Consequences of PDGFRα+ fibroblast reduction in adult murine hearts

Fibroblasts produce the majority of collagen in the heart and are thought to regulate extracellular matrix (ECM) turnover. Although fibrosis accompanies many cardiac pathologies and is generally deleterious, the role of fibroblasts in maintaining the basal ECM network and in fibrosis in vivo is poorly understood. We genetically ablated fibroblasts in mice to evaluate the impact on homeostasis of adult ECM and cardiac function after injury. Fibroblast-ablated mice demonstrated a 60-80% reduction in cardiac fibroblasts, which did not overtly alter fibrillar collagen or the ECM proteome evaluated by quantitative mass spectrometry and N-terminomics. However, the distribution and quantity of collagen VI, a microfibrillar collagen that forms an open network with the basement membrane, was altered. In fibroblast-ablated mice, cardiac function was better preserved following angiotensin II/phenylephrine (AngII/PE)-induced fibrosis and myocardial infarction. Analysis of cardiomyocyte function demonstrated weaker contractions and slowed calcium decline in both uninjured and AngII/PE infused fibroblast-ablated mice. Moreover, fibroblast-ablated hearts had a similar gene expression profile to hearts with physiological hypertrophy after AngII/PE infusion. Our results indicate that the adult mouse heart tolerated a significant degree of fibroblast loss with potential beneficial impacts on cardiac function. Controlled fibroblast reduction may have therapeutic value in heart disease by providing cardioprotective effects.


Introduction 48
Fibroblasts are considered to be the primary source of extracellular matrix (ECM) in the 49 heart (1, 2). The main physiological role of the cardiac fibroblast is maintaining the ECM 50 by balancing deposition and degradation of structural and nonstructural matricellular 51 proteins (3). This ECM network serves as a scaffolding to mechanically support 52 cardiomyocytes and permit transmission of lateral force by regulating mechanical signals 53 (4,5). It is thought that ECM-cytoskeletal connections are essential for proper stability 54 and contraction of the cardiomyocyte -and disruptions in these interactions underlie a 55 wide range of cardiomyopathies (6). Although the ECM network is considered to be 56 primarily synthesized and organized by cardiac fibroblasts and integral to proper 57 cardiomyocyte function (1, 2, 7), there is limited data focusing on homeostatic, fibroblast-58 specific ECM production in vivo and the potential impact of fibroblast reduction after the 59 formation of the ECM scaffold but prior to injury. 60 61 In response to cardiac injury, fibroblasts rapidly adopt an activated phenotype resulting in 62 their increased proliferation and deposition of a collagen-rich ECM (8)(9)(10)(11)). An initial 63 adaptive response maintains structural integrity of the damaged myocardium and 64 prevents rupture (12,13). However, prolonged ECM deposition can impair cardiac 65 compliance due to ventricular wall stiffening (14,15). Persistent ECM accumulation 66 resulting in fibrosis can also disrupt electrical transmission between cardiomyocytes 67 leading to contractile dysfunction (16). Despite its clinical and pathophysiological 68 significance, no interventions currently exist to directly treat or reverse cardiac fibrosis 69 (17). Notwithstanding, the cardiac fibroblast has emerged as an ideal candidate to 70 regulate fibrosis that accompanies cardiac injury, but its role remains relatively obscure. 71 72 Recent studies have focused on genetically targeting myofibroblasts or activated 73 fibroblasts as a potential therapeutic treatment for heart disease because of their 74 contribution to fibrotic scar formation and subsequent reduced heart function (10,(18)(19)(20)(21). 75 When chimeric antigen receptor (CAR) T cells were used to specifically target an 76 endogenous cell-surface glycoprotein on activated fibroblasts, fibrosis was reduced 77 leading to better cardiac function (22). Similarly, ablation of activated fibroblasts (21) or 78 activated fibroblast-specific depletion of Grk2, a downstream effector of G protein-couple 79 receptors that is known to be elevated in patients with heart failure (23), was shown to be 80 cardioprotective after injury. While these studies suggest beneficial outcomes from 81 fibroblast reduction, others have observed increased lethality and ventricular wall rupture, 82 presumably due to reduced ECM deposition (18,20), which was also observed when 83 Hsp47 (19), Fstl1 (24,25), or Smad3 (26) were disrupted in activated fibroblasts. Taken 84 together, these data suggest that manipulation of fibroblast numbers and matrix 85 deposition may require a more nuanced approach based on improved fundamental 86 knowledge. Considering that anti-fibrotic therapies are currently being proposed as 87 treatments for heart pathologies, a thorough evaluation and understanding of fibroblast 88 activities during tissue homeostasis is necessary. 89 90 Given the lack of data specifically addressing fibroblast functions in the uninjured heart, 91 we reduced fibroblast numbers by 60-80% in the adult murine heart using an inducible 92 fibroblast-specific Cre line. Surprisingly, no mortality resulted even though fibroblast loss 93 up to 80% was sustained 7 months post-induction. Moreover, heart function, protein 94 composition, and proteolytic turnover were largely well-maintained despite the level of 95 fibroblast depletion. Analysis of matrix components, such as type I, IV, and VI collagen 96 demonstrated that the type I collagen fibrillar network appeared relatively unchanged, 97 whereas changes in microfibrillar collagen and basement membrane components were 98 observed. Fibroblast-ablated hearts also showed improved function after MI and 99 angiotensin II/phenylephrine (AngII/PE) infusion, suggesting that an underlying reduction 100 in fibroblast numbers or cardiac adaptation to fewer fibroblasts may be protective. Our 101 findings reveal that fibroblast reduction before injury does not negatively affect the heart 102 and may provide beneficial effects when cardiac remodeling or repair are provoked by 103 injury. 104

