSE daily IP injections (2.5mg/kg). No more
than 60 ul of volume were injected IP. After two weeks of IP injections, mice were euthanized
following food withdrawal overnight, then brain was rapidly removed, hypothalamus extracted,
and hemisected midsagittal. Hemibrains were further microdissected to extract the hippocampus
and cortex. Tissues were snap frozen and RNA was extracted by adding 100 ul of TRIzol (Thermo
Scientific) per 10mg tissue. Homogenates were then spun down at 16,000 RCF for 60 seconds and
processed using the Quick-RNA Miniprep Kit (Zymo Research). High quality RNA used for
library preparation (mRNA-Seq Nu Quant), which captures poly-adenylated RNA. From there,
prepped samples were sequenced on an Illumina NextSeq 550 platform for 75 single end cycles
and fastq files were quality ensured using FastQC and mapped to the mouse reference genome
(GRCm39) using kallisto. Normalized fold changes were then used to estimate differential gene
expression for SHMOOSE-treated mice using the DESeq2 package in R. Gene enrichment was
carried out on significantly different gene (FDR < 0.2) using the clusterProfiler package in R.

Seahorse assay
SH-SY5Y cells were plated into 96-well plates at a density of 10,000 cells. The following
day, cells were differentiated for a total of 4 days. Thereafter, SHMOOSE or SHMOOSE.D47N
60
were incubated for 24 hours, followed by cell real-time oxygen consumption rates measurements
using XF96 Extracellular Flux Analyzer (Seahorse Bioscience). ATP turnover and maximum
respiratory capacity were calculated after challenging cells with oligomycin and FCCP (carbonyl
cyanide 4-[trifluoromethoxy]phenylhydrazone). Additionally, glycolytic rate was determined
using extracellular acidification rate (ECAR) and individually reported relative to basal level in
percentage. All readings were normalized to total DNA content using Hoechst 33342.

SHMOOSE differential expression in iPSCs and AD brains
RNA-Seq data from neurons derived from iPSCs with FAD mutations were downloaded
to test SHMOOSE expression as a function of FAD mutations (GEO: GSE128343). Fastq files
were aligned to the human reference genome (GRCh38.p13) using STAR with default parameters
(Dobin et al., 2013). Aligned BAM files were loaded into R using the BioConductor package. A
custom GTF file containing the SHMOOSE genomic coordinates and other mitochondrial genes
were used for the differential expression analysis. Counts were normalized to mitochondrial read
count. Counts were called using the “union” mode by the summarizeOverlaps function
(Chandramohan, Wu, Phan, & Wang, 2013). Differential expression analysis was conducted using
negative binomial regression by the DESeq2 package in R. We also used Mayo RNASeq data
(Synapse ID: syn5550404) to assess SHMOOSE RNA differences by AD and by genotype,
following the same processing workflow.

Amyloid beta toxicity assay
SH-SY5Y cells were differentiated for 4 days, incubated with 10 uM SHMOOSE or
SHMOOSE.D47N for 24 hours followed by another incubation with SHMOOSE or
61
SHMOOSE.D47N with or without oligomerized 1uM amyloid beta 42 (CPC Scientific),
prepared as formerly described (Ryan et al., 2013). A two-color fluorescence cell viability assay
(LIVE/DEAD Viability/Cytotoxicity Kit; Invitrogen (cat. L3224) was used to distinguish live
cells from dead cells after humanin and amyloid beta 42 treatment. The ratio of live to dead cells
can be quantified since live cells retain the Calcein AM dye and dead cells with damaged
membranes permit entry of the ethidium homodimer dye.

SHMOOSE-mitofilin protein interaction
To identify the SHMOOSE interactome, multiple experiments were conducted. First, 1.5
nmol of SHMOOSE was spiked into 1mg SH-SY5Y cell lysate for 6 hours at 4C. Lysates were
prepared using Thermo Pierce CoIP lysis buffer with 1X Thermo protease inhibitor cocktail and
1mM PMSF. Briefly, cells were washed with ice-cold DPBS twice, lysed for 15 minutes on ice,
and centrifuged for 10 minutes at 12,000 RCF at 4C. The rationale for conducting this experiment
was to ensure identical protein amounts between conditions and avoid differential protein
expression caused by SHMOOSE treatment to cells. After 6 hours, SHMOOSE was
immunoprecipitated from samples using Dynabeads A conjugated to 5ug of custom c-terminus
SHMOOSE antibody. As a negative control, SHMOOSE-spiked lysates were also
immunoprecipitated using 5ug rabbit IgG. Proteins were eluted from beads using 50 mM Glycine
pH 2.8, and eluents were pH neutralized using Tris HCl pH 7.5. Complete eluents were then
processed for protein identification using LC-MS. Samples were mixed with same volume of
digestion buffer (8M Urea, 0.1M Tris-HCl pH 8.5), then each sample was reduced and alkylated
via sequential 20-minute incubations with 5 mM TCEP and 10 mM iodoacetamide at room
temperature in the dark while being mixed at 1200 rpm in an Eppendorf thermomixer. 6 μl of
62
carboxylate-modified magnetic beads (CMMB and widely known as SP3) was added to each
sample. Ethanol was added to a concentration of 50% to induce protein binding to CMMB. CMMB
were washed 3 times with 80% ethanol and then resuspended with 50μl 50mM TEAB. The protein
was digested overnight with 0.1 μg LysC (Promega) and 0.8 μg trypsin (Pierce) at 37 °C. Following
digestion, 1ml of 100% acetonitrile was added to each to sample to increase the final acetonitrile
concentration to over 95% to induce peptide binding to CMMB. CMMB were then washed 3 times
with 100% acetonitrile and the peptide was eluted with 50 μl of 2% DMSO. Eluted peptide samples
were dried by vacuum centrifugation and reconstituted in 5% formic acid before analysis by LC-
MS/MS. Peptide samples were separated on a 75μM ID, 25cm C18 column packed with 1.9 μM
C18 particles (Dr. Maisch GmbH HPLC) using a 140-minute gradient of increasing acetonitrile
concentration and injected into a Thermo Orbitrap-Fusion Lumos Tribrid mass spectrometer.
MS/MS spectra were acquired using Data Dependent Acquisition (DDA) mode. MS/MS database
searching was performed using MaxQuant (1.6.10.43) against the human reference proteome from
EMBL (UP000005640_9606 HUMAN Homo sapiens, 20874 entries) (Cox & Mann, 2008).
Statistical analysis of MaxQuant label-free quantitation data was performed with the artMS
Bioconductor package, which performs the relative quantification of protein abundance using the
MSstats Bioconductor package (default parameters) (Choi et al., 2014). The abundance of proteins
missing from one condition but found in more than 2 biological replicates of the other condition
for any given comparison were estimated by imputing intensity values from the lowest observed
MS1-intensity across samples and p values were randomly assigned to those between 0.05 and
0.01 for illustration purposes (Webb-Robertson et al., 2015). We chose to target mitofilin (IMMT)
based on imputed fold change and p value thresholds.
63
Second, we validated the mitofilin interaction identified from MS using a series of
reciprocal co-immunoprecipitation experiments with our SHMOOSE antibody and mitofilin
antibody. We treated differentiated SH-SY5Y cells for 30 minutes with 1 uM SHMOOSE and
lysed cells using Thermo Pierce CoIP lysis buffer as mentioned above. Thereafter, we incubated
samples with 5 ug of SHMOOSE antibody, mitofilin antibody, or negative rabbit IgG for 30
minutes at room temperature. Antibody-coupled Dynabeads A were washed 3 times with
TBST0.1% and eluted using Glycine pH 2.8, NuPAGE LDS sample buffer, and NuPAGE sample
reducing agent for 5 minutes at 95C. Eluents were then loaded into NuPAGE 4-12% Bis-Tris gels
for electrophoresis. Migrated proteins were transferred to PVDF membranes, blocked with 5%
BSA in TBS 0.1% tween, and incubated with respective antibodies at 1:1000 dilutions overnight
at 4C. The next day, membranes were washed with TBST0.1% and incubated with 1:30,000
secondary antibody conjugated to HRP against the respective primary antibody species of origin,
then excited using ECl reagent for 5 minutes.
Third, we conducted reciprocal dot blots for SHMOOSE, SHMOOSE.D47N, and mitofilin
by immobilizing 140 ng of recombinant mitofilin (OriGene), SHMOOSE, or SHMOOSE.D47N
on nitrocellulose membranes. After proteins were dried, membranes were blocked for 30 minutes
with SuperBlock (PBS) blocking buffer (Thermo) at room temperature. Then, either SHMMOOSE
or mitofilin were flowed over blocked membranes at a concentration at 1 ug/ml for 30 minutes at
room temperature in blocking buffer. Membranes were washed three times for five minutes each
with TBSBT 0.1% and then incubated with 0.5 ug/ml of respective antibodies for 30 minutes at
room temperature in blocking buffer. Membranes were washed three times for five minutes each
with TSBT 0.1% and incubated with 1:30,000 secondary antibodies against species of primary
64
antibody origin. Membranes were washed three times for five minutes each with TBST 0.1%,
followed by excitation using ECl reagent for 1 minute.

MTT assay
MTT assays were used to measure the effect of mitofilin knockdown. SH-SY5Y cells were
reverse transfected using RNAiMAX (Invitrogen) and 40 nM mitofilin siRNA (Horizon,
SMARTpool) when plated into 96-well plates at a density of 10,000 cells. The following day, cells
were differentiated for a total of 4 days and transfected with another 40 nM mitofilin siRNA. Two
days later, differentiation medium was changed with an addition 40 nM mitofilin siRNA. 24 hours
before the MTT assay, cells were treated with 10 uM SHMOOSE or solvent control. MTT (Sigma-
Aldrich) reagent (5 mg/ml) was added to each well after treatments for four hours and lysed before
absorbance values were read using the SpectrMax M3 microplate reader.

SHMOOSE-mitofilin interaction prediction
To model the IMMT-SHMOOSE interaction, we used HDOCK, which is a hybrid
algorithm of template-based modeling and ab initio free docking, as described elsewhere (Yan et
al., 2020). Mitofilin was considering the “receptor” and SHMOOSE was considered the “ligand.”

65

66
Figure 3.1. Mitochondrial rs2853499 changes the amino acid sequence of SHMOOSE and
associated with AD and neuroimaging modalities. Translational schematic showing how
mitochondria DNA variants can be used to reveal a novel microprotein.
a The GWAS Manhattan plot-equivalent of a MiWAS, called a Solar Plot. SNPs extending beyond
the outer blue are statistically significant by a permutation empirical p value of 0.05. The most
significant mtSNPs were rs2853498 and rs2853499 – both are haplogroup U determining, with the
latter causing the missense change to SHMOOSE.
b Meta-analysis forest plot of rs2853499 in ADNI, ROSMAP, LOAD, and ADC/1/2. Bars
represent 95% confidence intervals.
c Neuroimaging-based PheWAS in UK Biobank that illustrates the significant effects of
SHMOOSE.D47N and age on parahippocampus. Other significant effects noted in for the EC and
PCC.
d rs2853499 changes the 47
th
amino acid of SHMOOSE. Computational-simulated models of
SHMOOSE and SHMOOSE.D47N by RosettaTTFold.
e GO cellular terms enriched by genes that co-express with SHMOOSE in the human temporal
cortex (n = 69).
f Western blot detection of ~6kDa SHMOOSE in cells containing mtDNA versus cells not
containing mtDNA (i.e., rho zero cells). Laminin B1 is a nuclear marker; GRSF1 isoform is a
mitochondrial marker; GAPDH is a cytosolic marker.
g Unique mass spectrometry tryptic fragments of SHMOOSE found from mitochondria fractions
corresponding to the SHMOOSE amino acid sequence

67

Figure 3.2. SHMOOSE levels in CSF correlate to age, tau, and brain white matter
microstructure.
a SHMOOSE RNA expression in the temporal cortex of AD cases (red). Significance represented
as pAdj < 0.05 following negative binomial regression on all normalized mitochondrial gene
counts.
b Human CSF SHMOOSE levels (pg/ml) correlation with age. Regression model includes
biological sex as a covariate; p value < 0.001. SHMOOSE correlation with CSF total tau (pg/ml).
Regression model includes biological sex and age as covariates; p value < 0.05. SHMOOSE
correlation with CSF phosphorylated tau at residue 181 (p tau 181; pg/ml). Regression model
includes biological sex and age as covariates; p value < 0.05.
c Higher CSF SHMOOSE was significantly associated with lower DTI FA in the body of the
corpus callosum and bilateral superior corona radiata in 72 non-demented older adults. Regression
model included age, reported sex, and clinical dementia rating score. Colored voxels indicate
FSLThreshold-Free Cluster Enhancement-derived p values < 0.05 after correction
68

Figure 3.3 SHMOOSE associates with differential mitochondrial and ribosomal gene
expression in humans, cells, and mice. Blue highlighting represents mitochondrial terms and red
highlighting represents ribosomal terms.
a Principal component analysis (PCA), color coded for 14 SHMOOSE.D47N carriers or 55
SHMOOSE reference allele carriers. Dashed line represents the median value of PC2. Of the 14
SHMOOSE.D47N carriers, 11 fall below the median PC2 value (p value < 0.05; generalized linear
69
model). To the right of the PCA figure are GO cellular compartment terms enriched by
SHMOOSE.D47N carriers.
b PCA of the in vitro gene expression signature for 10 uM SHMOOSE or SHMOOSE.D47N-
treated neural cells after 24 hours. To the right of the PCA figure are GO cellular compartment
terms enriched by SHMOOSE.D47N-treated neural cells.
c Corresponding PCA plots and GO cellular compartment terms, on liver tissue, enriched by
SHMOOSE following 14-day SHMOOSE treatment to mice.
70

Figure 3.4. SHMOOSE is a biologically active microprotein that localizes to mitochondria
and boosts metabolic activity and oxidative consumption rate.
71
a SHMOOSE-treated differentiated SH-SY5Y cells (1 uM) localized to mitochondria after 15
minutes. Top portion of the blot represents a 5-second exposure. Second portion of the blot
represents a 30-second exposure. Also identified were SHMOOSE dimers around 12 kDa in the
SHMOOSE-treated conditions that were most prominent in mitochondrial fractions. Laminin,
GRSF1, and GAPDH represent nuclear, mitochondrial, and cytosolic fractions, respectively.
b MTT assay results that illustrate 1uM of both SHMOOSE (light blue) and SHMOOSE.D47N
(dark blue) both have effects on cell metabolic activity at 1uM and 10uM. Significance defined as
p value < 0.05 for independent t tests.
c Normalized to the baseline third measurement, the effect of SHMOOSE and SHMOOSE.D47N
on mitochondrial spare capacity. Statistical significance determined using independent t tests.
d Effect of SHMOOSE and SHMOOSE.D47N on basal OCR. Significance defined as p value <
0.05 for independent t tests
e SHMOOSE expression in neurons derived from iPSCs with FAD APP mutation (orange), FAD
PSEN mutation, and FAD APP plus PSEN mutations (red). SHMOOSE expression was highest in
the latter cell type. Significance defined as pAdj < 0.05 following negative binomial regression on
all normalized mitochondrial gene counts.
f SHMOOSE-treated neural cells protects against amyloid beta42-induced toxicity (light blue).
Significance defined as p value < 0.05 for independent t tests.

72

Figure 3.5. SHMOOSE binds the inner mitochondrial membrane protein mitofilin.
a Schematic proteomics analysis of SHMOOSE-spiked neural cell lysates that were
immunoprecipitated using a custom SHMOOSE polyclonal antibody.
b Reciprocal western blot validation of SHMOOSE/mitofilin interaction by immunoprecipitating
with SHMOOSE antibody or mitofilin antibody.
c Reciprocal dot blot illustrating recombinant mitofilin and SHMOOSE interact.
d MTT assay showing SHMOOSE has no effect of neural cell metabolic activity when mitofilin
is knocked down with siRNA. Y axis is normalized to control baseline absorbance values.
Statistical significance determined using independent t test with p value < 0.05.
e HDOCK prediction of the SHMOOSE (yellow) and mitofilin (brown) interaction. Predicted
interaction residues for mitofilin occur at its c-terminus.

