Therapeutic effects of a small molecule agonist of the relaxin receptor ML290 in liver fibrosis

Abstract

Fibrosis is an underlying cause of cirrhosis and hepatic failure resulting in end stage liver disease with limited pharmacological options. The beneficial effects of relaxin peptide treatment were demonstrated in clinically relevant animal models of liver fibrosis. However, the use of relaxin is problematic because of a short half-life. The aim of this study was to test the therapeutic effects of recently identified small molecule agonists of the human relaxin receptor, relaxin family peptide receptor 1 (RXFP1). The lead compound of this series, ML290, was selected based on its effects on the expression of fibrosis-related genes in primary human stellate cells. RNA sequencing analysis of TGF-β1–activated LX-2 cells showed that ML290 treatment primarily affected extracellular matrix remodeling and cytokine signaling, with expression profiles indicating an antifibrotic effect of ML290. ML290 treatment in human liver organoids with LPS-induced fibrotic phenotype resulted in a significant reduction of type I collagen. The pharmacokinetics of ML290 in mice demonstrated its high stability in vivo, as evidenced by the sustained concentrations of compound in the liver. In mice expressing human RXFP1 gene treated with carbon tetrachloride, ML290 significantly reduced collagen content, α-smooth muscle actin expression, and cell proliferation around portal ducts. In conclusion, ML290 demonstrated antifibrotic effects in liver fibrosis.—Kaftanovskaya, E. M., Ng, H. H., Soula, M., Rivas, B., Myhr, C., Ho, B. A., Cervantes, B. A., Shupe, T. D., Devarasetty, M., Hu, X., Xu, X., Patnaik, S., Wilson, K. J., Barnaeva, E., Ferrer, M., Southall, N. T., Marugan, J. J., Bishop, C. E., Agoulnik, I. U., Agoulnik, A. I. Therapeutic effects of a small molecule agonist of the relaxin receptor ML290 in liver fibrosis.

Chronic liver fibrosis plays a crucial role in the progression of most liver diseases, including chemical intoxication, viral infection, alcoholic liver disease, and nonalcoholic steatohepatitis. Liver fibrosis is accompanied by changes in the extracellular matrix (ECM) with abnormal deposition of collagen, leading to the formation of permanent fibrous scar, inflammation, and subsequent degeneration of the liver tissue structure. Hepatic stellate cells (HSCs) and Kupffer cells are important mediators in the development of liver fibrosis (1). In normal liver, HSCs are quiescent; they function as a major storage location for vitamin A and are responsible for normal turnover of the basement membrane-like ECM in the perisinusoidal space of Disse (2). In damaged and diseased liver, the HSCs become activated and differentiate into myofibroblast-like cells (3). This process is accompanied by high rates of cell proliferation and activation of fibrotic genes. HSC activation is controlled by cytokines and growth factors released by Kupffer cells, hepatocytes, and HSCs themselves, with the TGFs playing critical roles. Activated HSCs propagate the inflammatory response and the remodeling of the liver ECM. When liver fibrosis progresses unchecked, it often leads to organ failure or cancer.

There is an increased need to develop more effective treatments to reverse liver fibrosis. One such agent is the relaxin peptide, which can potentially act as a multitarget antifibrotic drug (4). It has been demonstrated that relaxin, which activates downstream signaling pathways through its G protein-coupled receptor, relaxin family peptide receptor 1 (RXFP1), has therapeutic effects in kidney, liver, heart, and other organs fibrosis (4, 5). Experiments conducted in rodent models of liver fibrosis such as carbon tetrachloride (CCl₄) induced; thioacetamide induced, nonalcoholic fatty liver disease, or bile duct ligation, have established that relaxin treatment resulted in decreased collagen accumulation, reduced HSC activation, and ameliorated portal hypertension (6–10). These studies also showed that relaxin treatment decreased TGF-β1 signaling, induced activation of several ECM degrading matrix metalloproteinases (MMPs), and promoted angiogenesis. On the other hand, ablation of the relaxin gene in mice led to fibrosis in various organs, and this effect was reversed by relaxin treatment (8, 11–17). The safety of relaxin treatment in patients has been demonstrated in several clinical trials (18). A major problem with the application of such therapy to chronic diseases is the low stability of the recombinant relaxin in vivo, requiring continuous drug delivery. Furthermore, a potential understudied problem is the immunologic response to the recombinant peptide, especially in the case of human recombinant peptide used in rodent studies. To overcome these challenges, we have recently identified potent and efficacious small molecule agonists of the human relaxin receptor RXFP1 (19–22). These molecules are specific for the human relaxin receptor and have favorable absorption, distribution, metabolism, and excretion properties (20, 23). They utilize allosteric sites on the receptor and thus do not interfere with natural hormone function (24). Here, we showed that the lead compound, ML290, has antifibrotic properties in human HSCs and liver organoids as well as in a CCl₄-induced liver fibrosis mouse model expressing human RXFP1 receptor.

