PGC1/PPAR Drive Cardiomyocyte Maturation
through Regulation of Yap1 and SF3B2
Sean Murphy
1,2,3
, Matthew Miyamoto
1,3
, Anais Kervadec
4
, Suraj Kannan
1,2,3
, Emmanouil Tampakakis
1,3
, Sandeep
Kambhampati
1,2,3
, Brian Leei Lin
1
, Sam Paek
5
, Peter Andersen
1,3
, Dong-Ik Lee
1
, Renjun Zhu
1,3
, Steven S. An
5
, David A.
Kass
1
, Hideki Uosaki
1,3,6
, Alexandre R. Colas
4
, and Chulan Kwon
1,3,7*
1
Division of Cardiology, Department of Medicine;
2
Department of Biomedical Engineering;
3
Institute for Cell Engineering, Johns Hopkins University School of Medicine,
Baltimore, MD 21205 USA;
4
Sanford Burnham Prebys Medical Discovery Institute, La Jolla, CA 92037 USA;
5
Rutgers Institute for Translational Medicine and Science, New
Brunswick, NJ 08901 USA;
6
Division of Regenerative Medicine, Center for Molecular Medicine, Jichi Medical University, Tochigi, Japan;
7
Lead Contact ckwon13@jhmi.edu
Cardiomyocytes undergo significant levels of structural and func-
tional changes after birth—fundamental processes essential for the
heart to produce the volume and contractility to pump blood to the
growing body. However, due to the challenges in isolating single
postnatal/adult myocytes, how individual newborn cardiomyocytes
acquire multiple aspects of mature phenotypes remains poorly un-
derstood. Here we implemented large-particle sorting and analyzed
single myocytes from neonatal to adult hearts. Early myocytes ex-
hibited a wide-ranging transcriptomic and size heterogeneity, main-
tained until adulthood with a continuous transcriptomic shift. Gene
regulatory network analysis followed by mosaic gene deletion re-
vealed that peroxisome proliferator-activated receptor coactivator-
1 signaling—activated in vivo but inactive in pluripotent stem cell-
derived cardiomyocytes—mediates the shift. The signaling regulated
key aspects of cardiomyocyte maturation simultaneously through
previously unrecognized regulators, including Yap1 and SF3B2. Our
study provides a single-cell roadmap of heterogeneous transitions
coupled to cellular features and unveils a multifaceted regulator con-
trolling cardiomyocyte maturation.
D
ecades of advances in cellular and developmental cardiol-
ogy have provided fundamental insights into understand-
ing myocardial lineage specification in vivo, and this knowledge
has been instrumental for producing cardiomyocytes (CMs)
from pluripotent stem cells (PSCs) (Devalla and Passier, 2018;
Evans et al., 2010; Kattman et al., 2011). However, while new-
born CMs continue to increase their volume and contractility
through extensive morphological, functional, and metabolic
changes until adulthood, PSC-derived CMs (PSC-CMs) are
mired in an immature state even after long-term culture (De-
Laughter et al., 2016; Kannan and Kwon, 2018; Uosaki et al.,
2015). The lack of maturity significantly limits scientific and
therapeutic applications of PSC-CMs. Furthermore, despite
a number of genes involved in CM maturation are associated
with cardiomyopathies, little is known about its relevance to
the initiation and progression of cardiac pathogenesis (Shenje
et al., 2014; Uosaki et al., 2015). Thus, there is a signifi-
cant need to understand biological processes underlying CM
maturation in vivo.
CM maturation is a complex process essential for the heart
to circulate blood to the rapidly growing body (Yang et al.,
2014). After terminal differentiation, CMs undergo binucle-
ation/polyploidization around the first week of birth in mice.
They gradually increase in size and become rectangular with
uniformly patterned sarcomeres (Hirschy et al., 2006). To
efficiently propagate electrical activity, the plasma membrane
invaginates into the cells and forms transverse tubules, en-
abling excitation-contraction coupling (Ziman et al., 2010).
The myocytes become tightly connected via intercalated discs
to allow simultaneous contraction. These events are accompa-
nied with functional and metabolic changes including mature
calcium handling, increased contractile force, and mitochon-
drial maturation and oxidative phosphorylation (Yang et al.,
2014). These multi-adaptive changes occur in the early postna-
tal period and continue until adolescent/adult stages. However,
it remains an open question whether these distinct processes
occur in a coordinated fashion. Factors and pathways me-
diating these individual processes are poorly understood as
well.
