Angiotensin II-induced cardiac fibrosis and dysfunction are exacerbated by deletion of cGKI in periostin⁺ myofibroblasts Open Access

Myocardial fibrosis, defined by excessive deposition of extracellular matrix (ECM) proteins, leads to fibrotic scarring of the myocardium and is a common pathologic feature associated with a wide spectrum of cardiovascular diseases (CVDs), including hypertensive heart disease, diabetic hypertrophic, and idiopathic dilated cardiomyopathy [1-4], which, together with other CVDs, are the leading cause of death worldwide [5,6]. Although scar formation, especially after myocardial infarction (MI), compensates for cardiomyocyte (CM) cell death and prevents myocardial rupture, the progression of myocardial fibrosis is associated with a decline in cardiac function, frequent arrhythmias, loss of cardiac elasticity, and eventually resulting in the development of heart failure (HF) with poor outcomes [7-10].

Extensive preclinical studies have identified cardiac myofibroblasts (CMFs) as the major effector cell type contributing to the structural and functional changes of the heart in response to stress [11,12]. CMFs do not appear in the healthy heart but instead arise from activated resident cardiac fibroblasts (CFs) following cardiac injury. Accordingly, MI or a chronic elevation of the angiotensin II (Ang II) levels [11,13-15], among other causes, frequently triggers myocardial fibrosis and also CM hypertrophy, which is another feature of the structural, geometrical, and functional remodeling process.

Ang II is one major effector peptide of the renin–angiotensin–aldosterone system (RAAS) that conveys its adverse actions in the cardiovascular system mainly via the AT₁-receptor subtype, which is found at high density in CMs [16] and blood vessels [17,18], as well as in cultured human and rodent CFs [19-21]. Ang II effectively stimulates the proliferation and migration of CFs, highlighting its potential to activate CFs to phenotypically and functionally ‘switch’ to CMFs [14,19,20,22,23]. Interestingly, Ang II infused chronically at pressor dosages induces a change in the phenotype of CFs, resulting in the accumulation of ECM in the myocardium within days of treatment [24]. However, additional cell types including pericytes, mesenchymal, and hematopoietic cells have been suggested as possible sources for CMFs in the injured heart, leading to an ongoing debate about the actual progenitor cell type of CMFs [8,13]. Interestingly, CMFs combine the cytoskeletal and contractile functions of vascular smooth muscle cells (VSMCs) with the ECM-producing abilities of CFs. At the molecular level, this results in high expression levels of the cytoskeletal protein alpha smooth muscle actin (α-SMA) and an excessive synthesis and secretion of ECM proteins such as type I and III collagen (COL I and COL III), fibronectin (FN1), and matricellular proteins like periostin (POSTN), which together contribute to the increase in chamber stiffness, impaired muscle relaxation, and ultimately in a decline in cardiac performance [25-28]. Given these features of CMFs, a modulation of their function should be a promising strategy to counteract or even prevent the development and progression of multiple heart diseases, especially HF.

The second messenger 3′,5′-cyclic guanosine monophosphate (cGMP) signaling axis is a key regulator of cardiovascular homeostasis [29-31]. Cyclic GMP is generated intracellularly either by nitric oxide (NO)-mediated activation of NO-sensitive (soluble) guanylyl cyclase (NO-GC aka GC or sGC) isoform 1 or 2 or by natriuretic peptides (NP)-dependent stimulation of membrane-bound particulate guanylyl cyclase (pGC) receptors, also designated GC-A and GC-B [32,33]. The resulting increase in cGMP in the heart causes activation of the cGMP-dependent protein kinase type I (cGKI), which is considered to be the major downstream effector of the NO/NP-cGMP cascade in multiple cardiac cell types [34,35]. Another key element of this cascade are cGMP-hydrolyzing phosphodiesterases (PDEs), which control the duration, amplitude, and spread of an intracellular cGMP signal. PDEs, that can hydrolyze cAMP in addition to cGMP, are responsible for the so-called cross-talk between this two signaling pathways [35].

Clinical implementation of sacubitril, a neprilysin inhibitor that prevents, besides other effects, the degradation of NPs, and vericiguat, a NO-GC stimulator, in the treatment of HF with reduced ejection fraction (HFrEF) emphasize that the cGMP elevation presumably resulting from these principles improves patient outcomes. However, neither the exact cell type(s) nor the downstream mechanisms targeted by both drugs are clear [36-38]. Besides the widely appreciated effects of cGMP on the survival and growth of CMs [39,40], growing evidence from cell- and animal-based studies supports the notion that cGMP counteracts the adverse cardiac remodeling process in HF models by modulating the activity and phenotype of CFs and/or CMFs [41-43]. Accordingly, the analysis of genetically modified mice globally lacking NO-GC1 [44], which is the major NO-GC isoform in the cardiovascular system, or cGKI specifically in CMs [45] showed a significant increase in collagen deposition due to chronic Ang II challenge. Moreover, a CM-specific deletion of the Ca²⁺- and voltage-activated K⁺ channel BK, an established downstream target of cGMP/cGKI signaling, caused an unusually high accumulation of fibrosis in post-MI hearts [46]. Sildenafil, a PDE5 inhibitor with widely reported anti-hypertrophic and anti-fibrotic potential [47], counteracted the Ang II-induced increase in collagen deposition and fibrotic marker gene expression but only when cGKI was present in all non-smooth muscle cell types of the heart [48]. Among other possible explanations, the latter study strongly suggests that CF/CMF cGKI signaling may be involved in mediating the beneficial effects of the PDE5/cGMP axis on fibrosis. In contrast, however, chronic cGKI activation, caused by a constitutive, cGMP-independent regulation of the kinase in mice, can be harmful to the heart, especially in the presence of Ang II [49].

Taken together, these studies imply a strong anti-fibrotic function for the endogenous cardiac cGMP/cGKI axis. However, due to the lack of suitable models, it is mechanistically still unclear how or whether the signaling pathway counteracts the detrimental remodeling processes through a function that originates directly from CFs and/or CMFs [50-54].

