University of Southern California Dissertations and Theses

Characterization of the STARS (STochastic gene Activation with Regulated Sparseness) mouse line

(USC Thesis Other)

Characterization of the STARS (STochastic gene Activation with Regulated Sparseness) mouse line

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Page 1 of 40

Characterization of the STARS (STochastic
gene Activation with Regulated Sparseness)
mouse line
By
Junxiang Huang
Mentor: Li Zhang

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfilment of the Requirements for the Degree
MASTER OF SCIENCE
(Biochemistry and Molecular Biology)
December 2017
Page 2 of 40

Acknowledgement
I would like to express my special appreciation and thanks to my adviser Dr. Li I Zhang
and Dr. Huizhong Whit Tao for your constant encouragement and guidance towards
research and writing this thesis. Thank you for giving me this opportunity to participate
in this amazing lab. I would also like to thank other thesis committee members Dr.
Jeanine Chen, Dr. Zoltan Tokes and Dr. Ralf Langen.
I would like to express my special appreciation to Dr. Leena Ali Ibrahim for her
kindness, support and lead with the project. I would like to thank all members in the lab
for being a wonderful supportive team. I would also thank Lei Peng from Bonaguidi lab
for her suggestion on tamoxifen administration.
Thank you everyone for your support!

Page 3 of 40

Table of Content

Acknowledgement ..................................................................................................................... 2
Table of Content ........................................................................................................................ 3
List of Figures ........................................................................................................................... 4
Abstract ..................................................................................................................................... 5
Introduction ............................................................................................................................... 6
Results ..................................................................................................................................... 13
The design of STARS mouse. ............................................................................................. 13
Sparseness quantification of STARS driven by constitutive Cre lines................................ 15
Sparseness quantification of STARS driven by inducible Cre lines. .................................. 19
Transform the STARS reporter into an adapter line. ........................................................... 25
Discussion ............................................................................................................................... 28
Leakiness of Camk2a-CreERT2 lines ................................................................................. 28
To improve specificity of FLP-DOG .................................................................................. 29
Other sparse labeling methods ............................................................................................. 30
Materials and Methods ............................................................................................................ 32
Contributions ........................................................................................................................... 37
References ............................................................................................................................... 38

Page 4 of 40

List of Figures

Figure 1. Flowchart shows the strategy for genetically targeted labeling. (Page 8)
Figure 2. Schematic figure shows the design of STARS transgene. (Page 11)
Figure 3. Summary of STARS construct with different loxP variants and spacers of various
lengths. (Page 15)
Figure 4. Images of a STARS mouse under different light conditions. (Page 17)
Figure 5. Brain images of Rbp4-Cre mice crossed with Ai14 or STARS mice. (Page 17)
Figure 6. Brain images of PV-Cre mice crossed with Ai14 or STARS mice. (Page 18)
Figure 7. Cochlea images of PV-Cre mice crossed with Ai14 or STARS mice. (Page 18)
Figure 8. Sparseness of STARS driven by constitutive Cre lines. (Page 19)
Figure 9. Brain images of Camk2a-CreERT2 mice crossed with Ai14 or STARS mice. (Page 21)
Figure 10. Camk2a-CreERT2;STARS mice injected with 4-hydroxyl-tamoxifen. (Page 22)
Figure 11. Sparseness of STARS driven by the CamK2a-CreERT2 line. (Page 23)
Figure 12. Example individual neurons labeled in CamK2a-CreERT2;STARS. (Page 23)
Figure 13. Transform STARS into an adapter mouse line. (Page 24)
Figure 14. Images of 488nm-excited fluorescent signal from FLP-DOG and fDIO-eYFP co-
injected mice. (Page 26)
Figure 15. Images from the wildtype mice injected with fDIO-eYFP only. (Page 27)

Page 5 of 40

Abstract
In order to keep track of the morphology and activity of individual cells in highly
compact tissues, such as the brain, genetic methods that could automatically achieve
sparse labeling are highly demanded. Here, we present the STARS transgenic mouse as a
promising tool to visualize the detailed morphology of neurons in a genetically
determined population. The STARS mouse is a Cre-dependent reporter line with two
fluorescent reporter proteins expressed in an exclusive manner. Unlike traditional strains,
the activation of genes in STARS is under the control of an unequal OR logic gate. Thus,
in the presence of Cre recombinase, the two transgenes with different lengths would have
different probabilities to be expressed. In our case, one of the transgenes, mYFP, would
stably label only about 10% of neurons when it is driven by various constitutive Cre-
expressing lines. When it is driven by an inducible Cre-expressing line, the sparseness
could range from a few labeled cells per section to about 6% in maximum. The labeling
ratio is consistent across neuronal types and brain regions tested. With the help of a FLP-
DOG virus, we could further express FLP recombinase in YFP-positive cells, allowing
later genetic manipulations coupled with morphological characterization.

Page 6 of 40

Introduction
The functionality of the brain builds on its architecture. To understand how the brain
works, it is the very first step to figure out the anatomical basis of the nervous system.
Given the fact that mammalian brains consist of numerous cells and various cell types, it
is still a great challenge to create a “brain map” illustrating the anatomical location and
projection of neurons with single-cell resolution. Although a comprehensive blueprint of
the brain is not yet available, many approaches have been developed to reveal the
principle of the brain organization.
The first milestone of neuroanatomy was achieved by Santiago Ramón y Cajal in the
early 20
th
century. By improving the silver staining method invented Camillo Golgi,
Cajal was able to randomly label a small subset of brain cells, and therefore demonstrate
that those cells are the basic building units of the brain, termed neurons. Neuron doctrine
thus serves as a basic theory of the modern neuroscience
1
. Neurons are interconnected to
each other with their fiber-like processes, called axons or dendrites. With Golgi stain,
people can characterize and classify neurons based on their morphological properties.
Later, thanks to the development of electrophysiology, genetics and molecular biology,
neurons are further classified according to their electrical firing properties and
transcriptomes. This largely fills in the gap between the morphology and functionality of
neurons.
Page 7 of 40

