Pathologic gene network rewiring implicates PPP1R3A as a central regulator in pressure overload heart failure

Figures

• 

  Fig. 1. Regulatory rewiring of coexpression networks in HF. a Principal component analysis of gene expression profiles for 177 failing hearts and 136 nonfailing, control, hearts showing clear segregation of HF (red) vs. control (gray) population. b Differential connectivity of known biological processes in HF. Normalized connectivity (sum of WGCNA weights divided by maximum network weight) between representative genes from four known processes that play critical roles in HF (sarcomeric and contraction genes (orange), EC coupling (red), cardiac remodeling (green), and metabolism (blue)) to all genes from those same processes in HF and controls. Genes of each process are rows and columns are process and cohort. For example, MYBPC3 in the cardiac remodeling process (third row in green heat map) is highly connected to sarcomeric and contraction genes and cardiac remodeling genes in the HF network (fifth and eighth columns respectively) compared to all the control processes. c River plot demonstrating changing modular assignments for genes in the HF vs. control networks. Pink lines represent individual genes, with left-sided grouping representing membership in control (left) and HF (right)…

• 

  Fig. 2. High-quality tissue expression reveals previously undetected cardiac eQTL associations. Transcription factor (a) and transcription start site (b) annotation to eQTL distance distributions for failing (red) and control hearts. c Number of cis eQTLs found for each group that overlapped with GTEx eQTLs. d Fraction of modules with genes found to be controlled by at least one eQTL in HF (red, left) and controls (gray, right). e Heat map indicators for variants controlling multiple genes in-cis in HF (red, left) and controls (gray, right). In the rows are SNPs controlling genes (columns) colored intensely if the SNP controls the gene. f Variants from one locus control a network of G-protein-coupled receptors TAS2R present in both the failing and control groups

• 

  Fig. 3. Gene prioritization through network topology. a Diagrammatic representation of roles of local and global connectivity in defining each gene’s coordinator status. Local connectivity (LC) is the per-gene change in coexpression edges in the HF vs. control network. Global connectivity (GC) represents the enrichment of HF-relevant pathways in a gene’s neighborhood between HF and control networks (see Methods). b Change in local and global connectivity for all genes between control and HF networks identified PPP1R3A (green font) as a central coordinator in HF, indicating its increased association with HF-relevant pathways as well as coexpression relationships in the HF vs. control networks. c Gene-pathway (rows-columns) differential connectivity matrix for genes ranked highest for global network connectivity that is differentially increased between non-heart-failure and heart-failure conditions. Differential connectivity is measured by the KS statistic between the distribution of ranks for pathway genes in HF and non-heart-failure conditions. In particular, metabolism, HF and cardiomyopathy pathways (yellow indicates an increase in connectivity to the given pathway (columns) in HF…

• 

  Fig. 4. PPP1R3A knockdown in vitro reveals metabolic regulation in cardiomyocytes. a Experimental design. NRVMs were isolated, transfected with siRNA 36 h later. Phenylephrine or vehicle treatment started at 48 h. RNA was collected at 36, 48, 72 and 96 h after isolation. b Clustered heat map of NRVM transcriptional expression of central coordinators in response to PPP1R3A knockdown (measured by RNAseq). Expression is shown from NRVMs at 72 and 96 h after isolation normalized to pretreatment expression, and displayed as per-gene z-scores. Data from cells with and without phenylephrine (PE) are shown on the left and right sides of the heat map with and without siRNA knockdown as indicated. Stars indicate central coordinators significantly differentially regulated by PPP1R3A knockdown (FDR < 0.05, red stars indicate significance in the PE-treated group (red at 72 h, dark red at 96 h) and blue stars indicate significance in the untreated group after PPP1R3A knockdown at 72 (light blue) and 96 (dark blue) hours). c PPP1R3A knockdown protects against hypertrophic stimulus of phenylephrine treatment. Upper panel: Cell size measurements of a sample of cells under phenylephrine and normal…

• 

  Fig. 5. Cardiac effects of PPP1R3A ablation in vivo. a Fractional shortening is preserved in Ppp1r3a−/− mice after TAC. At 6 weeks (n = 5 animals for all groups), Ppp1r3a−/− TAC 37.6 ± 4.3%, Ppp1r3a−/− Sham 36.5 ± 4.7%, Ppp1r3a+/+ TAC 26.7 ± 9.5%, Ppp1r3a+/+ Sham 41.2 ± 8.4%, p = 0.03 ANOVA and p = 0.03 for Ppp1r3a−/− TAC vs. Sham by Bonferroni post test. At 8 weeks (n = 5 for all groups), Ppp1r3a−/− TAC 37.9 ± 2.8%, Ppp1r3a−/− Sham 41.6 ± 6.8%, Ppp1r3a+/+ TAC 24.4 ± 9.0%, Ppp1r3a+/+ Sham 38.5 ± 2.8%, p = 0.002 (ANOVA) and p = 0.01 for Ppp1r3a−/− TAC vs. Sham (Bonferroni post test). b Gene expression of HF markers is not increased in Ppp1r3a−/− animals after TAC: PPP1R3a: p < 0.0001 (ANOVA), p < 0.0001 Ppp1r3a+/+ vs Ppp1r3a−/− TAC and Ppp1r3a+/+ vs. Ppp1r3a−/− Sham, but p = 0.41 Ppp1r3a+/+ TAC vs. Sham (Bonferroni post test). MYH7/MYH6: p = 0.11 (ANOVA), p = 0.45 Ppp1r3a+/+ TAC vs. Sham and p = 1.0 Ppp1r3a−/− TAC vs. Sham (Bonferroni post test). Nppa: Ppp1r3a+/+ Sham 1 ± 0.18 (mean fold change ± SD), Ppp1r3a+/+ TAC: 4.8 ± 2.6, Ppp1r3a−/− Sham: 1.5 ± 1.1, Ppp1r3a−/− TAC: 2.5 ± 1.1. p = 0.006 (ANOVA), p = 0.006 Ppp1r3a+/+ TAC vs. Sham and p = 1.0 Ppp1r3a−/− TAC vs. Sham (Bonferroni post…