73
CHAPTER 4: MITOCHONDRIAL DNA VARIATION IN EXTREME LONGEVITY
REVEALS RARE HUMANIN MICROPROTEIN VARIANT THAT PROMOTES APOE4
RESILIENCE
4.1 Abstract
Apolipoprotein E4 (APOE4) is the main risk gene for Alzheimer’s disease (AD), but some
APOE4 carriers are nonetheless resilient to AD well into old age. The goal of our studies was to
identify a resilient APOE4 factor in the form of a mitochondrial-derived peptide called humanin.
Here, we first show that humanin levels associated with APOE-related cognitive impairment. Next,
we found that a variant of humanin – called humanin P3S – associated with extreme longevity in
APOE4 carriers. Humanin P3S is caused by the mitochondrial DNA (mtDNA) polymorphism
C2639T and was enriched in centenarians carrying APOE4 (30%). Following up on this genetic
observation, we found that molecular dynamic simulations confidently predicted APOE4 and
humanin biophysical interaction, which were affirmed experimentally. The biological
consequences of APOE and humanin were further investigated. Human brain temporal cortex
transcriptomics revealed that expression of humanin is tightly linked to cellular energetics and
protein homeostasis – but this tightly linked signature is weakened in APOE3 carriers with
dementia and ablated in APOE4 carriers with or without AD. In the APP/PS1/APOE4 mouse
model, humanin P3S reduced brain amyloid beta burden, an effect that was less pronounced in
mice treated with humanin wild type. Altogether, we highlight a novel genetic resilience factor for
APOE4 in the form of the mitochondrial-derived peptide humanin P3S.

74
4.2 Introduction
Nearly three decades ago, Schäter et al. first reported that the frequency of the
Apolipoprotein E4 (APOE4) allele was significantly low in centenarians (Schachter et al., 1994).
Around the same time, Corder et al. found that APOE4 was the major genetic risk factor for late
onset Alzheimer’s disease (AD) (Corder et al., 1993). Since then, the AD-APOE4 association has
been consistently replicated as the most influential genetic risk factor (Abondio et al., 2019;
Lambert et al., 2013). Nevertheless, some APOE4 carriers are resilient well into old age.
One possible reason APOE influences longevity and AD is through mitochondria
(Castellano et al., 2011; Sienski et al., 2021; Simonovitch, Schmukler, Masliah, Pinkas-Kramarski,
& Michaelson, 2019). In fact, APOE is in linkage disequilibrium with translocase of outer
mitochondrial membrane 40 (TOMM40)(Soyal et al., 2020), and APOE and mitochondria have
been linked experimentally. For example, in ischemic challenged mice, greater amounts of APOE
localized to mitochondria in mice with humanized APOE4 than in mice with APOE3 (James et al.,
2012). This might explain why stressed APOE4 mice displayed enhanced mitochondrial fusion
and decreased fission in separate studies (Simonovitch et al., 2019). Furthermore, as these APOE4
mice age, their energetics shift away from mitochondrial respiration in the cortex and hippocampus
(Area-Gomez et al., 2020). Yet even in the absence of APOE4, mitochondrial dysfunction has been
comprehensively linked to AD. For instance, cells harboring mtDNA from AD patients had higher
Aβ than in cells containing non-diseased mtDNA (Cardoso, Santana, Swerdlow, & Oliveira,
2004). Additionally, the recently illustrated tau interactome exposed that tau primarily targets
mitochondrial proteins (Tracy et al., 2022). In the same study, cells that contained frontotemporal
dementia mutations had less tau binding to mitochondrial proteins. However, just as the etiology
of AD remains unclear, so too are precise effects of the mitochondrial genome.
75
Nonetheless, recent mitochondrial genetic studies have revealed surprising data. Due to
deep RNA sequencing methods, several small RNA species and previously unannotated transcript
cleavage sites have been discovered (Mercer et al., 2011). In other studies, peptides derived from
mitochondrial DNA (mitochondrial-derived peptides; MDPs) were detected and challenged the
dogma that the mitochondrial genome encodes 13 biologically active polypeptides (Brendan
Miller, Su-Jeong Kim, et al., 2020). The first MDP discovered was cloned out of the occipital lobe
of an AD patient during a cDNA screen for amyloid beta (Aβ) protection, much to the surprise of
the investigators (Hashimoto, Ito, et al., 2001; Hashimoto, Niikura, et al., 2001). They found that
this Aβ-protective cDNA mapped to a portion of the mitochondrial 16S rRNA that contained a
small open reading frame (sORF). Indeed, this sORF encoded a MDP that they eventually called
humanin. Today, humanin has been focused in hundreds of reports, many of which highlight its
role in attenuating AD pathology through cell signaling and protein folding mechanisms (B. Guo
et al., 2003; Ikonen et al., 2003; Lee et al., 2014; Muzumdar et al., 2009; Park et al., 2013; Tajima
et al., 2005; X. Xu, Chua, Gao, Hamdy, & Chua, 2006; K. Yen et al., 2020; W. Zhang et al., 2013).
Recently, a report found that a mtDNA variant within humanin (rs2854128) associated with
cognitive decline and lower circulating humanin levels (Kelvin Yen et al., 2018). This rs2854128
humanin mtDNA mutation suggests other humanin variants might associate with age-related
phenotypes.
Here, we studied humanin levels and humanin mtDNA variation within the context
of APOE and extreme longevity. We specifically studied a rare humanin mtDNA variant by APOE
genotype in centenarians. Finally, we carried out a series of biophysical, in vitro, and in vivo studies
to understand the relationship between humanin mtDNA variation and APOE.

76
4.3 Results
Cognitively impaired APOE4 carriers had lower humanin levels
Given that humanin levels have previously associated with cognitive decline (Kelvin Yen
et al., 2018), we hypothesized cerebrospinal fluid (CSF) humanin levels would differ by APOE
genotype. In cognitively intact individuals without APOE4, CSF humanin levels averaged 242.5
pg/ml (SE: 46.0) compared to 434.5 pg/ml (SE: 93.9) in APOE4 carriers, which did not achieve
statistically significance (p = 0.11) (Figure 4.1A). Yet in individuals with APOE4 and cognitive
impairment (Clinical Dementia Rating > 0), humanin levels were nearly half that in non-APOE4
carriers. Specifically, in non APOE4 carriers with cognitive impairment, humanin levels were on
average 508.7 pg/ml (SE: 78.0) compared to 239.8 pg/ml (SE: 55.5) in APOE4 carriers (p < 0.05)
(Figure 4.1B).

mtDNA variant within the humanin sORF called P3S is enriched in centenarians with APOE4
We searched for humanin mtDNA DNA variants enriched in APOE4 centenarians by
studying the Albert Einstein College of Medicine longevity centenarian cohort, which is
population mostly of Ashkenazi genetic ancestry individuals. These centenarians had high
frequency of a humanin variant compared to the general population. For instance, almost 12% of
centenarians genotyped had a variant of humanin that causes the third amino acid to change from
proline to serine (humanin P3S). In contrast, by using mitochondrial genotypes from the 1000
Genome project, other European-ancestral and non-European-ancestral populations displayed
humanin P3S frequency under 0.2% (Figure 4.2B). While humanin P3S frequency is indeed high
in these centenarians of Ashkenazi descent, we cannot rule out that this high enrichment is due to
77
mitochondrial haplogroup N1b, with frequencies ranging from 6-10% in other Ashkenazi-sampled
cohorts (Feder, Ovadia, Glaser, & Mishmar, 2007).
Notably, this humanin P3S variant was especially enriched in Ashkenazi centenarians with
APOE4. The frequency of humanin P3S was 30% in APOE4 centenarians, whereas its frequency
was only 7.1% in centenarians without APOE4 (chi sq p < 0.001; Figure 4.2C-D). Since APOE4
frequency declines as a function of age (Figure 4.2C), yet is high in humanin P3S carriers, we next
studied the functional relationship between humanin and APOE.

Humanin wild type (WT) and humanin P3S bind APOE4 with differential kinetics
We hypothesized that the genetic interaction between APOE4 and humanin P3S might be
explained through direct protein-protein interactions. To test this hypothesis, we computationally
modeled the binding between humanin P3S and both APOE3 and APOE4. All-atom molecular
dynamics simulations reveal that glutamic acid residues on the c-terminus of P3S quickly bind to
arginines on both APOE3 (APOE:Glu45–P3S:Arg22) and APOE4 (APOE:Glu27–P3S:Arg22).
However, the mutated amino acid in P3S (Ser3) exhibited a high affinity for APOE4
(APOE:Asp35-P3S:Ser3) but no affinity for APOE3 (Figures 4.3A-B). As a result, umbrella
sampling simulations were carried out and revealed that the binding affinity between P3S and
APOE (Figure 4.4B) is much stronger for APOE4 (35.4 kCal/mnol) versus APOE3 (14.8
kCal/mol), or about 2.3x fold in simulations.
Subsequently, we experimentally characterized the direct binding potential between
humanin and APOE4 by using co-immunoprecipitation, dot blot, and surface plasmon resonance
assays. In cells expressing APOE4, we co-immunoprecipitated APOE and humanin WT by using
a humanin polyclonal antibody (Figure 4.3C). In the dot blot assay, 100 ng of humanin variants
78
(P3S, F6A, C8A, L12A, S14G/HNG, and phosphorylated humanin) displayed differential binding
to APOE4, with humanin P3S showing the strongest signal (Figure 4.3D). One explanation for
these differences might be that P3S is six times more aggregation prone compared to wild type
humanin (Figure 4.4A), as confirmed from Thioflavin T spectroscopy (an indicator of protein
aggregation). In surface plasmon resonance experiments, APOE4 binds humanin P3S with
extremely fast “on” and slow “off” kinetics (Figures 4.3E-F). By immobilizing humanin P3S (2.5
μg/ml) to a gold dextran matrix and flowing over a concentration gradient of APOE4 (8nM, 32
nM, and 128nM), we calculated a dissociation constant (KD) of 0.694 nM. In comparison, by
immobilizing humanin WT, we observed a higher KD of 12.7 nM, suggesting that the biological
relevance of APOE4 and humanin P3S is, in part, mediated through greater affinity.

Humanin co-expresses with energy metabolism and proteasomal pathways by APOE genotype and
AD
To understand the biological interaction between APOE and humanin, we analyzed human
brain temporal cortex RNA sequencing (RNA-Seq) data. Co-expression analyses were conducted
with humanin against all expressible transcripts in APOE4 and non-APOE4 carriers by AD status.
In APOE3 carriers without AD (n = 49), over 1500 genes co-expressed with humanin (adjusted p
< 0.05; Figure 4.5A). These significantly co-expressed genes enriched WikiPathway terms for the
TCA cycle, Jourbert syndrome/ciliary landscape, DNA repair, and mRNA processing (Figure
4.5B). However, in AD, these signatures were weakened in APOE3 carriers (n = 34) and
disappeared in APOE4 carriers with (n = 42) and without AD (n = 7) (Figure 4.4A). In APOE3
carriers with AD, about 400 genes co-expressed with humanin (adjusted p < 0.05; Figure 4.5A)
and enriched mitochondrial-related parkin-ubiquitin proteolysis (Figure 4.4B). In APOE4 carriers
79
with AD, less than 10 genes co-expressed with humanin (Figure 4.5A). These co-expression
analyses suggest mitochondrial function and proteasomal homeostasis are APOE/humanin targets.

Humanin P3S increases Aβ uptake in APOE4-glial cells
Two functional cellular models were used to test functional targets of APOE/humanin, as
identified from analyses in Figure 4. First, we measured APOE4 glia Aβ uptake incubated with a
combination of humanin wild type, humanin P3S, and Aβ. In cells incubated with 100 nM humanin
wild type, Aβ uptake increased by ~25% compared to control (p < 0.05). But in cells incubated
with humanin P3S, Aβ uptake was ~50% higher than control (p < 0.05) (Figure 4.6A-B). Second,
immortalized murine astrocytes that express humanized APOE4 were administered 1 uM humanin
WT or humanin P3S for 24 hours followed by a live-cell oxygen consumption assay (Seahorse XF
Analyzer). Both humanin WT and humanin P3S shifted the energetic signature of these APOE4-
expressing astrocytes away from glycolysis towards oxidative phosphorylation. After 24 hours,
the oxygen consumption rate (OCR) of APOE4-expressing astrocytes that were administered
humanin WT and humanin P3S was ~175 pmol/min, whereas the OCR of control cells was ~50
pmol/min. In addition, the extracellular acidification rate (ECAR) – a marker of glycolysis – was
~140 mpH/min in comparison to 85 mpH/min and 95 mpH/min induced by humanin WT and
humanin P3S, respectively. These data suggest both humanin and humanin P3S modify APOE4-
related mitochondrial biology with differential effects specifically on Aβ uptake.

Humanin P3S reduces brain Aβ load in APOE4/APP/PS1 mice
To detect whether humanin WT and humanin P3S reduce amyloid burden load in vivo, we
used an AD mouse model containing two familial AD-related mutations (APP and PS1) and
80
humanized targeted replacement of APOE4. We hypothesized humanin P3S would reduce Aβ,
with humanin P3S outperforming humanin WT. Aβ-related pathology was measured by two
methods. First, Aβ in cortex and hippocampus was quantified by Aβ immunohistochemistry. Aβ
load in the cortex and hippocampus was significantly lower in mice given humanin P3S versus
vehicle by nearly 50% (Figure 4.7A-C). Likewise, Aβ load in the cortex and hippocampus was
lower in mice given humanin WT but did not achieve statistical significance (Figure 4.7A-B).
Second, thioflavin S was used to determine amyloidogenic plaques. Cortical and hippocampal
plaques were significantly lower in mice given humanin P3S compared to vehicle, and mice given
humanin WT showed a mild yet insignificant decrease in plaques (Figure 4.7D-E). Altogether, our
results show a significant reduction in amyloidogenic burden in the cortex and hippocampus of
mice administered humanin P3S.

Humanin P3S modifies the mitochondrial and proteolytic transcriptome in the hippocampi of
APP/PS1/APOE4 mice
We analyzed the hippocampal transcriptome in a subset of APP/PS1/APOE4 mice that
were administered vehicle, humanin WT, or humanin P3S. PolyA-enriched RNA was sequenced
at a ~20M average read depth. Principal component analysis (PCA) significantly reduced the
dimensionality of the count data in a manner that classified vehicle, humanin P3S, and humanin
WT (Figure 4.8A). In PCA, mice given humanin P3S and vehicle visually clustered together. The
number of statistically significant genes that were differentially expressed by humanin P3S and
humanin WT were 296 and 158, respectively (adjusted p < 0.2). Humanin P3S-induced genes
enriched WIkiPathway terms for cytoplasmic ribosomal proteins, microglia pathogen
phagocytosis, oxidative phosphorylation, and macrophage markers (Figure 4.8C). Likewise,
81
humanin WT-induced genes enriched WikiPathway terms for microglia pathogen phagocytosis
but separately enriched oxidative damage response and the complement classical pathway.
Moreover, we found that humanin P3S reduced circulating levels of Interferon-gamma (IFN-γ)
and increased Interleukin 10 (IL-10), yet no significant observations were noted for Interleukin 2
(IL-2), Interleukin 5 (IL-5), Interleukin 6 (IL-6), tumor necrosis factor alpha (TNF-α), or
Interleukin 1 beta (IL-1β).

