Inhibition of Eicosanoid Degradation Mitigates Fibrosis of the Heart

All supporting data, analytical methods, and study materials developed from our group will be made available to other researchers for purposes of reproducing results or replicating the procedures. A detailed description of the methods and materials are provided in the Supplemental Methods in the Supplemental Material section.

Animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus, Aurora, Colorado. Human hearts were obtained from a tissue bank, which was reviewed and approved by the Colorado Multiple Institutional Review Board (COMIRB 01-568), and is maintained by the Division of Cardiology at the University of Colorado Anschutz Medical Campus, Aurora, Colorado. All patients were followed up by the University of Colorado Heart Failure Program and offered participation in the research protocol. Hearts were collected at the time of orthotopic cardiac transplantation.

All data are presented as mean±SEM. GraphPad Prism 8 was used for statistical analyses. Unpaired t test was utilized for comparisons between 2 groups: mixed-effect modeling, 1-way, 2-way, or repeated measures 2-way ANOVA followed by Tukey or Sidak multiple comparisons tests were used to determine the statistical significance among multiple groups. For instances in which unequal variances were detected using a Brown-Forsythe test, data were instead analyzed using a Brown-Forsythe 1-way ANOVA with a Dunnett T3 multiple comparisons test. The Kruskal-Wallis 1-way ANOVA on ranks followed by the 2-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli was utilized for data involving separate animals. As noted limitations, statistical analysis of studies involving in vitro cell work relied upon the central limit theorem, and furthermore, while statistical analysis involved post-hoc Tukey, Sidak, or Dunnett multiple comparisons testing, no additional multiple testing correction was performed. A P value of <0.05 was considered statistically significant.

In order to develop assays to screen chemical libraries for antifibrotic compounds, cell-based high-content–imaging approaches were developed (Figure 1A). Adult rat ventricular fibroblasts (ARVFs), normal human lung fibroblasts, or normal rat kidney fibroblasts were plated on black 96-well plates with optically clear bottoms; flow cytometry analysis confirmed that the primary ARVF preparations consisted of >95% CFs, with minimal endothelial cell or leukocyte contamination (Figure S1A through S1C). Cells were treated with the profibrotic agent, TGF-β₁ (2 ng/mL [normal rat kidney fibroblasts, normal human lung fibroblasts]‚ 10 ng/mL [ARVFs]) for 72 hours in the absence or presence of 546 distinct compounds (Selleckchem Target Selective Inhibitor Library, 10 μM screening dose). Cells were subsequently fixed and stained with antibodies against α-SMA and FN (fibronectin) or collagen type I (Col I)‚ and nuclei were stained with Hoechst 33342; FN was used as the second activation marker for ARVFs and rat kidney fibroblasts‚ while Col I was used as the second marker for lung fibroblasts. High content imaging was used to simultaneously quantify these signals (Figure 1B). Using HCS Studio image analysis software, cells were identified using the nuclear stain, and nuclear counts in each well were used to flag compounds with overt toxicity. Two cellular regions were used to quantify the fluorescence readouts of the above-mentioned fibrotic biomarkers (α-SMA, FN, Col I). One region contained the perinuclear cytoplasmic area (Ring), and the other included this area as well as the nuclear area (Circ).

Percent inhibition values for α-SMA expression induced by TGF-β were derived from normalizing the test compound wells against the –/+TGF-β control wells within each plate (Figure 1C through 1E). Nuclei counts for each well were normalized to the no TGF-β control wells. Compounds exhibiting >60% inhibition of TGF-β induced expression of α-SMA were considered hits if they did not exhibit overt toxicity (Figure 1C through 1E). A compound was considered toxic if the nuclei counts for that well fell below 80% of the average nuclei counts of the no TGF-β control wells.