MATERIALS AND METHODS

Cell culture experiments

The spontaneously immortalized human HSC cell line LX-2 (provided by Dr. Robert G Bennett, University of Nebraska, with the permission of Dr. Scott Friedman, Icahn School of Medicine at Mount Sinai, New York, NY, USA) (25) was authenticated by American Type Culture Collection (Manassas, VA, USA) using short tandem repeat analysis and matched to the published short tandem repeat loci of LX-2 (26). Cells were seeded in 6-well plates in DMEM (Corning, Corning, NY, USA) supplemented with 2% fetal bovine serum (Atlanta Biologicals, Flowery Branch, GA, USA). After overnight attachment, cells were treated with recombinant human TGF-β1 (2.5 ng/ml; MilliporeSigma, Burlington, MA, USA) to induce an activated HSC phenotype in the presence of either DMSO or 5 µM ML290 dissolved in DMSO for 72 h, after which RNA was extracted from cell lysates. To examine cytotoxicity of ML290, LX-2 cells were seeded in media containing 2% fetal bovine serum onto 96-well plates. Cell viability was assessed by measuring ATP content using the CellTiter-Glo Luminescent Assay (Promega, Madison, WI, USA) after 24 and 48 h incubation with 11 concentrations of ML290 (from 1 nM to 100 µM). Primary human HSCs (Zen-Bio, Research Triangle Park, NC, USA) were cultured in 35 mm poly-l-lysine coated dishes in hepatic stellate growth medium (Zen-Bio) supplemented with 3% fetal bovine serum. These cells exhibit activated phenotype after culture on plastic dishes (27). Cells were incubated with DMSO or 1 and 5 µM of the compounds in ML290 series (20) for 72 h, after which the cells were lysed for RNA collection. These concentrations were about 1–10 times higher than EC₅₀ of the compounds in cAMP assay in various cell lines (28–30). The following 5 small molecule compounds were tested: ML290 (PubChem SID: 134225125), 6 (134225094), 9 (134225114), 10 (134225112), and 11 (134225113).

RNA sequencing

RNA (RIN 9.9–10, determined by Agilent 2100 Bioanalyzer; Santa Clara, CA, USA) from LX-2 cells treated with TGF-β1 + DMSO (n = 4) and TGF-β1 + ML290 (n = 4) were used to construct libraries using the Illumina HiSeq platform PE150 at Novogene (Sacramento, CA, USA). Sequencing data were deposited in the Gene Expression Omnibus (GSE 122710). Bioinformatics analysis was performed on Partek Flow (St. Louis, MO, USA). FASTQ files containing pair-end sequence reads were mapped to the human reference genome GRCh38. Gene set analysis after alignment with the Spliced Transcripts Alignment to a Reference (STAR) aligner (https://research.csc.fi/-/star-aligner) generated gene counts, which were then ranked by false discovery rate (<0.01) with fold-change cutoff of 2. Differentially expressed genes filtered by the criteria listed above were matched to databases to perform gene ontology and pathway analysis.

Real-time quantitative PCR

Total RNA was isolated from cells or mouse liver by column purification using a GeneJet RNA Purification Kit (Thermo Fisher Scientific, Waltham, MA, USA) following the manufacturer’s protocol. First strand cDNA synthesis was done with the Verso cDNA Synthesis Kit (Thermo Fisher Scientific). Gene expression was assessed by quantitative PCR using the 2^−ΔΔCt method with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the housekeeping gene and the Go Taq Q-PCR master mix (Promega) in a Mastercycler RealPlex² System (Eppendorf, Westbury, NY, USA). Primer sequences are shown in Table 1.

TABLE 1.

Primers used for quantitative RT-PCR

	Sequence, 5′–3′
Gene	Forward	Reverse
Human
ACTA2	AAGCACAGAGCAAAAGAGGAAT	ATGTCGTCCCAGTTGGTGAT
CAV1	GTAGACTCGGAGGGACATC	CACTTGCTTCTCGCTCAG
COL1A1	AGACAGTGATTGAATACAAAACCA	GGAGTTTACAGGAAGCAGACA
COL5A1	GATGTCGCTTACAGAGT	TTCACAGTTGTTAGGATGG
GAPDH	AGCCACATCGCTCAGACAC	GCCCAATACGACCAAATCC
MMP1	GCACTGAGAAAGAAGACAAAGG	CTAAGTCCACATCTTGCTCTTG
PPARGC1A	TGAGAGGGCCAAGCAAAG	ATAAATCACACGGCGCTCTT
RXFP1	TGACATCTGGTTCTGTCTTCTTCT	CAGTCGTCCACACCGTTACA
TGFB1	CACTCCCACTCCCTCTCTC	GTCCCCTGTGCCTTGATG
TIMP1	GCAATTCCGACCTCGTCATC	TCTTGATCTCATAACGCTGGTATAA
TIMP3	GCTGGAGGTCAACAAGTACCA	CACAGCCCCGTGTACATCT
Mouse
Acta2	ACCACCCACCCAGAGTG	GTCTTCCTCTTCACACATAGC
Cav1	TAAATCACAGCCCAGGGAAA	GACCACGTCGTCGTTGAGAT
Col1a1	CCTCAGGGTATTGCTGGACAAC	ACCACTTGATCCAGAAGGACCTT
Ctgf	TGACCTGGAGGAAAACATTAAGA	AGCCCTGTATGTCTTCACACTG
Des	CGGCTAAGAACATCTCTGAGG	TAGGACTGGATCTGGTGTCG
Gapdh	AACGACCCCTTCATTGAC	TCCACGACATACTCAGCAC
Mmp13	GCCAGAACTTCCCAACCAT	TCAGAGCCCAGAATTTTCTCC
Mmp2	AACTTTGAGAAGGATGGCAAGT	TGCCACCCATGGTAAACAA
Pparg	GCTGTCATTATTCTCAGTGGAGAC	GAACAGCTGAGAGGACTCTGG
Rxfp1	CTGTGCTGGATTCCCATCTT	GTTGTGCCAGAGTTGATGGA
Tgfb1	TGGAGCAACATGTGGAACTC	CAGCAGCCGGTTACCAAG
Timp1	GCAAAGAGCTTTCTCAAAGACC	AGGGATAGATAAACAGGGAAACACT
Timp2	CGTTTTGCAATGCAGACGTA	GGAATCCACCTCCTTCTCG
Vim	CCAACCTTTTCTTCCCTGAA	TGAGTGGGTGTCAACCAGAG