Previous large-scale meta-analyses provided a transcrip-
tomic atlas of cardiac maturation, allowing us to determine
gene regulatory networks and pathways involved in cardiac
maturation (Uosaki et al., 2015). The scope was, however,
largely focused on prenatal stages, leaving the postnatal tran-
scriptome dynamics unclear. Moreover, cell-to-cell varia-
tions—poorly understood in the field—could not be deter-
mined with bulk analysis. This issue can be addressed by
single-cell RNA-sequencing (scRNA-seq) that enables compre-
hensive analysis of developmental and cellular trajectory and
heterogeneity (Grun and van Oudenaarden, 2015). However,
scRNA-seq is rarely utilized in myocyte biology due to techni-
cal difficulties associated with single-cell isolation of healthy,
mature CMs.
We have recently demonstrated that large-particle
fluorescence-activated cell sorting (LP-FACS) enables high-
quality scRNA-seq and functional analysis of mature adult
CMs (Kannan et al., 2019). Based on this, here we present
high-quality scRNA-seq analysis of CMs isolated from neonatal
Significance Statement
How the individual single myocytes achieve full maturity re-
mains a ‘black box’, largely due to the challenges with the iso-
lation of single mature myocytes. Understanding this process
is particularly important as the immaturity and early develop-
mental arrest of pluripotent stem cell-derived myocytes has
emerged a major concern in the field. Here we present the first
study of high-quality single-cell transcriptomic analysis of car-
diac muscle cells from neonatal to adult hearts. We identify a
central transcription factor and its novel targets that control key
aspects of myocyte maturation, including cellular hypertrophy,
contractility, and mitochondrial activity.
bioRxiv | February 6, 2020 1–16
.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 February 7, 2020. ; https://doi.org/10.1101/2020.02.06.937797doi: bioRxiv preprint
to adult hearts. We demonstrate that newborn CMs are highly
heterogeneous and progressively change expression of genes reg-
ulating cellular hypertrophy, contractility, and metabolism un-
til adulthood. By combining gene regulatory network analysis
with mosaic gene deletion and chromatin immunoprecipitation
sequencing (ChIP-seq), we identify peroxisome proliferator-
activated receptor (PPAR) coactivator-1 (PGC1) signaling as
a multi-faceted regulator coordinating CM maturation via its
novel targets Yap1 and SF3B2.
Results
CMs Exhibit High Levels of Transcriptomic Heterogeneity
During Postnatal Maturation.
Despite serving as a powerful
tool in biology and medicine, the large size and fragility of
mature CMs have limited the use of scRNA-seq in study-
ing CM growth and disease (Ackers-Johnson et al., 2018).
Conventional sorting or microfluidic platforms result in dam-
aged/ruptured myocytes due to inappropriate nozzle size/ flow
rate, leading to abnormal transcript reads and cell death. To
address this, we recently tested LP-FACS and found that it
allows isolation of healthy, mature myocytes, enabling both
high-quality scRNA-seq and functional analysis (Kannan et
al., 2019). Utilizing LP-FACS, we asked how postnatal CMs
become mature cells at the single cell transcriptome level. We
harvested hearts from postnatal day (p) 0, p7, p14, p21, and
p28 mice and dissociated CMs using standard Langendorff
perfusion, followed by isolation of viable single myocytes with
LP-FACS (Kannan et al., 2019) (Figure 1A). Curiously, tSNE
(van der Maaten and Hinton, 2008)-based clustering of p0–p28
single cell samples show partial segregation (Figure 1B), sug-
gesting that significant numbers of cells at each stage may
have similar transcriptome profiles as those of cells present at
the other stages. We then used Monocle2 (Qiu et al., 2017)
to organize transcriptome profiles of CMs at different devel-
opmental stages based on transcriptomic similarities. The
Monocle-based analysis produced a single pseudotime trajec-
tory showing a progressive, unidirectional pattern of CMs over
the course of maturation (Figure 1C). Strikingly, individual
cells from the same stages were found distributed broadly over
the trajectory (Figure 1D). When comparing the maturation
score (as indicated by pseudotime) among different timepoints,
we see that while average maturation scores increase with
age, transcriptomic heterogeneity is maintained even at p28
myocytes (Figure 1D). In agreement with this, similar levels
of cell size heterogeneity were found in postnatal CMs (Figure
1E).