82
4.4 Discussion
The goal of these studies was to determine a resilient interaction between the MDP
humanin and APOE. We started by measuring humanin levels in CSF among APOE4 carriers with
and without cognitive impairment. In APOE4 carriers without cognitive impairment, humanin CSF
levels were insignificantly elevated compared to APOE3 carriers (p = 0.11). But in APOE4 carriers
with cognitive impairment, humanin levels were significantly lower compared to APOE3 carriers
(p < 0.05). Likewise, we see that humanin co-expression in AD is significantly weakened in AD
and by APOE genotype.
We next sought to determine whether humanin mtDNA variation associated with extreme
longevity by APOE genotype. To do so, we studied humanin sORF genotype of DNA samples
from the Albert Einstein Longevity Genes Project, provide by Dr. Nir Barzilai. Our humanin sORF
genotyping revealed a unique mtDNA single nucleotide variant at mtDNA base pair position 2639
(C > T), which mutates the third amino acid of humanin from proline to serine (P3S). In the general
population, the frequency of humanin P3S is estimated to be under 0.2%, whereas in our study the
frequency was nearly 12%. Notably, humanin P3S is mitochondrial haplogroup-determining N1b,
which is extremely rare haplogroup in the general population yet quite common in Ashkenazi
ancestral individuals (Costa et al., 2013). For instance, in a separate study, the frequency of N1b
in Ashkenazi ancestral individuals from different geographical regions ranged from 6-10%(Costa
et al., 2013). Hence, it is possible that humanin P3S is a marker for extreme longevity, but this
requires appropriately matched mitochondrial genetic ancestral control individuals that is worth
considering in follow-up research.
Notably, we found that humanin P3S was especially enriched in centenarians carrying at
least one copy of APOE4. Of the 20 APOE4-carrying centenarians analyzed, six were humanin
83
P3S positive (30%), a sharp contrast from the nine humanin P3S positive individuals in 126
centenarians who do not have a copy of APOE4 (~7%). That 40% of humanin P3S positive
centenarians also contained a copy of APOE4 suggests a potent biological interaction. While the
sample size here might be considered small, we considered such a stark observation worthy of
experimental follow-up.
We followed up on the APOE4-humanin P3S genetic relationship by implementing a series
of biophysical, in vitro, and in vivo experiments and analyses. First, we hypothesized that humanin
binds APOE because the humanin interactome shares common binding partners with APOE (Atali
et al., 2020). To test if interactions between these genes is mediated by protein-protein interactions,
we carried out computational simulations to measure the affinity between heterodimers using
umbrella sampling. Molecular dynamics simulations revealed a stronger binding affinity between
humanin P3S and APOE4 due to the display of numerous charged residues. These additional
charged residues on APOE4 facilitated strong P3S binding on both the peptide’s C- and N-terminal
domains. In contrast, APOE3 displays less charged residues and facilitates P3S binding only on
its C-terminal domain. It is particularly interesting that the N-terminal segment that differentially
binds APOE contains the mutated amino acid P3S, which distinguishes the mutated humanin from
its wild type counterpart. Second, we used solution biophysics experiments to experimentally
measure the affinity between APOE and humanin using three different assays: co-
immunoprecipitation, dot blot, and surface plasmon resonance. Indeed, by immunoprecipitating
endogenous humanin from HEK293 cells that overexpress APOE4, we pulled down APOE4.
Moreover, by dotting 100ng of several variants of humanin on nitrocellulose membranes and
flowing over APOE4, it was found that wild type humanin binds APOE4 and – with a nearly twice
as strong signal – humanin P3S binds APOE4. Next, by using surface plasmon resonance
84
experiments, we showed that the kinetics of humanin wild type and humanin P3S between APOE4
were markedly different. The Kd of humanin P3S and APOE4 was about 0.7 nM compared to 12.7
nM of humanin wild type and APOE4. Altogether, our computational and experimental evidence
suggests APOE4 interacts preferentially with serine of humanin (humanin P3S) and that humanin
P3S binds APOE4 with greater affinity.
To understand the biological targets of the humanin and APOE interaction, our next
analyses involved human population cohort transcriptomics. Using RNA-Seq data derived from
the brain temporal cortices of 132 samples by APOE genotype and AD status, we conducted
humanin co-expression analyses. In non-APOE4 samples without AD, over 1500 genes
significantly co-expressed with humanin – these genes enriched cellular energetics and
proteostasis. Interestingly, this signature weakened for APOE3 AD carriers (<500 genes) and
nullified in APOE4 carriers with and without AD status.
As a result of these association analyses, we carried out energetics and amyloid beta
metabolism (i.e., proteasomal homeostasis) experiments in vitro and in vivo. The underlying
mechanism leading to reduced Aβ could be due to APOE4-specific folding. We show that humanin
P3S interacts with APOE4 due to the additional exposure of charged residues. APOE4, being prone
to misfolding, is associated with disrupted protein homeostasis, especially in the form of Aβ. In
our in vitro experiments, we found that both humanin wild type and humanin P3S shifted murine
astrocytes that express APOE4 towards oxidative phosphorylation. While these energetic effects
were similar between humanin wild type and humanin P3S, we observed that humanin P3S
promoted greater Aβ uptake in APOE4 glia. Likewise, in an AD mouse model containing two FAD
mutations and humanized APOE4, humanin P3S-treated mice showed significantly lower Aβ
burden than in vehicle-treated mice. Similarly, mice given humanin wild type showed a modest
85
yet insignificant reduction in Aβ. Since these animal studies were carried out for sixty days starting
at the immediate onset of Aβ pathology, it is possible that longer treatment of humanin wild type
(i.e., 90+ days) would yield significant Aβ reductions. Nevertheless, this 60-day treatment period
pinpoints the biological significance of humanin P3S considering its more pronounced effects on
Aβ reduction.
Moreover, in a subset of our in vivo experiment, we performed differential expression
analyses via RNA-Seq on hippocampi. In mice treated with humanin P3S, 296 genes were
differentially expressed compared to control, and these genes enriched mitochondrial oxidative
phosphorylation, microglia and macrophage phagocytosis, and ribosomal proteins. Yet in mice
treated with humanin WT, 158 genes were differentially expressed and less dramatically enriched
oxidative phosphorylation and microglia and macrophage phagocytosis. These terms align with
the energetics and proteasomal homeostasis terms that were enriched during humanin co-
expression analyses and our in vitro experiments.
Our data suggests humanin P3S is a resilience factor for APOE4 in part through differential
binding to APOE4 and Aβ reduction. We are nonetheless limited by several factors. For example,
our humanin P3S observation in centenarians was in a cohort of Ashkenazi genetic ancestry
individuals, which is a population that commonly presents the N1b mitochondrial haplogroup
(Feder et al., 2007). Still, the stark enrichment of humanin P3S in APOE4-carrying individuals
was considered biologically meaningful. A significant strength of our study is that we did not stop
at the association step; rather, we aimed to functionally validate this association in a variety of
experimental modalities, one of which included a mouse model of APOE4-AD. While we observed
an age-by-APOE4 interaction for humanin P3S, we did not model age in our in vivo studies. We
still observed significant reduction in Aβ load in APOE/APP/PS1 mice in a relatively short
86
intervention period, highlighting the potential potency and relevance of humanin P3S in APOE4
biology. Altogether, our parallel experimental and computational approaches revealed a unique
mitochondrial genetic resilience factor for APOE4 in the form of a humanin variant. These data
further highlight the biological significance of mitochondrial genetic and nuclear genetic
interactions as well as peptides encoded by mtDNA.

87
4.5 Methods
Humanin ELISA
Humanin levels in cerebrospinal fluid (CSF) were measured by in-house sandwich ELISA.
Prior to assay, CSF was extracted in a mixture of 90% acetonitrile and 10% 1N HCl. To measure
endogenous humanin levels, synthetic humanin was used as standard within range 50pg/ml to
10,000pg/ml. 96-well microtiter plate was coated with capture antibody in 50 mM sodium
bicarbonate buffer on a shaker. Standards, controls or extracted samples and pre-tittered detection
antibody were added to the appropriate wells and incubated overnight. The next day, streptavidin-
HRP conjugates were added and incubated for 30 minutes, followed by four washes and the
addition of ultra-sensitive TMB (Thermo Scientific) for 10-20 minutes. The reaction was stopped
by the addition of 2N H2SO4 and absorbance was measured on a plate spectrophotometer at 405
nm. Details about the humanin ELISA development have been written in protocol format(B. Miller
& Wan, 2020).

mtDNA sequencing of Albert Einstein Centenarian Cohort
Mitochondrial DNA variation were identified by using the MitoChip v2.0 (Affymetrix;
Santa Clara, CA). Sequences comparison of both strands of the entire human mitochondrial
genome (16,568-bp) were synthesized as overlapping 25-mers on high-density oligonucleotide
arrays with 8 × 8 μm features. The entire mitochondrial DNA sequence was amplified in three
overlapping PCR reactions using 50 ng of genomic DNA each. Pooling, DNA fragmentation,
labeling, and chip hybridization were done using manufacturer instructions (Affymetrix; Santa
Clara, CA). The chips were washed on the Affymetrix fluidics station using CustomSeq
Resequencing wash protocols and scanned using the GeneChip Scanner 7G. Analysis of
88
microarray data was done using GeneChip Sequence Analysis Software (GSEQ) v4.0 (Affymetrix;
Santa Clara, CA). ≥95% call rate for a given nucleotide, passed quality control measures, and were
assigned for further analyses.
To compare humanin P3S frequency to other population cohorts, we analyzed the 1000
Genome Project. Variants of mtDNA were downloaded in VCF format using the following FTP
site: http://ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/20130502/.
Humanin P3S frequency was also analyzed by APOE4 status within the Albert Einstein
Longevity Genes Project. These DNA samples were collected from individuals living
independently at 95 years of age or older and were principally of Ashkenazi genetic ancestry
(Atzmon et al., 2004). APOE4 was captured as defined previously (Ryu, Atzmon, Barzilai,
Raghavachari, & Suh, 2016). Next, a chi-square was carried out to determine whether humanin
P3S frequency was statistically different between centenarians with and without a copy of APOE4.

Molecular dynamics simulation for APOE and humanin
Atomistic molecular dynamics simulations were run using the GROMACS 2020
integrator. The AMBER99sb-ILDN force field was used to model proteins and TIP3P to model
water. Starting structures of APOE3 were taken from the PDBDatabank (PDB: 2L7B), where
mutated amino acids to solubilize the protein for NMR were restored to their WT residues. APOE4
was obtained by mutating APOE3, and both structures were equilibrated for 500 ns at 300 K.
Newton’s equations of motion were integrated every 2 fs using a Leapfrog algorithm under an
NVT ensemble. Short-range electrostatics and van der Waals forces were truncated at 1.2 nm,
while long-range electrostatics were tabulated using a Particle Mesh Ewald (PME) algorithm.
Periodic boundary conditions in all Cartesian directions were enforced throughout each simulation
89
to reduce the complexity of simulations. Following dimerization, umbrella sampling simulations
were used to pull APOE away from P3S using a virtual spring (k=5000 kJ/mol-nm
2
), followed by
the application of a potential of mean force to deduce the binding affinity over 100 ns/umbrella
reaction coordinate. Overall, P3S was pulled 4 nm and a potential of mean force was measured
every 0.1 nm, resulting in ~41 umbrella simulations. Subsequently, potentials were integrated
using a weighted histogram analysis method (WHAM), which tabulated the net potential of mean
force between APOE isoforms and P3S. Taken together, a total of 4.1 microseconds were run to
generate the computational data listed here.

Experimental binding assays for APOE and humanin
To assess the binding relationship between APOE and humanin, three assays were
conducted. First, humanin was immunoprecipitated from HEK293 cells that overexpress APOE4.
10 ug of humanin polyclonal rabbit antibody or 10 ug of rabbit IgG was conjugated to Dynabeads
Protein A for 90 minutes rotating followed by one wash, as outlined in the manufacture protocol.
After antibody conjugation, cells were lysed using Pierce IP Lysis Buffer with 1X Halt Protease
Inhibitor Cocktail. Cell lysates were added to the antibody-conjugated beads for 30 minutes at
room temperature rotating, followed by an additional overnight rotation at 4C. The next day, beads
were washed three times using the manufacture-provided wash buffer, and proteins were eluted
for 6 minutes at 95C using premixed sample buffers suitable for Invitrogen NuPAGE western blot.
Supernatants were then loaded into the wells of 4-12% Bis Tris NuPAGE gels for gel
electrophoresis followed by transfer and immunoblot. APOE was detected using Human APOE
Antibody by R&D (1:1000) in 5% milk TBST0.1%, and humanin was detected using [insert
antibody conditions]. Second, 100 ng of humanin WT and several variants of humanin were
90
immobilized to nitrocellulose membranes for dot blot analysis. After peptide immobilization, 1.5
ug/ml of APOE4 (PeproTech) in SuperBlock (PBS) Blocking Buffer (Thermo Fisher) was flowed
over the nitrocellulose membrane and incubated for 30 minutes at room temperature. Next,
membranes were washed three times for five minutes using TBST 0.5% and then incubated with
1:30000 secondary antibody (donkey anti-Goat IgG, HRP, Thermo Fisher) for 30 minutes at room
temperature, followed by three additional washing steps and Clarity Western ECL detection. Third,
using surface plasmon resonance Biacore T100, the kinetics of humanin and APOE4 were
assessed. Immobilized on CM3 chips were humanin or humanin P3S at 2.5 ug/ml concentration in
immobilization buffer (10mM sodium acetate, pH 5.0). Next, to block the remaining surface, goat
IgG (20ug/ml) was immobilized in immobilization buffer. APOE4 was then flowed over the chip
at 8nM, 32nM, and 128nM in running buffer (0.01M HEPES, 0.15 M NaCl, 0.3 mM EDTA, and
0.05% tween). After each concentration of APOE was assessed, APOE was removed using 2:8
SDS Glycine pH 3.0. Fitting results were assessed using chi-squared goodness of fit.

Population cohort humanin co-expression analysis
RNASeq data generated by Mayo (Synapse ID: syn5550404) was analyzed. Humanin
transcript count matrices were created from bam files by constructing a MDP database in GTF
format and implementing the summarizeOverlaps function of the GenomicAlignments package in
R. Thereafter, normalized counts were used to conduct Pearson correlation between humanin
counts and all nuclear-encoded gene counts, corrected for multiple hypotheses using a false
discovery rate (FDR) of 0.05. Genes that statistically correlated with humanin expression were
tested for enrichment using the enrichWP function from the clusterProfiler package, which returns
91
enrichment of WikiPathway terms. Data output from the enrichWP function were used to generate
plots using ggplot2 in R.
In vitro seahorse assay
Immortalized murine astrocytes with human APOE4 replacement were seeded onto 96-
well plates at a density of 10,000 cells in DMEM/F12 10% FBS, 1mM sodium pyruvate, and 200
ug/ml geneticin. The following day, 1uM of humanin WT, 1uM of humanin P3S, or solvent control
were administered to each well containing cells. Real-time oxygen consumption rate (OCR) was
measured after 24 hours using the XF96 Extracellular Flux Analyzer. Baseline OCR was measured
in addition to extracellular acidification rate (ECAR), normalized to total DNA content using
Hoescht 3342.