Organoid studies

Human liver organoids were produced as previously described in refs. 31, 32. To examine the effects of ML290 in preventing collagen accumulation, organoids were cotreated with 10 µg/ml LPS (LPS from Escherichia coli O111:B4; MilliporeSigma) (32, 33) with various concentrations of ML290 (from 1 nM to 10 µM) or DMSO (control). LPS concentration was selected to be effective to induce collagen deposition while showing minimal to no toxicity (unpublished results). After 4 d, organoids were harvested and processed for whole-mount staining with rabbit anti-collagen 1 antibody (1:150, ab34710) followed by goat anti-rabbit Alexa Fluor 647 (1:300, ab150079; Abcam, Cambridge, MA, USA) secondary antibody and DAPI (Thermo Fisher Scientific) to stain the nuclei. Macroconfocal organoid images were then taken, and red and blue pixel counts were determined using in-house designed software (RGBQuant, see Supplemental Data). A fibrotic index was determined by calculating the ratio of red (collagen) to blue (total cells) pixels.

In the second set of experiments, organoids were treated with LPS for 7 d to induce fibrosis, followed by DMSO (control) or 3 concentrations of ML290 (5, 50, or 500 nM) for an additional 7 d in the absence of LPS. Culture medium was refreshed every 72 h. After a total of 14 d in culture, organoids were processed for whole-mount immunofluorescence and simultaneously stained with rabbit anti-collagen 1 (1:150, ab34710; Abcam) antibody and mouse anti-vimentin antibody (1:150, ab8978; Abcam), followed by goat anti-rabbit Alexa Fluor 488 and goat anti-mouse Alexa Fluor 594 (1:300, A11070 and A11020 respectively; Thermo Fisher Scientific) secondary antibodies. After DAPI staining (blue nuclei), this resulted in green-stained collagen and red-stained stellate cells. All experiments with organoids were repeated 3 times in quadruplicates.

Animal studies

All animal studies were approved by the Institutional Animal Care and Use Committee at Florida International University under protocol AN16-003 and conducted in accordance with the Guide for the Care and Use of Laboratory Animals [National Institutes of Health (NIH), Bethesda, MD, USA]. Expression of Rxfp1 in mouse liver was analyzed using 4-mo-old Rxfp1^LacZ/+ males in which the LacZ reporter was inserted into the Rxfp1 locus (34). ML290 does not activate rodent RXFP1 receptor (35); therefore, to test its therapeutic effects, we used humanized RXFP1 mice (23). All mutant strains were produced on C57BL/6 background. Homozygous humanized RXFP1 (hRXFP1/hRXFP1) males with human RXFP1 cDNA inserted into inactivated mouse Rxfp1 locus (23) were used in immunohistochemistry (IHC) experiments and in CCl₄-induced liver fibrosis model. The number of animals analyzed is indicated in figure legends.

In the acute model, 3-mo-old hRXFP1/hRXFP1 males were injected intraperitoneally with a single dose of CCl₄ diluted in sesame oil (1 CCl₄: 7 oil; 1 μl/g body weight; MilliporeSigma). Animals were randomly divided into control and experimental groups, which received a single injection of the vehicle or ML290 at 37 mg/kg i.p. This dosage of ML290 was effective to cause physiologic response (23). The vehicle for injections contained 7.6% DMSO (with or without ML290), 9.2% N-methyl-2-pyrrolidone (NMP; Thermo Fisher Scientific), 9.2% Kolliphor HS15 (MilliporeSigma), 9.2% PEG400 (Thermo Fisher Scientific), and 64.6% PBS. Mice were euthanized, and liver samples were collected for RNA analysis 20 h after ML290 injection.

