YAP-Induced Glycolysis Drives Fibroinflammation and Disrupts Fibroblast Fidelity

Abstract

Background:

Separation of the pulmonic and systemic circulation is essential for terrestrial life and mammalians have evolved distinct cardiac chambers with specialized structures and functions. Transcriptomics profiling revealed cellular heterogenicity between heart chambers. However, the mechanisms underlying chamber-specific transcriptomic and metabolic differences—and their functional significance—remain poorly understood. The Hippo/YAP pathway is a conserved signaling network that regulates diverse cellular processes. The Hippo kinases inhibit YAP in cardiac fibroblasts (CF) to restrict fibrosis and inflammation. Nonetheless, how YAP regulates the metabolic microenvironment during homeostasis and fibroinflammation remains unclear.

Methods:

We investigated YAP and glycolysis activity in the four cardiac chambers by scoring the expression of YAP target genes and glycolysis genes in human single-nucleus RNA sequencing (snRNAseq) data. To compare glucose uptake between the left and right atria, we measured isotope-labeled glucose uptake in isolated mouse atria. To study the role of YAP in CFs, we inactivated the Hippo kinases, Lats1 and Lats2, in mouse CFs and performed metabolic studies, snRNA-seq, single-nucleus Assay for Transposase-Accessible Chromatin with sequencing, and spatial transcriptomics.

Results:

Metabolic and sequencing approaches revealed that Hippo-deficient CFs activated glycolysis to promote fibroinflammation. Inhibition of glycolysis or lactate production suppressed Hippo-deficient CF-induced fibrosis. Elevated YAP activity disrupted fibroblast lineage fidelity by inducing an osteochondroprogenitor (OCP) cell state. Blocking macrophage expansion pharmacologically reduced Hippo-deficient CF proliferation and fibrosis. Sequencing and functional studies showed that macrophages secreted insulin-like growth factor 1 (IGF1) to activate IGF1 signaling in Hippo-deficient CFs to increase cell proliferation and fibrosis.

Conclusion:

We discovered that right atrial CFs are more glycolytic and have higher YAP activity than CFs in other heart chambers. YAP activation in CFs induces glycolysis to drive fibrosis. YAP disrupts fibroblast lineage fidelity, driving them to a SOX9-expressing OCP cell state. Mechanistically, YAP activates the secretion of colony stimulating factor 1 (CSF1) to promote macrophage expansion. Blocking macrophage expansion reduces Hippo-deficient CF proliferation, OCP differentiation, and fibrosis, revealing that macrophages signal reciprocally to regulate CF cell states. Genomic and functional studies revealed that upregulated IGF1 receptor in Hippo-deficient CFs enables them to receive macrophage-secreted IGF1, thereby further enhancing CF proliferation and fibrosis.

Introduction

The evolution of a multi-chambered heart capable of separating pulmonary and systemic circulation represents a key adaption for terrestrial life. Single cell and spatial transcriptomics profiling revealed cellular heterogenicity between chambers ^1,2. It remains unclear how cell state heterogeneity between chambers is established and if there are implications for heart disease. Chamber specific transcriptomic differences, metabolism, cell fate, and cardiac microenvironment remain poorly understood in part because most work is focused on the ventricles.

The Hippo pathway, a conserved signaling pathway is composed of core Hippo kinases including the MST1/2 and LATS1/2 kinases, which phosphorylate and inhibit the transcription co-activators YAP and TAZ by promoting their cytoplasmic retention ^3,4. In a Hippo-low context, YAP and TAZ enter the nucleus and bind to transcription factor partners, such as TEA domain family members (TEAD1–4), to promote gene expression. In previous work, we deleted the Lats1/2 kinases in resting CFs, with a focus on ventricular CFs, and discovered that increased Yap function in CFs induced fibrosis, inflammation, and spontaneous induction of the myofibroblast cells state ⁵. Consistent with our data, other groups have shown that loss of YAP and TAZ reduces fibrosis and inflammation following cardiac injury ^6–8. In human cardiac fibrosis, the TEAD family has been implicated in human interstitial cardiac fibrosis ⁹. Nonetheless, how YAP regulates the fibroinflammatory and metabolic microenvironment is still poorly understood.

Here, we discovered that human and mouse right atrial (RA) CFs are more glycolytic and have higher baseline YAP activity than CFs in other chambers. YAP activation in CFs induces glycolysis to drive fibroblast activation and fibrosis. YAP disrupts fibroblast lineage fidelity, driving them to differentiate into a SOX9-expressing osteochondroprogenitor (OCP) cell state, where glycolysis is the preferred metabolic pathway ¹⁰. Mechanistically, YAP in CFs induces CSF1 secretion to promote Mac expansion within the microenvironment. Inhibiting CSF1 signaling, which suppresses Mac expansion, reduces Hippo-deficient CF proliferation, OCP differentiation, and fibrosis, revealing that Macs signal reciprocally to shift CF cell states. Within Hippo-deficient CFs, YAP induces Insulin-like Growth Factor 1 Receptor (IGF1R) expression, which enables them to receive outgoing IGF1 signaling from Macs, thereby further enhancing CF proliferation, fibrosis and OCP fate acquisition.

Methods

Detailed methods are provided in the Supplemental Materials.

Data availability

All data generated or analyzed in this study are included in this published article and its supplementary information files. Source data are provided with the manuscript. All raw and processed sequencing data are deposited at the National Center for Biotechnology Information’s Gene Expression Omnibus (GEO): GSE261643. Previously published datasets used in this study include data obtained from the Human Cell Atlas data portal (https://www.heartcellatlas.org). Code availability In-house code for reproducing all bioinformatics analyses is available at GitHub (https://github.com/XL-Genomics/2024_YAP_in_Atrial_Fibrosis).

Animals

All mouse experiments were approved and performed under Institutional Animal Care and Use Committee protocol #5713 and #5719 at Baylor College of Medicine.

Statistical Analysis

All statistical analyses were performed using GraphPad Prism 9, Python or RStudio. For sample sizes <6, the Mann–Whitney exact test (rank-based) and/or the permutation test¹¹ (median-based, implemented in Python) were applied; the code for the permutation test is provided in the Supplemental Materials. For sample sizes ≥6, normality was assessed with the Shapiro–Wilk test (p>0.05 indicating normality). Normally distributed data were analyzed using an unpaired t test (two groups) or one-way or two-ways ANOVA (with Tukey’s multiple comparisons test or, when variance was unequal, Brown–Forsythe and Welch ANOVA with Dunnett’s T3 multiple comparisons). Non-normally distributed data were analyzed with the Mann–Whitney test (two groups) or the Kruskal–Wallis test with Dunn’s multiple comparisons (more than two groups). One-tailed tests were used when a directional decrease was hypothesized; otherwise, two-tailed tests were performed. Error bars in scatter plots represent either the 95% confidence interval or the standard error of the mean, as specified in the figure legends.

