Early enforcement of cell identity by a functional component of the terminally differentiated state

TL;DR: A control principle for robust cell identity is suggested whereby a core component of the differentiated state also promotes differentiation from its own progenitor state, thus providing an explanation for why maintenance of the mature adipocyte state is so robust.

Abstract: How progenitor cells can attain a distinct differentiated cell identity is a challenging problem given that critical transcription factors are often not unique to a differentiation process and the fluctuating signaling environment in which cells exist. Here we test the hypothesis that a unique differentiated cell identity can result from a core component of the differentiated state doubling up as a signaling protein that also drives differentiation. Using live single-cell imaging in the adipocyte differentiation system, we show that progenitor fat cells (preadipocytes) can only commit to terminally differentiate after upregulating FABP4, a lipid buffer that is highly enriched in mature adipocytes. Upon induction of adipogenesis, we show that after a long delay, cells first abruptly start to engage a positive feedback between CEBPA and PPARG before then engaging, after a second delay, a positive feedback between FABP4 and PPARG. These sequential positive feedbacks both need to engage in order to drive PPARG levels past the threshold for irreversible differentiation. In the last step before commitment, PPARG transcriptionally increases FABP4 expression while fatty-acid loaded FABP4 binds to and increases PPARG activity. Together, our study suggests a control principle for robust cell identity whereby a core component of the differentiated state also promotes differentiation from its own progenitor state.

Summary

INTRODUCTION

• Terminal cell differentiation is fundamental for developing, maintaining, and regenerating tissues in all multicellular organisms and is the process by which specialized cells such as adipocytes (fat cells), osteoblasts, neurons, and muscle cells are generated from progenitor cells [1].
• Since negative feedback mostly prevents cells which have chosen one path from differentiating into alternative cell types, additional regulatory mechanisms must exist that selectively drive cells onto a unique path and allow cells to robustly assume and maintain a specific differentiated cell identity.
• These observations motivated their study here that FABP4 could provide such a self-reinforcement mechanism for unique cell identity.
• It is well-established that the expression of FABP4 is induced by PPARG, the master regulator of adipocyte differentiation [22].

RESULTS

• FABP4 is needed for adipogenesis in vitro and in vivo Previous studies suggested that FABP4 and PPARG are in a positive feedback relationship [18,20,23,25] (Fig 1A).
• This analysis showed that cells that go on to differentiate according to their PPARG levels (blue traces in Fig 2C, left) also started to increase their FABP4 levels before they reach the threshold and then increase much more strongly after they reach the threshold (red traces in Fig 2C, right).
• Since FABP4 can directly bind to PPARG in the presence of lipid , it is conceivable that the binding may increase PPARG expression indirectly by stabilizing PPARG.
• In condition 1 (Fig 6D), the authors compared wildtype OP9 preadipocyte cells which expressed FABP4 versus to OP9 cells in which FABP4 had been knocked out (FABP4-KO cells from Fig. 1D).

DISCUSSION

• The authors study focuses on a general question how cells ensure that they assume and maintain a specific terminally differentiated state given that differentiating cells typically have multiple cell-fate options and there is also cellular plasticity with cells being able to transition between different reversible progenitor states.
• The authors rationale for focusing on FABP4 was previous evidence that FABP4 may regulate PPARG expression and/or activity [18,20,23,25] and that FABP4 is highly expressed specifically in adipocytes [15].
• CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
• The authors study further addressed the question of how FABP4 can regulate PPARG since previous studies have yielded contradictory results.
• This first appeared to contradict earlier results which showed that WT cells, FABP4/FABP5 KO cells, FABP4/FABP5 KO cells can still effectively differentiate when rosiglitazone was added along with the DMI.

AUTHOR CONTRIBUTIONS

• F.B.K. designed experiments and contributed to the discussion.
• Z.B., T.C., and M.N.T. wrote the manuscript with input from all authors.

FIGURE LEGENDS

• (A) Schematic of proposed cell-identity positive feedback.
• Wildtype 3T3-F442A cells that were transfected and sorted at the same time but that did not harbor FABP4 or FABP5 knockdown were used as a control (Control-KO).
• (B) Changes in citrine-PPARG and FABP4-mKate expression measured in the same cell over the timecourse of adipogenesis.
• The timepoint when the feedbacks start to engage is reflected in a change in PPARG slope and occurs many hours before a cell will cross the PPARG threshold level at which differentiation is irreversibly triggered.
• Plotted lines are population median traces with shaded regions representing 25th and 75th percentiles of approximately 700 cells per condition, representative of 6 independent experiments.