To close this gap in knowledge, the present study aimed to explore the role of cGKI in Postn-positive (Postn⁺) CMFs, as Postn⁺ CMFs have been identified as the key effector cells contributing to myocardial fibrosis in neurohumoral, i.e., Ang II-induced cardiac stress models [13]. To generate CMF-specific conditional cGKI knockout (cmfKO) and corresponding control (CTR) littermates, we intercrossed floxed cGKI mice to a tamoxifen (TAM)-inducible PostniCre strain [50]. This strategy allowed us to investigate putative functions of cGKI in a defined CMF population that develops exclusively in response of the heart to pathological stresses, while prior to the TAM challenge, ‘pre-mutant’ cGKI animals should develop normally. By assessing cardiac outcome in terms of fibrosis development, extent of hypertrophy, and multiple determinants of cardiac function, as well as by monitoring the proliferative behavior of isolated CF/CMF primary cell cultures, we ultimately wanted to clarify if the cGMP/cGKI signaling cascade in CMFs affects the adverse remodeling induced by chronic Ang II exposure.

The online supplement contains further descriptions of the materials and methods.

All animal experiments performed were authorized by the local Ethics Committee for Animal Experiments (Regierungspräsidium Tübingen, PZ02/22 G) and are in compliance with the European Directive 2010/63/EU on the protection of animals used in scientific research. Animals were housed in a standardized cage system at defined room temperature (RT) and humidity conditions and had ad libitum access to food and water during a cycle of alternating 12 h of light and 12 h of darkness.

To inactivate cGKI specifically in the CMF cell population, a transgenic PostniCre^tg/+ (Tg(Postn-icre/ERT2)#Wet) mouse line was employed expressing a TAM-inducible Cre recombinase under the control of Postn promoter [50]. CMF-specific conditional cGKI knockout (cmfKO; genotype: PostniCre^tg/+ x cGKI^fl/fl) and corresponding littermate controls (CTR; genotype: PostniCre^tg/+ x cGKI^+/+) were obtained by intercrossing PostniCre^tg/+ to heterozygous floxed cGKI (cGKI^fl/+) [55-58] parental animals. Male mice, weighing 20–30 g and aged 10–16 weeks, were used for experiments. Tissue specificity and efficiency of Cre-recombination were assessed by intercrossing the transgenic PostniCre^tg/+ mouse line with the ROSA^mT/mG (B6.129(Cg)-Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,-EGFP)Luo}/J) double-fluorescent Cre reporter mouse line [59], resulting in the generation of experimental 10- to 16-week-old male double-transgenic ROSA^mT/mG × PostniCre^tg/+ and corresponding ROSA^mT/mG control mice.

For post-mortem analyses, experimental mice were euthanized in their home cage with CO₂ or under deep isoflurane anesthesia followed by cervical dislocation. All animal studies are reported in compliance with the ARRIVE guidelines [60].

Male mice, weighing 20–30 g and aged 10–16 weeks, were used for experiments. Our study examined male mice to avoid potential interactions between the female sex hormones in our tamoxifen-inducible model and because male animals usually exhibited less variability in the cardiac stress response. The pathophysiological effects of Ang II on the structure and function of the left ventricle have been extensively studied in male mice. Also, by focusing on one sex, the total number of subjects included in the study was lower, but this clearly limits any conclusions of our study for the female gender.

For induction of cardiac hypertrophy and cardiac fibrosis, osmotic minipumps (ALZET^® pump model 1004 #0009922) releasing Ang II (Sigma-Aldrich, #A9525 at 2 µg/g/d; diluted in 0.9% saline/0.01 M acetic acid) continuously for 28 days were implanted subcutaneously (s.c.) in the left flank of mice as previously described [50,61]. Initial anesthesia was achieved by intraperitoneal (i.p.) injection of a combination of medetomidine (500 µg/kg), midazolam (5 mg/kg), and fentanyl (50 µg/kg) and maintained throughout the surgical intervention with continuous inhalation anesthesia of 1.0–2.0 Vol% isoflurane/oxygen. At the end of the surgery, mice received flumazenil (0.5 mg/kg) and atipamezole (2.5 mg/kg) via s.c. application to antagonize anesthesia. Animal analgesia was intraoperative provided by administration of fentanyl i.p., followed by postoperative treatment with oral metamizole (1.25 mg/ml drinking water). Consistent with Kaur et al. (2016), Cre recombinase was activated by i.p. administration of 1 mg TAM (Sigma-Aldrich #T5648; dissolved in 50 µl Miglyol^®812, Caelo #3274) once daily, starting on the first postoperative day, for five consecutive days [50]. Animals received only a daily injection of TAM for five consecutive days in the corresponding control experiments without Ang II. The examination of the hearts was performed 28 days after the first TAM injection.

To assess changes in BP, heart rate and locomotory activity in CTR and cmfKO mice as a result of prolonged Ang II infusion, implantable transmitters (Data Science International TA11PA-C10) allowing data acquisition by a telemetry system (Data Science International-Dataquest A.R.T.3.1 software), were employed described as previously described [61,62]. Briefly, anesthesia was induced by i.p. injection of a combination of medetomidine (500 µg/kg), midazolam (5 mg/kg), and fentanyl (50 µg/kg) and was supplemented after 30 min by continuous inhalation anesthesia with 1.0–2.0 Vol% isoflurane/oxygen. After the mice were placed in supine position on a heating plate to maintain body temperature at 37°C throughout the procedure, a ventral midline incision was made to allow the fluid-filled catheter to be inserted through a small incision into the left common carotid artery and advanced toward the thoracic aorta. The catheter was secured in this position with two additional sutures, with the transmitter unit placed subcutaneously on the right flank of the mice. At the end of the surgery, anesthesia was antagonized by flumazenil (0.5 mg/kg) and atipamezole (2.5 mg/kg) s.c. and analgesia was provided by perioperative administration of buprenorphine (0.05–0.01 mg/kg) s.c and by postoperative treatment with paracetamol p.o. (1.3 mg/ml drinking water). Following a seven-day recovery period, basal BP values were recorded at 15 min intervals for a duration of 5 min for three consecutive days. Immediately after implantation of Ang II releasing osmotic minipumps, BP was monitored at the previously indicated intervals for the following seven days.