For snRNA-seq gene expression scores, box plots show the median (central line), 25^th-75th percentiles (box), and whiskers extending up to 1.5× the interquartile range. For imaging quantification, box plots show the median (central line), interquartile range (box), minimum and maximum (whiskers), with all individual data points displayed. A full summary of statistics is provided in the Summary of Statistics.

Results

Increased YAP and Glycolytic Activity in Right Atrial Cardiac Fibroblasts

We investigated YAP activity in the four human cardiac chambers using a YAP target gene expression module (YAP score, Table S1), interrogating available snRNAseq data ¹. As previous work revealed that YAP activity is increased by mechanical strain ¹², we predicted that YAP activity is highest in the LV. Unexpectedly, YAP activity is highest in CFs of the RA, which receives deoxygenated venous inflow in right-sided circulation and is the chamber with the lowest mechanical strain (Figure 1A). We used a similar approach to evaluate glycolysis gene expression in human data and observed that RA CFs exhibit the highest glycolysis score (Figure 1B; Table S2). We assayed whether RA imports more glucose than LA by measuring isotope-labeled glucose uptake in isolated mouse atria. Indeed, glucose uptake is higher in the RA than in LA (Figure 1C). Since myofibroblasts reprogram their metabolism to increase glycolysis, these data are consistent with our previous finding that Lats1/2 deletion resulted in spontaneous transition to a myofibroblast cell state ⁵. Furthermore, our findings indicate that YAP activity is elevated in human resting CFs of the RA compared to the other cardiac chambers, consistent with the hypothesis that YAP promotes glycolysis in RA CFs, which we test below.

Figure 1. Hippo-deficient CFs induce right atrial glycolysis, inflammation, and fibrosis.

(A, B) Single nucleus RNA sequencing (snRNAseq) analysis of published human heart data. (A) YAP activity score based on YAP target expression and (B) glycolysis score based on glycolytic gene expression in human CFs from the left ventricle (LV), right ventricle (RV), left atrium (LA), and right atrium (RA). (A, B) Left panels, a total of 37,179 single fibroblast nuclei from 13 donors were included in the analyses. The number of nuclei per donor ranged from 1,029 to 9,320, with a mean of 2,860. Box plots show the median (central line), interquartile range (box), and the whiskers extend to a maximum of 1.5× interquartile range beyond the boxes. Right panels, each connected-dot pair represents LA and RA from a single human donor. One donor was excluded due to unavailability of paired RA (final 12 donors). Paired Wilcoxon signed-rank test. (C) Glucose uptake of mouse LA and RA. Each connected-dot pair represents LA and RA from a single mouse (n=8). Wilcoxon matched-pairs signed rank test. (D) Experimental scheme. (E) Sirius red staining (collagen) of heart sections from control, Lats1/2^ΔCF, and Lats1/2YapTaz^ΔCF mice collected two weeks after Cre activation. Upper panels, low power images. Lower panels high magnification images. (F) Quantification of fibrotic area (defined as the collagen area over total tissue area) for control (n=4), Lats1/2^ΔCF (n=5), and Lats1/2YapTaz^ΔCF (n=11) mice. The box plot shows the median (central line), interquartile range (box), minimum and maximum (whiskers), with all data points displayed. Kruskal-Wallis test, multiple comparisons. (G-I) snRNAseq analysis showing the (G) uniform manifold approximation and projection (UMAP) of major cell types in control and Lats1/2^ΔCF atria. Major cell types include epicardial cells (EpiC), cardiac fibroblasts (CF), endothelial cells (Endo), cardiomyocytes (CM), and immune cells (IC). (H) Differentially expressed genes in indicated cell types. (I) Increased cell types in Lats1/2^ΔCF LA and RA compared to control. (J-M) Spatial transcriptomics (ST) analyses. (J) Control and Lats1/2^ΔCF heart sections stained with hematoxylin and eosin and anatomically annotated. Inter-atrial septum (IAS); interventricular septum (IVS); mitral valve (MV); tricuspid valve (TV); pulmonary artery (PA); aorta (Ao); aorta valve (AoV). (K) Deconvolution of major cell types from ST data. (L) ST data-derived YAP Score and Glycolysis Score in control and Lats1/2^ΔCF hearts and (M) their correlation in different samples and heart regions. (N) Glycolysis gene expression in CFs in snRNAseq dataset. (O-P) YAP binding in 118 YAP target genes versus 21,782 background genes (O) and 72 glycolysis genes vs 21,828 background genes (P) in YAP CUT&RUN (cleavage under targets and release using nuclease) in NIH3T3 cells Box plots show the median (central line), interquartile range (box), and the whiskers extend to a maximum of 1.5× interquartile range beyond the boxes. Unpaired Wilcoxon rank-sum test. (Q) Histone 3 lysine 27 acetylation (H3K27ac) and YAP signals in Hexokinase 2 (Hk2) locus in NIH3T3 cells.

Hippo Pathway Loss of Function Results in Atrial Arrythmias, Fibrosis and Inflammation

To discern YAP function in the RA and LA, we used the inducible Cre transgenic mouse line, Tcf21^iCre, to inactivate the Hippo pathway kinases Lats1 and Lats2 (Lats1/2) in resting mouse CFs (Lats1/2^ΔCF mice, Figure 1D). We carefully determined that Cre activity and Lats1/2 deletion efficiency were equivalent in RA and LA, ruling out variation in Cre activity as a trivial cause of distinct atrial phenotypes (Figure S1). We previously used this mouse model to investigate ventricular CFs ⁵ and in the current study we examine atrial phenotypes before ventricular phenotypes emerge.

Two weeks post-Cre induction, histology revealed elevated cellularity in the RA of Lats1/2^ΔCF mice compared to LA, as well as control atria (Figure S2A). Sirius red staining revealed dramatically increased collagen deposition in Lats1/2^ΔCF RA compared to LA and control tissue (Figure 1E and F). Importantly, we did not observe hyperplasia or fibrosis in Lats1/2^ΔCF LA or ventricles at any time points examined in this study (Figure S2A, B). Since atrial fibrosis has been linked to arrythmia¹³, we investigated whether Hippo-deficient CFs induce arrythmias by implanting telemetry devices into the mice. Electrocardiographic analysis of Lats1/2^ΔCF mice showed an irregular RR interval, an inverted P-wave, and increased PP interval (Figure S3A-D), which was associated with a slower heart rate and is consistent with sinus node dysfunction (Figure S3A-E) ¹⁴. Thus, Hippo-deficient CFs promote atrial arrythmia.