Cell culture and differentiation

• OP9 cells were cultured according to previously published protocols [20,23,37,46].
• Forty-eight hours after initiating differentiation, Stimulus I was removed from the cells and was replaced with Stimulus II consisting of growth media plus 10% FBS and 1.75 nM insulin for another 48 hours.
• As noted in some experiments, rosiglitazone (Cayman, USA) was added to the media to result in a final concentration of 1 μM. .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
• Page 27 3T3-F442A preadipocytes were grown and differentiated according to established protocols[30].
• After day 4, the media was replaced with just growth media.

Oil Red O staining

• To determine lipid accumulation 3T3-F442A preadipocytes were differentiated for 10 days and stained with Oil Red O. Briefly, different guide RNAs targeting FABP4 or FABP5 were designed (crispr.mit.edu) and oligos including the targeting sequences were annealed and cloned into pSpCas9n(BB)-2A-GFP (PX461; Addgene #48140), used for targeting FABP4, or pSpCas9(BB)-2AmiRFP670 (Addgene #91854), used for targeting FABP5.
• At day 0, the cells were treated with doxycycline 1ug/ml and adipogenic cocktail in MEM with Lglutamine, 10% FBS, and 100 U/ml penicillin/streptomycin.
• CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
• Next, western blot analysis of mKate-FABP4 clone2 was performed using anti-RFP antibody and anti-FABP4 to verify protein expression and to check for the correct molecular weight of the tagged protein (Supp Figs S3C and S3D).

FABP4 Gain-of-Function in pre-adipocytes via CRISPRa SAM

• Murine mesenchymal C3H10T1/2 cells stably expressing the CRISPRa-SAM complex were used for gain of function experiments following a previously established protocol [33].
• The resulting cell line (C3H10T1/2CRISPRa-SAM) can be used for activation of endogenous genes via chemical transfection with a single guide RNA -containing plasmid.
• Page 33 experiments, 2000 cells were seeded per well into a 96-well Costar Plastic plate.
• At Day 2, the media was replaced with media containing only insulin.
• PCR primer sequences were synthesized by Elim Biopharmaceuticals Inc (USA, CA) and are listed in Supplementary Table S8.

Fluorescent imaging

• Imaging was conducted using an ImageXpress MicroXL (Molecular Devices, USA) with a 10X Plan Apo 0.45 NA objective.
• A camera bin of 2x2 was used for all imaging condition.
• Living cells were imaged in FluoroBrite DMEM media with 10% FBS, 1% Penicillin/Streptomycin and insulin to reduce background fluorescence.
• Four, non-overlapping sites in each well were imaged.
• .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.

Imaging data processing

• Data processing of fluorescent images was conducted in MATLAB .
• Unless stated otherwise, fluorescent imaging data were obtained by automated image segmentation, tracking and measurement using the MACKtrack package for MATLAB.
• If cells were binned according to their PPARG expression, cells were binned based on their mean nuclear PPARG expression at the described timepoints.
• Briefly, PPARG values at the end of a differentiation experiment typically exhibit a bimodal distribution.
• Measuring protein decay rates using cycloheximide Protein decay rates were quantified as previously described [20,27].

Statistics

• Unless specified otherwise, data are expressed as mean +/- standard error of the mean (S.E.M).
• CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
• BioRxiv preprint A Adipogenic stimulus (DMI) CEBPB CEBPA FABP4 PPARG.
• Plotted lines are population median traces with shaded regions representing 25th and 75th percentiles of approximately 700 cells per condition, representative of 6 independent experiments.
• During Phase 2, both external stimulation and internal self-amplification are needed to increase PPARG levels up to the threshold.

Figures

Figure 3. The early, small increase in FABP4 is needed to push PPARG levels to the threshold to differentiate.

Figure 4. During adipogenesis, the PPARG-FABP4 feedback engages after the PPARG-CEBPA feedback engages.

Figure 5. The lipid-binding activity of FABP4 is needed to increase PPARG expression and induce adipogenesis.