In the chronic model of liver fibrosis, 4-mo-old hRXFP1/hRXFP1 males received the same dose of CCl₄ twice per week (Tuesday and Friday) for 6.5 wk (total of 13 injections). Four weeks after the first CCl₄ injection, mice were randomly allocated to receive either vehicle or ML290 as previously described 3 times per week (Tuesday, Wednesday, and Friday) for 2.5 wk (total of 8 injections). Mice were euthanized 72 h following the last CCl₄ injection and 24 h after the last ML290 or vehicle injection, and liver and blood samples were collected. Liver function markers were analyzed in mouse serum at the Baylor College of Medicine Center for Comparative Medicine (Houston, TX, USA).

Pharmacokinetics studies

The study was performed by Pharmaron (Irvine, CA, USA; www.pharmaron.com). The following vehicle was used to dissolve ML290: 60% PEG400, 5% DMSO, and 35% of 45% HP-β-CD in water. Two doses (15 and 30 mg/kg, i.p.) of ML290 were tested in this pharmacokinetic study, in which wild-type 1.5–2-mo-old C57BL/6J male and female mice were treated once a day for 14 d. Plasma and liver samples were collected at different time points after the last injection for analysis of ML290 concentration using HPLC DGU-20A5R (Shimadzu, Kyoto, Japan) and mass spectrometry AB API 6500 QTrap LC/MS/MS (AB Sciex, Framingham, MA, USA) techniques.

Histology

To localize Rxfp1 gene expression, Rxfp1^LacZ/+ males were treated with a single intraperitoneal injection of CCl₄. Forty-eight hours later, mice were euthanized, liver samples were frozen in optimal cutting temperature (OCT) compound, and 15-μm cryosections were prepared and stained using the Senescence β-Galactosidase Staining Kit (Cell Signaling, Danvers, MA, USA) as previously described (34). Slides were examined with a Carl Zeiss Axio A1 microscope equipped with an AxioCam MRc5 CCD camera (Oberkochen, Germany).

Collagen content in the left lateral lobe was analyzed in a blinded manner using 0.037% picrosirius red (Electron Microscopy Sciences, Hatfield, PA, USA) staining in 4.5-μm paraffin-embedded sections (6). At least 10 nonoverlapping bright field images at ×20 magnification were analyzed for each animal. IHC was performed using primary anti-RXFP1 pAb (1:2000, A.9227.1; Immundiagnostik, Bensheim, Germany), rabbit anti–α-smooth muscle actin (α-SMA) pAb (1:300, ab5694; Abcam), and rabbit anti-Ki67 pAb (1:500, RB-1510-P0; Thermo Fisher Scientific). The IHC staining was visualized with VECTASTAIN ABC Kits (Vector Laboratories, Burlingame, CA, USA) and ImmPACT DAB Peroxidase Substrate Kit (Vector Laboratories), producing positive brown staining. The nuclei were counterstained with Hematoxylin Harris Alum (MilliporeSigma). Staining was quantified with ImageJ software (NIH) (36) and presented as percentage of stained area within the analyzed image.

Immunofluorescence double staining was first performed with the anti-Ki67 antibody (1:500, overnight) and detected with biotin-conjugated goat anti-rabbit secondary IgG (1:200, 1 h) and Fluorescein Avidin DCS (A-2011; Vector Laboratories), followed by anti–α-SMA antibody (1:300), biotin-conjugated goat anti-rabbit secondary IgG, and Texas Red Avidin D (A-2006; Vector Laboratories). Nuclei were counterstained with DAPI, and images were analyzed using a fluorescent Carl Zeiss Axio Imager M2 microscope with an AxioCam MRm camera set.

Statistics

Data were analyzed using Prism v.5.02 (GraphPad Software, La Jolla, CA, USA). Comparisons between 2 groups were evaluated for significance using a Student’s t test, and between multiple groups by 1-way ANOVA with Dunnett’s multiple comparison post hoc test. All data are presented as means ± sem.

RESULTS

Relaxin receptor RXFP1 is expressed in fibrotic liver

Previously, RXFP1 expression was detected in activated but not in quiescent HSCs (37). To examine the RXFP1 expression pattern in mouse liver, we applied 2 approaches. First, we used mice heterozygous for Rxfp1-LacZ knockin reporter allele. Transcriptional activity of the Rxfp1 gene can be examined in such mice through the analysis of LacZ activity of the reporter in target tissues (34). To induce HSC activation, mice were treated with a single intraperitoneal CCl₄ injection. In livers isolated from CCl₄-treated Rxfp1^LacZ/+ animals 48 h after injection, strong staining was detected in the space of Disse and around the portal tract but not in hepatocytes (Fig. 1A), suggesting that Rxfp1 expression was activated in HSCs. In control liver sections isolated from nontreated Rxfp1^LacZ/+ males, there was almost no detectable LacZ activity (Fig. 1A). Second, we examined RXFP1 expression using transgenic mouse model with human internal ribosome entry site (IRES)-RXFP1 cDNA cassette inserted into inactivated mouse gene (23). IHC with human-specific anti-RXFP1 antibody showed similar expression pattern as in Rxfp1^LacZ/+ animals. Strong RXFP1-positive staining was localized mostly in the space of Disse and around portal tract 48 h after CCl₄ treatment (Fig. 1B). Liver sections obtained from similarly treated wild-type mice and preimmune serum negative control did not produce any staining validating the specificity of the antibody to human RXFP1 receptor (Fig. 1B). There was a dramatic increase of human RXFP1 mRNA detected by quantitative RT-PCR in liver samples from hRXFP1/hRXFP1 mice treated with CCl₄ compared with nontreated control livers (Fig. 1C).