Non-invasive echocardiography was conducted using a Vevo2100 imaging system (FUJIFILM Visual Sonics) in CTRs and cmfKOs treated either with or without Ang II at day 28, as previously described [46,61,63]. In brief, animals were anesthetized by continuous inhalation anesthesia with 1.0–2.0 Vol% isoflurane/oxygen, restrained in supine position on a heating plate to maintain a body temperature of 37°C and depilated in the left thoracic region to avoid ultrasound artifacts. Limb electrodes permitted continuous ECG monitoring. Following the application of the ultrasound gel, high-resolution two-dimensional echocardiographic images were acquired using a 30-MHz transducer in the left parasternal long-axis view (PLAX). M-Mode recordings were acquired from three distinct selected positions for quantification of three cardiac cycles, respectively, allowing the determination of global cardiac function parameters such as ejection fraction (EF) and fractional shortening (FS), as well as wall dimensions from the mean of nine cardiac cycles.

B-Mode recording in PLAX enabled the evaluation of both further cardiac function parameters and, in particular, regional left ventricle deformation capacities using speckle-tracking echocardiography (STE) [64]. After acquisition of left ventricular (LV) wall motion by semiautomatic tracing of the endocardium and epicardium, a total of three consecutive cardiac cycles were evaluated and averaged for the evaluation of the global longitudinal strain (GLS). For the assessment of regional deformation capacity, the algorithm of the software (Vevo^® LAB, Vevo^® Strain Analysis) subdivided the LV along the long axis into six distinct regions (AB: anterior base, AM: anterior mid, AA: anterior apex, PB: posterior base, PM: posterior mid, PA: posterior apex). Strain [%], strain rate [1 /s], and velocity [cm/s] in LV longitudinal motion direction were obtained for each segment by LV wall motion tracing. Besides the peak values for each segment, the average strain values were additionally determined from the average curve generated by software. Three consecutive cardiac cycles were selected and averaged for analysis.

For the isolation of CF/CMF primary cell cultures, CTR and cmfKO hearts obtained from Ang II-treated mice were harvested as quickly as possible and placed in ice-cold Dulbecco's Phosphate-Buffered Saline (DPBS). The atria along with the aorta were removed by a transversal incision through the heart, and ventricles were cleared of blood debris using DPBS, and then dissected into ~1 mm pieces. To extract the cells, pieces were digested for 80 min at 37°C in a water bath using a collagenase type II-based solution (ddH₂O containing 85 mM Na-Glutamate, 60 mM NaCl, 10 mM HEPES, 5.6 mM KCl, and 1 mM MgCl₂.6H₂O supplemented with 1 mg/ml collagenase type II and 1 mg/ml bovine serum albumin, pH 7.4). In order to support this process and to preserve cell viability, the supernatant consisting of the digestion solution containing cells detached from the tissue was transferred to 26 ml of culture medium (DMEM+GlutaMAX^TM supplemented with 1% Penicillin Streptomycin solution (PenStrep), 10% Fetal bovine serum, and 1% Insulin–Transferrin–Selen) every 20 min to protect them from further digestion, while fresh digestion solution was added to the not yet completely dissolved heart pieces. Using a 40-µm cell strainer (Greiner bio-one #542040), as well as a centrifugation step (5 min at 300 rpm), non-digested heart segments and isolated CMs were separated from the CF/CMF cell fraction. Purification of CFs/CMFs was achieved by a further centrifugation step for 7 min at 1400 rpm. The obtained cell pellet was resuspended in 1 ml of culture media, counted using a Neubauer haemocytometer (Millipore #MDH-2N1) and plated out accordingly to the required cell number for each experiment.

The GraphPad Prism 9.4.1 software was employed for the statistical analysis. All data are presented as mean + standard error of mean (SEM). Normal Gaussian distribution was confirmed by Shapiro–Wilk or Kolmogorov–Smirnov test. For the comparison of two groups (CTR versus cmfKO), the unpaired student t-test (α = 0.05) was performed in the presence of Gaussian normal distribution, and the Mann–Whitney U test was used as non-parametric test. In case of more than two groups (CTR+TAM; CTR+TAM+Ang II; cmfKO+TAM; cmfKO+TAM+Ang II), either the two-way ANOVA followed by a Šídák’s or Tukey’s multiple comparison test was performed or a Kruskal–Wallis test followed by the Dunn test for multiple comparisons was employed as non-parametric test. The log-rank test (Mantel-Cox test) was used to assess the survival rate. A more precise description of the statistical analyses conducted for the respective data sets is provided in the figure legends. P<0.05 were considered statistically significant with *P<0.05, **P<0.01, and ***P<0.001, indicating differences between the genotypes (CTR versus cmfKO). Comparison of the distinct treatments (+TAM versus +TAM+Ang II) within the same genotype was represented as followed: †P<0.05; §P<0.01; #P<0.001.

For all in vivo experiments, group sizes were equal by design, and G*Power (version 3.1.5) was used to calculate the required size of each experimental group of mice a priori with a power of 80% and an α level of 0.05. Randomization of the animals into different treatment groups was carried out before minipump implantation by generating random numbers for each individual animal. The different test conditions and experimental units were always carried out in parallel and with both animal genotypes at the same time. Genotype and treatment of the mice were not known to the experimenter and data analyst, so that an unbiased evaluation of the results was possible.

Statistics for main figures are listed together with raw data in Supplementary Table S1-S4.