Figure 1: FABP4 is needed to drive adipogenesis in vitro and in vivo.

Figure 6. Rosiglitazone, a PPARG agonist which binds to the same binding pocket in PPARG as fatty acid, bypasses the need for FABP4 upregulating PPARG expression during adipogenesis.

Figure 2. There is a delayed, small increase in FABP4 early in adipogenesis.

Full text

Bahrami-Nejad,et,al.! Page%1!!
Title: Early enforcement of cell identity by a functional component of the terminally
differentiated state
!
Authors: Zahra Bahrami-Nejad
1,
*
, Tinghuan Chen
1,2,
*
, Stefan Tholen
1
, Zhi-Bo Zhang
1,2
,
Atefeh Rabiee
1
, Michael L. Zhao
1
, Fredric B. Kraemer
3
, and Mary N. Teruel
1,2, #
Affiliations:
1
Dept. of Chemical and Systems Biology, Stanford University, Stanford, CA
2
Dept. of Biochemistry and the Drukier Institute for Children’s Health, Weill Cornell Medicine, New
York, NY
3
Dept. of Medicine/Division of Endocrinology, Stanford University, Stanford, CA and VA Palo Alto
Health Care System, Palo Alto, CA
*
Equal Contribution
!!
#
Corresponding Author: mnt4002@med.cornell.edu
ABSTRACT
How progenitor cells can attain a distinct differentiated cell identity is a challenging problem given that
critical transcription factors are often not unique to a differentiation process and the fluctuating
signaling environment in which cells exist. Here we test the hypothesis that a unique differentiated cell
identity can result from a core component of the differentiated state doubling up as a signaling protein
that also drives differentiation. Using live single-cell imaging in the adipocyte differentiation system,
we show that progenitor fat cells (preadipocytes) can only commit to terminally differentiate after
upregulating FABP4, a lipid buffer that is highly enriched in mature adipocytes. Upon induction of
adipogenesis, we show that after a long delay, cells first abruptly start to engage a positive feedback
between CEBPA and PPARG before then engaging, after a second delay, a positive feedback
between FABP4 and PPARG. These sequential positive feedbacks both need to engage in order to
drive PPARG levels past the threshold for irreversible differentiation. In the last step before
commitment, PPARG transcriptionally increases FABP4 expression while fatty-acid loaded FABP4
binds to and increases PPARG activity. Together, our study suggests a control principle for robust cell
identity whereby a core component of the differentiated state also promotes differentiation from its own
progenitor state.
HIGHLIGHTS
• Fatty-acid loaded FABP4 binds to and increases PPARG expression, thereby turning on PPARG
positive feedback loops that further increase PPARG expression.
• FABP4 critically controls the second phase of adipogenesis between activation of the feedback
loops and reaching the threshold to differentiate.
• Only a small fraction (~10%) of the FABP4 levels typically attained in mature fat cells is needed to
commit cells to the differentiated state, thus providing an explanation for why maintenance of the
mature adipocyte state is so robust.
.CC-BY-NC-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 14, 2020. ; https://doi.org/10.1101/2020.01.03.894493doi: bioRxiv preprint

Bahrami-Nejad,et,al.! Page%2!!
INTRODUCTION
Terminal cell differentiation is fundamental for developing, maintaining, and regenerating
tissues in all multicellular organisms and is the process by which specialized cells such as adipocytes
(fat cells), osteoblasts, neurons, and muscle cells are generated from progenitor cells [1]. However,
how progenitor cells can reach a specific and unique terminally differentiated state is not well
understood. In many differentiation processes, the critical transcription factors and signaling elements
driving cell fate are not unique to a specific differentiation process [2–5]. Furthermore, single cell
RNAseq and other single-cell technologies have provided direct evidence that cell-to-cell variability in
protein expression and signaling activities can cause differentiating cells to pass through stochastic
intermediate states that could misdirect cells to alternative final fates [6–9]. Given the ubiquitous use
of overlapping signaling and transcription programs and the significant signaling variability between
individual cells in a population, this raises the question how a differentiating cell avoids becoming lost
in intermediate states and reliably finds its way to the desired specific terminal cell fate.
As one mechanism supporting terminal cell fate identity, negative feedback has been shown
to play a role in regulating the differentiation decision between multiple fates whereby one fate
suppress the programs that drive differentiation of the other fates [10]. However, since negative
feedback mostly prevents cells which have chosen one path from differentiating into alternative cell
types, additional regulatory mechanisms must exist that selectively drive cells onto a unique path and
allow cells to robustly assume and maintain a specific differentiated cell identity. A number of studies
focusing on directed and transdifferentiation processes showed that differentiation can proceed by
multiple routes and yet converge onto similar transcriptional states [11,12], consistent with the view
that terminally differentiated cell states are ‘attractor basins’ in a transcriptional and signaling
differentiation landscape. The finding that the same differentiated state could be reached by passing
through different intermediate states motivated us to ask the question whether a unique attractor basin
requires a unique cell identify factor that would allow for the same terminally differentiated state to be
reached from different intermediate states. Specifically, we considered that cells may assume a robust
.CC-BY-NC-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 14, 2020. ; https://doi.org/10.1101/2020.01.03.894493doi: bioRxiv preprint