Genetic ablation of fibroblasts during cardiac homeostasis 106
After heart injury, fibroblasts are essential for the generation of replacement fibrous ECM 107 and multiple studies have suggested that disruption of fibroblast expansion may lead to 108 wall rupture (18,20). Although it is thought that fibroblasts are indispensable for 109 maintaining ECM structure during cardiac homeostasis, we recently demonstrated that a 110 loss of up to 50% of resident fibroblasts was inconsequential for the architecture and 111 function of the heart (27). These results led us to investigate whether the heart is able to 112 tolerate a further reduction in fibroblast numbers. As PDGFRa is expressed in a majority 113 of murine adult cardiac fibroblasts (8, 28), we induced diphtheria toxin A (DTA) expression 114 using a PDGFRa-CreER T2/+ mouse line to deplete resident fibroblasts (29). After Cre 115 induction by tamoxifen, DTA expression led to fibroblast loss within several days. We 116 refer to these mice as fibroblast-ablated or ablated. 117

118
A tdTomato Cre recombination reporter demonstrated that a 60-80% reduction in 119 PDGFRa-expressing cells occurred by 14 days post-induction ( Figure 1A, D). To 120 investigate fibroblast reduction independent of Cre expression, we assayed for vimentin 121 expression (30) and found fewer vimentin-positive cells in ablated hearts compared to 122 controls ( Figure 1B). To determine whether other cell populations were capable of 123 replacing fibroblast collagen production, we assessed cells actively expressing type I 124 collagen using a transgenic collagen reporter, collagen1a1-GFP (Col1a1-GFP). This 125 reporter expresses GFP under the control of the Col1a1 promoter (10, 28, 31-33). A 126 similar reduction in fibroblasts and collagen-expressing cells was observed ( Figure 1C, 127

Collagen fiber organization and basement membrane protein distribution 166
Despite a significant reduction in fibroblasts, ablated hearts were histologically and 167 functionally indistinguishable from controls. Although body weight was reduced in long-168 term fibroblast-ablated mice, heart weight to body weight ratio was proportionate in aged 169 females and increased slightly in aged males ( Taken together these data signify that fibroblast loss 176 is well tolerated in the heart in the absence of injury. 177 178 Because fibroblasts are responsible for the majority of type I collagen production in the 179 heart (1, 2), we evaluated collagen levels by immunostaining and hydroxyproline content 180 and found that type I collagen and hydroxyproline quantity remained constant in ablated 181 hearts (  2B-C). Of these, 15 high confidence internal peptides corresponding to 8 proteins were 260 more abundant in fibroblast-ablated hearts as compared to controls, implying higher 261 proteolytic modification of these proteins, whereas 54 internal peptides corresponding to 262 30 proteins had lower abundance in fibroblast-ablated hearts (Supplemental Tables 1 and  263 2). The most internal peptides were identified in myosin-6 (