To investigate the putative function of CMF-specific cGKI in Ang II-induced cardiac remodeling, we utilized a transgenic PostniCre (genotype: PostniCre^tg/+) mouse line expressing a TAM-inducible Cre-recombinase under the control of the Postn promoter in CMFs [50]. First, the cell specificity and inducibility of PostniCre activity were verified under patho-/physiological conditions. Therefore, PostniCre^tg/+ mice were intercrossed with a double-fluorescent ROSA^mT/mG Cre reporter mouse strain expressing membrane-targeted tandem dimer Tomato (mT) prior to Cre-mediated excision and membrane-targeted green fluorescent protein (mG) after excision of the loxP flanked mT cassette [59]. In unchallenged adult double-transgenic ROSA^mT/mG × PostniCre^tg/+ as well as ROSA^mT/mG mice used as controls, Cre-mediated recombination was induced by daily i.p. TAM-injection for five consecutive days (Supplementary Figure S1A). Four weeks after the first TAM injection, ubiquitous expression of the cell membrane-targeted mT protein was detected in the myocardium (Supplementary Figure S1B), as well as in multiple other organs including lung, aorta, liver, spleen, and kidney (Supplementary Figure S1C) of both genotypes. Importantly, we found no evidence of Cre-mediated excision of the mT transgene, which would lead to mG expression, confirming previous reports suggesting that Postn⁺ CMF develops only under pathophysiological conditions [8,13,50]. Thus, we next analyzed PostniCre-mediated recombinase activity in the ROSA^mT/mG reporter strain in response to chronic Ang II released from osmotic minipumps. As Ang II-induced mRNA and protein expression of periostin is stimulated within hours in CFs [65], Cre-activation was induced by TAM on the first day after minipump implantation for five consecutive days (Supplementary Figure S2A). This previously established protocol [50] resulted in PostniCre-mediated excision of the loxP-flanked DNA sequence encoding mT, which was evident due to mG expression exclusively in cells occurring in fibrotic heart areas of double transgenic ROSA^mT/mG x PostniCre^tg/+ mice (Supplementary Figure S2B). Except for small fractions of cells in the lungs of double-transgenic mice, mG expression was restricted to myocardial CMFs and not detectable in CMs, as well as in non-fibrotic cardiac regions or any other organ systems analyzed including liver, spleen, kidney, and aorta, all of which exhibited persistent expression of the red fluorescent mT-protein (Supplementary Figure S2B and C). In order to confirm these findings at a cellular level, we isolated primary CFs/CMFs from ROSA^mT/mG × PostniCre^tg/+ and ROSA^mT/mG mice after Ang II treatment. In contrast with ROSA^mT/mG CF/CMF, which exclusively expressed the red fluorescent mT-protein, we obtained a mixed culture of primary CFs/CMFs from the double-transgenic hearts. These primary cultures consisted of cells stimulated in vivo by Ang II to form Postn⁺ CMFs, characterized by the expression of the green fluorescent mG protein, and mT-positive CFs/CMFs (Supplementary Figure S2D). Due to the highly reliable TAM-controlled activation of the CMF-specific Cre-recombinase in Ang II exposed hearts, we conclude that the PostniCre^tg/+ mouse strain is suitable to assess the in vivo functions of cGKI in Postn⁺ CMFs in adverse cardiac remodeling.

CMF-specific KO mice (cmfKO; genotype: PostniCre^tg/+ x cGKI^fl/fl) lacking exon 10 of the cGKI gene required for proper kinase activity, as well as respective control mice (CTR; genotype: PostniCre^tg/+ x cGKI^+/+) (Supplementary Figure S3A), were subjected to the herein established TAM and Ang II treatment protocol (Supplementary Figure S2A). Consistent with the previous findings (Supplementary Figure S2B-D), genomic PCR analysis confirmed efficient conversion of the floxed Prkg1 allele into the KO allele in both primary CF/CMF cultures and, to a lesser extent, in the lungs obtained from Ang II-treated cmfKO mice (Figure 1A). Neither in other organs isolated from Ang II-treated cmfKO nor in the DNA samples obtained from the corresponding Ang II-treated CTR mice, we observed a recombination of the floxed cGKI gene locus (Figure 1A, Supplementary Figure S3B). Furthermore, immunoblots using a validated cGKI antibody [67] exhibited significantly reduced cGKI protein levels in primary CFs/CMFs obtained from Ang II-treated cmfKO animals compared with corresponding Ang II-treated CTR mice (Figure 1B and C). CMs isolated from an alternative and highly efficient CM-restricted cGKI-KO model (CM-cKO; genotype: αMHC-Cre^tg/+ x cGKI^fl/fl) used as negative control confirmed the high specificity of this approach. Next, we evaluated the cGKI expression pattern in CTR and cmfKO heart sections obtained from Ang II-treated mice. In CTR sections, cGKI expression correlated with the fibrotic myocardium and with Postn⁺ cells, while cmfKO hearts exhibited a substantial reduction in cGKI expression in myofibroblast-rich cardiac sections (Figure 1D).

Depletion of cGKI in CMF resulted in a significantly reduced survival of TAM-treated cmfKO mice during the Ang II challenge, whereas TAM in the absence of Ang II did not affect the overall survival of age- and littermate-matched conditional mutants (Figure 1E). Although TAM in combination with Ang II slightly decreased the survival of CTR animals, this trend was not significantly different from either the cmfKO treatment group or the CTR animals receiving TAM only (Figure 1E). With regard to this excess in mortality in both +TAM+Ang II treatment groups, we identified perivascular and also tubulointerstitial fibrosis in the kidney but no genotype-related adverse remodeling events (data not shown). Accordingly, the integrity of the elastic fibers as well as the thickness of the muscular layer of the thoracic and abdominal aorta showed neither massive abnormalities nor differences between +TAM+Ang II-treated CTR and cmfKO mice (data not shown), although reportedly the architecture of the vessel wall can be disrupted by the development of aortic aneurysms in response to Ang II [68].We next monitored the BP of CTR and cmfKO mice because Ang II causes hypertension through different mechanisms, whereas the contribution of the cGMP/cGKI signaling pathway in CMF to BP control is less clear. Prior to TAM-induced Cre recombination and the Ang II treatment, neither mean arterial pressure (MAP) nor systolic and diastolic BP were statistically different between the two genotypes. This was accompanied by nearly identical pulse pressure (PP) and heart rate (HR) values as well as similar activity patterns of he mice throughout the day and night (Supplementary Figure S4A-F). Infusion of Ang II resulted in a significant and time-dependent elevation of the MAP in CTR and cmfKO mice; however, the extent of this increase was identical in both genotypes (Figure 1F). Also, HR, systolic and diastolic BP, as well as PP elevations, and locomotor activity were identical for CTR and cmfKO mice receiving Ang II (Supplementary Figure S4G-L). Thus, a genotype-related difference in the absolute response of MAP, systolic BP, diastolic BP, and PP, the latter being an important predictor of cardiac complications [69,70], to chronic Ang II exposure could be excluded (Figure 1G). Combined, these results strongly suggest that the poor survival and higher myocardial vulnerability of cmfKO may be due to local effects of the neurohormone on the myocardium, rather than Ang II-related changes in systemic BP control.