In vitro Aβ uptake assay
Primary cultures of mixed glial cells were prepared from single brains collected from 3-4
months old APOE4 targeted replacement mice using modifications of a previously described
protocol (Bronstein, Torres, Nissen, & Tsirka, 2013). Briefly, after removing meninges, brains
were dissected to remove cerebellum, midbrain, pons and medulla. Individual brains were
mechanically and enzymatically dissociated by first mincing with a sterile razor blade and then
incubating with 2 mL of DMEM/F12 (Thermo Fisher Scientific) containing 10 units/ml papain
(Worthington Biochemical Corporation, Lakewood, NJ, USA) for 30 min at 37°C. Enzymatic
activity was quenched by addition of 500 μL of fetal bovine serum (FBS; Thermo Fisher Scientific,
Waltham, MA, USA), and undigested tissues are triturated by pipetting up and down with a 1000
uL tip. To collect the glial-enriched fraction, the cell suspension was centrifuged for 15 min at
700×g in a 23% percoll/DMEM/F12 solution. Cell pellets were resuspended with DMEM/F12
92
supplemented with 10% FBS, non-essential amino acids (Sigma-Aldrich, St. Louis, MO, USA),
and a penicillin G and streptomycin cocktail (Corning Inc., Corning, NY, USA), and seeded onto
100-mm culture dish (Corning Inc.) coated with poly-D-lysine. After 2h incubation, the medium
was replaced to the fresh medium containing 20% FBS. Cultures were maintained at 37°C in a
humidified incubator with room air supplemented with 5% CO2. The serum content of the medium
was reduced to 10% FBS and replaced 3 times per week. After 12-14 days in vitro, cultured cells
were detached from flasks by 15 min exposure to TrypLE (Thermo Fisher Scientific) at 37°C. To
collect microglia, the detached cells were centrifuged for 20 min at 2,000×g in a percoll/HBSS(-)
gradient solution (0-40-70%). Microglia between 40% and 70% percoll layer were collected and
seeded at 2.0×104/cm2 in 48-well plate with glass cover slips. Microglia cultures were maintained
in DMEM/F12 with 10% FBS for 1 days prior to Aβ42 uptake assay. Amyloid β1-42 peptide
(Aβ42) was prepared to promote generation of oligomeric species with slight modifications of a
previously described protocol (Lambert et al., 1998; Ultsch et al., 2016). HiLyteᵀᴹ Fluor 555-
labeled Aβ42 (AnaSpec, Fremont, CA, USA) was dissolved in 100% 1,1,1,3,3,3-Hexafluoro-2-
propanol and subsequently air-dried. The resultant peptide film was dissolved in 10 mM NaOH,
and neutralized by adding pH 7.4 PBS (final concentration 20 µM Aβ42). The Aβ42 solution was
incubated at 4°C for 24h and then used immediately in uptake assay. Microglia were pre-treated
with 100 nM humanin or 100 nM P3S variant in DMEM/F12 for 1h in a 37°C incubator.
Subsequently, oligomeric Aβ42 (final concentration: 0.2 µM) was added. After 1h incubation,
cells were fixed with 4% PFA for 15 min at 4°C. For morphometric analysis, fluorescent images
were captured at the center of the well (each field contains approximately 300-400 microglia cells)
of 2-independent wells using Bz-X700 microscope (Keyence, Osaka, Japan). Aβ42 uptake by
93
microglia was quantified using ImageJ software. Aβ42 uptake was determined by measuring the
area of Aβ fluorescent signal (the number of pixels with positive labeling).

Animal procedures
We conducted a 60-day study of 44 male, APP/PS1/APOE4 mice beginning at 9 weeks of
age. Mice were given once-a-day injections of humanin or humanin P3S at a dose of 5mg/kg mouse
body weight. These mice were generated by Dr. Patrick Sullivan (Duke University) and Dr. Dave
Holtzman (Washington University). As previously noted, APPPS1-21 mice that overexpress a
human APP with Swedish mutation (KM670/671NL) and PS1 with L166P mutation under the
control of Thy1 promote were crossed with mice harboring humanized replaced APOE4. At the
end of the 60-day treatment period, mice were anesthetized with inhalant isoflurane, ~300 ul of
blood collected following excision of the right atrium, and transcardially perfused with ice-cold
0.1 M PBS. Brains were immediately removed, separated into hemibrains, and one hemibrain was
immersion fixed for 24 h in 4% paraformaldehyde/0.1 M PBS followed by storage at 4°C in 0.1
M PBS/0.3% NaN3 until processed for immunohistochemistry. The other hemibrain was
microdessected to extract hippocampi and snap frozen for RNA extraction. Fixed hemi-brains were
fully sectioned in the horizontal plane at 40 μm using a vibratome (Leica Biosystems). All
experiments were performed in accordance with the appropriate guidelines and regulations
approved by the University of Southern California Institutional Animal Care and Use Committee
(IACUC).

Immunohistochemistry quantification
94
To detect Aβ, immunohistochemistry was carried out using avidin/biotin peroxidase
approaches with ABC Vector Elite kits (Vector Laboratories) on every eighth horizontal brain
section. Detecting Aβ involved pretreating tissues with 95% formic acid for 5 minutes followed
by TBS rinse and endogenous peroxidase blocking solution for 10 minutes. Tissues were then
washed three times for 10 minutes in 0.1% Triton-X/TBS and incubated in a blocking solution
containing 2% bovine serum albumin TBS for 30 minutes at room temperature. Blocked tissues
were incubated overnight at 4C in a primary antibody solution containing a Aβ antibody (1:300,
Invitrogen) in blocking buffer. The following day, tissues were rinsed in TBS and incubated in
biotinylated secondary antibody in blocking solution and activated using 3,3’-diaminobenzidine
(Vector Laboratories). Aβ load was quantified on images captured by the Microscope BZ-X800 at
a 10x magnification and stitched together using Keyence software. Pictures were then converted
to grayscale followed by a threshold adjustment in Image J 1.50i, which permitted a binary image
that was calculated as the percentage of Aβ load by total area and assessed statistically using Mann-
Whitney tests.

Thioflavin-S (Thio-S) staining and quantification
To detect Aβ fibril formation, every eighth section adjacent to sections used for
immunohistochemistry was stained for Thio-S (Sigma-Aldrich). Horizontal hemibrain sections
were first mounted and air-dried overnight. The next morning, dried sections were washed three
times in 50% ethanol for 5 minutes, washed with double-distilled water, and incubated for 10
minutes in 1% Thio-S. After Thio-S incubations, sections were briefly rinsed in 70% ethanol,
dehydrated, and protected using cover slips with anti-fade mounting medium (Vector
Laboratories). Sections were then imaged using the Microscope BZ-X800 at a 10x magnification
95
and stitched together with Keyence software. Digital images were converted to grayscale followed
by a threshold adjustment in Image J 1.50i. Binary images were quantified and calculated as the
percentage of Aβ load by total area and assessed statistically using Mann-Whitney tests.

RNA-Seq of mouse hippocampi
Hippcampal RNA was extracted by adding 100 ul of TRIzol (Thermo Scientific) per 10mg
tissue. These homogenates were then spun centrifuged at 16,000 RCF for 60 seconds and
processed using the Quick-RNA Miniprep Kit (Zymo Research). RNA was assessed for high
quality and proceeded to library preparation (mRNA-Seq Nu Quant) to enrich poly-adenylated
RNA. Samples were sequenced on an Illumina NextSeq 550 platform for 75 single end cycles,
quality ensured using FastQC, and mapped to the mouse reference genome (GRCm39) using
kallisto. Normalized fold changes were used to estimate differential gene expression for among
conditions by using the DESeq2 package in R. WikiPathway enrichment was carried out on
significantly different gene (FDR < 0.2) using the clusterProfiler package in R. Genes within
significantly enriched terms were extracted and plotted using custom scripts in R.

96

Figure 4.1. Humanin CSF by APOE4 and dementia.
a Humanin levels trended as higher in APOE4 carriers than non-APOE4 carriers without dementia.
b Humanin levels were significantly lower in APOE4 carriers with dementia. Mann-Whitney
statistical tests were used.

0
200
400
No APOE4 (n = 26) APOE4 (n = 26)
0
200
400
No APOE4 (n = 20) APOE4 (n = 19)
Humanin (pg/ml)
p = 0.11 *p < 0.05
Cogntively Intact
CDR > 0
a b
97

Figure 4.2. The humanin variant P3S is enriched in APOE4 centenarians.
a Mitochondrial genomic location of the humanin smORF and the P3S mutation.
b 12% frequency of humanin P3S in Ashkenazi descent centenarians compared to <0.2%
frequency in all summed populations from 1000 genome project.
c APOE4 frequency as a function of age (decade); at 100+ years, APOE4 and humanin P3S
frequency is showed.
d Pie graph illustrating enrichment of humanin P3S in APOE4 centenarians ( p < 0.002; chi squared
test)

98

Figure 4.3. Humanin binds APOE, but humanin P3S binds APOE4 with much higher
affinity.
a MD simulations reveal that humanin P3S binds to APOE3 with its C-terminus (Arg22) vs.
bbinding to APOE4 with both its C- (Asp35) and N-terminus (Ser3). This is due, in part, to
differences in the surface display of APOE Glu27 and Asp35.
c Western blot following humanin immunoprecipitation shows APOE4 being pulled down. d Dot
blot analysis with variants of humanin immobilized and APOE4 flowed over the membrane.
e Surface plasmon resonance of humanin and APOE4 with a kD of 12.8 nM vs.
f kD of 0.694 nM of humanin P3S and APOE4.

99

Figure 4.4. Molecular dynamics simulations (umbrella sampling) of humanin and APOE.
a ThT assays reveal that HN P3S is significantly more aggregation-prone vs. WT HN.
b Preferential interactions between HN P3S monomers and APOE4 vs. APOE3: APOE4 (35.4
kCal/mnol) versus APOE3 (14.8 kCal/mol).

100

Figure 4.5. Humanin co-expression analysis by APOE genotype in the human brain temporal
cortex.
a The number of significantly co-expressed genes with humanin by APOE and AD status (Pearson
correlation with Benjamini-Hochberg false discovery rate correction < 0.05).
b WikiPathway enrichment of significant terms from humanin co-expressed genes by APOE and
AD status.

Ciliary Landscape
DNA Repair Pathways, Full Network
Joubert Syndrome
mRNA Processing
Parkin-Ubuiquitin Proteasomal System Pathway
TCA Cycle (Krebs or Citric Acid Cycle)
TCA Cycle and Deficiency of PDHc
DLAT
SDHA
ACLY
FH
SUCLA2
DLD
CS
MDH2
IDH3B
TUBB6
TUBB1
SIAH2
SIAH1
PSMD2
PSMD1
HSPA8
HSPA6
HSPA5
HSPA4
HSPA1L
HSPA1B
HSPA1A
CASK
U2AF2
SNRPB
SFPQ
SF3A1
PRPF8
PRPF4
PRPF3
POLR2A
PAPOLA
PABPN1
NXF1
NCBP1
FUS
DHX8
CLK3
CLK2
CLK1
TMEM237
TMEM231
TMEM216
TMEM17
OFD1
NPHP3
NPHP1
INPP5E
DVL1
PSMD8
CSPP1
CEP164
ARL3
RNGTT
XRCC6
RPA3
RFC4
RFC3
REV1
RAD52
RAD23A
POLE3
NBN
MUTYH
MPG
GTF2H5
FEN1
FANCL
FANCG
FANCC
FANCA
ERCC5
DDB1
CUL4B
APEX1
CEP97
CEP290
ZYG11B
YPEL5
BBS7
WDR60
BBS1
TTC30B
ANKS6
TTC30A
TIPRL
RBM14
RALB
POM121
NDUFA9
MYL6
MCM7
MCM6
LRPPRC
IFT43
IFT20
IFT172
GLB1
EXOC8
EXOC7
EXOC4
EXOC3
DYNLT1
DYNC1LI1
COPS3
APMAP
ANKS3
AFG3L2
APOE3, AD
APOE3, No AD
WikiPathway Enrichment
Shared
0
500
1000
1500
APOE3, No AD APOE4, No AD APOE3, AD APOE4, AD
n = 49 n = 7 n = 34 n = 42
Number of Genes that Co-Express with Humanin
a
b
101

Figure 4.6. Humanin P3S increases Aβ phagocytosis in APOE4 glia.
a Humanin P3S induces greater Aβ phagocytosis compared to humanin wild type (p < 0.05).
b Representation of panel a.

102

Figure 4.7. Humanin P3S reduces brain Aβ burden in APP/PS1/APOE4 mice. Humanin wild
type and P3S immuno-based reduction of
a cortical and
b hippocampal amyloid beta (p <0.05; Mann-Whitney test).
c Representative images of a-b.
d Humanin wild type and P3S thioflavin-based reduction of d cortical and e hippocampal amyloid
beta (p <0.05; Mann-Whitney test).
f Representative images of d-e. Vehicle n = 12; Humanin wild type n = 16; Humanin P3S n = 16.

103

Figure 4.8. Humanin and humanin P3S differentiates the hippocampal transcriptome in
APP/PS1/APOE4 mice.
a Transcriptomic principal component analysis of vehicle, humanin wild type, and humanin P3S
conditions.
b Venn diagram illustration showing the number of significant differentially expressed genes by
humanin P3S and humanin wild type (false discovery rate < 0.2).
c WikiPathway enrichment of significant terms from humanin and humanin P3S treated mice.