Resident fibroblast ablation mitigates cardiac impairment after myocardial 273 infarction 274
Our data demonstrate that fibroblast loss does not lead to obvious structural or 275 compositional deficiencies at baseline. However, depletion of activated fibroblasts after 276 myocardial infarction (MI) has been shown to reduce collagen production and increase 277 morbidity due to ventricular wall rupture (18). It is unclear how hearts with pre-existing 278 resident fibroblast loss might respond to damage. To address this question, we induced 279 MI by permanently ligating the left anterior descending (LAD) artery >7 weeks post-280 induction. At 10 weeks post-MI, ~77% of control mice (7/9) and 100% of ablated mice 281 (5/5) survived. Compared to controls, fibroblast ablation did not affect key parameters 282 such as cardiac mass or lung weight, a measure of left-heart failure ( Figure 4A significance was determined by an unpaired t-test. ns: not significant, P > 0.05; *P 302 ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001. 303 304

Cardiac responses after Angiotensin II/phenylephrine infusion 305
The fibrotic response to injury is variable and disease-dependent (40). MI typically results 306 in cardiomyocyte death and replacement fibrosis, while reactive fibrosis is thought to be 307 induced by cardiac stress and inflammation (15). We, therefore, employed a second 308 disease model to determine the impact of fibroblast loss on reactive fibrosis. We used 309 angiotensin II and phenylephrine (AngII/PE) infusion to induce arterial hypertension, 310 adaptive cardiac hypertrophy, and remodeling. Fibroblast levels remained relatively while control mice had a modest, sustained decrease in LV ejection fraction, the ejection 317 fraction of hearts in ablated mice recovered to near baseline levels ( Figure 5E). 318 319 AngII stimulates hypertrophy in cardiomyocytes through angiotensin type 1 (AT1) 320 receptors (41). While we did not observe an increase in heart to body weight ratio, we 321 found an increase in cardiomyocyte CSA in AngII/PE treated, fibroblast-ablated hearts at 322 both 14 and 28 days compared to control hearts ( Figure 5F). We also observed a 323 significant reduction in collagen deposition in adventitial and interstitial regions of the LV 324 in fibroblast-ablated mice ( Figure 5G-H). Taken together, these data suggest that ablating fibroblasts before injury results in greater cardiomyocyte hypertrophy and reduced fibrosis 326 leading to normalization of ejection fraction. are mean ± SEM. Statistical significance was determined by an unpaired t-test. ns: 337 not significant, P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ****P ≤ 0.0001. 338 339