Based on the assumption that Ang II directly and independent of BP elevation promotes cardiac remodeling, we next assessed the extent of myocardial fibrosis and hypertrophy in CTR, as well as cmfKO hearts following the TAM protocol and a sustained Ang II treatment. Untreated CTR and cmfKO mice were also included in this analysis to exclude genotype-specific differences in the histological properties of the heart under physiological conditions. Again, all mice received +TAM to induce PostniCre-mediated recombination in CMFs. For the evaluation of fibrosis development, hearts were divided into eight equidistant regions, proceeding from the apex to the base of the heart. Thereby, picrosirius red (PSR) staining, highlighting collagen fibers in the connective tissue, identified a significant increase in global amount of cardiac fibrosis in cmfKO compared with corresponding CTR mice (Figure 2A and B). This increase was seen in each individual cardiac segment obtained from cmfKO hearts (Figure 2C) and reached significance in segments II–VI also for the comparison with the CTR. Accordingly, the loss of cGKI in CMF was associated with a higher type I and III collagen content of myocardial segments IV–VI (Supplementary Figure S5A-D), which is a significant finding, as these two collagens are the main isoforms of this ECM protein that accumulate in response to Ang II [14,71,72]. In contrast, our parallel examination of cardiac sections obtained from TAM but not Ang II challenged mice yielded no genotype-specific differences in the PSR-stained collagen fiber content (Figure 2B and C).

Prolonged Ang II treatment resulted in pathological cardiac enlargement in both genotypes. However, heart weight (HW) and the heart weight-to-tibia length (HW/TL) ratio, considered a surrogate parameter for the development of cardiac hypertrophy [73], were not different between cmfKO and respective CTR mice. Accordingly, the amount of cardiac mass gained did not differ between both genotypes, as further confirmed by the percentage increase in HW/TL ratio. Compared with the respective TAM-treated animals, however, both HW and HW/TL ratios of Ang II-treated mice were significantly higher for both genotypes, which again confirms the validity of the methodology and protocols used (Figure 2E–G). Because CM hypertrophy, in addition to changes in ECM composition, is responsible for the abnormal increase in myocardial mass, we determined CM sizes in hematoxylin and eosin (H&E) stained heart slices from untreated or Ang II-treated CTR and cmfKO mice [74]. Compatible with the cardiac enlargement and mass gain, we observed a significant increase in CM cross-sectional areas in both Ang II-treated genotypes as compared with their respective control groups that received TAM only. Consistent with our data quantifying fibrosis, this increase was more pronounced in Ang II-treated cmfKO versus CTR mice (Figure 2H and I). In a complementary approach, we used fluorescently labeled wheat germ agglutinin (WGA) to mark the borders of the CMs (Supplementary Figure S5E). Again, cross-sectional areas of randomly chosen cells were determined. Although the absolute values of the randomly selected cells differed slightly from those determined by examining the H&E-stained heart sections, significant size differences between the two genotypes were found only after +TAM+Ang II exposure (Supplementary Figure S5F).

To clarify the apparent discrepancy regarding the massive increase in fibrosis (Figure 2A and B) and the enlarged CMs in the presence of identical heart weights compared with the controls (Figure 2D–G), we quantified CM cell viability by TUNEL staining (Figure 3). Because the number of apoptotic CMs was significantly increased in CMF-specific cGKI KO hearts, we conclude that a functional cGMP/cGKI cascade in CMF improves the survival of CMs in the presence of Ang II. In addition, the higher cell death rate in the cmfKO explains why the total weight of the heart, despite the greater fibrosis of the cardiac muscle and the larger CMs, does not differ between conditional cGKI mutant and CTR mice. Next, we investigated whether these unique alterations in CM size, collagen density, and total organ weight were caused by an unexpected change in the cardiac periostin expression pattern. In the absence of Ang II and fibrotic lesions, we were unable to detect periostin in cardiac section of either genotype (Supplementary Figure S5G), while transcript levels (Supplementary Figure S6A and B) and the number of periostin-positive primary CF/CMFs obtained from +TAM+Ang II treated cmfKO and CTR mice were identical (Supplementary Figure S6C). This indicates that the matricellular protein periostin, which may actively contribute to tissue fibrosis and remodeling processes, is not responsible for the in vivo phenotype we observed in the absence of cGKI in CMFs.

To test whether the enhanced collagen deposition in cmfKO hearts was due to an accelerated proliferation rate of CMFs lacking cGKI [42,75,76], we visualized fibrotic areas by an anti-type I collagen immunofluorescence staining in Ang II-treated cmfKO and CTR hearts and used Ki-67 expression as marker for CMF proliferation. Thereby, we detected a significantly increased number of Ki-67-positive nuclei in cmfKO compared with CTR hearts, indicative of a higher cellular proliferation rate in the absence of CMF-specific cGKI (Figure 4A and B). To attribute the higher proliferation rate to impaired cGMP/cGKI signaling in cmfKO-derived CMFs, primary CFs/CMFs were isolated from TAM- and Ang II-treated CTR and cmfKO animals and cultured in a grid-based well system. The purity of cell cultures obtained was routinely verified by quantifying transcript levels of marker genes preferentially expressed either by CMs (Tnni3/cTNI) or endothelial cells (Icam1/ICAM-1) and by assessing Pecam1/PECAM-1, a broad cell surface marker of hematopoietic and immune cells including platelets (Supplementary Figure S6A and B). While these cell markers were expressed at very low levels, mRNA transcripts of periostin and α-SMA (Acta2), both representing widely used and specific markers of myofibroblast differentiation, were present at several orders of magnitude higher and, importantly, did not differ between the genotypes (Supplementary Figure S6B). This enabled us to monitor the growth of highly pure CF/CMF cultures containing >85% of periostin-positive cells (Supplementary Figure S6C and D) in vitro over a duration of five days (Supplementary Figure S6A). Consistent with the Ki-67-staining of the myocardium (Figure 4A and B), primary CF/CMF cultures obtained from +TAM and +Ang II cmfKO hearts exhibited significantly increased growth rates compared with the respective CTR cell cultures (Figure 4C and D), a finding that was further confirmed by counting the respective cell nuclei at the indicated time points (Supplementary Figure S6E and F).