The number of hit compounds per cell type is indicated in the Venn diagram, with only 9 of 546 compounds sharing a common ability to block TGF-β–induced activation of cardiac, pulmonary, and renal fibroblasts (Figure 1F); all screening data are provided (Excel File 1 in the Supplemental Material). Five of the compounds are kinase inhibitors (BI-D1870, GSK429286A, PLX7904, Sotrastaurin, and Tyrphostin AG-879), 2 inhibit epigenetic regulatory proteins (PF-CBP1 and PTC-209), 1 inhibits telomerase (BIBR1532), and 1 targets 15-PGDH (SW033291) (Figure 1G). 15-PGDH inhibition has garnered attention as a therapeutic approach to promote tissue regeneration,¹⁷ and SW033291 has recently been shown to inhibit lung fibrosis in preclinical models.^18,19 However, because the impact of SW033291 on fibroblast activation has not been previously addressed, this compound was chosen for follow-up investigation. SW033291 inhibits 15-PGDH-mediated oxidation and subsequent degradation of prostaglandins (Figure 1H) such as prostaglandin E2 (PGE2) and prostaglandin F2 alpha (PGF2α), as well as other eicosanoids.²⁰

To facilitate a detailed evaluation of SW033291, subsequent studies focused on CFs and cardiac fibrosis. SW033291 dose-dependently suppressed α-SMA and FN protein expression in ARVFs in the 96-well assay, with low micromolar half maximal effective concentrations, and was only cytotoxic at 30 μM, as determined by reduced nuclei count (Figure 2A). By dividing the effective concentration for inhibition of α-SMA expression by the approximate effective concentration for toxicity (29 μM), we estimate that SW033291 has a therapeutic index of 6.2 in this assay. Importantly, SW033291 did not block TGF-β–mediated nuclear import of SMAD2/3 transcription factors, suggesting that the compound does not directly block TGF-β receptor signaling, and instead functions through a distinct, downstream mechanism to suppress CF activation; the ALK5 TGF-β receptor inhibitor, SB525334, served as a positive control (Figure S2A). SW033291 also blocked α-SMA protein expression, in a concentration-dependent manner, in TGF-β–stimulated normal human CFs (Figure S2B), illustrating the translational significance of the findings. Although 10 μM SW033291 was well-tolerated by CFs (Figure 2B), follow-up studies were performed with a 2.5 μM final concentration of the compound, which effectively suppressed TGF-β–induced Acta2 (α-SMA) and Postn (periostin) mRNA expression in ARVFs and normal human CFs (Figure 2C and 2D). SW033291 also reversed activation of ARVFs that were pre-stimulated with TGF-β for 3 days, highlighting the potential of 15-PGDH inhibition to ameliorate pre-existing fibrosis (Figure 2E).

To address the contribution of off-target actions of SW033291, lentiviruses encoding short hairpin RNAs (shRNAs) that effectively knocked down endogenous 15-PGDH expression in ARVFs were used (Figure 2F). One of these shRNAs blocked TGF-β–mediated induction of α-SMA and periostin mRNA expression in ARVFs, while effects of the other failed to reach statistical significance (Figure 2G). SW033291 enhanced secretion of the 15-PGDH substrate, PGE2, in TGF-β–treated CFs similarly to 15-PGDH knockdown with shRNAs, suggesting engagement of the target enzyme by the compound (Figure 2H). While off-target actions of SW033291 in ARVFs cannot be ruled out, the data suggest that the compound blocks cardiac fibroblast activation, at least in part, by suppressing the activity of 15-PGDH.

To begin to address the therapeutic potential of 15-PGDH inhibition for cardiac fibrosis, studies were performed with fibroblasts isolated from explanted hearts from heart failure patients undergoing transplantation (Table S1). Consistent with prior findings, compared with cells from nonfailing controls, cultured CFs from failing human hearts maintained an activated state marked by high level α-SMA expression within stress fibers and, remarkably, treatment of the cells with SW033291 dramatically reversed this activated state (Figure 3A). These findings were corroborated with results from gene expression analyses using RNA from cultured failing human CFs from 3 independent patients (Figure 3B). Of note, CFs from patients 1 and 2 could not be further activated by TGF-β, while those from patient 3 were still responsive to the stimulus, likely reflecting a higher level of constitutive activation of the former fibroblasts.