Gene expression profiling of fibroblast-ablated hearts after AngII/PE infusion 340
To identify potential molecular mechanisms that underlie the cardioprotective aspect of 341 resident fibroblast loss during pathology, we identified differentially expressed genes by 342 microarray analysis in fibroblast-ablated hearts compared to controls after 14 days of 343 AngII/PE infusion. Forty-five genes were upregulated, and 204 genes were 344 Our knowledge of fibroblast biology in the injured heart is rapidly expanding. Because 415 fibroblast-targeting therapies are being contemplated as a mitigation strategy in fibrosis-416 associated cardiovascular diseases, an understanding of the homeostatic roles of 417 fibroblasts is important. Fibroblasts are thought to be the primary source of type I collagen, 418 and homeostatic collagen turnover and synthesis in the rat heart is estimated to be 419 between 6-9% per day (47). Yet, our study reveals that the murine adult heart is 420 functionally capable of sustaining a 60-80% reduction in fibroblasts for up to 7 months 421 after ablation. Even though a complete ablation of fibroblasts was not obtained, the 422 remaining fibroblasts did not expand and repopulate the heart. Despite this substantial 423 loss of fibroblasts, proteomics of the decellularized ventricles did not detect massive 424 changes in protein abundance, with type I collagen protein remaining relatively normal 425 and fibrillar organization retained. Proteomics analysis is challenged by the abundance 426 of cardiomyocyte components, post-translational modifications, and limited extraction of 427 cross-linked components. Nonetheless, the collective data suggests that ECM in the 428 unchallenged adult heart, once formed, is stable and has a low turnover rate, consistent 429 with results from multi-isotope imaging mass spectrometry (MIMS), suggesting that ECM 430 proteins have half-lives on the order of years (48). 431 432 The significant changes identified by TAILS in fibroblast-ablated hearts were surprisingly 433 few in number and did not include any ECM-derived peptides, despite extensive offline 434 fractionation prior to LC-MS/MS and identification of a large number of internal protein N-435 termini, suggesting that fibroblast depletion may not dramatically impact ECM proteolysis. 436 One explanation is that although fibroblasts are primarily responsible for ECM 437 degradation and deposition, there may be relatively low turnover at baseline in healthy 438 hearts, and a 60-80% loss of these cells may not result in a significant change in protein 439 composition. Alternatively, it is possible that the high sample complexity introduced by 440 abundant cardiomyocyte proteins may have reduced the yield of ECM peptides, 441 consistent with the finding that of the 1834 peptides, only 4.3% originated in 442 secreted/extracellular matrix proteins. Nevertheless, this approach, in addition to LFQ-443 proteomics and cardiac function analysis, also supports a relatively mild impact of 444 fibroblast ablation overall on the heart, although significant changes in several contractile 445 proteins and metabolic enzymes were identified. 446