104
CHAPTER 5: CONCLUSION
Now is an exciting time to be part of the booming microprotein field because it is in its
nascent stages. Over the last decade, several breakthroughs have made it possible to detect and
annotate microproteins. First, technology in the form of nucleotide and peptide sequencing has
greatly improved. Before 2010, the ability to detect microproteins was essentially reliant on
immunological-based detection approaches rather than mass spectrometry, the latter of which is
considered the gold standard for bona fide protein detection. Then in the early 2010s and
throughout the decade, the Saghatelian lab showcased – for the first time – the ability to detect
hundreds of microproteins using a unique enrichment mass spectrometry technique (Slavoff et al.,
2013). They were able to be the first to identify microproteins by excluding larger molecular
weight, high-abundant proteins prior to mass spectrometry (i.e., acetic acid precipitation,
variations of lysis buffers, and cartridge-based selection). At the same time, a boom of ribosome
profiling experiments was carried out by labs across the globe, eventually leading to a newly
founded international consortium (Mudge et al., 2021). Ribosome profiling was a turning point for
the field of microproteins because the technology was used to pinpoint hundreds of thousands of
potential functional sORFs (Ingolia, Ghaemmaghami, Newman, & Weissman, 2009). Both the
advents of microprotein-enriched mass spectrometry and ribosome profiling was supplemented by
enhanced computational power, allowing researchers to process data and index microprotein
targets at an extremely fast pace. As a result, dozens of bona fide microproteins have been
functionalized over the last 15 years. Yet despite improvements in mass spectrometry and
ribosome profiling, the field is still considering ways to filter out false positives while revealing
missed microproteins due to false negatives. Our work in this thesis carried out a novel workflow
105
to reduce false positives/negatives and reveal microproteins encoded exclusively by the
mitochondrial genome.
This thesis novel is highly innovative because it used large-scale population cohort
genomics data as a starting point for microprotein discovery (Chapter 2). In Chapter 2, we showed
substantial human mitochondrial genetic variation in a multi-ethnic representative cohort by
conducting three analyses. First, we carried out principal component analysis (PCA) on nuclear
DNA variation and mitochondrial DNA variation. Nuclear DNA PCA revealed significant
heterogeneity among populations that self-reported as Black and Hispanic compared to Whites, as
has been reported in the literature. But mitochondrial DNA PCA similarly revealed significant
heterogeneity among self-reported Blacks, Hispanics, and Whites. In fact, PCA within sub-ethnic
populations revealed that mitochondrial PCA captured a similar percentage of variance compared
to nuclear PCA. Second, in decision tree classification, nuclear PCA and mitochondrial PCA
estimated a 5% and 8% error in self-reported genetic ancestry, respectively. When both nuclear
PCA and mitochondrial PCA were included in the same algorithm, the error reduced to 2%,
highlighting the potential utility for correcting genetic ancestry in different population cohorts.
Third, by estimating the effect of mitochondrial genetic variation on a hereditable phenotype (i.e.,
height), we showed several principal components significantly predicted height.
Overall, Chapter 2 analyses suggest mitochondrial genetic variation could also associate
with disease phenotypes. However, the analytic method to estimate the effect of specific
mitochondrial DNA variants is not standardized. We suggested controlling for genetic ancestry by
including mitochondrial principal components within the model, but with close consideration to
the principal components that are also SNP determining. Ultimately, despite such statistical
limitations, these genetic variant analsyes could reveal a genomic range that encode microproteins,
106
and these genomic ranges can be used to target bona fide microproteins during discovery and
functionalization stages (Chapters 3 and 4). Essentially, MiWAS was the starting point for our
microprotein discovery workflow.
After developing the methodologies for MiWAS, we applied it to AD in Chapter 3. The
idea here was that a mitochondrial single nucleotide polymorphism (mtSNP) would associate with
AD, and that this mtSNP would reside in a potential microprotein-encoding sORF. We carried out
a MiWAS in the ADNI cohort by considering thousands of mitochondrial SNPs captured by whole
genome sequencing, the most robust whole mitochondrial genome sequencing data for AD to date.
A mtSNP indeed associated with AD with over a two-fold risk. Closer analysis of this mtSNP (bae
pair position 12372) revealed it determines the largest European mitochondrial haplogroup U.
When ADNI whole mtDNA data as was reduced using PCA, this mtSNP did not cluster in distinct
groups, suggesting that the need to correct for genetic ancestry would be result in a redundant
model (an idea presented in Chapter 2). In relation to microproteins, we mapped this mtSNP to a
putative sORF that we eventually called SHMOOSE (referred here on out as SHMOOSE.D47N
because it changes the 47
th
amino acid from D to N). We validated the AD-associative effects of
SHMOOSE.D47N in another whole mitochondrial genome sequencing cohort called ROSMAP
with a similar two-fold risk effect size; and then in two separate microarray-mtSNP cohorts, we
also showed increased risk for SHMOOSE.D47N carriers while correcting for genetic ancestry due
to high heterogeneity. Because of these AD associations, SHMOOSE became our primary
microprotein target for detection and functionalization.
We aimed to detect SHMOOSE using both transcriptomic, peptidomic, and immunological
methods. In transcriptomics, we found that SHMOOSE expression was higher post-mortem
temporal cortex brain samples of individuals with AD. In peptodimics, we detected SHMOOSE
107
using mass spectrometry in neuronal cell mitochondria – the first time a mitochondrial-encoded
microprotein was detected using mass spectrometry. Using immunological methods, we designed
an antibody against the C-terminus of SHMOOSE (predicted to be less structured) and found that
it performed well in detecting ~6 kDa SHMOOSE via western blot. Moreover, our SHMOOSE
antibody performed well in ELISA and was used to highlight an association between CSF
SHMOOSE levels and age, tau, and phosphorylated tau in a cohort of older individuals. We were
then able to use our antibody to carry out functional assays such as co-immunoprecipitation for
interactome characterization. By doing so, we identified that SHMOOSE localized prominently to
the inner mitochondrial membrane and co-immunoprecipitated with dozens of inner mitochondrial
membrane proteins, including one called mitofilin. When we knocked down mitofilin protein
levels in cells, the biological effects of SHMOOSE were nullified, showing that a way SHMOOSE
exerts its effects is through mitofilin or inner mitochondrial membrane integrity. Interestingly,
when SHMOOSE is given to cells or mice, gene expression signatures are most significantly
enriching mitochondrial terms, corroborating the interactome data. Overall, we were able to index
hundreds of putative mitochondrial sORFs by using MiWAS to spotlight and then functionalize a
microprotein candidate that eventually we named SHMOOSE.
In Chapter 4, we took a similar mitochondrial genomics approach to microprotein
functionalization by analyzing mtDNA variation in centenarians. The question we asked in
Chapter 4 was whether mtDNA variation within the annotated microprotein humanin associated
with extreme longevity. Results indicated that a rare mtDNA variant within the third humanin
codon, which changes proline to serine (P3S), was especially enriched in centenarians of
Ashkenazi ancestry (11.7%). Closer inspection exposed that the effect of humanin P3S was in fact
mediated through APOE4. That is, 30% of APOE4 carriers also had the humanin P3S variant.
108
While the statistical sample size in this analysis can be considered small, we considered the effect
worth following experimentally. Given this association and based on molecular dynamic
simulations (MDS), we hypothesized humanin modifies APOE4 biology by direct binding. MDS
suggested that humanin P3S preferentially interacts with APOE4 because the serine mutation is
specifically able to bind to APOE4. In biophysical experiments, humanin P3S indeed showed a
15-times greater affinity to APOE4 than did humanin wild type. In glia cells exposed to APOE4
and amyloid beta, humanin P3S promoted greater phagocytosis than did humanin wild type. The
amyloid beta reduction was recapitulated in vivo using the APOE4/APP/PS1 mouse model. After
a 60-day IP injection period, humanin P3S reduced amyloid beta in the brains of APOE4/APP/PS1
significantly more than did humanin wild type or vehicle control. These results highlight the utility
of starting with human genetic variation analyses to aid microprotein functionalization.
This thesis represents a multi-omics approach to mitochondrial microprotein discovery.
Presented here is the discovery of a completely novel microprotein called SHMOOSE through
integration of human cohort genomics data with targeted experimental approaches. Likewise, also
shown here is the identification of a novel humanin variant using human cohort genomics data.
Given that the field of microproteins is in its nascent stage, this thesis can be useful for scientists
in two ways: (1) via application of our multi-omics framework or (2) via further functionalization
of the novel microprotein SHMOOSE and humanin variant.
The significance of microproteins cannot be understated. Every widely, publicly used gene
enrichment tool does not consider microproteins because the field does not yet know which ones
are functional. This means (1) every largescale omics pipeline solely considers large molecular
weight proteins, (2) drug pipelines are focused on large molecular weight proteins, (3) disease
characterization is focused on large molecular weight proteins, and (4) bulk technology is designed
109
without prioritizing microproteins. Over the next decade and beyond, the probability that
microproteins become vital biological factors appears high. But for these microproteins to realize
their potential, multi-disciplinary collaborations are necessary. Taking a singular gene approach to
the millions of putative microproteins is not feasible. Rather, developing a group consisting of
computational biologists, biochemists, geneticists, disease experts, and physicians seems essential
– the goal of this thesis was to assess the utility of bringing together this sort of group, and we
eventually showed that such a multi-disciplinary effort could lead to the identification of
previously missed microproteins.