By using the membrane-permeant cGMP-analogue 8-Br-cGMP on primary CF/CMF cultures, we next tested whether an activation of the cGMP/cGKI axis would also interfere with the production of pro- and/or anti-fibrotic factors in these cells. To address this, CF/CMF cultures were stimulated with 8-Br-cGMP for 24 h and transcript analysis was performed on total RNA isolated from CTR and cmfKO cells (Supplementary Figure S6A). 8-Br-cGMP, in a cGKI-dependent manner, interfered with the expression of pro-fibrotic marker genes such as Acta2, Col1a1 and Fn1, whereas levels of IL-6 and TGFβ1 were not affected by 8-Br-cGMP/cGKI in cells obtained from +TAM- and +Ang II-treated CTR hearts. As we did not detect significant differences between genotypes, we conclude that the effects of cGKI on the pro-fibrotic gene expression profile contribute to a lesser extent than the effects of the pathway on the proliferative behavior of CF/CMF (Figure 4C and D and Supplementary Figure S6E) to the observed in vivo phenotype.

Finally, we assessed the phosphorylation of the vasodilator-stimulated phospho-protein (VASP) to demonstrate a proper activation of cGKI in response to the 8-Br-cGMP treatment. Indeed, 30 min of exposure increased the phospo-VASP (pVASP) to VASP ratio only in CTR cells, while pVASP and VASP levels, as well as the respective ratio, remained at the basal level in the absence of CMF cGKI (Figure 4F and G). This finding once again emphasizes a proper pharmacological modulation of cGKI in CTR versus cmfKO CMF.

Cardiac fibrosis interferes with a proper cardiac contractility and relaxation eventually limiting cardiac function [2]. Consequently, we applied non-invasive echocardiography technique to investigate structure- and function-related cardiac dysfunctions of CTR and cmfKO hearts at baseline and upon in vivo Ang II exposure. Consistent with the histochemical alterations, we detected a deteriorated global cardiac function in cmfKO hearts, characterized by a significant reduction in the ejection fraction (%EF) and fractional shortening (%FS) (Figure 4H–J). The extent of dysfunction was significant compared with corresponding TAM-treated cmfKO, as well as Ang II-treated CTR animals. The latter even showed the ability to functionally compensate for the cardiac damage induced by Ang II as %EF and %FS remained unchanged compared with untreated control animals. Additionally, end-systolic and end-diastolic measurements of the intraventricular septum (IVS) and LV posterior wall (LVPW) were evaluated in order to provide detailed insights into LV chamber dimension, i.e., wall thickening and motion. LV mass estimation by echocardiography identified an increase in both genotypes, which reached a statistical significance only for the cmfKO compared with their respective TAM-treated control group (Supplementary Figure S7A). Quantification of wall dimensions revealed a significant thickening of both the end-systolic LVPW in CTR hearts and of the end-diastolic IVS and LVPW in CTR and cmfKO hearts (Supplementary Figure S7B-J). Furthermore, we found evidence for a disturbed wall motion in the LVPW segment of Ang II-treated cmfKO as end-systolic to end-diastolic wall diameters were significantly reduced only in this group of animals (Supplementary Figure S7J). To investigate this in more detail during the cardiac cycle, strain analyses based on STE were carried out. STE measures not only wall endpoints but also LV wall motion and deformability, providing a highly sensitive and validated quantitative measure of myocardial contractile function [64,77]. LV wall motion, here expressed by velocity vectors, was reduced, as indicated by the shortened length of the vectors, in Ang II-treated cmfKO compared with CTR mice in both systole and diastole (Figure 4K). Moreover, GLS, referring to myocardial shortening and lengthening of the entire LV in a longitudinal direction throughout a cardiac cycle [78], revealed a deterioration of the overall deformation capacity of cmfKO hearts due to the prolonged Ang II infusion (Supplementary Figure S8A). These findings prompted us to further investigate specific regional deformation capacity of the cardiac muscle within six distinct cardiac segments automatically separated by the VevoStrain2100 software (Supplementary Figure S8B). Assessment of the regional longitudinal peak strain, as well as longitudinal peak strain rate, with the latter providing a measure of the deformation per time [79], exhibited a significantly impaired endocardial longitudinal deformation capacity in almost all cardiac regions of Ang II-treated cmfKO hearts (Supplementary Figure S8B-F). Besides other segments especially in the AA region, myocardial deformation was strongly aggravated in the absence of CMF-specific cGKI.

In conclusion, consistent with the underlying morphological and structural changes in response to the Ang II-induced cardiac injury, cmfKO hearts developed a severe loss of cardiac function as reflected by a decrease in %EF and %FS and a regional dysregulation of LV motion.

CMFs are the major source of ECM-proteins such as fibronectin, type I, and type III collagens in the injured heart. While the majority of quiescent CFs are characterized by the expression of transcription factor 21 (TCF21), CMFs exhibit elevated levels of markers such as α-SMA and periostin, which are considered important profibrotic cues for myofibroblast activation [80]. Periostin, induced by Ang II in CF within hours [24,65], represents a secreted matricellular protein enriched in collagen-rich connective tissue contributing to collagen fibrillogenesis and thus ECM re-/organization [81,82]. Accordingly, acquired Postn expression correlates highly with the activated CF phenotype and is postulated to be a common marker of CMFs, irrespective of their origin [13,50]. Although it has long been known that CMFs play a crucial role for the clinical course of HF, the specific analysis and manipulation of this cell type has been difficult due to the lack of specific cell markers [83]. By employing an innovative TAM-inducible Cre-recombinase under control of the Postn-promotor [50], we herein generated a CMF-restricted cGKI KO mouse model and investigated whether previously recognized antifibrotic properties attributed to cGMP/cGKI are due to the in vivo actions of this pathway in Postn⁺ CMFs (Supplementary Figure S3A).