447
In support of previous findings (27), there was not a mechanism within the heart that 448 signals fibroblast replenishment even with reductions up to 75%. It has been suggested 449 that cardiac fibroblasts are heterogeneous (34, 35), and there is a possibility that a 450 PDGFRa-negative fibroblast population compensates for fibroblast loss. However, 451 analysis of total heart RNA argues that type I collagen transcript and reporter transgene 452 levels remain reduced. Thus, in an undamaged heart, a loss in collagen synthesis might 453 be balanced by an equal loss in collagen degradation, another action that has been 454 primarily attributed to fibroblasts. 455 456 While severe changes to fibrillar collagen were not observed, there were more notable 457 differences in the basement membrane and microfibrillar collagen surrounding 458 cardiomyocytes. Therefore, these data suggest that fibroblasts produce factors, including 459 matricellular proteins that may stabilize the fibrillar collagen network via cross-linking or 460 proteolytic enzymes that regulate turnover of fibrillar collagens. Because type VI collagen 461 is secreted by fibroblasts and connects the basement membrane to fibrillar collagen (49), 462 we suspected that the observed basement membrane alterations in fibroblast-ablated 463 hearts may be directly due to fibroblast loss. Alternatively, it is also possible that the 464 observed basement membrane changes may be indirectly caused by disruption of 465 collagen matrix organization or that the basement membrane and cell-proximate networks 466 are not as well protected from turnover as cross-linked collagen. Collagen VI was reported 467 to be overexpressed in hypertension, diabetes, and post-MI (49, 50). Interestingly, a 468 recent study demonstrated that Col6a1 knockout mice had improved heart function after 469 MI (51). Therefore, our results are consistent with previous data in that the observed 470 reduction in collagen VI at baseline and after 14 days of AngII/PE could, in part, explain 471 the cardioprotective effect in fibroblast-ablated hearts after injury. 472 473 Because force generation is dependent on cardiomyocyte adhesion to the ECM which is 474 mediated by the basement membrane, we expected to observe a disruption of 475 cardiomyocyte contraction in fibroblast-ablated hearts. Indeed, our study is among the 476 first to demonstrate that in vivo modulation of the ECM by fibroblasts affects 477 cardiomyocyte function by abating contractile force and calcium efflux. One explanation 478 is that the disrupted basement membrane in ablated hearts could decrease myofilament 479 calcium sensitivity, which could be a result of reduced free calcium or force of contraction 480 by the myofilaments (52). While we did not examine free calcium in our hearts, the 481 disrupted basement membrane could alter myofilament adhesion leading to reduced 482 contraction of the sarcomeres. Changes in contraction efficiency could also be a result of 483 the observed altered proteolysis of proteins that are part of the contractile machinery, 484 such as myosin-6. Thin filament proteins can undergo modifications, such as proteolysis 485 leading to thin filament deactivation and slowed myocardial relaxation (53). In response 486 to AngII/PE infusion, weaker contractions and slowed calcium decline in cardiomyocytes, 487 as well as sustained normal cardiac function were also observed in fibroblast-ablated 488 hearts. Therefore, we hypothesized that fibroblast loss predisposes the heart to 489 physiological, rather than pathological hypertrophy in response to drug-induced fibrosis. 490 However, because we used a relatively early time point for analysis of AngII/PE infusion, 491 long-term injury models will be required to determine whether physiological hypertrophy 492 is sustained over time in fibroblast-ablated hearts. 493 494 Because activated fibroblasts are key contributors to scar formation, fibroblasts have 495 increasingly become a target of interest in combating heart disease. However, studies 496 that have focused on specifically reducing activated fibroblasts or targeting genes in 497 activated fibroblasts to attenuate pathological fibrosis have produced conflicting results. 498 In some circumstances, fibroblast disruption leads to rupture while in other scenarios 499 manipulations of fibroblasts appears to be protective (18-26). Our work further 500 demonstrates that resident fibroblast reduction prior to injury may have beneficial 501 outcomes under certain pathological conditions. Similarly, depletion of the resident 502 fibroblast population prior to AngII/PE infusion reduced reactive fibrosis, however 503 replacement fibrosis was not affected after MI. The minimal scar in the MI model could 504 explain why depletion of fibroblasts prior to injury does not result in rupture of the 505 ventricular wall, in contrast to what others have reported (18,20). Because we only 506 observe a 60-80% deletion of fibroblasts, the remaining 20-40% may be able to maintain 507 the damaged myocardium after MI. This suggests that lowering level of initial fibroblasts 508 could have beneficial effects in pathological conditions particularly in response to reactive 509 fibrosis. 510 511 While we did not evaluate the ECM protein composition of the fibrotic scar, we predict 512 that the ECM prior to injury may be altered resulting in subtle changes in matrix 513 organization and stiffness. Although our proteomics data did not detect large differences 514 in ECM components of fibroblast-ablated hearts at baseline, the differentially abundant 515 ECM proteins that were observed could contribute to the protective response after injury. 516 Because deletion of fibroblasts occurs several weeks before injury, the heart may have 517 adapted prior to insult. Moreover, fibroblast loss increased cardiomyocyte hypertrophy in 518 response to injury implicating a dynamic interaction between fibroblast and 519 cardiomyocytes during pathology. Further investigation of the cardioprotective effect of 520 resident fibroblast depletion will provide insights into the potential efficacy of anti-fibrotic 521 therapies and delineate the long-term effects on the ECM and cardiac function. 522

523
In summary, these studies demonstrate the surprising finding that a significant reduction 524 in cardiac fibroblasts is not detrimental to basal heart function. These observations 525 suggest that cardiac tissue and especially the ECM may be very resilient. Further studies 526 to determine the level of fibroblast loss that can be tolerated by cardiac tissue are 527 warranted. Fibroblast loss prior to injury potentially resulted in a type of pre-conditioning 528 that may be cardioprotective after injury. Further examination of the effect of fibroblast 529 ablation after injury will provide insight into whether manipulation of the fibroblasts at 530 specific stages of the pathological response may also have therapeutic value. Our study 531 reinforces the idea that controlled fibroblast reduction may be a potential strategy in 532 reducing maladaptive fibrosis in heart failure and other sustained cardiac diseases. 533 534