Based on our previous findings, we hypothesized that YAP and TAZ are 1) the primary LATS1/2 substrates, 2) are phosphorylated and inactivated by LATS1/2, and 3) deleting Yap and Taz would suppress the Lats1/2^ΔCF RA phenotype ⁵. Indeed, concurrent Yap and Taz loss of function in Lats1/2^ΔCF suppresses fibrosis and hyperplasia in Lats1/2^ΔCF RA (Figure 1E and 1F; Figure S2A, lower panels).

YAP Activation in Cardiac Fibroblasts Induces Glycolysis and a Fibroinflammatory Microenvironment in Right Atrium

Ten days after Cre activation, we performed single nucleus RNA sequencing (snRNAseq, Figure 1D). Using unbiased cell clustering, we identified eight major clusters (Figure 1G). Integrating differential gene expression with cell type-specific markers (Figure 1H), we identified CFs, cardiomyocytes (CMs), epicardial cells (EpiCs), endothelial cells (ECs), immune cells (ICs), Mural cells, Schwann cells, and adipocytes. In comparing cell type composition between samples, we discovered that CFs and ICs are enriched in Lats1/2^ΔCF RA, but not Lats1/2^ΔCF LA or controls (Figure 1I, Figure S4A), supporting the notion that YAP activity is more potent in the RA.

To obtain unbiased, spatially resolved cell state data to interrogate the Lats1/2^ΔCF microenvironment, we performed ST analysis two weeks after Cre induction with high and low dose tamoxifen concentrations to induce varying levels of Cre activity (Figure 1D). We used hematoxylin and eosin staining to annotate heart anatomy (Figure 1J and Figure S4B), integrated snRNAseq and ST datasets, and deconvoluted ST data using major cell type markers identified in the snRNAseq dataset. We also performed unbiased ST spot clustering and found that it recapitulated our anatomical clustering (Figure S5). We observed that CFs and ICs are enriched, while CMs are reduced in Lats1/2^ΔCF RA compared with controls (Figure 1K and Figure S4C). Indeed, CM area over total tissue was significantly reduced in Lats1/2^ΔCF RA (Figure S6A and S6B) as CFs expanded (Figure S6A and S6C; Figure S7 and S8). To test if CMs undergo cell death, we performed Terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay and found that CM death was increased in Lats1/2^ΔCF RA (Figure S6D and S6E). Thus, CM reduction in Lats1/2^ΔCF RA is due to both non-CM expansion and CM death.

To determine whether YAP promotes glycolysis in CFs, we scored ST data using the YAP target gene expression module (YAP score) and glycolysis gene expression module (glycolysis score). Compared with controls, Lats1/2^ΔCF RA displayed increased YAP and glycolysis scores (Figure 1L, M and Figure S4D), consistent with our observations from human RA. Our snRNAseq data revealed that glycolytic gene expression is increased in Lats1/2^ΔCF RA CFs compared to LA CFs and controls (Figure 1N). To determine whether YAP directly upregulates glycolytic genes, we reanalyzed our previous YAP CUT&RUN (Cleavage Under Targets and Release Using Nuclease) NIH3T3 data⁵. Consistent with our hypothesis, we observed YAP binding to known YAP targets and glycolysis genes (Figure 1O, P), supporting the conclusion that YAP/TEAD directly regulate glycolysis genes. Indeed, Histone 3 lysine 27 acetylation (H3K27ac), a marker for active transcription, colocalized with YAP binding in Hexokinase 2, a key glycolysis enzyme (Figure 1Q). Together, these results support the conclusion that Hippo-deficiency in CFs, with increased YAP activity, promotes glycolysis in CFs.

Glycolysis and Lactate Production are Required for Fibrosis in Hippo-deficient Hearts

Bulk RNAseq from CFs in Lats1/2^ΔCF hearts revealed upregulation of genes encoding glycolysis enzymes, as well as other genes that promote glycolysis (Figure 2A). Among these are Bisphosphoglycerate mutase (Bpgm), Phosphoglycerate Mutase 1 (Pgam1), Enolase 1 (Eno1), the glucose transporter Slc2a6, cMyc, and Insulin-like growth factor 1 receptor (Igf1r). As further validation, in situ hybridization revealed that the rate-limiting glycolytic enzyme hexokinase 2 (Hk2) and the glucose transporter Slc2a1 were upregulated in Lats1/2^ΔCF RA CF compared to control (Figure 2B-D).

Figure 2. Hippo-deficient CFs induce glycolysis and fibrosis.

(A) Previously published bulk RNAseq data⁵ of control and Lats1/2-deleted CFs sorted from adult mice. Selected glycolysis-related genes were z-scored and plotted. n=3, each group. (B) Hk2 and Slc2a1 expression in RA of indicated genotypes visualized by in situ hybridization. Cre lineage cardiac fibroblasts were labeled by a perinuclear GFP reporter. Scale bar, 50 μm. (C, D) Quantification of Hk2 (C) and Slc2a1 (D) expression in Cre lineage CFs. Hk2 and Slc2a1 levels were measured from 15 randomly selected CF nuclei per mouse from three control and four Lats1/2^ΔCF right atria. Scatter plots show the mean with 95% confidence intervals; each dot represents one CF nucleus. Control (n=45) and Lats1/2^ΔCF (n=60). Mann-Whitney test, two-tailed. (E) Seahorse Glycolysis Stress Test was used to measure the glycolysis capacity of primary cultured control or Lats1/2-deleted CFs. Error bar, standard error of the mean. n=16, each group. (F) Differential glycolysis metabolites from control and Lats1/2-deleted CFs. FDR<0.25. (G) Illustration of differential metabolites in glycolysis. (H) Sirius red-stained sections from the RA of Lats1/2^ΔCF mice treated with vehicle or glycolysis inhibitor, 2-deoxyglucose (2-DG). Lower panels: inset images of upper panel boxed regions. (I) Quantification of collagen over total tissue area in (H). n=5, each group. Mann-Whitney test, two-tailed. (J) Proliferating Lats1/2-deleted CFs in the RA of Lats1/2^ΔCF mice treated with vehicle or glycolysis inhibitor were labeled with EdU staining (red). Tcf21^iCre lineage CFs (GFP, green). White arrowheads, examples of proliferating Tcf21^iCre lineage CFs. Yellow arrows, examples of proliferating non-Tcf21^iCre CFs. (K) Quantification of the EdU-positive Tcf21^iCre lineage RA CFs of Lats1/2^ΔCF mice treated with vehicle (n=4) or glycolysis inhibitor (n=3). Mann-Whitney test, one-tailed. (L) Sirius red-stained sections from the RA of Lats1/2^ΔCF mice treated with vehicle or lactate dehydrogenase A inhibitor, oxamic acid (OXA) 750 mg/kg daily for two weeks. (M) Quantification of collagen over the RA tissue area from Lats1/2^ΔCF mice treated with vehicle (n=4) or OXA (n=3). Mann-Whitney test, one-tailed. (I,K,M) The box plot shows the median (central line), interquartile range (box), minimum and maximum (whiskers), with all data points displayed.