First, we verified the temporal, spatial, and cell type-specific control of the PostniCre recombinase utilizing the ROSA^mT/mG Cre reporter strain [46,59]. We were unable to detect TAM-induced Postn-mediated recombination under physiological conditions, suggesting tight regulation of recombinase activity (Supplementary Figure S1). However, after chronic Ang II exposure and TAM-treatment for five days in vivo (Supplementary Figure S2), the excision of the loxP-flanked DNA sequence encoding mT resulted in mG expression in fibrotic but not in non-fibrotic heart regions. Furthermore, we were able to detect recombination, albeit to a much lesser extent, in the double-transgenic ROSA^mT/mG x PostniCre^tg/+ lungs, which is consistent with reports suggesting that Postn plays a key role in lung fibroblast function [84]. In line with these observations, lineage tracing analyses using an alternative TAM-inducible Postn mouse model based on MerCreMer, a Cre fusion protein containing two modified ER ligand binding domains, revealed Postn expression in less than 1% of interstitial cells in various healthy tissues [13]. Kanisicak et al. (2016) and others, in turn, showed that cardiac damage to different stresses results in an accumulation of Postn⁺ cells, triggering muscle fibrosis [50]. Accordingly, targeted ablation of Postn⁺ CMFs prevented the cardiac remodeling in response to Ang II and after MI [50]. Thus, the herein observed PostniCre-mediated recombination in the lung and in the myocardium of Ang II-treated CTR and cmfKO mice following Ang II exposure confirms numerous early findings. In our hands, cardiac Postn expression was restricted to fibrotic areas [50,81], while the expression of cGKI was efficiently ablated from collagen-rich myocardium in PostniCre-driven cmfKO hearts (Figure 1D). In the absence of Ang II, however, and irrespective of the genotype of the mice, periostin was not detectable in the myocardium (Supplementary Figure S5G). Together, these findings allowed us to conclude that the primary source for cGKI expression in fibrotic lesions of the myocardium is attributable to Postn⁺ cells. Consistent with this post-Ang II expression pattern of myocardial cGKI, cGKI protein levels were significantly lower in mixed Postn⁺ CF/CMF cultures (Supplementary Figure S6B and C) obtained from cmfKO versus CTR hearts after the pathophysiological Ang II stimuli. This finding is consistent with other reports suggesting a functionally relevant role of the cGMP/cGKI axis in both CFs and CMFs [54,63].

Under physiological conditions, i.e., in the absence of Ang II, our in-depth analyses of CTR and cmfKO mice revealed no differences in survival, cardiac morphology, or heart function following TAM-induced PostniCre-recombination (Figures 2 and 4H–K, Supplementary Figure S4A-F, S5E-G, S7 and S8). Conversely, in response to sustained Ang II stimulation, PostniCre activation resulted in worse survival rates, CM hypertrophy, and an exaggeration of myocardial fibrosis in cmfKO hearts (Figures 1E and F and 2, S5A-F). Mice lacking cGKI globally develop hypertension likely due to a disruption of the NO-cGMP-cGKI cascade in VSMCs [85], while the contribution of CFs/CMFs to BP regulation is less clear. We, therefore, measured BP in freely moving cmfKO and CTR mice receiving TAM and Ang II according to the protocols described herein. Mean BP elevations by Ang II were significant as compared with the corresponding mice assessed prior to the Ang II infusion but yielded no gross differences between both genotypes (Figure 1F and G, S4K), suggesting that the higher susceptibility of CMF-specific cGKI KO hearts is due to local, i.e., intra-cardiac effects of the neurohormone Ang II, which promotes, besides its other adverse actions, the pro-fibrotic signaling in CF/CMF [86,87]. We were unable to clarify whether these cardiac effects of Ang II ultimately also explain the higher mortality in the +TAM+Ang II-treated cmfKO group (Figure 1E). However, we can rule out genotype-dependent effects on vascular remodeling processes in the aorta and in connection with renal fibrosis.

We were able to demonstrate a strong antiproliferative function of cGKI in CF as (i) the Ki-67 index in collagen-rich regions of the heart in vivo and (ii) the cell counts in vitro of mixed CFs/CMF cultures were amplified in the absence of the kinase. Thus, we conclude that the presence of cGKI is sufficient to counteract the expansion of Postn⁺ cells under Ang II-stimulated conditions. This increase in cell proliferation rates explains the higher extent of myocardial fibrosis and an accumulation of type I and III collagens in the absence of CMF-specific cGKI in vivo. This is also confirmed by our in vitro analysis of the pro-fibrotic gene expression profile (Figure 4E), although the significance of these experiments is somewhat lower as all transcript quantifications were performed only after 24 h of the 8-Br-cGMP treatment in CF/CMF cultured in the absence of Ang II.

In our study, we used a moderate Ang II infusion protocol. Thus, it is not surprising that no gross signs of cardiac dysfunction, displayed trough unaltered %EF and %FS readings, were observed in the CTR mice. In contrast, both conventional M-mode and STE-based imaging uncovered a severe structure-related decline in global systolic function (%EF and %FS) and GLS, a measure of longitudinally oriented fibers highly prone to wall stress [88,89] in cmfKO mice. In both rat models and patients with HF, in particular, GLS was a reliable measure of LV function and a superior predictor of myocardial fibrosis and adverse cardiac outcomes [90-92]. Accordingly, the overall deformation capacity of the LV, as well as its regional deformation capacities and their rates, all of which are indices of myocardial stiffness and fibrosis, were markedly impaired in cmfKO hearts, which again fits their pro-fibrotic cardiac phenotype.