Figure 1.

Expression of relaxin receptor RXFP1 in activated hepatic stellate cells. A) Expression of Rxfp1-LacZ reporter visualized with β-gal staining (arrows) in mouse liver. Strong staining appears in Rxfp1^LacZ/+ males treated with a single CCl₄ injection around portal tract and in perisinusoidal space. No staining was detected in nontreated livers. B) IHC with human-specific anti-RXFP1 antibody reveals human RXFP1 expression in perisinusoidal space and around portal veins (arrows) in humanized hRXFP1/hRXFP1 mice after a single CCl₄ injection. No RXFP1 staining was detected in wild-type (Wt) CCl₄-treated animals or in hRXFP1/hRXFP1 liver sections without primary antibody (negative control). C) Quantitative RT-PCR analysis of hRXFP1 expression in livers of hRXFP1/hRXFP1 mice shows a significant increase after CCl₄ treatment. Scale bars, 100 µm. **P < 0.01 (n = 5/group).

Selection of RXFP1 agonist ML290 as a potent antifibrotic agent in primary HSCs

We have previously described several small molecule RXFP1 agonists with variable activity and efficacy (20, 22). To select the most efficient antifibrotic compound, we evaluated the expression of 3 RXFP1 target genes involved in fibrogenesis in activated primary human HSCs treated with 2 concentrations of agonists at 1 or 5 µM for 72 h (Fig. 2). Primary HSCs become activated when cultured on plastic (27). Collagen type I-α 1 (COL1A1) is a major component of ECM in liver fibrosis, MMP1 is an interstitial collagenase, and peroxisome proliferator–activated receptor-γ coactivator 1-α (PPARGC1A) is a direct transcriptional target of relaxin/RXFP1 signaling (38). Out of the 5 compounds tested, only ML290 significantly affected expression of all these genes at higher concentration (5 µM) by decreasing COL1A1 expression and stimulating MMP1 and PPARGC1A expression (Fig. 2).

Figure 2.

Selection of lead RXFP1 agonist for antifibrotic properties in activated primary human HSCs. Shown are the effects of 1 and 5 µM of 5 RXFP1 agonists (see Materials and Methods) on the COL1A1, MMP1, and PPARGC1A gene expression after 72 h measured by quantitative RT-PCR. Data are presented as percentage of control (DMSO) treatment ± sem; n = 3/treatment. *P < 0.05, **P < 0.01, ***P < 0.001

Pharmacological properties of ML290

Previously, it was shown that ML290 did not have cytotoxic effects in several cell lines (20). Here, we tested ML290 cytotoxicity in human HSC cell line LX-2 (Fig. 3A). At 24 h, no significant differences in cell viability were detected after treatment with ML290. At lower concentrations up to 1 µM, ML290 increased the number of cells after 48 h treatment compared with DMSO-treated control, whereas at higher ML290 concentrations, no differences in cell viability were detected.

Figure 3.

Analysis of ML290 cytotoxicity and pharmacokinetics. A) The LX-2 hepatic stellate cell’s viability in medium containing various concentrations of ML290. Opened and closed squares denote 24- and 48-h treatment; n = 3 for each time point. Cell viability shown as a percentage of DMSO-treated control group ± sem. B) Pharmacokinetics of ML290 in wild-type male (circles) and female (squares) mouse liver (top segment) and plasma (bottom segment) after 2 wk of daily ML290 injection at 30 mg/kg (closed symbols, solid lines) or 15 mg/kg (open symbols, dashed line); n = 2 for each time point.

We have previously reported pharmacokinetics of ML290 after single intraperitoneal, intravenous, or oral administration as well as the maximum tolerated dose in a 5-d study in wild-type mice (22). To analyze the effects of multiple injections on pharmacokinetic profile of ML290, a 2-wk study was performed. In this experiment, we observed a high level of ML290 in plasma 72 h after the final injection in both male and female mice (Fig. 3B). In liver samples, the detected concentration of ML290 remained stable for at least 3 d after the final injection (Fig. 3B). Importantly, no signs of distress, behavioral abnormalities, or mortality were noticed in these mice. Finally, we analyzed 17 serum markers in male hRXFP1/hRXFP1 mice after 3 wk of daily injections of ML290 at 30 mg/kg i.p. (Table 2). No significant differences were observed in mice receiving ML290 compared with vehicle injections.