Up to date, the strategies for genetic targeting are still simple and straightforward (Figure
1A): it requires a driver, which could be a promoter, transcription factor or recombinase,
to initiate the expression of a reporter, which could be a fluorescent protein, ion channel,
enzyme or gene cassette. Traditionally, the drivers are the promoters of key enzymes for
the synthesis of certain neurotransmitters, such as ChAT (Choline acetyltransferase) and
Gad (Glutamate decarboxylase)
2
. This strategy has been a great success for grouping and
targeting neurons based on their innate functionality. However, unlike other somatic
cells, neurons even with a similar transcriptomic background, could have distinct
functions within the same neural circuits due to their distinct connectivity patterns. This
phenomenon poses challenges to scientists who want to keep track of the morphology
and projection pattern of individual neurons. Because the conditions that can be utilized
to satisfy the requirement for the expression of the reporter is always limited, usually less
than 3, the reporter is usually activated in a large number of neurons within a small area.
This would simply lead to the scenario that scientists usually find themselves lose track
of the number or projections of neurons, for the simple reason that neuronal processes are
easily densely packed under the microscope. Furthermore, for those functional imaging
studies, for instance calcium imaging, the spatially overlapping neurons under the scope
could be a major source of confounding signals. Because the Golgi stain is purely a
chemical staining method, it could not be further combined with genetic manipulation.
Page 8 of 40

With the need to distinctively label neurons, a genetic method that could intrinsically
achieve sparse labeling is highly demanded.

Figure 1. Flowchart shows the strategy for genetically targeted labeling.
So, the question we need to address becomes how to create sparseness genetically? Many
efforts have been put into developing sparse labeling and there are multiple methods
available. These include CreER
3
, SLICK
4
, Brainbow
5,6
, MADM
7
, Supernova
8
,
SLENDER
9
and etc. They could be grouped according to the transcription/translation
level at which sparseness is determined.
Post-translationally, sparseness could be created by CreER, an engineered Cre
recombinase fused with modified estrogen receptor domains. The fusion protein would
only translocate into the nucleus and achieve recombination in the presence of its
Page 9 of 40

artificial ligand, 4-hydroxyl-tamoxifen (4-HT)
10
. The mouse model utilizing this strategy
was first created in 2002 and later became the first widely applied tool for sparsely
labeling
3
. It enables researchers to control both the ratio and the timing of labeling event.
The shortcoming of this tool comes from the fact that the percentage of labeled cells is
not linearly dependent on the dose of 4-HT. In many cases, an ideal sparseness is hard to
be predicted and repeated. Also, because of the instability of 4-HT, researchers usually
treat animals with tamoxifen (TMX), which would be later metabolized into 4-HT by the
liver. The variability of the liver function among individual animals thus further hinders
the repeatability.
At the transcriptional level, strategies usually aim to reduce the possibility of mRNA
synthesis. SLICK
4
and Supernova
8
are two of those. SLICK takes advantage of the
“position-effect variegation” of transgene to reduce expression level, while Supernova
relies on the intrinsic low level of leaky expression from tetracycline-inducible promoter.
The apparent drawback of these approaches is that the sparseness could not be regulated
and its level among cell types could be drastically different. To partially overcome this
limitation, Supernova is recommended to be introduced by in utero electroporation (IUE)
so that the copy number of the transgene could be regulated to a certain degree. So far
neither of these techniques has been widely applied.
Page 10 of 40

To some extent, the strategy focusing on generating sparseness transcriptionally or
translationally would always be limited by the intrinsic randomness and variability
among these processes. Therefore, we want to develop methods that create sparseness at
the DNA level. The idea behind this type of methods is that if it is tricky to manipulate
the level of output, why don’t we try to vary the number of possible outputs (Figure 1B).
The methods MADM
7
and Brainbow
5,6
are therefore proposed. The idea behind these
methods is that neighboring neurons could be assigned different output values if the
number of possible output values is great enough. Then, individual cells could be
identified and characterized unambiguously. MADM is to recombine a pair of loci
enabling fluorescent protein expression between sister chromosomes and therefore would
generate 4 types of possible output. With this technique, researchers could trace the
lineage of individual progenitor cells. As for the renowned Brainbow method, it generates
8 possible spectral colors by excising/silencing random number of its 3 fluorescent
reporters. The limited number of color outputs is an obvious reason that MADM is more
powerful for developmental studies in which the parental relationship of cells is critical.
The shortcoming of Brainbow comes from more of a technical aspect: the variation
among section thickness, depth of fluorescent source, exposure time and etc. could
significantly affect the reliability of color judgement. Furthermore, an average sparseness
of about 10% is not high enough for the detailed characterization of neuronal morphology
in some cases.
Page 11 of 40

With the above issues in mind, we wondered whether it is possible to create sparse
labeling starting from the beginning of transcription cascade and later reported a strategy
called STARS (STochastic gene Activation with Regulated Sparseness) in mammalian
cell lines (Figure 2)
11
. Here, we would like to describe and characterize a transgenic
mouse line, of which ROSA26 locus is inserted with the STARS transgene.

Figure 2. Schematic figure shows the design of STARS transgene.