Bahrami-Nejad,et,al.! Page%3!!
differentiated cell identity by using a type of positive feedback whereby a core component that is
uniquely expressed and needed in the differentiated state doubles-up as a signaling co-factor that
drives an irreversible step in the differentiation process.
Such a self-reinforcing mechanism that promotes robust cell identity (1) would have to control
a late step before commitment to ensure that cells select the correct differentiation path and cell
identity, and (2) once cells have committed, would have to robustly lock the cell into its differentiated
state. To test if such a positive-feedback mechanism exists and how it may reinforce cell identity of a
differentiated state, we used the adipocyte differentiation system since it is a well-characterized and
experimentally accessible terminal cell differentiation process [13]. We were also intrigued by a
previous observation that increased levels of fatty acids can promote differentiation of precursor cells.
As a candidate for such a self-reinforcing mechanism for cell identity, the fatty acid binding protein,
FAPB4, is one of the most abundant proteins in adipocytes, where it makes up between 0.5-6% of
soluble protein [14]. FABP4 is normally expressed at high levels only in adipocytes [15,16] and is a
critical core component of adipocytes that has important cell-internal functions such as binding to
hormone sensitive lipase (HSL) and buffering lipid release [17]. However, there is also evidence that
FABP4 may have an additional role in positively regulating the transition from progenitor cells into
mature adipocytes [18–20]. These observations motivated our study here that FABP4 could provide
such a self-reinforcement mechanism for unique cell identity. We note that FABP4 can also be
released from mature adipocytes and has cell-external roles in different cell types including
preadipocytes and other adipocytes where it has been shown to reduce PPARG expression as a
pathological and possibly also normal regulatory function [21].
It is well-established that the expression of FABP4 is induced by PPARG, the master regulator
of adipocyte differentiation [22]. However, whether FABP4 has a role in regulating PPARG activity
during the differentiation process has been difficult to resolve since very little FABP4 is expressed
early in adipogenesis [16,23] and that cells in the population differentiate at different times making it
challenging to use traditional bulk cell assays. Also, it is not clear how FABP4 regulates PPARG and
.CC-BY-NC-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 14, 2020. ; https://doi.org/10.1101/2020.01.03.894493doi: bioRxiv preprint

Bahrami-Nejad,et,al.! Page%4!!
if and how FABP4 and PPARG may reinforce each other’s expression. Finally, if FABP4 is indeed
uniquely important for adipogenesis, it is not clear why genetic studies in FABP4-knockout mice failed
to show a suppression of adipogenesis [24].
Here, using live single-cell analysis of fluorescently-tagged endogenous PPARG and FABP4,
we show that FABP4 and PPARG build up only very slowly during a first phase of the adipogenic
program until they transition to a second phase marked by engagement of positive feedback between
each other. This feedback starts after about 24 hours when the FABP4 level is very low and ends
approximately 12 hours later when the levels of PPARG and FABP4 rapidly build up and pass a critical
threshold for differentiation. We show that the feedback-activation of PPARG can be repressed by a
mutant FABP4 that is deficient in binding fatty acid and that FABP4 can directly interact with PPARG.
This suggests that FABP4 has - at much lower levels than seen in differentiated cells - a transport
function to enhance fatty acid binding to PPARG, a mechanism which is known to increase PPARG
activity [22]. Finally, we show that FABP5 may compensate for a loss of FABP4 and still allow cells to
differentiate. Together, our study provides support for a likely more general model that robust cell
identity can be initiated and reinforced by having a unique core component of the differentiated state
double-up as a signaling factor that initiates and then reinforces the path to this unique differentiated
state.
RESULTS
FABP4 is needed for adipogenesis in vitro and in vivo
Previous studies suggested that FABP4 and PPARG are in a positive feedback relationship
[18,20,23,25] (Fig 1A). One arm of this positive feedback is well-established since there are PPARG
binding sites on the FABP4 promoter, and PPARG activity has been shown to strongly upregulate
FABP4 mRNA and protein levels [14]. However, the relevance of FABP4 in regulating PPARG has
been controversial as FABP4 knockout mice are not defective in adipogenesis [24]. We therefore
performed a series of experiments to address if and how FABP4 can regulate PPARG expression and
.CC-BY-NC-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 14, 2020. ; https://doi.org/10.1101/2020.01.03.894493doi: bioRxiv preprint