To obtain functional evidence that YAP induced glycolysis in CFs, we performed a glycolytic stress test using isolated CFs from mouse hearts (Figure 2E). CFs were infected with adenoviral Cre to delete Lats1/2, and we infected CFs with adenoviral GFP as controls. Lats1/2-deleted CFs displayed higher extracellular acidification, indicating increased glycolysis and lactic acid production into the medium compared to controls (Figure 2E, Figure S9A and S9B). Glycolysis metabolite analyses revealed that Lats1/2-deleted CFs were depleted for glucose and had increased glycolysis intermediates: fructose 1,6-bisphosphate (FBP), glyceraldehyde 3-phosphate, and glycerol 3-phosphate (Figure 2F and 2G). Indeed, lactate in the culture medium of Lats1/2-deleted CFs was higher than in control CF cultured medium (Figure S9C), indicating that anaerobic glycolysis with lactate production was increased in Lats1/2-deleted CFs, which had increased YAP activity.

To determine whether glycolysis is required for YAP-induced fibroinflammation in vivo, we inhibited glycolysis by injecting 2-deoxy-d-glucose (2-DG), a competitive inhibitor of hexokinase, into Lats1/2^ΔCF mice. Lats1/2^ΔCF RA fibrosis and CF proliferation were suppressed by 2-DG, supporting the conclusion that glycolysis is essential for the Lats1/2^ΔCF RA fibroinflammation phenotype (Figure 2H-K). To test if YAP induces glycolysis to promote fibroblast activation, nuclear size and the expression of α-smooth muscle actin (α-SMA) were measured. Increased nuclear size and α-SMA protein level were observed in Hippo-deficient CFs compared to control CFs, while glycolysis inhibition suppresses these phenotypes (Figure S10), indicating that glycolysis is required for YAP-induced fibroblast activation. To determine whether lactate production drives YAP-induced fibrosis, we inhibited lactate dehydrogenase A by injecting oxamic acid (OXA) into Lats1/2^ΔCF mice. OXA treatment also reduced Lats1/2^ΔCF RA fibrosis, revealing that lactate production is required for YAP-induced fibrosis (Figure 2L and 2M). Together, these data indicate that increased YAP activity in Lats1/2^ΔCF CFs induces glycolysis and lactate production, which promotes a metabolic microenvironment permissive for fibroinflammation.

YAP Induces an Osteochondroprogenitor Cell State in Adult Cardiac Fibroblasts

In comparing cell type composition in ST data, Lats1/2^ΔCF RA was enriched for chondrocyte, osteoblast, and pancreatic stellate cell states compared to controls (Figure 3A). ST data further revealed that chondrocyte and osteoblast markers were upregulated in Lats1/2^ΔCF RA (Table S3, Figure 3B and S11A). snRNASeq evaluation of osteoblast and chondrocyte markers revealed that CFs displayed the highest scores, supporting the data that Lats1/2^ΔCF RA CFs transition to chondrocyte and osteoblast-like cell states (Figure S11B and S11C). Enriched osteoblast and chondrocyte marker genes are co-expressed, consistent with the notion that Lats1/2^ΔCF CFs acquire an OCP cell state (Figure S11D). Alcian blue staining revealed that chondrocyte-secreted proteoglycans in the cartilage matrix are upregulated in Lats1/2^ΔCF RA (Figure 3C and 3D). This staining was reduced when Lats1/2 and Yap/Taz were simultaneously deleted (Figure 3C and 3D), further supporting the conclusion that YAP and TAZ are required LATS1/2 substrates in this context.

Figure 3. Differentiation of Hippo-deficient RA CFs into OCPs.

(A) In the ST data, top enriched cell types in the Lats1/2^ΔCF RA compared with controls. (B) chondrocyte and osteoblast scores (marker expression) in the ST data. (C) Alcian blue (proteoglycan) staining of RA sections from control (n=4), Lats1/2^ΔCF (n=6), and Lats1/2YapTaz^ΔCF (n=3). (D) Percentage of Alcian blue area as defined by Alcian blue over RA tissue area. The box plot shows the median (central line), interquartile range (box), minimum and maximum (whiskers), with all data points displayed. Kruskal-Wallis test, multiple comparisons. (E-J) snRNAseq analysis of CF subclusters. UMAP (E), differentially expressed genes (F), and sample sources (G) of CF subclusters from control and Lats1/2^ΔCF atria. (H) YAP target gene expression (YAP score) and OCP marker expression (OCP score) and their correlation (I). OCP marker expression in the CF subclusters in the snRNAseq (J) and ST (K). (L) SOX9 (red) immunostaining of control and Lats1/2^ΔCF RA sections, GFP (white, Cre lineage cells), DAPI (blue, nuclei), Scale bar, 50 μm. (M) Quantification of SOX9-positive Tcf21^iCre lineage cells (white). n=3, each group. Mann-Whitney test, one-tailed. (N) In situ hybridization of Col2a1 mRNA (OCP marker, red) and immunostaining of GFP (Cre lineage, white) in control and Lats1/2^ΔCF RA. DAPI (blue, nuclei). Scale bar, 50 μm. (O) Quantification of Col2a1-positive Tcf21^iCre lineage cells in control and Lats1/2^ΔCF. n=3, each group. Mann-Whitney test, one-tailed. (M,O) The scatter plots show the mean ± standard error of the mean; each dot represents one animal.

We next pooled and reclustered CFs from control and Lats1/2^ΔCF atria into four major subclusters: CF1–4 (Figure 3E). Differential gene expression revealed that CF1–2 are primarily derived from control and LA of Lats1/2^ΔCF (Figure 3F and 3G). CF1–2 marker genes, such as Gelsolin (Gsn), are characteristic of resting CFs (Figure S11E). Moreover, ST revealed that CF1–2 are enriched in control hearts and Lats1/2^ΔCF LA (Figure S11F). CF4 is exclusively derived from Lats1/2^ΔCF RA, expresses canonical YAP targets, such as Amotl2 and Vgll3, and has the highest YAP target score (Figure 3G and 3H; Figure S11E; Table S1). From these data, we concluded that CF4 comprises Lats1/2-deleted CFs (Figure 3H). ST revealed that known YAP targets, Amotl2 and Vgll3 expression (Figure S11F) and YAP target scores (Figure 1L) were increased in Lats1/2^ΔCF RA compared to control mice RA.