The STARS mouse is a Cre-dependent reporter line. In the current version of STARS
mouse, the reporter genes are mCherry and membrane-bound yellow fluorescent protein
(mYFP). Unlike traditional reporter strains, the activation of gene of interest in STARS is
under the control of an unequal OR logic gate. The OR gate is achieved by flanking two
sets of gene of interest with two pairs of lox sites, which are interlocked (Figure 2). In the
presence of Cre recombinase, one of the reporter gene would be incised, and the
recombination between the other pair is excluded. Because the two flanked cassettes have
Page 12 of 40

different lengths, the two reporter genes would have different probabilities to be
expressed. The current version of STARS transgene incorporated contains an overall 8
kilo-bases (kb) of spacer, whereas mYFP is about 1 kb. Thus, when the STARS mouse is
crossed with constitutive Cre lines (PV-cre, Vglut2-cre, Math2-cre), the sparseness of
mYFP positive cells is about 10%. Moreover, we explored the possibility that sparseness
could be further decreased by crossing the STARS with a CreER line. Expectedly, the
sparseness of labeled cells could be further fine-tuned down to less than 1% of the
population. This is especially useful to characterize those neuronal types that is enormous
in number, for example in our experiment with CamK2a-positive neurons. Thanks to the
low labeling ratio, we are able to track individual type II spiral ganglion fibers in the
cochlea.
Moreover, we came to realize that it is not good enough to just label cells with
fluorescent indicators, which do not have any biological impact. In order to broaden the
application of STARS with various types of effector protein, we tried to transform the
STARS mouse into a what we call “Adapter” line by injecting adeno-associated virus
(AAV) expressing a GFP-dependent FLP recombinase. In other words, only a small
subset of cells expressing Cre can also co-express FLP. Researchers can further introduce
FLP-dependent transgenes encoding (e.g. ion channels, enzymes, receptor) for functional
studies.
Page 13 of 40

Results
The design of STARS mouse.
First discovered in 1981
12
, thanks to its robustness across species, the Cre recombinase
has become a predominant “gene tailor” in numerous transgenic animals
13
. Up to date,
there are nearly 700 Cre-expressing and more than 250 Cre-reporter mouse lines created
and distributed all over the world (Data from the website of The Jackson Laboratory). For
a long time, the Cre-loxP system is one of the most popular gene manipulation
technologies and widely used in various research fields including neuroscience. The
primary usage of Cre-loxP system for neuroscientists is to specifically access to a group
of neuron with identified molecular profile. The Cre is most commonly used as a “key” to
switch on/off the expression cassette built in the reporter mouse lines. And this key is
under the control of a promoter of interest. Due to the common transcriptomic expression
among the selected group of neurons, many neurons would be labeled at the same time.
This property is especially helpful for the studies mapping functional brain circuit,
because it usually requires many neurons acting in orchestra to give an output that is
sufficiently strong to drive its downstream target. However, this also turns into a trouble
when it comes to study the anatomical properties of neurons because the processes of
labeled neurons tend to overlap with each other under a regular optic microscope. In
order to reliably trace the projection of single neurons, it would take significant amount
of efforts when using higher magnification imaging methods, e.g. electron microscopy
Page 14 of 40

(EM)
14
, due to the huge amount of data and computationally intense data analysis. The
simplest way to resolve this problem is to achieve sparse labeling in the tissue. As it is
mentioned above, there is not many reliable and flexible genetic methods for this
purpose.
Here, we would like to introduce a knock-in transgenic mouse line, also named STARS,
that carries a STARS transgene that would only sparsely label cells with eYFP in a Cre-
dependent manner. In our previous study, we have demonstrated that the STARS
transgene could expression either of the reporter genes with distinct designed
probabilities. Because the recombination mediated by Cre recombinase depends on both
the sequence of loxP site variants and distance between loxP sites, the probability of each
gene being expressed in different STARS constructs was first tested in HEK293 cells. As
shown in the figure 3, the ratio of cells expressing mYFP (assuming that the efficiency of
transfecting Cre is close to 100%) drops as the length of the spacer increases. When the
length reaches over 8kb, it becomes possible to reach a high level of sparseness. On the
other hand, the loxP variants also have a noticeable effect on the mYFP-positive cell
ratio. Based on these in vitro data, and to balance the technical difficulty of inserting long
transgenes, the STARS transgene for making the transgenic mouse was designed to have
a 7.2 kb spacer together with mCherry between the first pair of loxP sites and with
lox3172 being the second pair of loxP variants.
Page 15 of 40

Figure 3. Sparseness of STARS construct with different loxP variants and spacers
lengths.
Later, the STARS mouse was generated by using the TARGATT technology (Applied
StemCell, Inc). STARS transgene was knocked into the ROSA26 locus that harbors an
attP site with the help of PhiC31 integrase in TARGATT embryonic stem cell line. The
STARS mouse was confirmed by PCR genotyping (Data not shown). To our surprise, the
mouse has a prominent feature that the mCherry expressed by most of the cells is so
bright that we could identify STARS-positive mice by shining green LED light on the
body. Strong red fluorescence could be observed from the regions not covered by fur,
such as the ears, tail and palms (Figure 4). Later genotyping of the STARS mouse can be
visually achieved and we have never found any case of false-positive mouse using this
method.

Sparseness quantification of STARS driven by constitutive Cre lines.
This version of STARS mouse is a Cre-dependent reporter line. So, we firstly tested
whether consistent sparse labeling could be achieved in STARS mice when transgene
Page 16 of 40

expression is driven by different tissue/cell type specific Cre lines. We want to test
whether STARS would work equally fine in different neuronal types, e.g. excitatory and
inhibitory neurons. We crossed the STARS mouse with various driver lines including
Rbp4-Cre (labeling excitatory layer 5 neurons), PV-IRES-Cre and VIP-IRES-Cre
(labeling subtypes of inhibitory neurons). Example images are shown in Figure 5 and 6.
As you can see, the cells expressing mYFP scattered in different tissues. Moreover,
because the STARS transgene is driven by a synthetic universal promoter (CAG
promoter), we also want to check whether beyond the central nervous system a similar
level of sparse labeling could also be achieved. Images of the cochlea were collected as
well (Figure 7). Ai14 mouse is a general Cre-dependent reporter line that would express
tdTomato. Sparseness was defined as the ratio of GFP-positive cell density in STARS to
tdTomato-positive cell density in Ai14. The data were collected and quantified together
with a former member from the lab, Dr. Leena Ali Ibrahim. Overall, the sparseness of
STARS mouse driven by a constitutive Cre-expressing line ranged from 8-12 percent
regardless of cell type or location (Figure 8).
Page 17 of 40

Figure 4. Images of a STARS mouse under different light conditions.
The left image was taken under regular sunlight lamp. The right image was illuminated
by green LED and with a red filter before the camera.