We next examined the expression profiles of maturation-
associated genes by plotting them along pseudotime, grouped
by function. Expression levels of structural genes known to be
upregulated in mature CMs, including Myh6, Tnni3, Myom2
(Yang et al., 2014), were gradually increased (Figure 1F). Cal-
cium handling genes and ion channels including Ryr2, Atp2a2,
Casq1, critical for contractility development, were continually
upregulated while genes involved in cell cycle, including Cdk1,
Cdk4, Ccnd1 (Mohamed et al., 2018; Yang et al., 2014), were
downregulated (Figure 1F). Conserved genes governing cel-
lular hypertrophy, including Mtor, Yap1, Igf1r (Lloyd, 2013;
Perez-Gonzalez et al., 2019), were modestly downregulated
or maintained in expression as they are known to regulate
cell volume (Figure 1F). Gene ontology (GO) analysis indi-
cated that genes related to muscle contraction and cellular
metabolism are highly regulated during the process (Figure
S1A). Consistently, hierarchical clustering showed that mi-
tochondrial gene expression is gradually increased over time
(Figure S1B). Our single-cell approach quantitatively shows
that postnatal CM maturation takes place in a continuous,
but highly heterogeneous fashion.
Gene Network Analysis Predicts PGC1/PPAR Signaling as A
Key Regulator of Cardiac Maturation .
The regulatory mecha-
nisms underlying cardiomyocyte maturation are largely un-
known. To gain mechanistic insights into upstream regulators
governing postnatal CM maturation, we used Ingenuity Path-
way Analysis that infers regulators of differentially expressed
genes by knowledge base of expected effects between tran-
scriptional regulators and their target genes (Kramer et al.,
2014). Using the scRNA-seq dataset, we first generated gene
regulatory networks with predicted upstream regulators by
p-value and activation scores. We found that transcriptional
cofactors and a group of nuclear receptors (PPARs, thyroid
hormone receptors, retinoid receptors, etc) are predicted to be
upstream transcriptional regulators significantly affecting the
overall gene expression changes during maturation (Figure 2A,
Figure S1C, Figure S1D, and Figure S1E). We further used a
cut off of 10e-5 for the false discovery rate and ranked them
by p-value. This analysis inferred the transcriptional cofactors
PGC1
α
/
β
—two PGC1 isoforms—as among the most influen-
tial factors (Figure 2C). Thus, we quantified their expression
levels in developing hearts and PSC-CMs. Expression of the
cofactors and nuclear receptors was gradually increased from
embryonic to adult stages in vivo, with more pronounced upreg-
ulation after birth (Figure 2B, d). An incremental expression
pattern was also observed in most of the genes in long-term
cultured PSC-CMs, but PGC1/PPAR
α
levels remained con-
stantly low (Figure 2B). This indicates that PGC1/PPAR
α
are misregulated in PSC-CMs and may be responsible for
their maturation arrest (DeLaughter et al., 2016; Uosaki et
al., 2015).
PGC1 Is Required Cell-Autonomously for Postnatal CM
Growth and Contractility Development.
PGC1
α/β
are con-
served transcriptional coactivators for PPARs and other nu-
clear receptors and known as central regulators of energy
metabolism (Finck and Kelly, 2006). The two isoforms show
extensive sequence homology with functional redundancy (Lai
et al., 2008; Rowe et al., 2010). In fact, embryonic deletion of
either PGC1
α
or PGC1
β
does not affect heart formation, but
the double knockout results in lethality soon after birth with
small hearts accompanied by mitochondrial defects (Lai et al.,
2008). However, these studies deleted the alleles globally or in
embryonic stages, leaving their cell-autonomous, postnatal role
unknown. Based on the single cell bioinformatics prediction
and low levels of expression in PSC-CMs, we hypothesized
that PGC1 intrinsically mediates postnatal maturation of
CMs. To test this, we generated conditional mosaic knockout
(cmKO) mice, which avoids non-cell-autonomous effects and
early lethality caused by global or conditional deletion (Figure
3A). For this, we generated PGC1/alpha//beta flox/flox; Ai9
mice and administered AAV vectors expressing Cre specifi-
cally in CMs (AAV9-cTnT-iCre) at p0. In this system, cmKO
cells are generated in neonatal CMs and identified by RFP
expression. We titrated AAV vector particles and injected
subcutaneously a dose of 2e10 genome copies per mouse that
2 | Murphy et al.