As validation, Amotl2 and Vgll3 in situ hybridization in control and Lats1/2^ΔCF RA, which includes co-staining for GFP (Tcf21^iCre lineage cells), revealed that these genes are upregulated in Lats1/2^ΔCF RA and are enriched in Tcf21^iCre lineage (Figure S11H), supporting the data that the CF4 subcluster is predominantly Lats1/2-deleted and YAP active. CF3, expressing markers of CF activation, including Postn, was also enriched in Lats1/2^ΔCF RA (Figure 3F; Figure S11I). In contrast to CF4 which has high YAP activity, CF3 exhibits a low YAP target gene score and low GFP reporter expression (Figure 3E and 3H; Figure S11I), supporting the conclusion that CF3 are CFs located in the Lats1/2^ΔCF RA that escaped Lats1/2 deletion due to Tcf21^iCre inefficiency (Figure S1C).

Evaluation of chondrocyte and osteoblast marker expression in CF subclusters revealed that CF4 cells had the highest OCP scores (Figure 3H). Moreover, OCP and YAP scores are highly correlated in both snRNA-seq and ST datasets, most notably in Lats1/2^ΔCF RA (Figure 3I; Figure S11G). Furthermore, OCP marker expression, including Sox9, Col2a1, Col11a1, Col12a1, Acan, and Runx2, were upregulated in Lats1/2^ΔCF RA compared with controls (Figure 3J and 3K; Figure S11J and S11K). Immunostaining revealed that protein levels of the OCP marker SOX9 are increased in GFP-positive Tcf21^iCre lineage cells in Lats1/2^ΔCF RA compared to controls (Figure 3L, M). Moreover, >90% of SOX9-positive cells are GFP-positive (Figure S11L), indicating that YAP activation autonomously promotes SOX9 upregulation and the OCP cell state. Expression of the SOX9 target Col2a1 ¹⁵, is increased in Lats1/2^ΔCF RA CFs (Figure 3N, O). Since OCPs exhibit senescence features following injury¹⁶, we scored gene expression of senescence and senescence-associated secretory phenotype (SASP) in our ST and snRNAseq datasets. Both senescence and SASP signature are increased in Lats1/2^ΔCF RA and in Hippo-deficient CFs (CF4) (Figure S12). In sum, these results support the conclusion that increased YAP activity in RA CFs induces the OCP cell state in resting CFs.

Outgoing CSF1 signaling from Hippo-deficient Cardiac Fibroblasts Promote Macrophage Expansion

We deconvoluted the ST data from Lats1/2-deleted CFs using the CF4 markers identified in the snRNAseq. The resulting cells, which we refer to as YAP^High CFs, are enriched in Lats1/2^ΔCF RA but not control RA (Figure S11M). In contrast, YAP^Low CFs, identified using CF1–3 markers, are enriched in control atria and Lats1/2^ΔCF LA (Figure S11M). Using the ST data, we compared cell types and found that immune cells (ICs) and YAP^High CFs are colocalized (see Figure 1K). Pooling and reclustering ICs from control and Lats1/2^ΔCF mice yielded six IC subclusters (Figure 4A). By comparing differential gene expression, we identified CCR2+ Macs (MP1), CCR2- Macs (MP2), proliferating Macs (MP3), dendritic cells (DC), T cells (TC) and B cells (BC). MP2 and MP3 are enriched in Lats1/2^ΔCF RA compared to control RA (Figure 4B-D). MP3 shares both MP1 and MP2 signatures although the MP2 signature score is higher than MP1 score (Figure S13). Furthermore, ST data revealed that MP2 colocalizes with YAP^High CF in Lats1/2^ΔCF RA (Figure 4E). To identify the outgoing intercellular signaling from YAP^High CFs, we performed ligand-receptor analysis (CellChat)¹⁷ using our snRNA-seq and identified multiple ligand-receptor pairs (Figure 4F).

Figure 4. Hippo-deficient CFs secrete CSF1 to promote Mac expansion.

(A-D) snRNAseq analyses of immune cell subclusters in Lats1/2^ΔCF and control RA. Macrophage 1–3 (MP1–3), dendritic cell (DC), T cells (TC), and B cell (BC). (A) UMAP. (B) Sample source. (C) Differentially expressed genes of different clusters. (D) Enriched immune cell types in Lats1/2^ΔCF RA (4,503 nuclei from 2 samples) compared to control RA (671 nuclei from 2 samples). Chi-squared test. (E) CFs with a high YAP score (YAP^High CF4) and MP2 were deconvoluted in the ST data using cell markers and the quantification of their colocalization. (F) Increased CF to MP2 signaling pathways in Lats1/2^ΔCF RA identified by ligand-receptor analysis of the snRNAseq data. (G) ST of Csf1 and Csf1r expression and their colocalization in control and Lats1/2^ΔCF mouse atria. (H) RNAscope of Csf1 (red) in Tcf21^iCre lineage cells (GFP, white), DAPI (blue, nuclei), yellow arrowheads (examples of GFP⁺ cells), Scale bar, 50 μm. (I) Quantification of Csf1 expression in Tcf21^iCre lineage cells in the RA. Each dot represents a Tcf21^iCre lineage cell (27 cells from four control mice and 107 cells from four Lats1/2^ΔCF mice). Mann-Whitney test, two-tailed. (J) Immunostaining of Tcf21^iCre lineage cells (anti-GFP, white), Macs (anti-CSF1R, red), DAPI (blue, nuclei), and EdU (proliferating cells, green) in control and Lats1/2^ΔCF RA. Yellow arrowheads (CSF1R-positive Macs). Scale bar, 50 μm. (K) Percentage of EdU-positive Tcf21^iCre lineage cells in the RA of control (n=5) and Lats1/2^ΔCF (n=4) hearts. Mann-Whitney test, two-tailed. (L) Experimental scheme. (M) CSF1R (red) immunolabeling of Macs in Lats1/2^ΔCF RA treated with CSF1R inhibitor or vehicle. DAPI (blue, nuclei), white arrowheads (CSF1R-positive Macs), scale bar, 50 μm. (N) Percentage of CSF1R-positive Macs among all cells (DAPI) in the RA of Lats1/2^ΔCF mice treated with CSF1R inhibitor or vehicle. n=4 per group. Mann-Whitney test, two-tailed. (I,K,N) The box plots show the median (central line), interquartile range (box), minimum and maximum (whiskers), with all data points displayed. (O) Fold-change of immune subclusters in the snRNAseq of RA from Lats1/2^ΔCF mice treated with CSF1R inhibitor (1,103 nuclei from two samples) or vehicle (920 nuclei from one sample). Chi-squared test. (D,O) Box plots show the median (central line), interquartile range (box), and the whiskers extend to a maximum of 1.5× interquartile range beyond the boxes.