When cultured on artificial collagen networks, activated fibroblasts possess the ability to contract these gels through an α-SMA-dependent mechanism. To further evaluate effects of 15-PGDH inhibition on fibroblast activation, failing or nonfailing human CFs were seeded on collagen gels and treated with SW033291 or vehicle control. Consistent with the α-SMA expression data, failing human CFs had significantly higher basal contractile activity than nonfailing CFs, and SW033291 significantly reduced failing CF-induced collagen gel contraction over a 5-day period (Figure 3C). Additionally, parallel studies of normal human CFs and ARVFs demonstrated that SW033291 suppressed TGF-β–induced collagen gel contraction (Figure 3D and 3E). These findings provide further evidence for a reduction in CF activation status upon 15-PGDH inhibition.

TGF-β stimulation of ARVFs led to enhanced proliferation, as evidenced by increased incorporation of EDU into DNA, which was blocked by SW033291 (Figure S3A). However, further evaluation of the data indicated a more complex impact of the 15-PGDH inhibitor, with SW033291 suppressing proliferation of α-SMA-positive activated CFs, while enhancing proliferation of α-SMA-negative CFs (Figure S3B).

A model of robust interstitial cardiac fibrosis driven by Ang II infusion in mice was used to assess efficacy of 15-PGDH inhibition in vivo. One day after implantation of Ang II osmotic minipumps, mice were dosed twice daily with SW033291 (5 mg/kg, IP) for 13 additional days prior to animal sacrifice (Figure 4A). SW033291 treatment did not alter animal body weight (Figure 4B), indicating that the compound was generally well-tolerated in vivo, consistent with prior findings using the same dosing paradigm.¹⁷ Left ventricular (LV) levels of PGE2 were quantified as a pharmacodynamic readout of 15-PGDH inhibition, and indicated that SW033291 treatment significantly increased the level of this eicosanoid in the LV compared with Ang II-infused mice administered vehicle control (Figure 4C). Ang II infusion resulted in increased 15-PGDH transcript expression in the LV, which was modestly reduced by SW033291 (Figure 4D). Picrosirius red dye staining of LV sections revealed that the compound effectively suppressed Ang II-mediated cardiac fibrosis (Figure 4E and 4F; Figure S4A); this held true regardless of whether the Picrosirius red stain was imaged using standard bright-field microscopy (Figure 4E, upper) or alternatively by evaluating the birefringence of the Picrosirius red signal using polarized light microscopy (Figure 4E, lower). SW033291 reduced Ang II-mediated stimulation of α-SMA mRNA and protein expression, as determined by qRT-PCR and immunofluorescence staining of LVs, respectively (Figure 4G; Figure S4B and S4C). Most α-SMA+ cells in LVs of Ang II-treated mice co-expressed the fibroblast marker, PDGFRα, and SW033291 treatment diminished both signals (Figure S4D and S4E), suggesting that 15-PGDH inhibition blocks both CF activation and expansion in response to pathological stress. Consistent with this, SW033291 administration also dampened Ang II-mediated expression of vimentin, which is commonly used to assess CF abundance in the heart (Figure S4F). Additionally, SW033291 blunted Ang II-dependent periostin protein deposition (Figure 4H and 4I) and Col1a1 (collagen type Iα1) mRNA expression (Figure 4J), and reduced the number of CD45⁺ leukocytes and Ccl2 transcript expression in the heart (Figure S4G and S4H). SW033291 treatment did not alter Ang II-mediated increases in LV mass or cardiomyocyte hypertrophy (Figure S5A through S5C), but did reduce expression of the mRNA encoding atrial natriuretic factor, which is a marker of pathological cardiac hypertrophy; the reduction in β-myosin heavy chain mRNA expression in mice treated with SW033291 was more modest (Figure S5D). Treatment of cultured neonatal rat ventricular myocytes with SW033291 failed to dramatically impact phenylephrine-induced expression of these markers (Figure S5E), suggesting that the observed reduction of the mRNAs in vivo was secondary to inhibition of 15-PGDH in a nonmyocyte cell type(s). Together, these data suggest that the primary mechanism by which SW033291 mitigated Ang II-induced cardiac fibrosis was by blocking CF activation and expansion, and possibly also through anti-inflammatory actions.