.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 February 7, 2020. ; https://doi.org/10.1101/2020.02.06.937797doi: bioRxiv preprint
Fig. 1. Postnatal CMs exhibit high levels of transcriptomic heterogeneity
A, Experimental design for scRNA-seq and computational analysis of CMs isolated from p0–p28
hearts. B, t-SNE plot representation of p0–p28 CMs. C, Monocle-based developmental trajectory of p0–28 CMs. D, Distribution of normalized pseudotimes (maturation scores)
by age. E, FACS-based cell size analysis with time of flight. F, Log expression of CM genes associated with structural maturation, calcium handling, cell cycle, hypertrophy, and
ion channels plotted over pseudotime.
results in a mosaic heart with 5–10% RFP+ myocytes. The re-
sulting RFP+ myocytes showed efficient deletion of PGC1
α
/
β
,
quantified by qPCR (Figure S2A, Figure S2B, and Figure
S2C). We made transverse sections and analyzed RFP+ cells
with
α
-actinin staining. RFP+ cells appeared smaller than
neighboring myocytes in size (Figure 3B). To precisely quan-
tify size, we dissociated CMs using Langendorff perfusion and
measured the areas of RFP+ and RFP- CMs after plating at
low density (Figure 3B). RFP- cells showed a heterogeneous
but progressive increase in size over time, consistent with
single-cell LP-FACS cell volume analysis (Figure 3C, cyan
columns). While RFP expression itself did not affect cell size
(Figure S2D), we observed that RFP+ cells remain persistently
smaller compared to RFP- cells. (Figure 3C, red columns).
This suggests the requirement of PGC1 in cellular hypertrophy.
Next, we tested whether PGC1 deficiency affects intact CM
contractile function by video microscopy. RFP+ cells showed
significantly lower fractional shortening and contraction veloc-
ity (Figure 3D, Figure 3E, Figure 3F, Figure S2E, Figure S2F,
Figure S2G, and Figure S2H). Their contractile properties
were further assessed by measuring calcium transients with
the ratiometric dye Fura-2 AM. Consistent with the sarcomere
shortening data, RFP+ cells showed significantly lower peak
Ca2+ amplitude and slower velocity than RFP- cells (Figure
3G, Figure 3H, Figure 3I, Figure S2I, Figure S2J , Figure S2K
, and Figure S2L). This kinetics suggest that PGC1-deficient
myocytes develop less mature calcium cycling apparatus for
Ca2+ release and re-sequestration. Together, our mosaic gene
Murphy et al. bioRxiv | February 6, 2020 | 3
.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 February 7, 2020. ; https://doi.org/10.1101/2020.02.06.937797doi: bioRxiv preprint
Fig. 2. PGC1/PPAR is a predicted key upstream regulator of CM maturation
A, Top transcriptional regulators plotted by p-value and IPA activation z-score with nuclear
receptors highlighted in red. B, Heatmap of gene expression of PGC1 and nuclear receptors in developing mouse hearts and cultured PSC-CMs, quantified by qPCR. C, P-value
ranking of top upstream regulators of CM maturation with two PGC1 isoforms highlighted in red. D, Expression trends over pseudotime of PGC1 and PPAR
α
in postnatal CMs.
deletion approach reveals a cell-autonomous, required role of
PGC1 in cellular hypertrophy and contractility development
of postnatal CMs at the single cell level.
PGC1/PPARα Regulates Genes Affecting Cell Size, Calcium
Handling, and Mitochondrial Activity .
Given the crucial role
of PGC1 in developing myocyte hypertrophy and contractil-
ity, we investigated how PGC1 mediates these processes at
the single-cell transcriptome level. After deleting PGC1 as
above, we isolated RFP+ (PGC1 cmKO) cells by LP-FACS
from p7–p28 hearts and conducted scRNA-seq analysis. The
resulting trajectory reconstructed by Monocle with control
and cmKO cells showed that PGC1-deficient CMs become
less heterogeneous at p7 (Figure 4A and Figure 4B). They
also maintained lower maturation scores throughout the stages
(Figure 4B), which is consistent with the failure to increase in
contractility and size.
To determine how PGC1 signaling mediates these processes,
we used fuzzy clustering to group the temporal trends of indi-
vidual gene expression in both control and PGC1 cmKO CMs
(Figure 4C and Figure 4D). We initially selected upregulated
and downregulated clusters for each group and subsequently
performed overlap analysis. Notably, the analysis showed that
less than 7.6% or 16.2% of genes overlap between control and
PGC1 cmKO CMs in upregulated or downregulated clusters,
respectively (Figure 4C and Figure 4D). This suggests that
gene regulatory networks associated with maturation have
been severely disrupted in PGC1-deficient CMs. GO anal-
ysis of differentially expressed genes (329 upregulated, 255
downregulated) showed that muscle fiber/cell development
and mitochondrial/electron transport chain processes are sig-
nificantly impaired in PGC1 cmKO cells (Figure 4E).
PGC1 binds to PPAR
α
to transcriptionally activate their
target genes (Vega et al., 2000). To identify their target genes
during postnatal CM development, we performed ChIP-seq
on p7 hearts with antibodies targeting PGC1
α
and PPAR
α
.