Methods 535
Mice 536 All animal protocols and experiments were approved by the University of Hawaii at Manoa 537 Institutional Animal Care and Use Committee. Both males and females on a mixed 538 C57Bl/6 background were used for these studies. R26R tdT (54) (Jackson labs, 007914) 539 and R26R DTA (55) (Jackson labs, 010527) mice were purchased from Jackson 540 Laboratory. Collagen1a1-GFP transgenic mice express cytoplasmic GFP under the 541 control of a col1a1 promoter/enhancer and were generated by Dr. David Brenner (33). 542 PDGFRa CreERT2/+ mice were kindly provided by Dr. Brigid Hogan (Duke University Medical 543 Center) (29). DTA expression was induced between 8 to 10 weeks of age by oral gavage 544 on two non-consecutive days using 0.2 mg/g of body weight of tamoxifen (AdipoGen, 545 CDX-T0200). The genotype for fibroblast reduction was PDGFRa CreERT2/+ ; R26R tdT/DTA . All 546 control mice used in these experiments were tamoxifen induced and age matched. 547 Control genotypes were PDGFRa CreERT2/+ ; R26R tdT/+ or PDGFRa +/+ ; R26R tdT/DTA 548 (littermate controls). A mix of male and female littermate (Cre -) and non-littermate (Cre + ) 549 controls were used. All experiments were performed on adult mice older than 8 weeks. 550 551

Screening for PDGFRa deletion 552
All mice used in these experiments were screened for fibroblast loss either by PDGFRa 553 expression, Col1a1GFP transgene expression, or PDGFRa transcript levels in the heart 554 or kidney. Kidney tissue was collected, stored in RNAlater Stabilization Solution 555 (Invitrogen, AM7021), and processed for RNA extraction as described below. Mice with 556 less than 45% reduction of either fibroblasts or PDGFRa expression were excluded from 557 studies (Supplemental Table 3). 558 559 Immunostaining and microscopy 560 Cardiac tissue was excised, washed with DPBS, and saturated with 3 M KCl to arrest the 561 heart in diastole. Hearts were bisected coronally and fixed with freshly prepared 4% PFA 562 for 2 h at room temperature or 10% neutral buffered formalin (NBF) for 24 h at 4°C. Tissue 563 fixed with 4% PFA was cryoprotected and frozen embedded. Immunostaining was 564 performed on 10 µm tissue cryosections as previously described (27). Primary antibodies 565 used for immunostaining are listed in Supplemental Table 4. Primary antibodies were 566 detected using secondary antibodies from Thermo Fisher at a 1:500 dilution for 1 h at 567 room temperature. Nuclei were stained with DAPI (Roche, 10-236-276-001). Tissue fixed 568 with 10% NBF was processed for paraffin embedding. Trichrome staining was performed 569 on 5 µm tissue sections using Gomori's Trichrome Stain Kit (VWR, 84000-308) according 570 to the manufacturer's protocol. A Zeiss Axiovert 200 microscope equipped with an 571 Olympus DP71 camera was used for imaging. Images and figures were edited in 572 Photoshop CS6 (Adobe). 573 574

Western blot 575
Atria and valves were removed from isolated hearts. Whole ventricle tissue was 576 homogenized in RIPA buffer with protease inhibitor cocktail (Bimake, B14001) using a 577 Dounce homogenizer. Samples were centrifuged at 16,000 x g for 20 min at 4°C, and 578 supernatant was collected. Blots were probed with primary antibody overnight at 4°C, and 579 then incubated with the corresponding Li-Cor IRDye secondary antibody for 1 h at room 580 temperature. Primary antibodies used for western blot are listed in Supplemental Table  581 4. An Odyssey CLx imaging system was used for detection and images were analyzed 582 using Image Studio version 5.2.5 software (Li-Cor Biosciences). 583 584 qRT-PCR 585 RNA isolation was performed on whole ventricle tissue using IBI Isolate DNA/RNA 586 reagent (IBI Scientific, IB47602) and PureLink RNA mini kit (Thermo Fisher, 12183025) 587 according to the manufacturer's instructions. RNA quality and concentration were 588 determined by spectrophotometry using a NanoDrop 2000 instrument (Thermo Fisher). 589 cDNA was synthesized using M-MLV Reverse Transcriptase (Sigma, M1302) and 590 random hexamer primer (Thermo Fisher, SO142). qPCR analysis was performed using 591 PowerUp SYBR Green Master Mix (Thermo Fisher, A25742) and a LightCycler 480 592 instrument (Roche). Samples were run in triplicate and normalized to 18s or Gapdh 593 expression. The 2 -DDCt method was used for determining relative gene expression levels. 594 Primer sequences used for qRT-PCR are listed in Supplemental Table 5. 595 596