As CSF1 is a direct YAP target ¹⁸, we focused on CSF1 receptor (CSF1R) signaling, which regulates Mac survival and proliferation ¹⁹. Our ST data revealed that Csf1 and Csf1r are upregulated in Lats1/2^ΔCF RA and strongly colocalize (Figure 4G). In situ hybridization revealed that Csf1 levels are upregulated in Lats1/2^ΔCF RA CFs (Figure 4H and 4I). In addition, CSF1R antibody immunofluorescence revealed that Macs are sparse in control RA and are more abundant in Lats1/2^ΔCF RA (Figure 4J), consistent with the snRNAseq and ST data. Pulse-chase EdU labeling revealed that proliferating Macs are increased in Lats1/2^ΔCF RA (Figure 4J and 4K). As Lats1/2^ΔCF mice carry the Cre reporter Rosa26^Sun1-GFP, we labeled Tcf21^iCre lineage cells using a GFP antibody (Figure 4J). The majority of GFP-positive RA Lats1/2-deleted CFs were surrounded by Macs, suggesting signaling between Lats1/2-deleted CFs and the surrounding Macs.

Macrophages Promote Hippo-deficient CF Proliferation and OCP Differentiation

To test the functional role of CSF1 signaling, we treated Lats1/2^ΔCF mice with the CSF1R inhibitor GW2580 and performed histology, snRNAseq, single nucleus Assay for transposase-accessible chromatin with sequencing (snATACseq) and ST (Figure 4L and Figure S14A-S14D). Compared to vehicle-treated Lats1/2^ΔCF mice, the GW2580-treated Lats1/2^ΔCF mice had fewer Macs in RA, indicating that CSF1 signaling is required for RA Mac expansion (Figure 4M and 4N). Furthermore, snRNAseq and ST revealed that MP2 and MP3 were reduced in Lats1/2^ΔCF hearts following GW2580 treatment compared to vehicle treated Lats1/2^ΔCF mice (Figure 4O; Figure S14B). These results support the conclusion that outgoing CSF1 signaling from Lats1/2-deleted CFs to Macs induces Mac proliferation. We performed ST on Lats1/2^ΔCF mice treated with GW2580 or vehicle, which revealed that colocalization of Lats1/2-deleted CFs and MP2 was drastically reduced following GW2580 treatment (Figure S14D). Strikingly, Sirius red staining was also dramatically reduced in Lats1/2^ΔCF RA after GW2580 treatment, supporting the conclusion that Macs enhances RA fibrosis (Figure 5A and 5B).

Figure 5. Macs promote differentiation of Hippo-deficient CFs into OCPs.

(A) Sirius red-stained sections from the RA of Lats1/2^ΔCF mice treated with vehicle or CSF1R inhibitor. Lower panels: high magnification of the boxed regions in upper panels. (B) Quantification of collagen over RA tissue area. Vehicle, n=9. CSF1R inhibitor, n=10. Mann-Whitney test, two-tailed. (C) Proliferating Lats1/2-deletd CFs in the RA of Lats1/2^ΔCF mice treated with vehicle or CSF1R inhibitor were labeled with EdU (green), Lats1/2-deleted CFs (GFP, white), scale bar, 50 μm. (D) Percentage of EdU-positive RA Lats1/2-deleted CFs in mice treated with vehicle (n=7) or CSF1R inhibitor (n=7). Welch’s test, two-tailed. (E) Csf1r expression in CF and macrophage clusters in the snRNAseq dataset. (F) YAP score, glycolysis score, and OCP score of YAP^High CF4 from indicated groups. Nuclear number: Lats1/2^ΔCF + Vehicle (n=129), Lats1/2^ΔCF + CSF1R inhibitor-1 (n=421) and Lats1/2^ΔCF + CSF1R inhibitor-2 (n=380). Kruskal-Wallis test, multiple comparisons. (G) Alcian blue/nuclear fast red staining of RA of indicated groups. (H) Quantification of Alcian blue-stained area over total tissue area in (G). Vehicle, n=6. CSF1R inhibitor, n=8. Unpaired t-test, two-tailed. (I) Immunostaining of SOX9 (red) and GFP (green, Lats1/2-deleted CFs) in RA of indicated groups, white arrowheads (SOX9-positive Lats1/2-deleted CFs), DAPI (blue, nuclei), scale bar, 50 μm. (J) Percentage of SOX9-positive cells within the RA Lats1/2-deleted CF (GFP+) population in the vehicle-treated (n=7) or CSF1R inhibitor-treated group (n=7). Welch’s t-test, two-tailed. (K) In situ hybridization of Col2a1 transcripts (red dots) and immunostaining of Tcf21^iCre lineage cells (GFP, green) in indicated groups, DAPI (blue, nuclei), white arrowheads (Col2a1-positive Lats1/2-deleted CFs), scale bar, 50 μm. (L) Quantification of Col2a1-positive cells within Lats1/2-deleted CF population in the vehicle-treated (n=7) and CSF1R inhibitor-treated group (n=7), Welch’s t-test, two-tailed. (M-P) ST and quantification of Sox9 (M, N) and Col2a1 (O, P) expression in Lats1/2^ΔCF hearts treated with vehicle or CSF1R inhibitor. (N,P) Unpaired Wilcoxon rank-sum test. (B,D,H,J,L) The box plots show the median (central line), interquartile range (box), minimum and maximum (whiskers), with all data points displayed. (F,N,P) Box plots show the median (central line), interquartile range (box), and the whiskers extend to a maximum of 1.5× interquartile range beyond the boxes.

To further investigate the hypothesis that Macs promote fibrosis in Lats1/2^ΔCF mice, we examined whether Macs in the local microenvironment enhances Hippo-deficient CF proliferation. Lats1/2^ΔCF hearts treated with GW2580 had reduced EdU-positive, GFP labelled Tcf21^iCre lineage Hippo-deficient CFs compared to vehicle treatment (Figure 5C and 5D). Importantly, CSF1R expresses in MP1–3 but not CF1–4 clusters arguing against a direct effect of GW2580 on CFs (Figure 5E). These results indicate that Macs in the tissue microenvironment are essential for RA Hippo-deficient CF proliferation and accelerated fibrosis.

To determine whether Macs promote YAP activity in RA Hippo-deficient CFs, we evaluated the YAP score in YAP^high CF4 in GW2580-treated mice and compared to vehicle treated Lats1/2^ΔCF mice. The YAP score was reduced in GW2580-treated YAP^high CF4 (Figure 5F), supporting the conclusion that Macs non-cell autonomously enhance YAP activity in YAP^high CF4. Moreover, glycolysis and OCP scores were also reduced in YAP^high CF4 but not YAP^Low CFs after GW2580 treatment (Figure 5F; Figure S14E).