Figure 5. Brain images of Rbp4-Cre mice crossed with Ai14 or STARS mice.
Excitatory layer 5 neurons were labeled by mYFP or tdTomato in the images. The upper
images shows the expression of tdTomato in Rbp4-Cre;Ai14 mice. The lower images
shows the expression of mYFP in Rbp4-Cre;STARS mice. V1: primary visual cortex.
Hippo: hippocampus. Scale bar: 100 µm.

Page 18 of 40

Figure 6. Brain images of PV-Cre mice crossed with Ai14 or STARS mice.
Parvalbumin (PV)-positive inhibitory neurons were labeled by mYFP or tdTomato in the
images. The upper images shows the expression of tdTomato in PV-Cre;Ai14 mice. The
lower images shows the expression of mYFP in PV-Cre;STARS mice. V1: primary
visual cortex. Hippo: hippocampus. Scale bar: 100 µm.

Figure 7. Cochlea images of PV-Cre mice crossed with Ai14 or STARS mice.
Spiral ganglions, inner and outer hair cells were labeled by mYFP or tdTomato in the
images. The upper images shows the expression of tdTomato in PV-Cre;Ai14 mice. The
lower images shows the expression of mYFP in PV-Cre;STARS mice. Scale bar: 100
µm.
Page 19 of 40

Figure 8. Sparseness of STARS driven by constitutive Cre lines.
The y-axis shows the sparseness level (defined as the percentage ratio of GFP-positive
cell density in STARS to tdTomato-positive cell density in Ai14). The x-axis and the title
of charts indicate the region and mouse line from which the sparseness is quantified. A1:
primary auditory cortex. V1: primary visual cortex. Hippo: hippocampus.

Sparseness quantification of STARS driven by inducible Cre lines.
Then, we asked whether it is possible to further decrease the labeling ratio by combing
STARS with other techniques, for instance CreER-expressing line. The motivation comes
from the fact that although the cell body observed is few in number, the neuronal
processes from multiple neurons would still intertwine with each other, making the
projection tracing and morphological characterization inconvenient. This is especially
true for the cases in the cortex where neurons exhibit long-range and complicated
neurites. We first crossed the STARS with the CamK2a-CreERT2 mouse line. The
CamK2a-CreERT2 line was generated by the Allen Institute and reported as “leaky” in
Page 20 of 40

the absence of tamoxifen (TMX) injection (Figure 9). We wished STARS could reduce
the labeling rate and thus enabling easy single-neuron tracing.
Surprisingly, without tamoxifen injection, there is almost no mYFP-positive neuronal cell
observed in the CamKII-CreERT2;STARS mouse (Figure 9, lower left). This suggests
that the sparseness of STARS mice driven by CreER lines could be adjusted to as low as
possible. Then we asked what the upper limit of labeling ratio is for CreER driven
STARS mice. The saturation dose of tamoxifen for CamK2a-CreERT2 mouse line was
reported to be 200µg/g body weight injected for 5 consecutive days. As shown in figure 9
(right columns), much more neurons were mYFP-positive and the maximum sparseness
is 6.51% (Figure 11). This illustrates that by crossing STARS with CreER lines a
relatively wide range of sparseness might be reached, by manipulating the amount of
tamoxifen injected. The users of this line thus could have more options and flexibly
adjust the sparseness for their own needs.
However, for morphological characterization, not only the sparseness but also the actual
number of labeled cells matters. As shown in figure 9, even if at a sparseness level of less
than 10 percent, the fluorescent neurons would still overlap with each other in some
densely-packed regions such as the dentate gyrus. This partially results from the great
absolute number of Camk2a-positive cells. And other cell types may have this problem as
Page 21 of 40

well. So next I tried to titrate the dosage of tamoxifen on Camk2a-CreERT2;STARS
mice.

Figure 9. Images of Camk2a-CreERT2 mice crossed with Ai14 or STARS mice.
The left two columns are from vehicle injected mouse, while the right two columns are
injected saturated dose of tamoxifen. The upper rows shows the expression of tdTomato
in Camk2a-CreER;Ai14 mice. The lower rows shows the expression of mYFP in
Camk2a-CreER;STARS. V1: primary visual cortex. Hippo: hippocampus. Scale bar: 100
µm.

I started the titration by using the 4-hydroxyl-tamoxifen (4-HT) because it is the actual
ligand for CreER and thus expected to have more reliable labeling outcome. Because 4-
HT is considered 3 times more efficient than TMX
15,16
, I used a single shot of 100µg/g
body weight 4HT as the maximum dose, taking both economical and efficiency factors
into account. Multiple brain regions, including primary visual cortex (V1), primary
Page 22 of 40

auditory cortex (A1) and hippocampal dentate gyrus (DG), were checked 5 days after the
injection (Figure 10). We did observe relatively uniform labeling ratios across different
brain regions under a given dose, but the labeled cell number is clearly not linearly
dependent on the dosage applied (Figure 11). Also, the overall sparseness is much lower
than 6%, the ratio we got with saturating TMX. This could be due to the 4HT’s fast-
metabolized property and its limited ability of passing blood brain barrier
10
.

Figure 10. Camk2a-CreERT2;STARS mice injected with 4-hydroxyl-tamoxifen.
The regions that images taken were shown on the left. The dose injected was shown on
the bottom. V1: primary visual cortex. Hippo: hippocampus. Scale bar: 100 µm.

Page 23 of 40

Figure 11. Sparseness of STARS driven by the CamK2a-CreERT2 line.
The y-axis shows the sparseness. The x-axis shows injected dose that the sparseness is
quantified. The brain regions were indicated by the legend on the right. TMX: tamoxifen.
4HT: 4-hydroxyl-tamoxifen. A1: primary auditory cortex. V1: primary visual cortex. DG:
dentate gyrus.