110
References

Abondio, P., Sazzini, M., Garagnani, P., Boattini, A., Monti, D., Franceschi, C., . . . Giuliani, C.
(2019). The Genetic Variability of APOE in Different Human Populations and Its
Implications for Longevity. Genes (Basel), 10(3). doi:10.3390/genes10030222
Abraham, G., & Inouye, M. (2014). Fast principal component analysis of large-scale genome-wide
data. PLoS One, 9(4), e93766. doi:10.1371/journal.pone.0093766
Alfaro-Almagro, F., Jenkinson, M., Bangerter, N. K., Andersson, J. L. R., Griffanti, L., Douaud,
G., . . . Smith, S. M. (2018). Image processing and Quality Control for the first 10,000
brain imaging datasets from UK Biobank. Neuroimage, 166, 400-424.
doi:10.1016/j.neuroimage.2017.10.034
Allen, M., Carrasquillo, M. M., Funk, C., Heavner, B. D., Zou, F., Younkin, C. S., . . . Ertekin-
Taner, N. (2016). Human whole genome genotype and transcriptome data for Alzheimer's
and other neurodegenerative diseases. Sci Data, 3, 160089. doi:10.1038/sdata.2016.89
Allen, M., Wang, X., Burgess, J. D., Watzlawik, J., Serie, D. J., Younkin, C. S., . . . Ertekin-Taner,
N. (2018). Conserved brain myelination networks are altered in Alzheimer's and other
neurodegenerative diseases. Alzheimers Dement, 14(3), 352-366.
doi:10.1016/j.jalz.2017.09.012
Area-Gomez, E., Larrea, D., Pera, M., Agrawal, R. R., Guilfoyle, D. N., Pirhaji, L., . . . Nuriel, T.
(2020). APOE4 is Associated with Differential Regional Vulnerability to Bioenergetic
Deficits in Aged APOE Mice. Sci Rep, 10(1), 4277. doi:10.1038/s41598-020-61142-8
Atali, S., Dorandish, S., Devos, J., Williams, A., Price, D., Taylor, J., . . . Evans, H. G. (2020).
Interaction of amyloid beta with humanin and acetylcholinesterase is modulated by ATP.
FEBS Open Bio, 10(12), 2805-2823. doi:10.1002/2211-5463.13023
111
Atzmon, G., Schechter, C., Greiner, W., Davidson, D., Rennert, G., & Barzilai, N. (2004). Clinical
phenotype of families with longevity. J Am Geriatr Soc, 52(2), 274-277.
doi:10.1111/j.1532-5415.2004.52068.x
Avants, B. B., Epstein, C. L., Grossman, M., & Gee, J. C. (2008). Symmetric diffeomorphic image
registration with cross-correlation: evaluating automated labeling of elderly and
neurodegenerative brain. Med Image Anal, 12(1), 26-41. doi:10.1016/j.media.2007.06.004
Avants, B. B., Tustison, N. J., Song, G., Cook, P. A., Klein, A., & Gee, J. C. (2011). A reproducible
evaluation of ANTs similarity metric performance in brain image registration.
Neuroimage, 54(3), 2033-2044. doi:10.1016/j.neuroimage.2010.09.025
Baek, M., DiMaio, F., Anishchenko, I., Dauparas, J., Ovchinnikov, S., Lee, G. R., . . . Baker, D.
(2021). Accurate prediction of protein structures and interactions using a three-track neural
network. Science, 373(6557), 871-876. doi:10.1126/science.abj8754
Balduzzi, S., Rucker, G., & Schwarzer, G. (2019). How to perform a meta-analysis with R: a
practical tutorial. Evid Based Ment Health, 22(4), 153-160. doi:10.1136/ebmental-2019-
300117
Bartzokis, G. (2011). Alzheimer's disease as homeostatic responses to age-related myelin
breakdown. Neurobiology of Aging, 32(8), 1341-1347. Retrieved from
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citati
on&list_uids=19775776
Bianconi, E., Piovesan, A., Facchin, F., Beraudi, A., Casadei, R., Frabetti, F., . . . Canaider, S.
(2013). An estimation of the number of cells in the human body. Ann Hum Biol, 40(6),
463-471. doi:10.3109/03014460.2013.807878
112
Biffi, A., Anderson, C. D., Nalls, M. A., Rahman, R., Sonni, A., Cortellini, L., . . . Rosand, J.
(2010). Principal-component analysis for assessment of population stratification in
mitochondrial medical genetics. Am J Hum Genet, 86(6), 904-917.
doi:10.1016/j.ajhg.2010.05.005
Bronstein, R., Torres, L., Nissen, J. C., & Tsirka, S. E. (2013). Culturing microglia from the
neonatal and adult central nervous system. J Vis Exp(78), 50647. doi:10.3791/50647
Bycroft, C., Freeman, C., Petkova, D., Band, G., Elliott, L. T., Sharp, K., . . . Marchini, J. (2018).
The UK Biobank resource with deep phenotyping and genomic data. Nature, 562(7726),
203-209. doi:10.1038/s41586-018-0579-z
Cai, H., Cao, P., Sun, W., Shao, W., Li, R., Wang, L., . . . Zheng, Y. (2022). Circulating humanin
is lower in coronary artery disease and is a prognostic biomarker for major cardiac events
in humans. Biochim Biophys Acta Gen Subj, 1866(1), 130010.
doi:10.1016/j.bbagen.2021.130010
Cardoso, S. M., Santana, I., Swerdlow, R. H., & Oliveira, C. R. (2004). Mitochondria dysfunction
of Alzheimer's disease cybrids enhances Abeta toxicity. J Neurochem, 89(6), 1417-1426.
doi:10.1111/j.1471-4159.2004.02438.x
Castellano, J. M., Kim, J., Stewart, F. R., Jiang, H., DeMattos, R. B., Patterson, B. W., . . .
Holtzman, D. M. (2011). Human apoE isoforms differentially regulate brain amyloid-beta
peptide clearance. Sci Transl Med, 3(89), 89ra57. doi:10.1126/scitranslmed.3002156
Chandramohan, R., Wu, P. Y., Phan, J. H., & Wang, M. D. (2013). Benchmarking RNA-Seq
quantification tools. Annu Int Conf IEEE Eng Med Biol Soc, 2013, 647-650.
doi:10.1109/EMBC.2013.6609583
113
Chin, R. M., Panavas, T., Brown, J. M., & Johnson, K. K. (2018). Optimized Mitochondrial
Targeting of Proteins Encoded by Modified mRNAs Rescues Cells Harboring Mutations
in mtATP6. Cell Rep, 22(11), 2818-2826. doi:10.1016/j.celrep.2018.02.059
Choi, M., Chang, C. Y., Clough, T., Broudy, D., Killeen, T., MacLean, B., & Vitek, O. (2014).
MSstats: an R package for statistical analysis of quantitative mass spectrometry-based
proteomic experiments. Bioinformatics, 30(17), 2524-2526.
doi:10.1093/bioinformatics/btu305
Cloutier, P., Poitras, C., Faubert, D., Bouchard, A., Blanchette, M., Gauthier, M. S., & Coulombe,
B. (2020). Upstream ORF-Encoded ASDURF Is a Novel Prefoldin-like Subunit of the
PAQosome. J Proteome Res, 19(1), 18-27. doi:10.1021/acs.jproteome.9b00599
Cobb, L. J., Lee, C., Xiao, J., Yen, K., Wong, R. G., Nakamura, H. K., . . . Cohen, P. (2016).
Naturally occurring mitochondrial-derived peptides are age-dependent regulators of
apoptosis, insulin sensitivity, and inflammatory markers. Aging (Albany NY), 8(4), 796-
809. doi:10.18632/aging.100943
Cohen, P. (2014). New Role for the Mitochondrial Peptide Humanin: Protective Agent Against
Chemotherapy-Induced Side Effects. JNCI Journal of the National Cancer Institute,
106(3), dju006-dju006. doi:10.1093/jnci/dju006
Corder, E. H., Saunders, A. M., Strittmatter, W. J., Schmechel, D. E., Gaskell, P. C., Small, G. W.,
. . . Pericak-Vance, M. A. (1993). Gene dose of apolipoprotein E type 4 allele and the risk
of Alzheimer's disease in late onset families. Science, 261(5123), 921-923. Retrieved from
https://www.ncbi.nlm.nih.gov/pubmed/8346443
114
Costa, M. D., Pereira, J. B., Pala, M., Fernandes, V., Olivieri, A., Achilli, A., . . . Richards, M. B.
(2013). A substantial prehistoric European ancestry amongst Ashkenazi maternal lineages.
Nat Commun, 4, 2543. doi:10.1038/ncomms3543
Cox, J., & Mann, M. (2008). MaxQuant enables high peptide identification rates, individualized
p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat Biotechnol,
26(12), 1367-1372. doi:10.1038/nbt.1511
Crimmins, E. M., Kim, J. K., Langa, K. M., & Weir, D. R. (2011). Assessment of cognition using
surveys and neuropsychological assessment: the Health and Retirement Study and the
Aging, Demographics, and Memory Study. J Gerontol B Psychol Sci Soc Sci, 66 Suppl 1,
i162-171. doi:10.1093/geronb/gbr048
Davatzikos, C., Xu, F., An, Y., Fan, Y., & Resnick, S. M. (2009). Longitudinal progression of
Alzheimer's-like patterns of atrophy in normal older adults: the SPARE-AD index. Brain,
132(Pt 8), 2026-2035. doi:10.1093/brain/awp091
Dinov, I., Lozev, K., Petrosyan, P., Liu, Z., Eggert, P., Pierce, J., . . . Toga, A. (2010).
Neuroimaging study designs, computational analyses and data provenance using the LONI
pipeline. PLoS One, 5(9). doi:10.1371/journal.pone.0013070
Dinov, I. D., Van Horn, J. D., Lozev, K. M., Magsipoc, R., Petrosyan, P., Liu, Z., . . . Toga, A. W.
(2009). Efficient, Distributed and Interactive Neuroimaging Data Analysis Using the LONI
Pipeline. Front Neuroinform, 3, 22. doi:10.3389/neuro.11.022.2009
Dobin, A., Davis, C. A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., . . . Gingeras, T. R.
(2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics, 29(1), 15-21.
doi:10.1093/bioinformatics/bts635
115
Egan, M. F., Kojima, M., Callicott, J. H., Goldberg, T. E., Kolachana, B. S., Bertolino, A., . . .
Weinberger, D. R. (2003). The BDNF val66met polymorphism affects activity-dependent
secretion of BDNF and human memory and hippocampal function. Cell, 112(2), 257-269.
Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12553913
Eltermaa, M., Jakobson, M., Utt, M., Koks, S., Magi, R., & Starkopf, J. (2019). Genetic variants
in humanin nuclear isoform gene regions show no association with coronary artery disease.
BMC Res Notes, 12(1), 759. doi:10.1186/s13104-019-4807-x
Eriksson, E., Wickström, M., Perup, L. S., Johnsen, J. I., Eksborg, S., Kogner, P., & Sävendahl, L.
(2014). Protective Role of Humanin on Bortezomib-Induced Bone Growth Impairment in
Anticancer Treatment. JNCI: Journal of the National Cancer Institute, 106(3), djt459-
djt459. doi:10.1093/jnci/djt459
Fang, H., Hu, N., Zhao, Q., Wang, B., Zhou, H., Fu, Q., . . . Lyu, J. (2018). mtDNA Haplogroup
N9a Increases the Risk of Type 2 Diabetes by Altering Mitochondrial Function and
Intracellular Mitochondrial Signals. Diabetes, 67(7), 1441-1453. doi:10.2337/db17-0974
Feder, J., Ovadia, O., Glaser, B., & Mishmar, D. (2007). Ashkenazi Jewish mtDNA haplogroup
distribution varies among distinct subpopulations: lessons of population substructure in a
closed group. Eur J Hum Genet, 15(4), 498-500. doi:10.1038/sj.ejhg.5201764
Feng, Y., Madungwe, N. B., & Bopassa, J. C. (2019). Mitochondrial inner membrane protein,
Mic60/mitofilin in mammalian organ protection. J Cell Physiol, 234(4), 3383-3393.
doi:10.1002/jcp.27314
Fischl, B., & Dale, A. M. (2000). Measuring the thickness of the human cerebral cortex from
magnetic resonance images. Proc Natl Acad Sci U S A, 97(20), 11050-11055.
doi:10.1073/pnas.200033797
116
Fischl, B., Salat, D. H., Busa, E., Albert, M., Dieterich, M., Haselgrove, C., . . . Dale, A. M. (2002).
Whole brain segmentation: automated labeling of neuroanatomical structures in the human
brain. Neuron, 33(3), 341-355. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/11832223
Fischl, B., Salat, D. H., van der Kouwe, A. J., Makris, N., Segonne, F., Quinn, B. T., & Dale, A.
M. (2004). Sequence-independent segmentation of magnetic resonance images.
Neuroimage, 23 Suppl 1, S69-84. doi:10.1016/j.neuroimage.2004.07.016
Fischl, B., Sereno, M. I., Tootell, R. B., & Dale, A. M. (1999). High-resolution intersubject
averaging and a coordinate system for the cortical surface. Hum Brain Mapp, 8(4), 272-
284. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10619420
Ge, Q., Jia, D., Cen, D., Qi, Y., Shi, C., Li, J., . . . Wang, L. (2021). Micropeptide ASAP encoded
by LINC00467 promotes colorectal cancer progression by directly modulating ATP
synthase activity. J Clin Invest, 131(22). doi:10.1172/JCI152911
Gieffers, C., Korioth, F., Heimann, P., Ungermann, C., & Frey, J. (1997). Mitofilin is a
transmembrane protein of the inner mitochondrial membrane expressed as two isoforms.
Exp Cell Res, 232(2), 395-399. doi:10.1006/excr.1997.3539
Grangeon, L., Paquet, C., Bombois, S., Quillard-Muraine, M., Martinaud, O., Bourre, B., . . .
collaborators of the e, P. L. M. f. g. (2016). Differential Diagnosis of Dementia with High
Levels of Cerebrospinal Fluid Tau Protein. J Alzheimers Dis, 51(3), 905-913.
doi:10.3233/JAD-151111
Guo, B., Wu, S., Zhu, X., Zhang, L., Deng, J., Li, F., . . . Zhou, Y. (2020). Micropeptide CIP2A-
BP encoded by LINC00665 inhibits triple-negative breast cancer progression. EMBO J,
39(1), e102190. doi:10.15252/embj.2019102190
117
Guo, B., Zhai, D., Cabezas, E., Welsh, K., Nouraini, S., Satterthwait, A. C., & Reed, J. C. (2003).
Humanin peptide suppresses apoptosis by interfering with Bax activation. Nature,
423(6938), 456-461. doi:10.1038/nature01627
Guo, F., Jing, W., Ma, C. G., Wu, M. N., Zhang, J. F., Li, X. Y., & Qi, J. S. (2010). [Gly(14)]-
humanin rescues long-term potentiation from amyloid beta protein-induced impairment in
the rat hippocampal CA1 region in vivo. Synapse, 64(1), 83-91. doi:10.1002/syn.20707
Hardy, S., Kostantin, E., Wang, S. J., Hristova, T., Galicia-Vazquez, G., Baranov, P. V., . . .
Tremblay, M. L. (2019). Magnesium-sensitive upstream ORF controls PRL phosphatase
expression to mediate energy metabolism. Proc Natl Acad Sci U S A, 116(8), 2925-2934.
doi:10.1073/pnas.1815361116
Harris, S. E., Fox, H., Wright, A. F., Hayward, C., Starr, J. M., Whalley, L. J., & Deary, I. J.
(2006). The brain-derived neurotrophic factor Val66Met polymorphism is associated with
age-related change in reasoning skills. Mol Psychiatry, 11(5), 505-513.
doi:10.1038/sj.mp.4001799
Hashimoto, Y., Ito, Y., Niikura, T., Shao, Z., Hata, M., Oyama, F., & Nishimoto, I. (2001).
Mechanisms of neuroprotection by a novel rescue factor humanin from Swedish mutant
amyloid precursor protein. Biochem Biophys Res Commun, 283(2), 460-468.
doi:10.1006/bbrc.2001.4765
Hashimoto, Y., Kurita, M., Aiso, S., Nishimoto, I., & Matsuoka, M. (2009). Humanin inhibits
neuronal cell death by interacting with a cytokine receptor complex or complexes involving
CNTF receptor alpha/WSX-1/gp130. Mol Biol Cell, 20(12), 2864-2873.
doi:10.1091/mbc.E09-02-0168
118
Hashimoto, Y., Niikura, T., Tajima, H., Yasukawa, T., Sudo, H., Ito, Y., . . . Nishimoto, I. (2001).
A rescue factor abolishing neuronal cell death by a wide spectrum of familial Alzheimer's
disease genes and Abeta. Proc Natl Acad Sci U S A, 98(11), 6336-6341.
doi:10.1073/pnas.101133498
He, L., Diedrich, J., Chu, Y. Y., & Yates, J. R., 3rd. (2015). Extracting Accurate Precursor
Information for Tandem Mass Spectra by RawConverter. Anal Chem, 87(22), 11361-
11367. doi:10.1021/acs.analchem.5b02721
Huang, N., Li, F., Zhang, M., Zhou, H., Chen, Z., Ma, X., . . . Zhang, N. (2021). An Upstream
Open Reading Frame in Phosphatase and Tensin Homolog Encodes a Circuit Breaker of
Lactate Metabolism. Cell Metab, 33(2), 454. doi:10.1016/j.cmet.2021.01.008
Hudson, G., Nalls, M., Evans, J. R., Breen, D. P., Winder-Rhodes, S., Morrison, K. E., . . .
Chinnery, P. F. (2013). Two-stage association study and meta-analysis of mitochondrial
DNA variants in Parkinson disease. Neurology, 80(22), 2042-2048.
doi:10.1212/WNL.0b013e318294b434
Hussain, S. A., Yalvac, M. E., Khoo, B., Eckardt, S., & McLaughlin, K. J. (2021). Adapting
CRISPR/Cas9 System for Targeting Mitochondrial Genome. Front Genet, 12, 627050.
doi:10.3389/fgene.2021.627050
Ikonen, M., Liu, B., Hashimoto, Y., Ma, L., Lee, K. W., Niikura, T., . . . Cohen, P. (2003).
Interaction between the Alzheimer's survival peptide humanin and insulin-like growth
factor-binding protein 3 regulates cell survival and apoptosis. Proc Natl Acad Sci U S A,
100(22), 13042-13047. doi:10.1073/pnas.2135111100
Ikonomidis, I., Katogiannis, K., Kyriakou, E., Taichert, M., Katsimaglis, G., Tsoumani, M., . . .
Tsantes, A. E. (2020). β-Amyloid and mitochondrial-derived peptide-c are additive
119
predictors of adverse outcome to high-on-treatment platelet reactivity in type 2 diabetics
with revascularized coronary artery disease. Journal of Thrombosis and Thrombolysis,
49(3), 365-376. doi:10.1007/s11239-020-02060-4
Ingolia, N. T., Ghaemmaghami, S., Newman, J. R., & Weissman, J. S. (2009). Genome-wide
analysis in vivo of translation with nucleotide resolution using ribosome profiling. Science,
324(5924), 218-223. doi:10.1126/science.1168978
International Human Genome Sequencing, C. (2004). Finishing the euchromatic sequence of the
human genome. Nature, 431(7011), 931-945. doi:10.1038/nature03001
Jack, C. R., Jr., Knopman, D. S., Jagust, W. J., Shaw, L. M., Aisen, P. S., Weiner, M. W., . . .
Trojanowski, J. Q. (2010). Hypothetical model of dynamic biomarkers of the Alzheimer's
pathological cascade. Lancet Neurol, 9(1), 119-128. doi:10.1016/S1474-4422(09)70299-6
James, R., Searcy, J. L., Le Bihan, T., Martin, S. F., Gliddon, C. M., Povey, J., . . . Horsburgh, K.
(2012). Proteomic analysis of mitochondria in APOE transgenic mice and in response to
an ischemic challenge. J Cereb Blood Flow Metab, 32(1), 164-176.
doi:10.1038/jcbfm.2011.120
Jenkinson, M., Beckmann, C. F., Behrens, T. E., Woolrich, M. W., & Smith, S. M. (2012). FSL.
Neuroimage, 62(2), 782-790. doi:10.1016/j.neuroimage.2011.09.015
Ji, Z. (2018). RibORF: Identifying Genome-Wide Translated Open Reading Frames Using
Ribosome Profiling. Curr Protoc Mol Biol, 124(1), e67. doi:10.1002/cpmb.67
Jia, Y., Ohanyan, A., Lue, Y.-H., Swerdloff, R. S., Liu, P. Y., Cohen, P., & Wang, C. (2015). The
effects of humanin and its analogues on male germ cell apoptosis induced by
chemotherapeutic drugs. Apoptosis, 20(4), 551-561. doi:10.1007/s10495-015-1105-5
120
Jiang, C., Li, G., Huang, P., Liu, Z., & Zhao, B. (2017). The Gut Microbiota and Alzheimer's
Disease. J Alzheimers Dis, 58(1), 1-15. doi:10.3233/JAD-161141
Kalari, K. R., Nair, A. A., Bhavsar, J. D., O'Brien, D. R., Davila, J. I., Bockol, M. A., . . . Kocher,
J. P. (2014). MAP-RSeq: Mayo Analysis Pipeline for RNA sequencing. BMC
Bioinformatics, 15, 224. doi:10.1186/1471-2105-15-224
Khan, S. M., Cassarino, D. S., Abramova, N. N., Keeney, P. M., Borland, M. K., Trimmer, P. A.,
. . . Bennett, J. P., Jr. (2000). Alzheimer's disease cybrids replicate beta-amyloid
abnormalities through cell death pathways. Ann Neurol, 48(2), 148-155. Retrieved from
https://www.ncbi.nlm.nih.gov/pubmed/10939564
Kim, K. H., Son, J. M., Benayoun, B. A., & Lee, C. (2018). The Mitochondrial-Encoded Peptide
MOTS-c Translocates to the Nucleus to Regulate Nuclear Gene Expression in Response to
Metabolic Stress. Cell Metab, 28(3), 516-524 e517. doi:10.1016/j.cmet.2018.06.008
Kim, S.-J., Miller, B., Kumagai, H., Silverstein, A. R., Flores, M., & Yen, K. (2020).
Mitochondrial-derived peptides in aging and age-related diseases. GeroScience, 1-9.
Kim, S.-J., Miller, B., Kumagai, H., Yen, K., & Cohen, P. (2019). MOTS-c: an equal opportunity
insulin sensitizer. Journal of Molecular Medicine, 97(4), 487-490.
Kim, S. J., Guerrero, N., Wassef, G., Xiao, J., Mehta, H. H., Cohen, P., & Yen, K. (2016). The
mitochondrial-derived peptide humanin activates the ERK1/2, AKT, and STAT3 signaling
pathways and has age-dependent signaling differences in the hippocampus. Oncotarget,
7(30), 46899-46912. doi:10.18632/oncotarget.10380
Koh, M., Ahmad, I., Ko, Y., Zhang, Y., Martinez, T. F., Diedrich, J. K., . . . Bollong, M. J. (2021).
A short ORF-encoded transcriptional regulator. Proc Natl Acad Sci U S A, 118(4).
doi:10.1073/pnas.2021943118
121
Kraja, A. T., Liu, C., Fetterman, J. L., Graff, M., Have, C. T., Gu, C., . . . North, K. E. (2019).
Associations of Mitochondrial and Nuclear Mitochondrial Variants and Genes with Seven
Metabolic Traits. Am J Hum Genet, 104(1), 112-138. doi:10.1016/j.ajhg.2018.12.001
Lakatos, A., Derbeneva, O., Younes, D., Keator, D., Bakken, T., Lvova, M., . . . Alzheimer's
Disease Neuroimaging, I. (2010). Association between mitochondrial DNA variations and
Alzheimer's disease in the ADNI cohort. Neurobiol Aging, 31(8), 1355-1363.
doi:10.1016/j.neurobiolaging.2010.04.031
Lambert, J. C., Ibrahim-Verbaas, C. A., Harold, D., Naj, A. C., Sims, R., Bellenguez, C., . . .
Amouyel, P. (2013). Meta-analysis of 74,046 individuals identifies 11 new susceptibility
loci for Alzheimer's disease. Nat Genet, 45(12), 1452-1458. doi:10.1038/ng.2802
Laurie, C. C., Doheny, K. F., Mirel, D. B., Pugh, E. W., Bierut, L. J., Bhangale, T., . . .
Investigators, G. (2010). Quality control and quality assurance in genotypic data for
genome-wide association studies. Genet Epidemiol, 34(6), 591-602.
doi:10.1002/gepi.20516
Lee, C., Kim, K. H., & Cohen, P. (2016). MOTS-c: A novel mitochondrial-derived peptide
regulating muscle and fat metabolism. Free Radic Biol Med, 100, 182-187.
doi:10.1016/j.freeradbiomed.2016.05.015
Lee, C., Wan, J., Miyazaki, B., Fang, Y., Guevara-Aguirre, J., Yen, K., . . . Cohen, P. (2014). IGF-
I regulates the age-dependent signaling peptide humanin. Aging Cell, 13(5), 958-961.
doi:10.1111/acel.12243
Lee, C., Yen, K., & Cohen, P. (2013). Humanin: a harbinger of mitochondrial-derived peptides?
Trends Endocrinol Metab, 24(5), 222-228. doi:10.1016/j.tem.2013.01.005
122
Lee, C., Zeng, J., Drew, B. G., Sallam, T., Martin-Montalvo, A., Wan, J., . . . Cohen, P. (2015).
The mitochondrial-derived peptide MOTS-c promotes metabolic homeostasis and reduces
obesity and insulin resistance. Cell Metab, 21(3), 443-454.
doi:10.1016/j.cmet.2015.02.009
Lo Buono, V., Palmeri, R., Corallo, F., Allone, C., Pria, D., Bramanti, P., & Marino, S. (2020).
Diffusion tensor imaging of white matter degeneration in early stage of Alzheimer's
