Rearrangement of the Protein Phosphatase 1 Interactome During Heart Failure Progression

Abstract

Background: Heart failure (HF) is a complex disease with a rising prevalence despite advances in treatment. Protein phosphatase 1 (PP1) has long been implicated in HF pathogenesis, but its exact role is both unclear and controversial. Most previous studies measured only the PP1 catalytic subunit (PP1c) without investigating its diverse set of interactors, which confer localization and substrate specificity to the holoenzyme. In this study, we define the PP1 interactome in cardiac tissue and test the hypothesis that this interactome becomes rearranged during HF progression at the level of specific PP1c interactors.

Methods: Mice were subjected to transverse aortic constriction and grouped on the basis of ejection fraction into sham, hypertrophy, moderate HF (ejection fraction, 30%-40%), and severe HF (ejection fraction <30%). Cardiac lysates were subjected to affinity purification with anti-PP1c antibodies followed by high-resolution mass spectrometry. PP1 regulatory subunit 7 (Ppp1r7) was knocked down in mouse cardiomyocytes and HeLa cells with adeno-associated virus serotype 9 and siRNA, respectively. Calcium imaging was performed on isolated ventricular myocytes.

Results: Seventy-one and 98 PP1c interactors were quantified from mouse cardiac and HeLa lysates, respectively, including many novel interactors and protein complexes. This represents the largest reproducible PP1 interactome data set ever captured from any tissue, including both primary and secondary/tertiary interactors. Nine PP1c interactors with changes in their binding to PP1c were strongly associated with HF progression, including 2 known (Ppp1r7 and Ppp1r18) and 7 novel interactors. Within the entire cardiac PP1 interactome, Ppp1r7 had the highest binding to PP1c. Cardiac-specific knockdown in mice led to cardiac dysfunction and disruption of calcium release from the sarcoplasmic reticulum.

Conclusions: PP1 is best studied at the level of its interactome, which undergoes significant rearrangement during HF progression. The 9 key interactors that are associated with HF progression may represent potential targets in HF therapy. In particular, Ppp1r7 may play a central role in regulating the PP1 interactome by acting as a competitive molecular "sponge" of PP1c.

Keywords: Ppp1r7 protein, mouse; heart failure; mass spectrometry; protein phosphatase 1; proteomics.

Figure 1. Establishing the PP1 interactome in the mouse heart

A. Overview of the experimental work-flow. LC-MS/MS=liquid chromatography-tandem mass spectrometry; XIC=extracted ion chromatogram; LFQ=label free quantitation. B. Volcano-plot showing that out of 267 identified and quantified proteins, 71 were significantly enriched in the PP1c vs. IgG-IP, including many known PP1 subunits (red). C. MA-plot showing the relative enrichment of proteins in the PP1c vs. IgG-IP where the highlighted proteins were statistically significant based on the volcano-plot in (B). Known binary interactors of PP1c based on yeast two-hybrid (Y2H) databases are in blue (besides the known PP1 subunits which are in red). Proteins with an infinite ratio (∞) have no LFQ signals from the IgG-IP. D–E. Binary interactions among the 70 enriched proteins based on Y2H databases. Green=PP1 catalytic subunits; pink=known PP1 R-subunits; blue=other PP1 interactors.

Figure 2. Proteins associated with HF progression

A. Heart tissues from mice in the four groups were processed in parallel. B. Quantified proteins from the PP1c AP-MS (692 proteins) were used to build a predictive model using partial least squares (PLS) regression to predict the EF of each animal. Individual dots represent individual samples, and dotted lines show 95% confidence interval for sample group placement. Sham (black) and hypertrophy (Hyp; grey) samples group together, while severe heart failure (sHF; purple) clusters furthest from these groups, with overlap with moderate heart failure (mHF; red). C. Repeat PLS regression model using a subset of 40 proteins that were the top predictors of HF progression, identified using forward selection (see Supplemental Figure 3 and Supplemental Table 5).

Figure 3. Rearrangement of the PP1 interactome during HF progression

A. Plot of all MS-quantified proteins ranked by relative abundance based on iBAQ values from wild-type heart lysates with PP1c-IP. Key proteins are based on PLS regression model. B. Venn diagram showing the overlap of proteins detected without PP1c-IP (“Proteome”), interactors based on PP1c vs. IgG-IPs (“Interactome”), and key proteins that are associated with HF progression as identified by PLS regression. C. Levels of binding to PP1c of the 9 key interactors (overlap between “Interactome” and “PLS regression key proteins” from (B) during HF progression. *p<0.05, **p<0.01, ***p<0.001.

Figure 4. Protein levels of Ppp1r7 during HF progression

Representative western blots (A), Coomassie staining showing total protein levels (B), and bar graphs (C) demonstrating an increase in Ppp1r7 and PP1c normalized to CSQ or total protein levels, but no change in Ppp1r7 normalized to PP1c. Numbers in bars represent numbers of samples. CSQ=calsequestrin. *p<0.05 and **p<0.01 vs. Sham. ^#p<0.05 and ^##p<0.01 vs. Hyp.

Figure 5. AAV9-mediated knockdown of Ppp1r7 impairs cardiac function in vivo

A. Representative M-mode echocardiography tracings from αMHC transgenic (TG) mice at baseline and 8 weeks after injection of either scramble shRNA (shScr)-AAV9 (top) or shRNA targeting Ppp1r7 (shPpp1r7)-AAV9 (bottom). Summary graphs showing (B) left ventricular ejection fraction (EF) and (C) left ventricular end systolic diameter (ESD) at baseline and 2, 4, 6 and 8 weeks after AAV9 injection. *p<0.05, ***p<0.001 vs. control.

Figure 6. AAV9-mediated Ppp1r7 knockdown impairs calcium handling in ventricular myocytes

A–B. Representative confocal line scan images showing increased number of Ca sparks in myocytes isolated from shPpp1r7-AAV9 treated mice. C. Bar graphs showing quantification of Ca sparks frequency (CaSF), (D) total sarcoplasmic reticulum Ca content (SR load) and (E) CaSF normalized to SR load. *p<0.05 vs. control.

Figure 7. Remodeling of the PP1 interactome with PPP1R7 knockdown

A. Experimental workflow. LC-MS/MS=liquid chromatography-tandem mass spectrometry; XIC=extracted ion chromatogram; LFQ=label free quantitation. B–C. Western blots and quantification showing significant knockdown (KD) of PPP1R7 in HeLa cells treated with siRNA. ***p<0.001 vs. control. D. Relative changes in the PP1 interactome with PPP1R7 KD. The interactors are ordered from the highest to lowest PP1c-bound based on the HeLa PP1 interactome from the top left to bottom right. Known PP1 subunits are in red and key interactors based on PLS regression are in blue. Proteins with the same symbols (#, $, ~, %, &, ^, @, !) are in the same complexes based on yeast two-hybrid (Y2H) databases. *p<0.05, **p<0.01, ***p<0.001 vs. control.