Figure 12. Example individual neurons labeled in CamK2a-CreERT2;STARS.
The legend indicates the cell type and location identified by the neuron morphology. The
images were taken using 40× objectives. V1: primary visual cortex. Scale bar: 50 µm.
Page 24 of 40

Moreover, with single 10µg/g 4-HT injection into CamK2a-CreERT2;STARS mice, there
were less than 3 neurons labeled per section (Figure 11). Thanks to this extremely high
sparseness, I was able to image individual neurons in a Golgi-stain like manner with no
or little interference of neuronal processes from neighboring neurons (Figure 12).
Because the fluorescent indicator (mYFP) in STARS is membrane bound, the entire cell
surface of single neurons could be visualized clearly and unambiguously compared to
Golgi stain. With high magnification images, the dendritic spines can be individually
characterized (Figure 12).

Figure 13. Transform STARS into an adapter mouse line.
A. Concept of antigen-controlled protein stabilization
17
. FP, fusion protein; An, antigen.
In my experiments, antigen of the destabilized nanobody is GFP and its variants. The
fusion protein is a FLP recombinase (FLPo). So the whole protein expressed by the AAV
virus is named FLP-Depend-On-GFP (DOG). B. Schematic diagram of testing the FLP-
DOG on STARS. FLP-DOG: AAV2 FLP-DOG virus. fDIO-eYFP: AAV DJ fDIO-eYFP
virus.

Page 25 of 40

Transform the STARS reporter into an adapter line.
So far, we have shown that our STARS reporter line could reliably realize sparsely
labeling neurons and make detailed morphological characterization much easier. But still,
it would be better if we could manipulate the neuronal transcriptomics or activity of these
labeled cells. It would be very valuable for researchers studying gene autonomous
function or local neural circuits. To this end, it is necessary to introduce an effector
protein into the system. With this in mind, I tried to deliver Flippase (FLP) into neurons
in a GFP/YFP-dependent manner. Luckily, we found the AAV2 FLP-DOG (FLP
recombinase-Depend-On-GFP), which is a viral vector expressing FLP protein fused to
destabilized GFP nanobody
17,18
. Only when the nanobody binds to GFP or its variants,
the fusion protein could save itself from fast degradation, thus later be transported into
the nucleus and activate downstream targets (Figure 13A).
AAV2 FLP-DOG and a FLP-dependent eYFP virus (AAV DJ fDIO-eYFP) were co-
injected into PV-IRES-Cre;STARS, in which part of the PV-positive neurons expressed
mYFP (Figure 13B). In this experiment, the fDIO-eYFP is an indicator for the presence
of functional FLP. Although the indicator is also YFP, it is much brighter than its
endogenous membrane-bound variant in STARS because it is expressed from the viral
vector. As shown in figure 14, there are neurons expressing eYFP that is much brighter
than the endogenous membrane-bound YFP. That indicates the FLP-DOG virus is indeed
Page 26 of 40

able to unlock the fDIO-eYFP virus. However, unfortunately in our control experiments,
in which both viruses were injected into wildtype C57BL/6J mice, there are also eYFP-
positive neurons observed (Figure 14).

Figure 14. Images of 488nm-excited fluorescent signal from FLP-DOG and fDIO-
eYFP co-injected mice.
The genotypes of injected mice are shown on the left. And the brain regions of the image
are shown at the bottom. M1: primary motor cortex. V1: primary visual cortex. Hippo:
hippocampus region. Scale bar: 100 µm.

An experiment to tested whether the leakiness comes from the fDIO-eYFP itself was
carried out. This time, fDIO-eYFP virus diluted to the same titer and mixed with a tiny
amount of AAV1 pSynI-tdTomato virus was injected. The tdTomato expressing virus
was used to indicate a successful injection. It turned out that fDIO-eYFP shows no
Page 27 of 40

detectable leakiness alone (figure 15). The visible signal in GFP channel was in fact
bleached signal from tdTomato because they were completely overlapping. Together,
these results indicate the PV-Cre;STARS mice could express functional FLP recombinase
after injected with AAV FLP-DOG. However, some FLP-DOG proteins may escape from
fast degradation when they are not bound with YFP . This may be prevented by
introducing a time interval (days) between the injections of the two viruses.

Figure 15. Images from the wildtype mice injected with fDIO-eYFP only.
The first row shows the images acquired from GFP channel. And the second row shows
the merge image of GFP and tdTomato channel. Brain regions of the image are shown at
the bottom. M1: primary motor cortex. V1: primary visual cortex. Hippo: hippocampus
region. Scale bar: 100 µm.

Page 28 of 40

Discussion
In this article, we demonstrated that STARS transgene is feasible to create stable
sparseness in the transgenic reporter mouse model across cell types and brain regions.
When STARS is driven by constitutive Cre-expressing lines, the mYFP would only
labeled around 10% of the Cre-positive cells. When it is driven by a CreER line, cells
being labeled could range from a few per section to about 6%. Also, STARS could
eliminate the leaky labeling caused by CreER line. More importantly, with titrated
tamoxifen administration, STARS could realize genetically determined “Golgi like”
stain. Moreover, with the help of virus injection, we showed that it is possible to
transform STARS from a reporter mouse line into a recombinase-expressing adapter line,
which will potentially be useful for functional imaging or activity intervention studies.
Leakiness of Camk2a-CreERT2 lines
In our CreER-driven STARS animals, one out of three animals injected with vehicle did
show sparse labeling, but labeling ratio is even higher than some tamoxifen injected
animals. It is impossible from cross contamination because they started to be kept from
the tamoxifen group in separate cages before the injection. So we reason that the
leakiness of CamKII-CreER line could be a result of transient robust expression of
CreER during development. The protein level might go too high and the translocation of
CreER fusion protein into nucleus happens spontaneously. Because the gene circuit built
Page 29 of 40