TABLE 2.

Serum markers in hRXFP1/hRXFP1 male mice treated daily with injections of vehicle or ML290 (30 mg/kg, i.p.) for 21 d

Marker	Vehicle (n = 7)	ML290 (n = 5)
ALT (U/L)	56.51 ± 6.86	43.99 ± 10.42
AST (U/L)	128.4 ± 6.55	120.3 ± 15.48
ALP (U/L)	31.25 ± 3.86	29.34 ± 2.57
Total bilirubin (mg/dl)	0.17 ± 0.01	0.15 ± 0.01
Direct bilirubin (mg/dl)	0.05 ± 0.00	0.05 ± 0.01
Indirect bilirubin (mg/dl)	0.12 ± 0.01	0.10 ± 0.01
BUN (mg/dl)	19.71 ± 1.49	29.81 ± 6.96
Cholesterol (mg/dl)	82.43 ± 7.45	81.19 ± 5.26
Creatine phosphokinase (U/L)	1942 ± 542.1	1261 ± 215.6
Glucose (mg/dl)	112.0 ± 13.6	94.39 ± 7.45
Lactate dehydrogenase (U/L)	345.4 ± 33.91	278.3 ± 24.91
Calcium (mg/dl)	8.74 ± 0.13	8.66 ± 0.20
Magnesium (mg/dl)	2.86 ± 0.09	2.95 ± 0.10
Phosphorous (mg/dl)	7.78 ± 0.33	8.02 ± 0.30
Triglycerides (mg/dl)	57.73 ± 5.39	54.72 ± 3.20
Total CO2 (mEq/L)	29.72 ± 1.58	28.98 ± 1.00
VLDL (mg/dl)	11.55 ± 1.08	10.94 ± 0.64

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BUN, blood urea nitrogen; VLDL, very low-density lipoprotein.

ML290 induces an antifibrotic response in activated HSCs

ML290-induced changes in the expression of several genes involved in profibrotic pathway were reported in activated primary HSCs as well as in LX-2 cells (22, 30). Here, we tested whether ML290 treatment results in more extensive changes in gene expression indicative of therapeutic action of the relaxin/RXFP1 signaling. RNA sequencing (RNA-seq) analysis on TGF-β1–activated LX-2 cells treated with ML290 for 72 h was performed (Fig. 4). Roughly equal numbers of down- (272) and up-regulated (225) genes with at least 2-fold differences and false discovery rate < 0.01 were detected (Fig. 4A). To validate the RNA-seq results and to evaluate genes with lower expression, we performed quantitative RT-PCR with a selected group of pro- and antifibrotic genes (Fig. 4B). Expression of ACTA2, COL1A1, COL5A1, TGFB1, Caveolin 1 (CAV1), and tissue inhibitor of metalloproteinase 3 (TIMP3) was significantly decreased when compared with DMSO-treated cells (control), whereas MMP1 and PPARGC1A expression was significantly increased. ML290 treatment for 72 h did not affect TIMP1 gene expression. Importantly, expression of RXFP1 was decreased in ML290-treated cells, suggesting a suppression of myofibroblast-like differentiation of HSCs. A hierarchical heatmap of 19 key genes directly involved in regulating fibrotic phenotype of activated HSCs demonstrated a clear antifibrotic effect of ML290 treatment (Fig. 4C). To investigate other biological functions affected by ML290, dysregulated genes were analyzed using ToppGene Suite (https://toppgene.cchmc.org/) (39). A partial list of biological processes, cellular components, and molecular functions revealed by Kyoto Encyclopedia of Genes and Genomes (KEGG; https://www.genome.jp/kegg/kegg1.html) pathway and Reactome Pathway (https://reactome.org/) analysis is presented in Fig. 4D. ML290 significantly affected ECM and collagen formation as well as the cellular pathways previously shown to be activated by relaxin or ML290/RXFP1 signaling, such as the MAPK, PI3K–protein kinase B, p53, cytokine, and IL signaling pathways (5, 28).

Figure 4.

Transcriptional analysis of TGF-β1–activated LX-2 stellate cells treated with ML290. A) A volcano plot shows significant differences in gene expression by RNA sequencing in cells treated with DMSO or 5 μM ML290 for 72 h (n = 4; log₂ fold change). A total of 225 genes were up-regulated (red dots) and 272 genes were down-regulated (green dots). B) Validation of the RNA-seq results by quantitative RT-PCR with selected group of fibrosis-related genes. Data are presented as percentage of DMSO (control) treatment and normalized to the GAPDH gene expression as an internal control. *P < 0.05, **P < 0.01, ***P < 0.001 (n = 4). C) Heatmap showing standardized values (z score) of 19 fibrosis-related genes in LX-2 cells treated with DMSO or ML290. Standardized expression levels depicted by the color gradient: up-regulated genes in red, down-regulated genes in green. D) Gene enrichment pathway analysis for biological processes in LX-2 cells treated with ML290.