Larger CM cross-sectional areas, massive increase in fibrosis, and high CMF proliferation rates in cmfKO hearts did not correlate with larger HW or HW/TL ratios compared with CTR. Chronic exposure of Ang II has been shown to induce CM death [93-95], an effect that ultimately results in a loss of myocardial mass [96,97], which we confirmed by a higher count of apoptotic CMs in the absence of cGKI in Postn⁺ cells. These conflicting actions of Ang II on cell growth, i.e., CM hypertrophy and CF/CMF proliferation versus cell death mechanisms apparently triggered a unique remodeling response within the cmfKO myocardium, ultimately resulting in an increase in muscle fibrosis, which could subsequently interfere with CM function and survival. Loss of structural integrity due to CM loss, in turn, also creates mechanical stress that may mediate CF-to-CMF activation. Thus, an urgent question that arises concerns the possible molecular interactions, paracrine and/or other mediators that are produced or released in a cGKI-dependent manner by Postn⁺ CMF to be exchanged with the CMs to influence the function and ‘fate’ of the latter cell type. Two recent studies utilizing a FRET-based biosensor for cGMP live-cell imaging in CMs suggest that cGMP produced in response to NPs or upon stimulation of NO-GC in CFs can enter CMs via gap junctions [53,54]. Interestingly, this transfer has been observed for TCF21⁺ CFs, which are a prominent source for CMFs, i.e., Postn⁺ CMF [13]. Although it is quite conceivable that the transfer of cGMP may contribute to the antifibrotic and antihypertrophic CF/CMF-to-CM cross-talk within the Ang II challenged heart, this hypothesis awaits experimental confirmation. While the extent of cGMP transfer under cardiac stress conditions may be stimulated by NO- and NP-dependent mechanisms, a contribution of cGKI seems to be rather unlikely because (i) the kinase acts exclusively downstream of cGMP production, and (ii) the intercellular transfer of cGKI itself from CF/CMF-to-CM via gap junctions seems neither very feasible nor likely. But how does the communication between myocardial CF/CMF and neighboring CMs works in vivo and how is cGKI involved in the transmission of the signals? Although this is obviously something that future studies must clarify in detail, it is well known that alterations in collagen levels, their disproportionate deposition within the myocardium, and even disbalances of collagen subtypes, promoted by CF/CMF, are directly associated with alterations in CM function. Besides these changes of the biomechanical environment (Figure 2A–C and S5A-D), the communication between both cell types might involve alterations in ionic/electrical coupling properties (transmitted through direct cell-to-cell, i.e., gap junctions) and biochemical factors that are secreted in a paracrine manner from CF/CMF [98,99]. Regarding the latter, our transcript analyses suggest that TGFβ1 and IL-6, two factors that can elicit CF-to-CMF conversion, ECM protein synthesis and CM hypertrophy, do not respond to cGMP/cGKI activation in Postn⁺ CMF. Future studies should therefore consider other paracrine mediators such as fibroblast growth factor-2, ANP or BNP and their modulation upon pharmacological cGMP/cGKI stimulation as well as an in-depth analysis of the possible functions of this pathway in CF/CMF on Ca²⁺ handling mechanisms that ultimately may also affect neighboring CMs.

In summary, the presented results highlight that the antifibrotic actions previously attributed to intrinsic NO/cGMP and NP/cGMP signaling pathways may require cGKI, specifically in Postn⁺ CMFs . Interestingly, all in vivo effects reported herein were seen already in the absence of pharmacologically enhanced cGMP. At least for the Ang II-induced hypertensive heart, we, therefore, conclude that the endogenous response of counterregulatory mechanisms, which in the short term are usually compensatory and maintain cardiac function and perfusion of vital organs, is able to prevent deleterious CMF behaviors through cGKI activation (Figure 5). Nevertheless, it will be very interesting to examine whether anti-fibrotic effects of cGMP-elevating compounds such as vericiguat, BNP, and CNP also act via CMF-specific cGKI to enable positive changes in cardiac structure and/or function. Such approaches will also show how and whether the increase in catalytic activity of cGKI translates to kinase-regulated mechanism(s) downstream that directly counteract the stress response triggered by Ang II in Postn⁺ CMFs.

We herein show that the loss of cGKI in the Postn⁺ subpopulation of CMFs in vivo leads to an exacerbation of the cardiac response to Ang II. These massive functional and structural changes were already observed after the genetic knockout of the kinase. From a clinical perspective, it will be interesting to investigate if an activation of the kinase in Postn⁺ CMFs contributes to the clinical benefit of cGMP-elevating drugs such as vericiguat and neprilysin, which are prescribed to HFrEF patients. As exciting as this hypothesis is, it is far from being experimentally proven and, importantly, we have not yet exposed the Ang II-challenged cmfKO model to a respective treatment regime, i.e., to cGMP-elevating drugs. Also, it is presently not clear to what extent our findings regarding the antifibrotic roles of cGKI in Postn⁺ CMF apply to post-MI hearts or to HF with preserved EF (HFpEF), which is a heterogeneous syndrome and proved to be a difficult entity to treat with many well-established HFrEF medications [100,101]. Yet, fibrosis is a major contributor to the pathogenesis of the ischemic myocardium and to HFpEF. We, therefore, anticipate that a better understanding of the cGMP/cGKI axis in Postn⁺ CMFs will lead to important insight into the interaction between CMF and CM, which may ultimately foster our efforts to positively shape the cardiac microenvironment in the presence of various stress cues.

Requests for resources and reagents can be addressed directly to the Lead contact, Robert Lukowski ([email protected]). All data generated or analyzed during this study that are not included in this published article and its supplementary information are available from the corresponding author on reasonable request.

All other authors have explicitly declared that there are no conflicts of interest related to this article.

M.C.S., L.B., T.P., M.G., and R.L. were supported by GRK2381: "cGMP: From Bedside to Bench", DFG grant number 335549539. F.I. is an associated member of this research consortium. R.L. further acknowledges support by the German Research Foundation grant number LU 1490/13-1.

M.C.S., L.B., and R.L.: Methodology. M.C.S., L.B., A.K., C.K.B., N.L., and L.U.: Investigation. M.C.S., L.B., L.U., and R.L.: Data curation and analysis. L.B., T.P., S.M., K.P., N.L., A.R., N.W., and M.G.: Resources and Tools. F.I., and R.L.: Conceptualization. M.C.S. and R.L.: Writing—original draft. R.L. supported by F.I.: Funding acquisition and supervision.

All experimental procedures were conducted in accordance with the animal protection laws in Germany and were approved by the local Ethics Committee for Animal Research (Regierungspräsidium Tübingen (PZ 02/22 G and PZ 06/23 M)).

We thank Franz Hofmann for providing the floxed cGKI^fl/fl strain, Lucas Matt for helpful discussions as well as Ilona Böhler, Leonie Tuminello, Jule Kolmer, Jule Reiss, and Michael Glaser for excellent technical support.

Ang II

angiotensin II

CF

cardiac fibroblast

CM

cardiomyocyte

CMF

cardiac myofibroblast

CTR

control

CVDs

cardiovascular diseases

HF

heart failure

Postn

periostin

TAM

tamoxifen

VSMCs

vascular smooth muscle cells

cGMP

3′,5′-cyclic guanosine monophosphate

cmfKO

CMF-specific cGKI knockout