in STARS is a XOR (exclusive OR) gate, the cell and its daughter cells would always
carry the same outcome from the recombination. If one progenitor cell goes “mYFP-
expressing” early in the development, its daughter cells would always be mYFP positive.
From this angle of aspect, STARS might even be more useful for lineage tracing during
development. Later, I would like to inject CamK2a-CreER;STARS mice with other
concentration of tamoxifen to generate a curve of sparseness versus dose.
To improve specificity of FLP-DOG
As for the AAV FLP-DOG injection experiments, we suspect that the leakiness results
from the transient high expression of FLP-DOG fusion protein during the early stage
after transduction. If the peak level of FLP-DOG protein before its homeostasis exceed
certain threshold, the recombinase may escape from degradation and diffuse into nucleus
and activate the reporter virus. In our test, it also forms a positive feedback loop, meaning
the unlocked fDIO-eYFP virus itself could further increase the level of functional FLP
recombinase. This make our system ultra-sensitive to FLP-DOG expression level and the
test could be considered the strictest test for the specificity of this virus. To overcome this
problem, I would like to do a two-step virus injection into the same brain region of
STARS mice: FLP-DOG first. After the first viral expression reaches its homeostatic
state, then the second reporter virus, fDIO-eYFP, could be injected. By doing this, we
hope to achieve a true eYFP-dependent FLP recombinase expression in STARS mice.
Page 30 of 40

Other sparse labeling methods
Here, we present the STARS method and this transgenic mouse as a tool to generate
sparseness. We believe it is the most easy-operating and reliable method. What is worth
to mention here, is that the most straightforward way to sparsely deliver transgene into
target cells is to inject virus with a proper titer. Although stable sparseness is doable after
practice, this method is limited by the capacity of viral vectors and the generation of scar
tissues after invasive injection. For these concerns, a transgenic animal model enabling
sparsely labeling is still considered a better option.
Despite these classic methods, the blooming CRISPR/Cas9 technique promotes a new
IUE based method called SLENDR, which has the potential to realize sparse labeling. By
introducing gRNA during the embryonic stage, the SLENDR system could creat in vivo
knock-in or knock-out of a large number of genes/transgenes. If a mouse model with
multi-color background is available, the SLENDR is possible to create even more color
outputs by randomly decreasing/increasing one of the color elements. Because there is no
theoretical limitation on how many genes CRISPR/Cas9 could edit, then the question
goes to how many distinguishable reporters could bioengineering scientists could create
for us. This is also indeed a very exciting path to look forward to. But the downside of
this method is also obvious. Because the CRISPR/Cas9 system achieve labeling by
Page 31 of 40

introduce mutation into the genome, let along the efficiency problem, the off-target
mutation is always a serious consideration for functional genetic studies.
Page 32 of 40

Materials and Methods
Generation of STARS constructs and stable transfected HEK293 cell lines
All constructs were made by traditional molecular cloning methods (PCR, restriction
endonuclease digestion and ligation). The CMV β-actin enhancer (CAG) promoter from
pCAGGS-ES plasmid (gift of Dr. Le Ma was subcloned into pBlue-script SK+/− first.
Then, lox-flanked (flox) fluorescent protein, lox2272-loxP pair was PCR and inserted by
using the following primer sets: membrane bound YFP (primer1:
agatatcctgcagataacttcgtataggatactttatacgaagttataccatgggatgtattaaatca, primer2:
ccatgggaagcttataacttcgtataatgtatgctatacgaagttatgaattcggccggccggcgcgccgaattcgtcgaggccg
cgaattaaaaaacc) and mCherry (primer1:
ccatgggaagcttataacttcgtataggatactttatacgaagttatcacgtgccaccatggtgagcaagggcgagga,
primer2: ggaattcctcgagataacttcgtataatgtatgctatacgaagttatagatctgtcgaggccgcg
aattaaaaaacc). Single polyA fragment (1 kb) was also PCR out and inserted between
mYFP and mCherry. Multi-copy polyA spacers were ligated by compatible restriction
sites. To make stable HEK293 cell lines (cultured in DMEM, containing 10% FCS, 1%
penicillin/streptomycin), the STARS cassettes were subcloned into pcDNA5/FRT
expression vector (Invitrogen). HEK293 cells transfected with STARS were ensured to
bear only single copy of transgene by using Flp-In™ system (Invitrogen). The induction
of recombination on STARS transgene was achieved by pCAG-Cre plasmid using
Page 33 of 40

lipofectamine 2000 (Invitrogen) according to the product manual. Images was taken by a
custom-built epi-fluorescence microscope under 10× objective (Olympus) five days after
the transfection.
Animals
All animal usage was approved by Institutional Animal Care and Use Committee
(IACUC) of University of Southern California. The STARS transgenic mouse line was
generated using TARGATT technology
20
by Applied StemCell Inc. with the vector we
provided. The entire cassette of single pA STARS was eventually cloned into the
TARGATT vector (Applied StemCell Inc.) and the spacer fragment (6X-pA=7.2kb) was
subsequently inserted. Single copy of the transgene was inserted into ROSA26 locus and
verified by genotyping and phenotyping described in result session.
For the sparseness quantification experiments, the STARS mice were crossed with
various Cre-expressing lines including Rbp4-Cre (MGI:4367067), PV-IRES-Cre (Stock
number at The Jackson Laboratory (JAX): 008069), VIP-IRES-Cre (Stock number at
JAX: 010908) and CamK2a-CreERT2 (Stock number at JAX: 012362). Additionally,
each Cre-expressing line was crossced with Ai14 (Stock number at JAX: 007914)
reporter line as the reference of cells were Cre recombinase positive. The genotyping of
the F1 progeny was determined following the protocol described on the JAX website
Page 34 of 40