ML290 reduces LPS-induced fibrosis in human liver organoids

To model the complex cell environment of hepatic tissues, we tested the effects of ML290 in 3-dimensional multicellular human liver organoids (31) (Fig. 5A). Quantification of the signal intensity clearly demonstrated that cotreatment of LPS with ML290 at concentration as low as 10 nM had a profound effect on collagen accumulation in the liver organoids, as evidenced by a reduction in red staining for collagen when compared with 1 nM ML290–treated group (Fig. 5B). We also tested whether ML290 can reduce collagen content in an established fibrosis caused by 7 d pretreatment with LPS (Fig. 5C). Treatment with ML290 for 7 d at 5, 50, or 500 nM in established fibrotic liver organoids resulted in a clear decrease of collagen content at the 2 higher concentrations of the compound when compared with DMSO-treated control organoids. Interestingly, there was no visible decrease in vimentin staining, a marker of mesenchymal cells, observed in ML290-treated organoids. Toxicity was not observed in any experiments, and all organoids remained viable for the entire treatment period.

Figure 5.

Antifibrotic effects of ML290 in human liver organoids. A) Effect of LPS treatment on collagen accumulation. Whole-mount immunofluorescence staining for COL1A1 (red) shows a sharp increase of collagen content in organoid treated with 10 μg/ml LPS for 4 d. B) A decrease of COL1A1 content in organoids treated simultaneously with 10 μg/ml LPS and various concentrations of ML290 for 4 d. Quantitative analysis of COL1A1 staining in 4 randomly chosen organoids per treatment concentration using confocal imaging is shown below. Fibrotic index is calculated based on the ratio of red and blue pixels. Experiments were performed on 3 independent occasions, and results are expressed as the mean of the fibrotic index ± sem. Significant differences were found between ML290 at 1 nM when compared to higher concentrations of ML290. *P < 0.001. C) Whole-mount immunofluorescence staining of COL1A1 (green) and vimentin (red) in liver organoids treated first with 10 μg/ml LPS for 7 d to induce fibrosis and then with various concentrations of ML290 for additional 7 d. Representative images are shown. Scale bars, 100 μm.

ML290 is a potential therapeutic agent for liver fibrosis

We next tested the effects of ML290 in a well-established CCl₄-induced mouse model of fibrosis previously shown to respond to relaxin treatment (6, 9). We first evaluated the acute effect of simultaneous treatment of CCl₄ and ML290 in hRXFP1/hRXFP1 mice. A single injection of CCl₄ in adult males resulted in an increase of profibrotic Acta2, Col1a1, and Tgfb1 (Fig. 6) gene expression in liver 20 h after treatment. Cotreatment with CCl₄ and ML290 (37 mg/kg, i.p.) attenuated this increase, indicating that the observed ML290 effects on activated HSCs in vitro were reproduced in vivo.

Figure 6.

Effects of ML290 on fibrotic gene expression in mouse liver after a single injection of CCl₄ and ML290. Gene expression was analyzed by quantitative RT-PCR in liver samples of RXFP1 humanized male mice 20 h after intraperitoneal injection of CCl₄ + vehicle (n = 5), CCl₄ + ML290 (n = 5), or nontreated group (n = 10). Gene expression was normalized to Gapdh expression. *P < 0.05, ***P < 0.001.

We then tested the therapeutic potential of ML290 to suppress multiple facets of established CCl₄-induced liver fibrosis. As in the acute treatment with CCl₄, multiple injections of CCl₄ resulted in an increase of RXFP1 expression around the portal tract (Fig. 7A). These areas were also positive for α-SMA staining (Fig. 7B). Analysis of collagen content using Sirius red staining followed by ImageJ quantification of stained areas revealed a strong antifibrotic effect of ML290 treatment when compared with vehicle-treated males (Fig. 7C, D). Overall, staining in ML290-treated livers was decreased by 30% compared with the vehicle group; however, it remained higher than in nontreated livers of age-matched males. Analysis of α-SMA staining at low magnification (X5) did not reveal any difference between vehicle- and ML290-treated mice (2.2 ± 0.2% in CCl₄ + vehicle vs. 2.0 ± 0.2% in CCl₄ + ML290) (Fig. 7B). We then focused on the analysis of α-SMA staining around portal veins where RXFP1 expression was most prominent (Fig. 7A). Using higher magnification (X20), we selected images with large portal veins. This approach showed that in these areas, ML290 caused more than 50% reduction in α-SMA staining (Fig. 7C, D). We observed a significant reduction of Ki67 immunostaining in cells surrounding portal tract but not in hepatocytes in ML290-treated mice (Fig. 7B–D). Double immunofluorescence staining in CCl₄-treated livers demonstrated a significant number of dividing cells positive for Ki67 also coexpressed α-SMA (Fig. 7B). These findings support the notion that ML290 not only reduces collagen content but also functions as a critical regulator of cell proliferation and myofibroblast differentiation. Gene expression changes in selected group of profibrotic genes in the livers of ML290-treated mice demonstrated its antagonistic effect (Fig. 8), including the reduction of Col1a1 and Acta2 and activation of vimentin (Vim), Timp1 and Timp2, and Pparg. No significant differences in expression of other markers of fibrosis such as connective tissue growth factor (Ctgf) or desmin (Des), Tgfb1, Mmp2, Mmp13, or Cav1 were detected. ML290 did not affect liver function in CCl₄-treated hRXFP1/hRXFP1 mice as evidenced by analysis of serum markers in the 2 groups (Table 3). No differences in the total body or liver weight were found between groups (unpublished results).