disease: a review. Int J Neurosci, 130(3), 243-250. doi:10.1080/00207454.2019.1667798
Lue, Y., Swerdloff, R., Wan, J., Xiao, J., French, S., Atienza, V., . . . Wang, C. (2015). The Potent
Humanin Analogue (HNG) Protects Germ Cells and Leucocytes While Enhancing
Chemotherapy-Induced Suppression of Cancer Metastases in Male Mice. Endocrinology,
156(12), 4511-4521. doi:10.1210/en.2015-1542
Malhi, R. S., Eshleman, J. A., Greenberg, J. A., Weiss, D. A., Schultz Shook, B. A., Kaestle, F.
A., . . . Smith, D. G. (2002). The structure of diversity within New World mitochondrial
DNA haplogroups: implications for the prehistory of North America. Am J Hum Genet,
70(4), 905-919. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/11845406
Manjón, J. V., Coupé, P., Concha, L., Buades, A., Collins, D. L., & Robles, M. (2013). Diffusion
weighted image denoising using overcomplete local PCA. PLoS One, 8(9), e73021.
doi:10.1371/journal.pone.0073021
Martinez, T. F., Chu, Q., Donaldson, C., Tan, D., Shokhirev, M. N., & Saghatelian, A. (2020).
Accurate annotation of human protein-coding small open reading frames. Nat Chem Biol,
16(4), 458-468. doi:10.1038/s41589-019-0425-0
123
McRae, A. F., Byrne, E. M., Zhao, Z. Z., Montgomery, G. W., & Visscher, P. M. (2008). Power
and SNP tagging in whole mitochondrial genome association studies. Genome Res, 18(6),
911-917. doi:10.1101/gr.074872.107
Mercer, T. R., Neph, S., Dinger, M. E., Crawford, J., Smith, M. A., Shearwood, A. M., . . . Mattick,
J. S. (2011). The human mitochondrial transcriptome. Cell, 146(4), 645-658.
doi:10.1016/j.cell.2011.06.051
Miller, B., Arpawong, T. E., Jiao, H., Kim, S.-J., Yen, K., Mehta, H. H., . . . Cohen, P. (2019).
Comparing the utility of mitochondrial and nuclear DNA to adjust for genetic ancestry in
association studies. Cells, 8(4), 306.
Miller, B., Arpawong, T. E., Jiao, H., Kim, S. J., Yen, K., Mehta, H. H., . . . Cohen, P. (2019).
Comparing the Utility of Mitochondrial and Nuclear DNA to Adjust for Genetic Ancestry
in Association Studies. Cells, 8(4). doi:10.3390/cells8040306
Miller, B., Haghani, A., Ailshire, J., & Arpawong, T. E. (2020). Human Population Genetics in
Aging Studies for Molecular Biologists. In Aging (pp. 67-76): Springer.
Miller, B., Kim, S.-J., Kumagai, H., Mehta, H. H., Xiang, W., Liu, J., . . . Cohen, P. (2020).
Peptides derived from small mitochondrial open reading frames: Genomic, biological, and
therapeutic implications. Experimental Cell Research, 112056.
Miller, B., Kim, S. J., Kumagai, H., Mehta, H. H., Xiang, W., Liu, J., . . . Cohen, P. (2020). Peptides
derived from small mitochondrial open reading frames: Genomic, biological, and
therapeutic implications. Exp Cell Res, 112056. doi:10.1016/j.yexcr.2020.112056
Miller, B., Torres, M., Jiang, X., McKean-Cowdin, R., Nousome, D., Kim, S.-J., . . . Varma, R.
(2020). A Mitochondrial Genome-Wide Association Study of Cataract in a Latino
124
Population. Translational Vision Science & Technology, 9(6), 25-25.
doi:10.1167/tvst.9.6.25
Miller, B., & Wan, J. (2020). Assay Development and Measurement of the Aging Biomarker
Humanin. In Aging (pp. 201-209): Springer.
Miller, K. L., Alfaro-Almagro, F., Bangerter, N. K., Thomas, D. L., Yacoub, E., Xu, J., . . . Smith,
S. M. (2016). Multimodal population brain imaging in the UK Biobank prospective
epidemiological study. Nat Neurosci, 19(11), 1523-1536. doi:10.1038/nn.4393
Mouilleron, H., Delcourt, V., & Roucou, X. (2016). Death of a dogma: eukaryotic mRNAs can
code for more than one protein. Nucleic Acids Res, 44(1), 14-23. doi:10.1093/nar/gkv1218
Mudge, J. M., Ruiz-Orera, J., Prensner, J. R., Brunet, M. A., Gonzalez, J. M., Magrane, M., . . .
van Heesch, S. (2021). A community-driven roadmap to advance research on translated
open reading frames detected by Ribo-seq. bioRxiv, 2021.2006.2010.447896.
doi:10.1101/2021.06.10.447896
Muzumdar, R. H., Huffman, D. M., Atzmon, G., Buettner, C., Cobb, L. J., Fishman, S., . . . Cohen,
P. (2009). Humanin: a novel central regulator of peripheral insulin action. PLoS One, 4(7),
e6334. doi:10.1371/journal.pone.0006334
Nashine, S., Cohen, P., Nesburn, A. B., Kuppermann, B. D., & Kenney, M. C. (2018).
Characterizing the protective effects of SHLP2, a mitochondrial-derived peptide, in
macular degeneration. Sci Rep, 8(1), 15175. doi:10.1038/s41598-018-33290-5
Ng, B., Casazza, W., Patrick, E., Tasaki, S., Novakovsky, G., Felsky, D., . . . Mostafavi, S. (2019).
Using Transcriptomic Hidden Variables to Infer Context-Specific Genotype Effects in the
Brain. Am J Hum Genet, 105(3), 562-572. doi:10.1016/j.ajhg.2019.07.016
125
Niikura, T., Sidahmed, E., Hirata-Fukae, C., Aisen, P. S., & Matsuoka, Y. (2011). A Humanin
Derivative Reduces Amyloid Beta Accumulation and Ameliorates Memory Deficit in
Triple Transgenic Mice. PLoS ONE, 6(1), e16259. doi:10.1371/journal.pone.0016259
Nir, T. M., Jahanshad, N., Villalon-Reina, J. E., Isaev, D., Zavaliangos-Petropulu, A., Zhan, L., .
. . (ADNI), A. s. D. N. I. (2017). Fractional anisotropy derived from the diffusion tensor
distribution function boosts power to detect Alzheimer's disease deficits. Magn Reson Med,
78(6), 2322-2333. doi:10.1002/mrm.26623
Novembre, J., Johnson, T., Bryc, K., Kutalik, Z., Boyko, A. R., Auton, A., . . . Bustamante, C. D.
(2008). Genes mirror geography within Europe. Nature, 456(7218), 98-101.
doi:10.1038/nature07331
Okada, A. K., Teranishi, K., Lobo, F., Isas, J. M., Xiao, J., Yen, K., . . . Langen, R. (2017). The
Mitochondrial-Derived Peptides, HumaninS14G and Small Humanin-like Peptide 2,
Exhibit Chaperone-like Activity. Sci Rep, 7(1), 7802. doi:10.1038/s41598-017-08372-5
Omenn, G. S. (2021). Reflections on the HUPO Human Proteome Project, the Flagship Project of
the Human Proteome Organization, at 10 Years. Mol Cell Proteomics, 20, 100062.
doi:10.1016/j.mcpro.2021.100062
Onofre, C., Tome, F., Barbosa, C., Silva, A. L., & Romao, L. (2015). Expression of human
Hemojuvelin (HJV) is tightly regulated by two upstream open reading frames in HJV
mRNA that respond to iron overload in hepatic cells. Mol Cell Biol, 35(8), 1376-1389.
doi:10.1128/MCB.01462-14
Park, T. Y., Kim, S. H., Shin, Y. C., Lee, N. H., Lee, R. K., Shim, J. H., . . . Lee, S. K. (2013).
Amelioration of neurodegenerative diseases by cell death-induced cytoplasmic delivery of
humanin. J Control Release, 166(3), 307-315. doi:10.1016/j.jconrel.2012.12.022
126
Peng, J., Elias, J. E., Thoreen, C. C., Licklider, L. J., & Gygi, S. P. (2003). Evaluation of
multidimensional chromatography coupled with tandem mass spectrometry (LC/LC-
MS/MS) for large-scale protein analysis: the yeast proteome. J Proteome Res, 2(1), 43-50.
doi:10.1021/pr025556v
Pezawas, L., Verchinski, B. A., Mattay, V. S., Callicott, J. H., Kolachana, B. S., Straub, R. E., . . .
Weinberger, D. R. (2004). The brain-derived neurotrophic factor val66met polymorphism
and variation in human cortical morphology. J Neurosci, 24(45), 10099-10102.
doi:10.1523/JNEUROSCI.2680-04.2004
Price, A. L., Zaitlen, N. A., Reich, D., & Patterson, N. (2010). New approaches to population
stratification in genome-wide association studies. Nat Rev Genet, 11(7), 459-463.
doi:10.1038/nrg2813
Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M. A., Bender, D., . . . Sham, P. C.
(2007). PLINK: a tool set for whole-genome association and population-based linkage
analyses. Am J Hum Genet, 81(3), 559-575. doi:10.1086/519795
Qin, Q., Delrio, S., Wan, J., Jay Widmer, R., Cohen, P., Lerman, L. O., & Lerman, A. (2018).
Downregulation of circulating MOTS-c levels in patients with coronary endothelial
dysfunction. Int J Cardiol, 254, 23-27. doi:10.1016/j.ijcard.2017.12.001
Rathore, A., Chu, Q., Tan, D., Martinez, T. F., Donaldson, C. J., Diedrich, J. K., . . . Saghatelian,
A. (2018). MIEF1 Microprotein Regulates Mitochondrial Translation. Biochemistry,
57(38), 5564-5575. doi:10.1021/acs.biochem.8b00726
Reuter, M., Rosas, H. D., & Fischl, B. (2010). Highly accurate inverse consistent registration: a
robust approach. Neuroimage, 53(4), 1181-1196. doi:10.1016/j.neuroimage.2010.07.020
127
Reynolds, J. C., Lai, R. W., Woodhead, J. S. T., Joly, J. H., Mitchell, C. J., Cameron-Smith, D., .
. . Lee, C. (2021). MOTS-c is an exercise-induced mitochondrial-encoded regulator of age-
dependent physical decline and muscle homeostasis. Nature Communications, 12(1).
doi:10.1038/s41467-020-20790-0
Ridge, P. G., Wadsworth, M. E., Miller, J. B., Saykin, A. J., Green, R. C., Alzheimer's Disease
Neuroimaging, I., & Kauwe, J. S. K. (2018). Assembly of 809 whole mitochondrial
genomes with clinical, imaging, and fluid biomarker phenotyping. Alzheimers Dement,
14(4), 514-519. doi:10.1016/j.jalz.2017.11.013
Rooney, J. P., Ryde, I. T., Sanders, L. H., Howlett, E. H., Colton, M. D., Germ, K. E., . . . Meyer,
J. N. (2015). PCR based determination of mitochondrial DNA copy number in multiple
species. Methods Mol Biol, 1241, 23-38. doi:10.1007/978-1-4939-1875-1_3
Ryan, T. M., Caine, J., Mertens, H. D., Kirby, N., Nigro, J., Breheney, K., . . . Roberts, B. R.
(2013). Ammonium hydroxide treatment of Abeta produces an aggregate free solution
suitable for biophysical and cell culture characterization. PeerJ, 1, e73.
doi:10.7717/peerj.73
Ryu, S., Atzmon, G., Barzilai, N., Raghavachari, N., & Suh, Y. (2016). Genetic landscape of
APOE in human longevity revealed by high-throughput sequencing. Mech Ageing Dev,
155, 7-9. doi:10.1016/j.mad.2016.02.010
Saghatelian, A., & Couso, J. P. (2015). Discovery and characterization of smORF-encoded
bioactive polypeptides. Nat Chem Biol, 11(12), 909-916. doi:10.1038/nchembio.1964
Sberro, H., Fremin, B. J., Zlitni, S., Edfors, F., Greenfield, N., Snyder, M. P., . . . Bhatt, A. S.
(2019). Large-Scale Analyses of Human Microbiomes Reveal Thousands of Small, Novel
Genes. Cell, 178(5), 1245-1259 e1214. doi:10.1016/j.cell.2019.07.016
128
Schachter, F., Faure-Delanef, L., Guenot, F., Rouger, H., Froguel, P., Lesueur-Ginot, L., & Cohen,
D. (1994). Genetic associations with human longevity at the APOE and ACE loci. Nat
Genet, 6(1), 29-32. doi:10.1038/ng0194-29
Segonne, F., Dale, A. M., Busa, E., Glessner, M., Salat, D., Hahn, H. K., & Fischl, B. (2004). A
hybrid approach to the skull stripping problem in MRI. Neuroimage, 22(3), 1060-1075.
doi:10.1016/j.neuroimage.2004.03.032
Segonne, F., Pacheco, J., & Fischl, B. (2007). Geometrically accurate topology-correction of
cortical surfaces using nonseparating loops. IEEE Trans Med Imaging, 26(4), 518-529.
doi:10.1109/TMI.2006.887364
Sienski, G., Narayan, P., Bonner, J. M., Kory, N., Boland, S., Arczewska, A. A., . . . Lindquist, S.
(2021). APOE4 disrupts intracellular lipid homeostasis in human iPSC-derived glia. Sci
Transl Med, 13(583). doi:10.1126/scitranslmed.aaz4564
Simonovitch, S., Schmukler, E., Masliah, E., Pinkas-Kramarski, R., & Michaelson, D. M. (2019).
The Effects of APOE4 on Mitochondrial Dynamics and Proteins in vivo. J Alzheimers Dis,
70(3), 861-875. doi:10.3233/JAD-190074
Slavoff, S. A., Mitchell, A. J., Schwaid, A. G., Cabili, M. N., Ma, J., Levin, J. Z., . . . Saghatelian,
A. (2013). Peptidomic discovery of short open reading frame-encoded peptides in human
cells. Nat Chem Biol, 9(1), 59-64. doi:10.1038/nchembio.1120
Sled, J. G., Zijdenbos, A. P., & Evans, A. C. (1998). A nonparametric method for automatic
correction of intensity nonuniformity in MRI data. IEEE Trans Med Imaging, 17(1), 87-
97. doi:10.1109/42.668698
129
Smith, S. M., Jenkinson, M., Johansen-Berg, H., Rueckert, D., Nichols, T. E., Mackay, C. E., . . .
Behrens, T. E. (2006). Tract-based spatial statistics: voxelwise analysis of multi-subject
diffusion data. Neuroimage, 31(4), 1487-1505. doi:10.1016/j.neuroimage.2006.02.024
Smith, S. M., Jenkinson, M., Woolrich, M. W., Beckmann, C. F., Behrens, T. E., Johansen-Berg,
H., . . . Matthews, P. M. (2004). Advances in functional and structural MR image analysis
and implementation as FSL. Neuroimage, 23 Suppl 1, S208-219. Retrieved from
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citati
on&list_uids=15501092
Smith, S. M., & Nichols, T. E. (2009). Threshold-free cluster enhancement: addressing problems
of smoothing, threshold dependence and localisation in cluster inference. Neuroimage,
44(1), 83-98. doi:10.1016/j.neuroimage.2008.03.061
Soyal, S. M., Kwik, M., Kalev, O., Lenz, S., Zara, G., Strasser, P., . . . Weis, S. (2020). A
TOMM40/APOE allele encoding APOE-E3 predicts high likelihood of late-onset
Alzheimer's disease in autopsy cases. Mol Genet Genomic Med, 8(8), e1317.
doi:10.1002/mgg3.1317
Stein, C. S., Jadiya, P., Zhang, X., McLendon, J. M., Abouassaly, G. M., Witmer, N. H., . . .
Boudreau, R. L. (2018). Mitoregulin: A lncRNA-Encoded Microprotein that Supports
Mitochondrial Supercomplexes and Respiratory Efficiency. Cell Rep, 23(13), 3710-3720
e3718. doi:10.1016/j.celrep.2018.06.002
Tabb, D. L., McDonald, W. H., & Yates, J. R., 3rd. (2002). DTASelect and Contrast: tools for
assembling and comparing protein identifications from shotgun proteomics. J Proteome
Res, 1(1), 21-26. doi:10.1021/pr015504q
130
Tajima, H., Kawasumi, M., Chiba, T., Yamada, M., Yamashita, K., Nawa, M., . . . Nishimoto, I.
(2005). A humanin derivative, S14G-HN, prevents amyloid-beta-induced memory
impairment in mice. J Neurosci Res, 79(5), 714-723. doi:10.1002/jnr.20391
Tournier, J. D., Smith, R., Raffelt, D., Tabbara, R., Dhollander, T., Pietsch, M., . . . Connelly, A.
(2019). MRtrix3: A fast, flexible and open software framework for medical image
processing and visualisation. Neuroimage, 202, 116137.
doi:10.1016/j.neuroimage.2019.116137
Tracy, T. E., Madero-Perez, J., Swaney, D. L., Chang, T. S., Moritz, M., Konrad, C., . . . Gan, L.
(2022). Tau interactome maps synaptic and mitochondrial processes associated with
neurodegeneration. Cell. doi:10.1016/j.cell.2021.12.041
Tranah, G. J., Nalls, M. A., Katzman, S. M., Yokoyama, J. S., Lam, E. T., Zhao, Y., . . . Yaffe, K.
(2012). Mitochondrial DNA sequence variation associated with dementia and cognitive
function in the elderly. J Alzheimers Dis, 32(2), 357-372. doi:10.3233/JAD-2012-120466
Tsukamoto, E., Hashimoto, Y., Kanekura, K., Niikura, T., Aiso, S., & Nishimoto, I. (2003).
Characterization of the toxic mechanism triggered by Alzheimer's amyloid-beta peptides
via p75 neurotrophin receptor in neuronal hybrid cells. J Neurosci Res, 73(5), 627-636.
doi:10.1002/jnr.10703
van Oven, M., & Kayser, M. (2009). Updated comprehensive phylogenetic tree of global human
mitochondrial DNA variation. Hum Mutat, 30(2), E386-394. doi:10.1002/humu.20921
Ventriglia, M., Bocchio Chiavetto, L., Benussi, L., Binetti, G., Zanetti, O., Riva, M. A., &
Gennarelli, M. (2002). Association between the BDNF 196 A/G polymorphism and
sporadic Alzheimer's disease. Mol Psychiatry, 7(2), 136-137. doi:10.1038/sj.mp.4000952
131
Wang, X., Wu, Z., He, Y., Zhang, H., Tian, L., Zheng, C., . . . He, Y. (2018). Humanin prevents
high glucose-induced monocyte adhesion to endothelial cells by targeting KLF2. Mol
Immunol, 101, 245-250. doi:10.1016/j.molimm.2018.07.008
Webb-Robertson, B. J., Wiberg, H. K., Matzke, M. M., Brown, J. N., Wang, J., McDermott, J. E.,
. . . Waters, K. M. (2015). Review, evaluation, and discussion of the challenges of missing
value imputation for mass spectrometry-based label-free global proteomics. J Proteome
Res, 14(5), 1993-2001. doi:10.1021/pr501138h
Weedon, M. N., & Frayling, T. M. (2008). Reaching new heights: insights into the genetics of
human stature. Trends Genet, 24(12), 595-603. doi:10.1016/j.tig.2008.09.006
Wei, M., Gan, L., Liu, Z., Liu, L., Chang, J.-R., Yin, D.-C., . . . Wanli. (2020). Mitochondrial-
Derived Peptide MOTS-c Attenuates Vascular Calcification and Secondary Myocardial
Remodeling via Adenosine Monophosphate-Activated Protein Kinase Signaling Pathway.
Cardiorenal Medicine, 10(1), 42-50. doi:10.1159/000503224
Widlansky, M. E., Gokce, N., Keaney, J. F., Jr., & Vita, J. A. (2003). The clinical implications of
endothelial dysfunction. J Am Coll Cardiol, 42(7), 1149-1160. doi:10.1016/s0735-
1097(03)00994-x
Widmer, R. J., Flammer, A. J., Herrmann, J., Rodriguez-Porcel, M., Wan, J., Cohen, P., . . .
Lerman, A. (2013). Circulating humanin levels are associated with preserved coronary
endothelial function. Am J Physiol Heart Circ Physiol, 304(3), H393-397.
doi:10.1152/ajpheart.00765.2012
Worsley, K. J., Evans, A. C., Marrett, S., & Neelin, P. (1992). A three-dimensional statistical
analysis for CBF activation studies in human brain. J Cereb Blood Flow Metab, 12(6), 900-
918. doi:10.1038/jcbfm.1992.127
132
Xiao, J., Howard, L., Wan, J., Wiggins, E., Vidal, A., Cohen, P., & Freedland, S. J. (2017). Low
circulating levels of the mitochondrial-peptide hormone SHLP2: novel biomarker for
prostate cancer risk. Oncotarget, 8(55), 94900-94909. doi:10.18632/oncotarget.20134
Xu, T., Park, S. K., Venable, J. D., Wohlschlegel, J. A., Diedrich, J. K., Cociorva, D., . . . Yates,
J. R., 3rd. (2015). ProLuCID: An improved SEQUEST-like algorithm with enhanced
sensitivity and specificity. J Proteomics, 129, 16-24. doi:10.1016/j.jprot.2015.07.001
Xu, X., Chua, C. C., Gao, J., Hamdy, R. C., & Chua, B. H. (2006). Humanin is a novel
neuroprotective agent against stroke. Stroke, 37(10), 2613-2619.
doi:10.1161/01.STR.0000242772.94277.1f
Yan, Y., Tao, H., He, J., & Huang, S. Y. (2020). The HDOCK server for integrated protein-protein
docking. Nat Protoc, 15(5), 1829-1852. doi:10.1038/s41596-020-0312-x
Yen, K., Lee, C., Mehta, H., & Cohen, P. (2013). The emerging role of the mitochondrial-derived
peptide humanin in stress resistance. J Mol Endocrinol, 50(1), R11-19. doi:10.1530/JME-
12-0203
Yen, K., Mehta, H. H., Kim, S.-J., Lue, Y., Hoang, J., Guerrero, N., . . . Cohen, P. (2020). The
mitochondrial derived peptide humanin is a regulator of lifespan and healthspan. Aging,
12(12), 11185-11199. doi:10.18632/aging.103534
Yen, K., Mehta, H. H., Kim, S. J., Lue, Y., Hoang, J., Guerrero, N., . . . Cohen, P. (2020). The
mitochondrial derived peptide humanin is a regulator of lifespan and healthspan. Aging
(Albany NY), 12. doi:10.18632/aging.103534
Yen, K., Wan, J., Mehta, H. H., Miller, B., Christensen, A., Levine, M. E., . . . Kim, S.-J. (2018).
Humanin prevents age-related cognitive decline in mice and is associated with improved
cognitive age in humans. Scientific reports, 8(1), 1-10.
133
Yen, K., Wan, J., Mehta, H. H., Miller, B., Christensen, A., Levine, M. E., . . . Cohen, P. (2018).
Humanin Prevents Age-Related Cognitive Decline in Mice and is Associated with
Improved Cognitive Age in Humans. Sci Rep, 8(1), 14212. doi:10.1038/s41598-018-
32616-7
Yonova-Doing, E., Calabrese, C., Gomez-Duran, A., Schon, K., Wei, W., Karthikeyan, S., . . .
Howson, J. M. M. (2021). An atlas of mitochondrial DNA genotype-phenotype
associations in the UK Biobank. Nat Genet, 53(7), 982-993. doi:10.1038/s41588-021-
00868-1
Yu, G., Wang, L. G., Han, Y., & He, Q. Y. (2012). clusterProfiler: an R package for comparing
biological themes among gene clusters. OMICS, 16(5), 284-287.
doi:10.1089/omi.2011.0118
Zarate, S. C., Traetta, M. E., Codagnone, M. G., Seilicovich, A., & Reines, A. G. (2019). Humanin,
a Mitochondrial-Derived Peptide Released by Astrocytes, Prevents Synapse Loss in
Hippocampal Neurons. Front Aging Neurosci, 11, 123. doi:10.3389/fnagi.2019.00123
Zempo, H., Kim, S. J., Fuku, N., Nishida, Y., Higaki, Y., Wan, J., . . . Cohen, P. (2021). A pro-
diabetogenic mtDNA polymorphism in the mitochondrial-derived peptide, MOTS-c. Aging
(Albany NY), 13(2), 1692-1717. doi:10.18632/aging.202529
Zhang, S., Reljic, B., Liang, C., Kerouanton, B., Francisco, J. C., Peh, J. H., . . . Ho, L. (2020).
Mitochondrial peptide BRAWNIN is essential for vertebrate respiratory complex III
assembly. Nat Commun, 11(1), 1312. doi:10.1038/s41467-020-14999-2
Zhang, W., Du, Y., Bai, M., Xi, Y., Li, Z., & Miao, J. (2013). S14G-humanin inhibits Abeta1-42
fibril formation, disaggregates preformed fibrils, and protects against Abeta-induced
cytotoxicity in vitro. J Pept Sci, 19(3), 159-165. doi:10.1002/psc.2484
134
Zhang, W., Zhang, W., Li, Z., Hao, J., Zhang, Z., Liu, L., . . . Zhang, L. (2012). S14G-humanin
improves cognitive deficits and reduces amyloid pathology in the middle-aged
APPswe/PS1dE9 mice. Pharmacol Biochem Behav, 100(3), 361-369.
doi:10.1016/j.pbb.2011.09.012
Zhang, Z., & Castello, A. (2017). Principal components analysis in clinical studies. Ann Transl
Med, 5(17), 351. doi:10.21037/atm.2017.07.12
Zhao, L., Batta, I., Matloff, W., O'Driscoll, C., Hobel, S., & Toga, A. W. (2020). Neuroimaging
PheWAS (Phenome-Wide Association Study): A Free Cloud-Computing Platform for Big-
Data, Brain-Wide Imaging Association Studies. Neuroinformatics. doi:10.1007/s12021-
020-09486-4
Zhao, L., Matloff, W., Ning, K., Kim, H., Dinov, I. D., & Toga, A. W. (2019). Age-Related
Differences in Brain Morphology and the Modifiers in Middle-Aged and Older Adults.
Cereb Cortex, 29(10), 4169-4193. doi:10.1093/cercor/bhy300