Alcian blue staining revealed that GW2580-treated Lats1/2^ΔCF mice displayed reduced proteoglycan deposition compared with vehicle-treated Lats1/2^ΔCF mice (Figure 5G and 5H). Additionally, SOX9 protein (Figure 5I and 5J) and Col2a1 RNA (Figure 5K and 5L) levels were reduced in Hippo-deficient CFs from GW2580-treated Lats1/2^ΔCF mice compared to vehicle-treated Lats1/2^ΔCF mice. Consistently, ST revealed that Sox9 (Figure 5M and 5N) and Col2a1 (Figure 5O and 5P) RA expression is decreased in GW2580-treated Lats1/2^ΔCF RA. Together, these results support the conclusion that Macs non-autonomously promote YAP activity, glycolysis, and the OCP cell state in RA Hippo-deficient CFs.

YAP Activates Igf1r in Right Atrial Cardiac Fibroblasts to Promote Fibrosis

To search for outgoing signal(s) from Macs to RA Lats1/2-deleted CFs, we observed that expression of Insulin-like growth factor receptor 1 (IGFR1) is upregulated in Lats1/2-deleted CFs in bulk RNAseq data ⁵ (see Figure 2A). Moreover, YAP^High CF4 also has the highest Igf1r expression in the snRNA-seq data indicating increased competence to receive the IGF signal (Figure S15A). In situ hybridization revealed that Igf1r transcript levels are enriched in RA Lats1/2-deleted CFs (labeled by Tcf21^iCre and the RFP TOMATO reporter Rosa26^Ai9) compared to controls (Figure 6A and 6B).

Figure 6. Hippo-deficient CFs upregulate IGF1 signaling to promote fibrosis.

(A) In situ hybridization of Igf1r transcripts (green dots) and immunostaining of Tcf21^iCre lineage cells (TOMATO, red) of indicated groups. DAPI (blue, nuclei), white arrowheads (Igf1r-positive Lats1/2-deleted CFs), scale bar: 50 μm. (B) Quantification of Igf1r-high Tcf21^iCre lineage CFs in the control (n=3) or Lats1/2^ΔCF RA (n=4). Mann-Whitney test, one-tailed. (C) In situ hybridization of Igf1 transcripts (white dots) and Mac immunostaining of CSF1R (red) in RA sections of indicated groups. DAPI (blue, nuclei), White arrowheads (Igf1-positive Macs). Scale bar: 50 μm. (D) Quantification of Igf1-positive macrophages in control (n=3) or Lats1/2^ΔCF RA (n=4). Mann-Whitney test, one-tailed. (E) Igf1 expression in immune cell subclusters in the snRNAseq dataset. (F) Bulk ATAC-seq data analysis of chromatin accessibilities in the Igf1r locus of indicated CF groups. Bottom panels, protection score and motifs in newly accessible sites in Lats1/2-deleted CFs compared to control. (G) Experimental scheme. (H) Sirius red-stained RA sections of indicated groups, lower panels: high magnification images of upper panel boxed regions. (I) Quantification of collagen over total tissue area. Vehicle (n=5), IGF1R inhibitor (n=4). Mann-Whitney test, two-tailed. (J) Proliferating RA Lats1/2-deleted CFs (red) of indicated groups labeled with EdU (green). Scale bar, 50 μm. White arrowheads, EdU-positive Lats1/2-deleted CFs. (K) Quantification of EdU-positive RA Lats1/2-deleted CFs in mice treated with vehicle (n=5) or IGF1R inhibitor (n=4), Mann-Whitney test, two-tailed. (B,D,I,K) The box plots show the median (central line), interquartile range (box), minimum and maximum (whiskers), with all data points displayed. (L) Working model.

We conducted in situ hybridization and immunostaining to test whether IGF1 signals originate in Macs. Lats1/2^ΔCF RAs display more Igf1-expressing Macs than control RAs (Figure 6C and 6D). Among macrophage clusters, MP2 shows the highest Igf1 expression (Figure 6E). These results support the conclusion that outgoing IGF1 signaling from MP2 induces IGF1R signaling in RA Lats1/2-deleted CFs. We determined whether Igf1r is a direct YAP target in CFs via bulk ATAC-seq, evaluating chromatin accessibility of the Igf1r locus in control and Hippo-deficient CFs ⁵. Multiple newly accessible sites upstream of Igf1r are present in Lats1/2-deleted CFs but not controls (Figure 6F). We generated a new snATAC-seq RA dataset and found that newly accessible peaks in the Igf1r locus are also present in RA Lats1/2-deleted CFs (Figure S15B). Using footprint analysis²⁰, we discovered that multiple TEAD motifs with high footprint protection scores are located within the newly accessible sites in both the bulk ATAC-seq and snATAC-seq datasets (Figure 6F and Figure S15C and S15D). These data support the notion that YAP induces chromatin accessibility at TEAD elements and induces activation of Igf1r transcription, which renders RA Lats1/2-deleted CFs competent to receive Igf1 signaling from Macs.

To functionally investigate IGF1 signaling, we treated Lats1/2^ΔCF mice with the IGF1R inhibitor picropodophyllin and collected hearts two weeks later (Figure 6G). Strikingly, Sirius red staining revealed that picropodophyllin treatment suppresses fibrosis in Lats1/2 ^ΔCF RAs compared to vehicle-treated controls (Figure 6H and 6I). Moreover, EdU staining revealed that RA Lats1/2-deleted CF proliferation is suppressed in picropodophyllin-treated mice compared to controls (Figure 6J and 6K). Together, these results support the conclusion that RA Lats1/2-deleted CFs upregulate IGF1R to receive IGF1 signaling from Macs, thereby promoting CF proliferation and fibrosis (Figure 6L).

Discussion

We discovered that human and mice RA CFs have higher baseline YAP and glycolytic activity compared to other chambers. Moreover, there is more glucose uptake in the RA compared to LA. YAP activation in CFs induces an accelerated fibroinflammatory response in the RA, which is mediated by YAP induction of glycolytic enzymes and glucose transporters. Increased glycolysis results in more lactate production, which is required for the RA fibroinflammatory response. Macs expand in the microenvironment, induced by outgoing CSF1 from YAP expressing CFs, and express IGF1 that reciprocally signals back to YAP expressing CFs. This signaling loop serves to amplify the CF fibroinflammatory cell state.