Previously, we demonstrated that mice subjected to uninephrectomy and implanted with pellets that release deoxycorticosterone acetate (DOCA) develop mild diastolic dysfunction with preserved ejection fraction, which is associated with low-level cardiac fibrosis and normal myocyte relaxation.²¹ To determine whether 15-PGDH inhibition is capable of reversing established diastolic dysfunction, mice were subjected to uninephrectomy/DOCA or uninephrectomy alone (sham control) and, after 4 weeks, 1 cohort of uninephrectomy/DOCA mice was treated with SW033291 (Figure 4K). Serial echocardiography confirmed that mice had reduced E/A and E′/A′ ratios 4 weeks after uninephrectomy/DOCA, which is indicative of diastolic dysfunction, and 4 weeks of SW033291 treatment led to a marked recovery of these measures of diastolic function (Figure 4L and 4M; Figure S6A and S6B; Table S3). In a cohort of mice subjected to uninephrectomy alone and analyzed after 8 weeks, SW033291 had no effect on E’/A’ or E/A compared with vehicle-treated controls (Figure S6C). Ejection fraction was preserved throughout the 8-week duration of the study (Figure S6D), and no changes in global longitudinal strain or heart rate were noted (Table S3). LV mass, but not cardiomyocyte hypertrophy, was modestly reduced by SW033291 treatment in the uninephrectomy/DOCA model (Figure S6E through S6G). However, consistent with findings in the Ang II infusion model, 15-PGDH inhibition blunted uninephrectomy/DOCA-induced expression of markers of CF activation and cardiac fibrosis (Figure S6H through S6N); we note that the impact of SW033291 on collagen 1a1 and 3a1 mRNA expression was more modest than observed with the other markers. Together, the data suggest that SW033291 ameliorated diastolic dysfunction, at least in part, through suppression of cardiac fibrosis in uninephrectomy/DOCA mice.

To address the mechanism(s) by which 15-PGDH inhibition blocks CF activation, transcriptomic profiling by RNA sequencing was performed using RNA from cultured ARVFs treated with TGF-β and SW033291, alone or in combination, for 48 hours (n=4 for each condition). Using a fold expression change >1.5× and adjusted P value of <0.05 as cutoffs, differential expression analysis showed that 146 genes upregulated by TGF-β were suppressed by SW033291, and 170 genes downregulated by TGF-β were augmented by SW033291, as shown in Figure 5A (see Excel File 2 for full differential expression results). Principal component analysis using the same set of genes shows clear segregations by treatment group (Figure 5B). Ingenuity Pathway Analysis of TGF-β–induced transcripts that were suppressed by SW033291 showed a strong enrichment for fibrosis and inflammatory signaling (Figure S7A and S7B). Conversely, genes downregulated by TGF-β and rescued by SW033291 were predominately associated with cell proliferation (Figure S7C). Further analysis, focusing on predicted upstream regulators that were inhibited or activated upon treatment of ARVFs with SW033291, strongly suggested that the 15-PGDH inhibitor was stimulating ERK1/2 (extracellular signal-regulated kinase 1/2) mitogen-activated protein kinase signaling in the CFs (Figure 5C). Indeed, multiple target genes that are known to be induced by ERK1/2 signaling were upregulated in SW033291-treated ARVFs (Figure 5D). Consistent with this, follow-up immunoblotting and indirect immunofluorescence studies demonstrated that ERK1/2 phosphorylation, which is a surrogate for activation, was downregulated in ARVFs treated with TGF-β for 48 hours, and rescued by SW033291 co-administration (Figure 5E through 5G). Evaluation of a subset genes that were differentially expressed in ARVFs confirmed that 15-PGDH inhibition also stimulated cardiac ERK1/2 target gene expression in mice infused with Ang II (Figure 5H).