Antibody specificities were validated by enrichment of their
known target genes (Figure S3A, Figure S3B, Figure S3C,
Figure S3D, and Figure S3E). Motif analysis revealed that
binding sites for ESRRA, NRF, THR
β
, and RORA are en-
riched in chromatin bound by PGC1/PPAR
α
which identified
known and predicted binding partners (Figure 4F). We further
compared genes differentially expressed in PGC1-deleted CMs
with ChIP-enriched peaks in the genome and used HOMER
to annotate peaks. This analysis identified 148 genes directly
regulated by PGC1/PPAR
α
(Figure 4G and Table S1). They
included several novel targets of PGC1/PPAR
α
with unknown
roles in CM maturation, including genes encoding Yap1, a
transcriptional effector of Hippo signaling with crucial roles
in cardiac regeneration (Xin et al., 2013), SF3B2/SAP18,
RNA splicing factors (Golas et al., 2003; Singh et al., 2010),
TIMM50, a mitochondrial translocase regulating mitochondrial
function (Tort et al., 2019), and STRIP1, a core component
of the striatin-interacting phosphatases and kinase complex
regulating cell contractility (Suryavanshi et al., 2018).
PGC1/PPARα Agonists Increase Size, Contractility, and Mito-
chondrial Activity of PSC-CMs .
Since PGC1 is required for
CM hypertrophy and contractility, and its levels and activ-
ity remain low in PSC-CMs (Uosaki et al., 2015), we further
investigated if increased levels of PGC1 can promote the mat-
uration of PSC-CMs. For this, we initially increased PGC1
levels by expressing GFP-PGC1 (Puigserver et al., 1998) in
PSC-CMs. We found that GFP+ CMs became significantly
larger than GFP- CMs after transfection (Figure S4A). The
hypertrophic growth was recapitulated by pharmacological
4 | Murphy et al.
.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 February 7, 2020. ; https://doi.org/10.1101/2020.02.06.937797doi: bioRxiv preprint
Fig. 3. PGC1 is required for CM hypertrophy and contractility development
A, Experimental scheme showing generation and analyses of a cmKO heart achieved by
injection of AAV9-cTnT-Cre into PGC1
α
/
β
flox/flox; Ai9 mice at p0. B, Heart slice showing a cmKO myocyte in myocardium (top) and dissociated control (middle) and cmKO
(bottom) myocytes. C, Violin plots of cell area distributions in control (blue) and cmKO (red) CMs at p7, p14, p28. n=44,11,90,49,522,132 (left to right). D–F, Sarcomere
shortening data with the average trace, fractional shortening and contraction velocity. Control n=13, cmKO n=19. G–I, Calcium handling with average calcium trace, peak height,
and departure velocity. Control n=12, cmKO n=9. p-value: *<0.5,**<0.1,***<0.01.
stimulation with pyrroloquinoline quinone (PQQ), a widely
used PGC1 activator (Chowanadisai et al., 2010) (Figure 5A,
Figure 5B, Figure 5F, and Figure S4C). A similar effect was
also observed when cells were treated with the PPAR
α
-specific
ligand WY14643 (Krey et al., 1997). Individual treatment
significantly improved contractile force, measured by traction
force microscopy, which was further increased when combined
(Figure 5C and Figure S4B). This phenotype was accompanied
with a significant increase in mitochondrial density and ac-
tivity, determined by electron microscopy and Seahorse assay
(Figure S4D, Figure S4E, Figure S4F, and Figure S4G). These
data suggest that PGC1/PPAR
α
have an instructive role in
PSC-CM maturation.
PGC1/PPARα Signaling Promotes CM Maturation by Regu-
lating Key Upstream Regulators of Cellular Hypertrophy and
Calcium Handling .
To determine how PGC1/PPAR
α
medi-
ates CM growth, we analyzed expression levels of conserved cell
size regulators Mtor, Yap1, Igf1. Among these, we found that
Yap1, one of the validated PGC1/PPAR
α
targets by ChIP-
seq, is markedly downregulated in PGC1 cmKO cells (Table
S2). To test if Yap1 affects CM size in vivo, we generated
Yap1 cmKO cells in postnatal hearts as shown in Figure 2E
and analyzed their growth. Notably, the cmKO CMs became
significantly smaller than normal CMs (Figure 5D and Figure
5E), indicating that Yap1 may mediate PGC1/PPAR
α
sig-
nals for CM hypertrophy. To test this, we chemically blocked
Yap1 transcriptional activity with the Yap1 inhibitor (R)-
Murphy et al. bioRxiv | February 6, 2020 | 5
.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 February 7, 2020. ; https://doi.org/10.1101/2020.02.06.937797doi: bioRxiv preprint