(www.jax.org) and our phenotyping method described above. Only the F1 progeny were
used in our experiments.
Tamoxifen Induction
Tamoxifen (TMX, T5648, Sigma) was dissolved in sunflower seed oil (S5007, Sigma)
with 10% ethanol to 20 mg/mL. And 4-hydroxyl-tamoxifen (ALX-550-095, VWR) was
first dissolved in pure ethanol and then mixed a mixture of sunflower seed oil and caster
oil (259853, Sigma) with a ratio of 4:1 to 10 mg/mL as described
16
. The remaining
ethanol was eliminated by SpeedVac (Thermo Fisher Scientific) later. The stock solutions
were diluted to corresponding concentration to ensure the volume being injected is
proportional to animal’s body weight. Solutions were kept in -20 degree Celsius up to 2
months. All solution was administrative intraperitoneally after the mice were anesthetized
by vaporized isoflurane briefly. Only mice over 2 month’s age were selected for the
experiments. Animals were perfused with PBS and fixed with 4% paraformaldehyde
(PFA) 5 days after their last injection.
Histology
Adult animals utilized for quantifying sparseness were perfused as described above.
Brain tissues were collected after perfusion. As for cochlea samples, they were collected
from pups after recapitulation. All tissues were fixed in 4% PFA overnight. Coronal
sections of brain were later sliced by vibratome (Leica VT1000). Brain sections with 100
Page 35 of 40

μm thickness or whole mount tissue of cochlea were mounted on glass slides. Images
were taken by a confocal microscope (Olympus).
For Ai14 progeny, samples were directly imaged after mounting. For STARS progeny,
sample were later stained with anti-GFP antibody (dilution 1:250, ab290, Abcam).
Procedure of immunohistochemistry is summarized below.
Samples were permeabilized with 0.3% TrionX-100 for 2 hours and followed by
incubation in blocking buffer (5% serum in PBST) for at least 2 hours at room
temperature. Primary antibody incubation 24 hours at 4°C was followed by secondary
antibody (dilution 1:500, Invitrogen) incubated overnight at 4°C.
Mice injected with viruses were not stained.
Image analysis and quantification of sparseness
Z-stack images were taken from samples and the analysis was done in Fuji software.
Multiple 100µm×100µm×100µm boundaries were drawn and number of cell bodies
inside the cube was counted manually within a particular brain region. Sparseness was
defined as the ratio of average GFP-positive cells per cube and average tdTomato-
positive cells per cube. In case of analysis of sparseness in the cochlea, the hair cells
could be simply counted because they were lined up uniformly and compared with the
total number of hair cells within that region.
Page 36 of 40

Stereotaxic injection
Adult PV-Cre;STARS mice were injected with 50 nL virus mixture in a stereotaxic setup.
The final titer of the AAV2 FLP-DOG is 1×10
12
pfu (Gene Transfer Vector Core,
Harvard and MIT) and the AAV DJ fDIO-eYFP (Salk Institute) is 1.5×10
12
pfu according
to supplier’s label. During surgery, animals were kept anesthetized with isoflurane and
pure oxygen. After surgery, ketoprofen (3-5 μg/g boday weight) was injected
subcutaneously for 3 days consecutively. Each animal was injected to M1, V1 and
hippocampus unilaterally. 14-21 days after the injection, animals were euthanized and
brain tissue was collected for further examination.

Page 37 of 40

Contributions
The design of the STARS mouse, data from cell lines, images of Ai14 and part of STARS
crossed with the constitutive Cre-expressing lines and the quantification of sparseness in
constitutive Cre-expressing STARS is done by Leena A.I. The rest of data and the figures
was acquired or drew/re-organized by Junxiang H.

Page 38 of 40

References
1. Baars, B. J., Gage, N. M., Baars, B. J. & Gage, N. M. Chapter 1 – Mind and brain.
in Cognition, Brain, and Consciousness 2–31 (2010). doi:10.1016/B978-0-12-
375070-9.00001-2
2. Linda Madisen, Theresa A Zwingman, Susan M Sunkin, Seung Wook Oh, Hatim
A Zariwala, Hong Gu, Lydia L Ng, Richard D Palmiter, Michael J Hawrylycz,
Allan R Jones, E. S. L. & H. Z. A robust and high-throughput Cre reporting and
characterization system for the whole mouse brain. Nat. Neurosci. 13, 133–140
(2010).
3. Hayashi, S. & McMahon, A. P. Efficient Recombination in Diverse Tissues by a
Tamoxifen-Inducible Form of Cre: A Tool for Temporally Regulated Gene
Activation/Inactivation in the Mouse. Dev. Biol. 244, 305–318 (2002).
4. Young, P. et al. Single-neuron labeling with inducible Cre-mediated knockout in
transgenic mice. Nat. Neurosci. 11, 721–8 (2008).
5. Livet, J. et al. Transgenic strategies for combinatorial expression of fluorescent
proteins in the nervous system. Nature 450, 56–62 (2007).
6. Cai, D., Cohen, K. B., Luo, T., Lichtman, J. W. & Sanes, J. R. Improved tools for
the Brainbow toolbox. Nat. Methods 10, 540–547 (2013).
Page 39 of 40

7. Zong, H., Espinosa, J. S., Su, H. H., Muzumdar, M. D. & Luo, L. Mosaic analysis
with double markers in mice. Cell 121, 479–492 (2005).
8. Luo, W. et al. Supernova: A Versatile Vector System for Single-Cell Labeling and
Gene Function Studies in vivo. Sci. Rep. 6, 35747 (2016).
9. Mikuni, T., Nishiyama, J., Sun, Y., Kamasawa, N. & Yasuda, R. High-
Throughput, High-Resolution Mapping of Protein Localization in Mammalian
Brain by in Vivo Genome Editing. Cell 165, 1803–1817 (2016).
10. Valny, M., Honsa, P., Kirdajova, D., Kamenik, Z. & Anderova, M. Tamoxifen in
the Mouse Brain: Implications for Fate-Mapping Studies Using the Tamoxifen-
Inducible Cre-loxP System. Front. Cell. Neurosci. 10, 1–12 (2016).
11. Wang, S. Z., Liu, B. H., Tao, H. W., Xia, K. & Zhang, L. I. A genetic strategy for
stochastic gene activation with regulated sparseness (STARS). PLoS One 4, 1–6
(2009).
12. Sternberg, N., Hamilton, D. & Hoess, R. Bacteriophage P1 site-specific
recombination. J. Mol. Biol. 150, 487–507 (1981).
13. Soriano, P. Generalized lacZ expression with the ROSA26 Cre reporter strain. Nat.
Genet. 21, 70–71 (1999).
14. Chklovskii, D. B., Vitaladevuni, S. & Scheffer, L. K. Semi-automated
Page 40 of 40