Flow cytometry 597
Adult hearts were transcardially perfused with DPBS containing 0.9 mM Ca 2+ , and atria 598 and valves removed. Single-cell suspensions were obtained as previously described (28

Proteomics of decellularized tissue 645
Lysates from decellularized heart tissue were obtained as described above, and reduced, 646 alkylated, and trypsin-digested into peptides. The peptides were cleaned using a Sep-647 Pak Vac C18 cartridge (Waters Corporation) and analyzed label-free by liquid 648 chromatography-tandem mass spectrometry using a Q Exactive mass spectrometer 649 (Thermo Fisher). A 15 cm Å~ 75 μm C18 column (5 μm particles with 100 Å pore size) 650 was used and the nano-UPLC ran at 300 nL/min with a 150 min linear acetonitrile gradient 651 (from 5 to 35% B over 150 min; A = 0.2% formic acid in water; B = 0.2% formic acid in 652 90% acetonitrile). Tandem  659 RAW files were analyzed using Proteome Discoverer 2.2 (Thermo Fisher). Precursor 660 mass tolerance was set at 10ppm and fragment mass tolerance was set at 0.6Da. 661 Dynamic modification was set to oxidation (+15.995 Da (M)) and static modification was 662 set to carbamidomethyl (+57.021 Da (C)). Samples were searched against the reviewed 663 mouse database downloaded from Uniprot (on November 2018, with 16977 sequences). 664 A strict false discovery rate (FDR) of 1% was applied. Label-free quantification was done 665 based on precursor ion intensity and normalization was done using the total peptide 666 amount (from all peptides identified). Proteins were included only if they were identified 667 by at least two high-confidence peptides. 668

669
For further visualizations such as multi-scatter plot, principle component analysis (PCA), 670 volcano plot, and heat map, normalized abundances from all the samples were imported 671 in Perseus 1.6.5.0. Values were log 2(x) transformed and data was filtered by excluding 672 proteins which were not identified in at least 50% of the samples. The missing values for 673 proteins present >50%, but not all samples were imputed from the normal distribution 674 feature. Multi-scatter plot and PCA analysis were performed using all proteins, whereas 675 only ECM proteins were used for generating the volcano plot and unsupervised 676 hierarchical clustering. 677 678

Heart degradomics using Terminal Amine Isotopic Labeling of Substrates (TAILS) 679
Protein extraction was performed using T-Per tissue protein extraction reagent (Thermo 680 Fisher) with protease inhibitor cocktail (Roche) and prepared for TAILS as previously 681 described (37). Briefly, heart tissue was homogenized in 0.5 mL of T-Per on ice and 682 centrifuged, supernatant was collected in a fresh tube (T-Per extract) and 4M GuHCl 683 containing a protease-inhibitor cocktail was added to the pellet and incubated further at 684 4 º C for 24 h on a rotary shaker (GuHCl extract). All subsequent processing was done 685 separately for T-Per and GuHCl extracts. Protein estimation was done using the Bradford 686 assay and 200 µg of protein from each heart was denatured, reduced, and alkylated as 687 per the iTRAQ labeling protocol (SciX). iTRAQ labels were reconstituted in DMSO and 688 samples were incubated with iTRAQ labels for 2 h in the dark. Excess iTRAQ reagent 689 was quenched by incubation with 100 mM ammonium bicarbonate for 30 min in the dark. analysis was done and N-termini identified with a false discovery rate < 1% were 705 annotated essentially as recently described (37). 706 707