ERK1/2 is activated upon phosphorylation by mitogen-activated protein kinase kinase enzymes. When ARVFs were treated with the mitogen-activated protein kinase kinase inhibitor, PD98059, in combination with TGF-β and SW033291, the ability of the 15-PGDH inhibitor to block α-SMA and periostin mRNA expression was lost (Figure 6A). Conversely, inhibition of JNK or p38 kinase, which control other arms of the mitogen-activated protein kinase cascade, had no impact on SW033291-mediated inhibition of ARVF activation (Figure 6A and 6B). PD98059 treatment also nullified SW033291-mediated suppression of TGF-β–induced ARVF α-SMA protein expression and stress fiber formation (Figure S8), as well as collagen gel contraction (Figure 6C). A second mitogen-activated protein kinase kinase inhibitor, U0126, reversed SW033291-mediated blockade of normal human CF activation (Figure 6D through 6G). These findings demonstrate that ERK1/2 signaling is required for SW033291 to inhibit CF activation.

Therapeutic effects of SW033291 in models of tissue regeneration and lung fibrosis have been attributed to elevated levels of PGE2 upon 15-PGDH inhibition.^17–19 Thus, we hypothesized that exogenous PGE2 would recapitulate the suppressive effects of SW033291 on CF activation. Surprisingly, however, PGE2 failed to block ARVF activation in the high-content–imaging assay, and only reduced TGF-β–induced α-SMA protein expression at the highest concentration tested, which was also cytotoxic (Figure 7A). Likewise, agonists of EP1, EP2, EP3, and EP4 receptors, which are G protein-coupled receptors through which PGE2 signals, also failed to block ARVF activation (Figure S9), and antagonists of these receptors had no effect on SW033291-mediated suppression on ARVF activation (Figure S10).

Next, we tested other exogenously added eicosanoids for their ability to block ARVF activation, and only 12(S)-HETE was efficacious (Figure 7B; Figure S11). Follow-up studies confirmed that 12(S)-HETE treatment blocked TGF-β–mediated induction of α-SMA and periostin mRNA expression in ARVFs (Figure 7C) and stimulated CF ERK1/2 signaling (Figure 7D and 7E). SW033291 increased secretion of 12(S)-HETE from the CFs (Figure 7F) and increased abundance of this eicosanoid in LVs of mice infused with Ang II (Figure 7G).

12(S)-HETE is known to stimulate the G protein-coupled receptor, GPR31.²² To address the possibility that signaling via GPR31 blocks CF activation, ARVFs were transfected with small interfering RNA to knockdown expression of endogenous GPR31 (siGPR31) or scrambled siRNA (siControl) and stimulated with TGF-β in the absence or presence of exogenous 12(S)-HETE. GPR31 knockdown markedly reduced the ability of 12(S)-HETE to suppress ARVF activation, as determined by evaluation of α-SMA expression (Figure 7H), in a manner that correlated with reduced expression of the ERK1/2 target gene, EGR1 (Figure 7I). Quantitative RT-PCR confirmed efficient knockdown of GPR31 expression by the siRNA, and also revealed a dramatic increase in GPR31 expression in ARVFs transfected with siControl and treated with 12(S)-HETE, suggestive of a positive feedback loop (Figure 7J).

Experiments were performed to address the translational significance of 12(S)-HETE signaling in the heart. 12(S)-HETE blocked TGF-β–mediated activation of normal human CFs in a manner that was dependent on ERK1/2 signaling, as evidenced by altered α-SMA protein expression and collagen gel contraction (Figure 8A through 8C). Furthermore, 12(S)-HETE profoundly reduced α-SMA expression in constitutively activated CFs from a patient with heart failure, similarly to SW033291 (Figure 8D and 8F).

The initial step in producing eicosanoids, which are 20 carbon-containing signaling molecules, involves cyclooxygenase-2 (COX-2)-mediated conversion of arachidonic acid into prostaglandin H2, which is subsequently transformed into a variety of eicosanoids, including PGE2 and 12(S)-HETE.²³ Countering the action of COX-2 is 15-PGDH, a nicotinamide adenine dinucleotide-dependent enzyme that promotes eicosanoid degradation through oxidation of a hydroxyl group on the signaling mediators.²⁰ Nonsteroidal anti-inflammatory drugs selectively targeting COX-2 have been shown to significantly increase the risk of developing cardiovascular complications, including heart failure,²⁴ and cardioprotective actions of COX-2 have been demonstrated in preclinical models.²⁵ Because 15-PGDH inhibition is opposite from COX-2 inhibition, causing an increase as opposed to a decrease in eicosanoid abundance, it is intriguing to speculate that compounds such as SW033291 will be cardioprotective. Consistent with this notion, we found that mice thrive when treated with SW033291 and exhibit less cardiac fibrosis in response to chronic Ang II infusion. Furthermore, even though COX-2-derived eicosanoids are known to promote inflammation, we observed decreased numbers of leukocytes in the hearts of mice treated with SW033291. Whether 15-PGDH inhibition elicits salutary effects in the heart beyond suppression of fibrotic remodeling and inflammation, such as through altering cardiomyocyte contractility, remains to be determined.