reconstruction of neural circuits using electron microscopy. Curr. Opin. Neurobiol.
20, 667–675 (2010).
15. Guenthner, C. J., Miyamichi, K., Yang, H. H., Heller, H. C. & Luo, L. Permanent
genetic access to transiently active neurons via TRAP: Targeted recombination in
active populations. Neuron 78, 773–784 (2013).
16. Ye, L. et al. Wiring and Molecular Features of Prefrontal Ensembles Representing
Distinct Experiences. Cell 165, 1776–1788 (2016).
17. Tang, J. C. Y. et al. Detection and manipulation of live antigen-expressing cells
using conditionally stable nanobodies. Elife 5, 1–27 (2016).
18. Tang, J. C. Y. et al. A nanobody-based system using fluorescent proteins as
scaffolds for cell-specific gene manipulation. Cell 154, 928–939 (2013).

Abstract (if available)

Abstract In order to keep track of the morphology and activity of individual cells in highly compact tissues, such as the brain, genetic methods that could automatically achieve sparse labeling are highly demanded. Here, we present the STARS transgenic mouse as a promising tool to visualize the detailed morphology of neurons in a genetically determined population. The STARS mouse is a Cre-dependent reporter line with two fluorescent reporter proteins expressed in an exclusive manner. Unlike traditional strains, the activation of genes in STARS is under the control of an unequal OR logic gate. Thus, in the presence of Cre recombinase, the two transgenes with different lengths would have different probabilities to be expressed. In our case, one of the transgenes, mYFP, would stably label only about 10% of neurons when it is driven by various constitutive Cre-expressing lines. When it is driven by an inducible Cre-expressing line, the sparseness could range from a few labeled cells per section to about 6% in maximum. The labeling ratio is consistent across neuronal types and brain regions tested. With the help of a FLP-DOG virus, we could further express FLP recombinase in YFP-positive cells, allowing later genetic manipulations coupled with morphological characterization.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Identification and characterization of the enhancer elements for lymphatic-specific expression of Prox1

PDF

Slit/Robo signaling underlies the spatial patterning of spiral ganglion neurons to shape the peripheral auditory circuitry assembly

PDF

Neurogenic placodes provide migratory enteric sensory neural progenitors in response to endothelin signaling pathway

PDF

Characterization of the progenitor cell zone in feather follicles

PDF

Phosphorylation of Synaptojanin differentially regulates synaptic vesicle endocytosis of distinct vesicle pools

PDF

MicroRNAs involved in the regulation of Endothelin-1 gene expression in endothelial cells

PDF

Neuroendocrine regulation of the transcription factor SKN-1/Nrf2 in oxidative stress response

PDF

Uncovering the influence of N-terminal phosphorylation on conformational dynamics of huntingtin exon 1 monomer

PDF

Structural characterization of the functional amyloid Orb2A using EPR spectroscopy

PDF

Development of ECM for the preservation of adult mouse pancreatic islet function in vitro

PDF

A synthetic lethal screen for NF-κB-dependent plasma cell disorders

PDF

DNA methylation markers for blood-based detection of small cell lung cancer in mouse models

PDF

Investigating the function and epigenetic regulation of ABCA3, a novel LUAD tumor suppressor gene

PDF

A novel construct to study the pulsatility of insulin secretion in single cells, islets and whole pancreas

PDF

Studies on the role of a novel protein, TMEM 56 in tumorigenic growth for MCF-7 cells

PDF

A functional genomic approach based on shRNA-mediated gene silencing to delineate the role of NF-κB and cell death proteins in the survival and proliferation of KSHV associated primary effusion l...

PDF

Generation and characterization of anti-CD138 chimeric antigen receptor T (CAR-T) cells for the treatment of hematologic malignancies

PDF

Exploring alternative roles of visual arrestin 1 in photoreceptor synaptic regulation and deciphering the molecular pathway of retinal degeneration using mouse knockout technology

PDF

Deletion of Krüppel-like factor 4 (KLF4) affects NPCs to embark on differentiative progression and leads to delayed neurogenesis in mouse embryonic cortex

PDF

Positive regulation of RNA polymerase III-mediated transcription of tRNA genes by the Mediator kinase submodule

Asset Metadata

Creator Huang, Junxiang (author)

Core Title Characterization of the STARS (STochastic gene Activation with Regulated Sparseness) mouse line

School Keck School of Medicine

Degree Master of Science

Degree Program Biochemistry and Molecular Biology

Publication Date 09/23/2018

Defense Date 07/21/2017

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

Tag Cre-dependent reporter mouse,OAI-PMH Harvest,Stars,stochastic sparse labeling

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Tao, Huizhong (committee chair), Zhang, Li (committee chair), Chen, Jeannie (committee member), Langen, Ralf (committee member), Tokes, Zoltan (committee member)

Creator Email 408145970@qq.com,junxianh@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-432977

Unique identifier UC11264362

Identifier etd-HuangJunxi-5759.pdf (filename),usctheses-c40-432977 (legacy record id)

Legacy Identifier etd-HuangJunxi-5759.pdf

Dmrecord 432977

Document Type Thesis

Rights Huang, Junxiang

Type texts

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

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

Repository Name University of Southern California Digital Library

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