Bahrami-Nejad,et,al.! Page%5!!
adipogenesis. We first carried out CRISPR-mediated genome editing to completely knock out FABP4
expression in OP9 cells (Fig 1B). We then induced adipogenesis using the standard 96-hour DMI
protocol and measured the percent of differentiated cells using a previously established single-cell
immunocytochemistry assay based on assessing whether PPARG expression is above or below a
threshold [20,23,26,27] (see Methods). As expected, control-KO OP9 cells differentiated robustly (Fig
1C). In contrast, FABP4-KO OP9 cells were strongly defective in increasing PPARG expression and
differentiating, with less than 10% of cells differentiating compared to control cells. Since FABP5 has
been shown to compensate for FABP4 knock-out both in vitro and in vivo [24,28,29], we also tested
whether we could further reduce the amount of adipogenesis in the FABP4-KO cells by also knocking
down FABP5 using siRNA. Indeed, we could reduce adipogenesis to almost nothing when we knocked
down both FABP4 and FABP5 (Fig 1C).
We further validated these findings in another commonly used preadipocyte cell system, 3T3-
F442A cells [30]. We used CRISPR-mediated genome editing to knockout FABP4 or both FABP4 and
FABP5 (Fig 1D). 3T3-F442A cells which had been subjected to the same FABP4 knockout protocol,
but which did not harbor any FABP4 knockout were used as control cells. We plated the 3T3-F442A
cells into 96-well plates and induced adipogenesis using the standard protocol in 3T3-F442A cells
which is to add insulin[30]. Indeed, when FABP4 was knocked out, 3T3-F442A preadipocytes showed
reduced differentiation, as measured by PPARG staining, as well as lipid accumulation, as measured
by Oil Red-O staining (Fig 1E and 1F). The latter is indicative that the KO cells are not functional
mature adipocytes. To test whether FABP4 was essential for adipogenesis in vivo, we used a
previously-established method in which 3T3-F442A preadipocytes are subcutaneously injected into
the sternum of immune-deficient mice, giving rise to fat pads resembling normal adipose tissue[31].
Since fat is not normally present at the sternum of mice, the fat pad formed at the sternum after
injection of preadipocyte cells is generated by de novo adipogenesis of the injected cells[31]. We
injected our preadipocyte cells into the sternum of 8-week mice, and a fat pad was allowed to form for
4 weeks. The FABP4-KO and FABP4/FABP5 double knockout (DBKO) preadipocyte cells indeed
.CC-BY-NC-ND 4.0 International licenseavailable under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprintthis version posted August 14, 2020. ; https://doi.org/10.1101/2020.01.03.894493doi: bioRxiv preprint

Frequently asked questions

Q1. What are the contributions in this paper?

Using live single-cell imaging in the adipocyte differentiation system, the authors show that progenitor fat cells ( preadipocytes ) can only commit to terminally differentiate after upregulating FABP4, a lipid buffer that is highly enriched in mature adipocytes. Upon induction of adipogenesis, the authors show that after a long delay, cells first abruptly start to engage a positive feedback between CEBPA and PPARG before then engaging, after a second delay, a positive feedback between FABP4 and PPARG. Together, their study suggests a control principle for robust cell identity whereby a core component of the differentiated state also promotes differentiation from its own progenitor state.