Prior to the current report, the role of 15-PGDH in cardiac homeostasis and pathogenesis had not previously been addressed. However, a rapidly emerging body of literature based on studies in noncardiac systems suggests widespread therapeutic potential for 15-PGDH inhibition. SW033291 was originally discovered in a high-throughput screen to identify small molecule regulators of 15-PGDH enzymatic activity, and was shown to enhance hematopoietic recovery after bone marrow transplantation and stimulate tissue regeneration in colon and liver injury in mouse models.¹⁷ Subsequent studies demonstrated that SW033291 promotes splenic niche hematopoietic regeneration, bone regeneration, and skeletal muscle regeneration.^26–29 Other observed beneficial effects of SW033291 include suppression of pulmonary fibrosis and protection from renal injury.^{18,19,30–32} Our findings establish a previously unrecognized cardioprotective action of SW033291, and further highlight the potential of 15-PGDH inhibitors to block fibrosis across organ systems. It is important to note that tissue regeneration has been linked to suppression of profibrotic pathways in multiple organs,^33–36 and thus the ability of SW033291 to block fibrosis could be due, in part, to stimulation of pro-regenerative signaling networks.

In all of the aforementioned noncardiac studies, beneficial effects of SW033291 treatment were ascribed to increased levels of PGE2 upon 15-PGDH inhibition. Cardioprotective effects of PGE2 signaling via EP receptors have been well-documented,³⁷ and we noted increased PGE2 levels in hearts of mice treated with SW033291, suggesting that this eicosanoid may contribute to suppression of Ang II-mediated cardiac fibrosis by the 15-PGDH inhibitor in vivo. However, our studies with cultured murine and human CFs failed to reveal an inhibitory effect of PGE2 signaling on activation of the cells. Specifically, neither ectopic PGE2 nor agonists of EP1, EP2, EP3, or EP4 receptors were able to block TGF-β–mediated CF activation. Furthermore, antagonists of these receptors failed to block the ability of SW033291 to suppress CF activation. A survey of other exogenously added eicosanoids revealed one, 12(S)-HETE, that was able to efficiently block TGF-β–induced activation of cultured murine or human CFs. 12(S)-HETE signals through GPR31,²² and knocking down expression of this G protein-coupled receptor blocked the ability of 12(S)-HETE to suppress CF activation. While 12(S)-HETE has not been reported to be a substrate for 15-PGDH,²⁰ we noted significantly increased levels of this eicosanoid in ARVFs treated with SW033291, as well as in hearts of mice treated with the 15-PGDH inhibitor. It is not known whether elevated 12(S)-HETE abundance upon SW033291 treatment is a direct or indirect consequence of 15-PGDH inhibition. Notwithstanding this uncertainty, the findings highlight the potential of developing agents that promote GPR31 signaling in CFs as antifibrotic therapies.