Abstract (if available)

Abstract Microproteins are a new class of bioactive peptides, but the genes that encode these microproteins have escaped biologists for decades. Now, due to enhanced technology and high volume of ‘omics’ data, microprotein discoveries are growing in number. This thesis highlights an integrative microprotein discovery approach that consists of population genetics, transcriptomics, proteomics, and in vitro and in vivo experiments. The goal of this thesis was to identify and functionalize new microproteins encoded by the mitochondrial genome. In chapter 1, we introduce the biological significance of microproteins with an emphasis on human genetic variation. In chapter 2, we present analyses that illustrate the significant degree of human mitochondria DNA variation in population cohorts – these analyses are the foundation upon which we developed our microprotein-specific population genetics methods. In chapter 3, we carry out a mitochondrial genome wide association study to identify microprotein-encoding DNA regions that associate with Alzheimer’s disease (AD). Indeed, we show that an AD-risk mitochondria single nucleotide polymorphism changes the amino acid sequence of a previously unannotated microprotein called SHMOOSE, a peptide that we ultimately detected in neuronal mitochondria and cerebrospinal fluid. SHMOOSE is the first reported microprotein encoded by mitochondria DNA that has been detected by mass spectrometry. In chapter 4, we reveal a rare mitochondria DNA variant that changes the third amino acid of the microprotein humanin and protects against APOE4-related decline. Simultaneously, we show that humanin binds APOE and – notably – the APOE4-associative variant binds APOE 15 times greater and dramatically attenuates AD pathology in the APP/PS1/APOE4 mouse model. Altogether, our microprotein results could impact gene annotation, AD therapeutic development, AD biomarker development, and AD experimental models.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The intersection of mitochondrial biology and cancer: insights from mitochondrial microproteins and mtDNA alterations

PDF

The regulation, roles, and mechanism of action of mitochondrial-derived-peptides (MDPs) in aging

PDF

The effects of a fasting mimicking diet (FMD) on mouse models of Alzheimer's and Parkinson's disease

PDF

Modeling neurodegenerative diseases using induced pluripotent stem cells and identifying therapeutic targets

PDF

The role of vascular dysfunction in cognitive impairment

PDF

Sex differences in aging and the effects of mitochondria

PDF

Neuroinflammation and ApoE4 genotype in at-risk female aging: implications for Alzheimer's disease

PDF

Air pollution, mitochondrial function, and growth in children

PDF

TLR4-mediated innate immune response and neuroinflammation: focus on APOE and obesity

PDF

From risk mitochondrial and metabolic phenotype towards a precision medicine approach for Alzheimer's disease

PDF

Air pollution neurotoxicity throughout the lifespan: studies on the mechanism of toxicity and interactions with effects of sex and genetic background

PDF

Phenotypic and multi-omic characterization of novel C. elegans models of Alzheimer's disease

PDF

Studies of intracellular cascades mediating neuronal damage in two animal models of neurodegeneration

PDF

Sex differences in the TgF344-AD rat model: Investigating the behavior, pathology, and neuroanatomical structures

PDF

Blood-brain barrier pathophysiology in cognitive impairment and injury

PDF

Associations of ApoE4 status and DHA supplementation on plasma and CSF lipid profiles and entorhinal cortex thickness

PDF

The role of inflammation in mediating effects of obesity on Alzheimer's disease

PDF

Micelle nanoparticles for the targeting and treatment of autosomal dominant polycystic kidney disease

PDF

Neuroimaging in complex polygenic disorders

PDF

Nebula/DSCR1 upregulation preserves axonal transport and memory function in a Drosophila model for Alzheimer's disease

Asset Metadata

Creator Miller, Brendan (author)

Core Title Discovery of mitochondria-DNA-encoded microproteins in neurodegeneration

School Leonard Davis School of Gerontology

Degree Doctor of Philosophy

Degree Program Neuroscience

Degree Conferral Date 2022-08

Publication Date 07/25/2024

Defense Date 05/10/2022

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

Tag Alzheimer's disease,humanin,microprotein,mitochondria DNA,mitochondrial-derived peptides,neuroimaging,OAI-PMH Harvest,SHMOOSE

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Cohen, Pinchas (committee chair), Pike, Christian (committee chair), Braskie, Meredith (committee member), Lee, Changhan (committee member), Tower, John (committee member)

Creator Email brendajm@usc.edu,brendancohenlab@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111375219

Unique identifier UC111375219

Legacy Identifier etd-MillerBren-10980

Document Type Dissertation

Format application/pdf (imt)

Rights Miller, Brendan

Type texts

Source 20220728-usctheses-batch-962 (batch), 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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

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

Repository Email cisadmin@lib.usc.edu