Myocardial infarction 708
Adult mice >22.0 g were subjected to MI as previously described (8). Briefly, a 709 thoracotomy was performed between the third and fourth rib to expose the LAD artery. 710 The LAD proximal artery was permanently ligated using a 7.0 silk suture. Ligation was 711 confirmed by visualization of LV blanching and ST elevation on the electrocardiogram. 712 713

Osmotic pump implantation 714
Adult mice >20.0 g were infused with AngII/PE to induce cardiac hypertrophy and fibrosis. 715 Mice were anesthetized with 1-2% isoflurane and mini-osmotic pumps (Alzet, 2001(Alzet, , 2002 or 2004) were implanted subcutaneously. A combination of 1.5 µg/g/day angiotensin II 717 (Calbiochem, 05-23-0101) and 50 µg/g/day phenylephrine hydrochloride (Sigma, P6126) 718 or saline, was infused for 7, 14, or 28 days. 719 720 Echocardiography 721 Echocardiography was performed using a Vevo 2100 system (VisualSonics) to analyze 722 cardiac function after MI and AngII/PE infusion in conscious mice. Briefly, hair removal 723 cream was applied to the chest until all fur was removed from the area. A layer of 724 ultrasound gel was applied to the animal's chest, and the probe was lowered at the 725 parasternal line, 90° between the probe and the heart. B-and M-modes were performed 726 to record 2-dimensional and 1-dimensional transverse cardiac measurements and used 727 to analyze LV function. Independent echocardiography on mice 7 months post-induction 728 at The Ohio State University also found no differences in cardiac measurements between 729 control and fibroblast-ablated hearts. 730 731

Pressure-volume (PV) loop 732
Cardiac hemodynamic measurements were assessed via a closed chest approach using 733 a 1.4 rodent pressure volume catheter (Transonic) advanced into the left ventricle through 734 the right carotid artery (56). In brief, mice were anaesthetized by ketamine (55 mg kg -1 ) 735 plus xylazine (15 mg kg -1 ) in saline solution and placed in supine position on a heat pad. 736 Following a midline neck incision, the underlying muscles were pulled to expose the 737 carotid artery. Using a 4-0 suture, the artery was tied and the pressure-volume catheter 738 was advanced through the artery into the left ventricle of the heart. After 5-10 min of 739 stabilization, values at baseline and stimulation at varying frequencies (4-10 Hz) were 740 recorded. To measure the beta-adrenergic response, 5mg kg -1 dobutamine was injected 741 intraperitoneal. All the measurement and analysis were performed on LabChart7 (AD 742 Instruments). 743 744 Cardiomyocyte cross-sectional area (CSA) quantification 745 Cardiomyocyte boundaries were identified by wheat germ agglutinin labelling in 10 µm 746 tissue sections that were processed as described above. Cardiomyocytes in cross-747 section were defined by having a circular to oval shape, surrounded by circular capillaries. 748 A total of 100 cardiomyocytes in a section were outlined and CSA was calculated using 749 ImageJ (NIH). 750 751

Cardiomyocyte isolation 752
Cardiomyocytes from whole ventricle tissue were isolated from adult mice using a 753 modified Langendorff-free collagenase digestion protocol (57). Briefly, mice were 754 anesthetized with an intraperitoneal (IP) injection of tribromoethanol (0.4 mg/g of body 755 weight), the thoracic cavity was opened, and descending aorta was severed. The right 756 ventricle was immediately flushed with EDTA buffer and the ascending aorta was 757 clamped. The heart was excised and the left ventricle was perfused with EDTA and 758 perfusion buffer. The heart was enzymatically digested with 0.5 mg/mL Collagenase Type 759 II (Worthington, LS004176), 0.5 mg/mL Collagenase Type IV (Worthington, LS004188), 760 and 0.05 mg/mL Protease Type XIV (Sigma, P5147). The atria and valves were removed 761 and tissue was teased apart. Cells were dissociated by gentle trituration with a wide-bore 762 pipette tip and cell suspension was filtered through a 100 µM nylon strainer. Cells were 763 the Database for Annotation, Visualization and Integrated Discovery (DAVID) version 6.8 787 program to search for enriched gene ontology (GO) terms. 788 789