RNA sequencing analysis suggested that increased ERK1/2 signaling after 15-PGDH inhibition may contribute to suppression of CF activation, and this possibility was experimentally corroborated by the demonstration that mitogen-activated protein kinase kinase inhibitors blocked the ability of SW033291 to suppress murine and human CF activation in response to TGF-β. These findings initially seemed paradoxical. Indeed, mitogen-activated protein kinase kinase inhibitors have been shown to block IL-11–mediated CF activation in culture and reduce cardiac fibrosis in vivo in a mouse model of Marfan syndrome subjected to pressure overload.^38,39 However, several other studies have demonstrated that ERK1/2 signaling inhibits conversion of fibroblasts into an activated state marked by expression of α-SMA,^40–43 and a more recent report defined a mechanism for ERK1/2 signaling, downstream of fibroblast growth factor 2, in mediating dedifferentiation of lung myofibroblasts.⁴⁴ ERK1/2-mediated dedifferentiation of lung myofibroblasts was associated with enhanced proliferation of the cells. In this regard, we noted a pro-proliferative gene signature in ARVFs treated with SW033291 and, while the 15-PGDH inhibitor blocked proliferation of the bulk of ARVFs, a subset of the cells (α-SMA-negative) actually exhibited enhanced proliferation. As such, it is possible that SW033291 suppresses CF activation, at least in part, by triggering a dedifferentiation program in the cells. Deletion of ERK1/2 in cardiomyocytes results in impaired cardiac function and early lethality in response to aging or pressure overload, whereas constitutive activation is cardioprotective.^45,46 Based on our findings, we posit that similar detrimental and favorable effects would be observed upon loss- and gain-of-function of ERK signaling in CFs, respectively.

Remarkably, SW033291 and 12(S)-HETE were able to not only prevent TGF-β–induced activation of normal human CFs, but also to reverse persistent activation of CFs derived from humans with heart failure. Constitutive activation of failing heart-derived CFs has been previously described,⁴⁷ and may represent a consequence of “epigenetic imprinting” that enables the cells to maintain a stimulated state despite being removed from the pathogenic environment. Our results indicate that SW033291 and 12(S)-HETE treatment are sufficient to rewire the transcriptional program in these cells to achieve a cell state transition that dampens ECM production, and thus suggest the possibility of achieving therapeutic effects clinically, wherein 15-PGDH inhibition may ameliorate preexisting fibrosis. Prior positive results with SW033291 formed the foundation for developing next-generation 15-PGDH inhibitors with enhanced drug-like properties in the biotechnology sector, highlighting the perceived promise of translating preclinical findings related to this pathway to the treatment of human diseases.⁴⁸ Given the data presented here, further evaluation of potential cardioprotective effects of 15-PGDH inhibition is warranted.

A recent study described the discovery of salinomycin as a hit in a screen of natural/botanical compounds and FDA-approved drugs for the ability to suppress expression of an α-SMA promoter-reporter construct in NIH-3T3 fibroblasts.⁴⁹ Salinomycin, an anti-bacterial and anti-parasitic ionophore with ill-defined molecular targets, was subsequently shown to have a remarkable ability to block CF activation and elicit multiple salutary effects in mouse models of pathological cardiac remodeling. More advanced phenotypic assays using normal human CF cultures and a machine learning algorithm,⁵⁰ or mixtures of human induced pluripotent stem cell-derived cardiomyocytes and nonmyocytes,⁵¹ have been developed and validated with known antifibrotic compounds. Further evaluation of the hit compounds described here using these assays could reveal new insights into molecular mechanisms of CF activation and the potential utility of the compounds for the treatment of cardiac fibrosis.

Because SW033291 was delivered systemically in vivo, we cannot rule out the possibility that 15-PGDH inhibition in cell types other than CFs, such as macrophages, contributes to suppression of cardiac fibrosis by the compound. Future evaluation of mice with conditional deletion of the gene encoding 15-PGDH in fibroblasts will help elucidate whether this enzyme serves CF-autonomous roles in the control of cardiac fibrosis. Furthermore, a comprehensive hemodynamic evaluation of the impact of SW033291 treatment on diastolic function in alternative settings, such as the more severe two-hit high fat diet plus L-NAME mouse model of HFpEF,⁵² is needed to fully address the potential utility of 15-PGDH inhibition as a therapeutic strategy to improve cardiac relaxation.

A phenotypic screening/chemical biology approach uncovered eicosanoid degradation in fibroblasts as a new therapeutic target for the treatment of cardiac fibrosis. Small molecule inhibitors of 15-PGDH have the potential to block pathological fibrosis across organ systems by triggering protective autocrine/paracrine signaling that suppresses fibroblast activation and culminates in reduced ECM production.

Expanded Materials and Methods

Figures S1–S12

Tables S1–S3

Excel Files 1–2

References 53–58