YAP Induces a Permissive Microenvironment in Right Atrium that is Prone to Accelerated Fibroinflammation

In Lats1/2^ΔCF mice, we observed a stronger fibroinflammatory phenotype in the RA than in the LA, despite having equivalent Cre activity. Our data are consistent with the conclusion that elevated baseline YAP and glycolytic activity in RA CFs makes the RA more prone to fibrosis. We found that human RA CFs have higher YAP activity and glycolysis than LA CFs, which is consistent with the observation that YAP and glycolysis activate each other in cancer ^21–23. Moreover, the relatively hypoxic environment of the RA, which receives deoxygenated blood from the body, likely promotes glycolysis and perhaps YAP activity ²⁴.

We uncovered multiple mechanisms that contribute to RA fibroinflammation. Increased YAP activation in Lats1/2^ΔCF RA autonomously induces a CF cell state that enhances cell cycle progression, a glycolytic metabolic state, and competence to receive IGF signaling while also signaling to Macs through CSF1. Mac proliferation and recruitment in the RA drastically alters the local RA microenvironment, which increases fibroinflammation. CF proliferation and the fibroinflammatory phenotype are promoted by Macs since the CSF1 inhibitor, which inhibits Mac expansion, suppressed the fibroinflammatory phenotype and reduced YAP activity, glycolysis, and OCP cell fate acquisition.

Our interrogation of ligands and receptor expression in CFs and Macs led us to test IGF1-IGF1R signaling as an outgoing signal from Macs to CFs. While it remains unclear how outgoing IGF1 signaling from Macs enhance YAP activity of the Hippo-deficient CFs there is precedent for our observation. IGF1 signaling is known to inhibit Hippo and thereby activate YAP activity ²⁵, this mechanism is less likely in our model, because Hippo signaling is genetically ablated in CFs. There are other plausible, although less direct mechanisms which may play a role. We speculate that Macs may increase tissue stiffness, which in turn enhances YAP activity, given that YAP responds to mechanical cues²⁶ and Macs are known to remodel the extracellular matrix^27–29. We note that weaknesses of the study include the possibility that GW2580 inhibits CFs directly via an off-target mechanism rather than directly inhibiting Mac signaling. While the preponderance of our data argues against this interpretation, more work needs to be done to definitively rule this out using genetic experiments.

YAP Induces a Glycolytic and Lactate Rich Metabolic Cell State in Right Atrial Fibroblasts

Our data support the conclusion that YAP directly activates genes in the glycolytic pathway and that YAP activity enhances the local glycolytic microenvironment to induce fibroblast activation. In cardiomyocytes it has been shown that YAP activates Slc2a1 to induce glucose import and this occurs through an interaction with HIF1α ²⁴. We previously showed that YAP directly activates cMyc, which is a potent inducer of glycolysis ^5,30. Using pharmacologic inhibition of glucose metabolism by 2-DG treatment, we revealed the importance of glucose metabolism in Lats1/2^ΔCF RA fibrosis phenotype. In addition, inhibition of lactate production also inhibits the phenotype supporting the conclusion that completion of glycolysis with lactate is required for the Lats1/2^ΔCF RA fibrosis phenotype. Moreover, Mac derived IGF1 signaling also induces glycolysis to further amplify the CF glycolytic metabolic state ³¹.

YAP Induces the Osteochondroprogenitor Cell State in Right Atrial Fibroblasts

Following injury, CFs transition into myofibroblasts to secrete extracellular matrix as a repair process, which drives fibrosis and impairs the heart’s pumping function ³². In the context of an injured tissue, adult fibroblasts differentiate into matrifibrocyte ³³, osteogenic ³⁴ and adipogenic fates ³⁵, which is considered to be a component of the tissue repair process. It is generally believed that in the absence of injury or aging, adult CFs maintain a stable resting CF state. Our data indicate that in the RA Lats1/2-deleted CF, increased YAP activity disrupts CF lineage fidelity and induces spontaneous acquisition of the SOX9-expressing OCP cell state, which is an unexpected finding.

Our data as well as previous work provide insight in the mechanisms by which YAP promotes OCP differentiation. Because SOX9 is a key transcriptional factor during chondrogenesis ^36,37, we hypothesize that YAP directly upregulates Sox9, similarly to that reported in cancer³⁸. OCPs reside in a hypoxic and glycolytic microenvironment during chondrogenesis, which is crucial for their differentiation during cartilage development. We hypothesize that the hypoxic environment of the RA promotes the differentiation of RA Lats1/2-deleted CFs to OCPs. Indeed, CFs increase SOX9 expression, although not the OCP cell state, in ischemic zone of myocardial infarction mouse models ^39–42. We observed that Macs are required for the expression of SOX9 protein in RA Lats1/2-deleted CFs, suggesting non-cell autonomous regulation of SOX9 in CFs by outgoing IGF1 from Macs, which was previously reported to upregulate SOX9 ⁴³. Future studies are warranted to clarify the mechanisms underlying YAP-SOX9 regulation and how Macs affect CF cell state transitions.

Supplementary Material

Expanded Materials & Methods

Supplemental Table S1-S5

Online Figures S1-S15

References 44–66

Major Resources Table

Summary of Statistics

Code for permutation test in Python

Novelty and Significance.

What is known?

• Single-cell and spatial transcriptomics have revealed molecular heterogeneity among heart chambers.

• Hippo/Yes-associated protein (YAP) pathway in cardiac fibroblasts (CFs) promote inflammation and fibrosis.

What new information does this article contribute?

• Right atrial CFs exhibit higher glycolytic and YAP activities compared to CFs in other chambers.

• YAP activation in CFs induces glycolysis and lactate production to drive right atrial fibrosis.

• YAP disrupts fibroblast lineage fidelity, promoting differentiation into a SOX9-expressing osteochondroprogenitor (OCP) state.

• YAP upregulates colony stimulating factor 1 (CSF1) in CFs to promotes macrophage expansion, which in turn enhances CF proliferation, OCP differentiation and fibrosis.

• YAP-induced macrophages secrete insulin-like growth factor 1 (IGF1) to activate IGF1R signaling in CFs to further promote their proliferation and fibrosis.

This study uncovers that YAP induces a reciprocal CSF1/IGF1 signaling loop between CFs and macrophages. Since CSF1 and IGF1 signaling pathways promote glycolysis, our findings suggest that YAP defines a glycolytic and fibroinflmmatory niches in the right atria.

Non-standard Abbreviations and Acronyms:

CF

cardiac fibroblast

YAP

Yes-associated protein

TAZ

transcriptional coactivator with PDZ-binding motif protein

TEAD

transcriptional enhanced associate domain transcription factor

OCP

osteochondroprogenitor

SOX9

SRY-box transcription factor 9

CSF1

colony stimulating factor 1

IGF1

insulin-like growth factor 1