Figure 7.

Therapeutic effects of ML290 in established CCl₄-induced liver fibrosis. A) Representative image of IHC for human RXFP1 shows intense positive brown staining around portal vein (arrows). Scale bar, 50 µm. B) Double immunofluorescence staining for α-SMA (red, arrows), Ki67 (green), and DAPI (blue). Scale bar, 20 µm. C). Liver sections from nontreated, CCl₄ + vehicle, and CCl₄ + ML290 treated RXFP1 humanized male mice stained with Sirius red (red staining for collagen), IHC for α-SMA, and Ki67 around portal veins (both brown staining). Insets in Ki67 panel show the differences in the number of proliferating cells. Scale bar, 100 μm; inset in Ki67, 20 μm. D) Quantification of percentage positive staining. Nontreated group, n = 8; CCl₄ + vehicle, n = 13; CCl₄ + ML290, n = 15. **P < 0.01, ***P < 0.001.

Figure 8.

Changes in gene expression in liver of RXFP1 humanized mice with CCl₄-induced liver fibrosis treated with ML290. Gene expression was normalized to Gapdh expression. Nontreated group, n = 10; CCl₄ + vehicle, n = 13; CCl₄ + ML290, n = 15. *P < 0.05. **P < 0.01, ***P < 0.001.

TABLE 3.

Serum markers in RXFP1 humanized male mice treated with CCl₄ and vehicle or ML290 (37 mg/kg)

Marker	CCl4 + vehicle (n = 5)	CCl4 + ML290 (n = 13)
ALT (U/L)	84.2 ± 11.0	103.8 ± 14.3
AST (U/L)	159.4 ± 12.8	167.5 ± 18.9
ALP (U/L)	73.6 ± 3.1	63.4 ± 2.4
LDH (U/L)	397.8 ± 76.7	461.5 ± 44.9
Direct bilirubin (mg/dl)	0.05 ± 0.01	0.05 ± 0.01
Indirect bilirubin (mg/dl)	0.12 ± 0.2	0.12 ± 0.3
Total bilirubin (mg/dl)	0.15 ± 0.03	0.17 ± 0.03
Albumin (g/dl)	3.7 ± 0.08	3.5 ± 0.11
Globulin (g/dl)	1.46 ± 0.08	1.49 ± 0.08
Albumin/globulin	2.58 ± 0.09	2.4 ± 0.07
Total protein (g/dl)	5.18 ± 0.16	4.65 ± 0.40

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; LDH, lactate dehydrogenase.

DISCUSSION

This study demonstrated that treatment with the lead RXFP1 agonist, ML290, resulted in an amelioration of fibrotic phenotype in activated HSCs in vitro and in vivo. Specifically, ML290-induced gene expression revealed by RNA-seq analysis in LX-2 cells demonstrated a suppression of collagen expression, increased turnover of ECM, suppression of TGF signaling, and myofibroblast differentiation. In human liver organoids, ML290 treatment prevented collagen accumulation in LPS-induced fibrosis and reduced collagen accumulation in organoids with established fibrotic phenotype. ML290 had no cytotoxicity in LX-2 cells and demonstrated sustained stability in liver after multiple injections in mice. No liver toxicity was found for the compound. Notably, this study revealed for the first time that ML290 treatment significantly reduced collagen content in CCl₄-induced liver fibrosis in mice, an effect that was associated with a reduction in the expression of α-SMA and reduced HSC proliferation.

Relaxin activation of RXFP1 results in activation of several downstream signaling pathways, such as cAMP, cGMP, various kinases, nitric oxide, and others depending on the cell type and, possibly, the receptor expression level. The ML290 series of agonists were identified through cAMP response in stably transfected human embryonic kidney 293–RXFP1 cells (20, 21). Further analysis revealed that in vascular cells, ML290 behaves as a biased allosteric agonist at RXFP1; it does not induce the ERK1/2 phosphorylation, a characteristic of relaxin stimulation (28). ML290 also shows higher potency to activate phosphorylated p38MAPK and cGMP accumulation than that of cAMP accumulation (28). Previously, we demonstrated that despite lower activity in the cAMP assay, ML290 showed higher efficacy in other relaxin-dependent cellular and gene activation tests than several other compounds in this series (20). Here, we tested 5 different compounds to select the agonist most effective at targeting expression of several genes involved in remodeling of ECM in primary HSCs (22, 30). ML290 consistently suppressed COL1A1 and TGFB1 expression and up-regulated the previously established relaxin target PPARGC1A (38). The data suggest that apart from cAMP, other signaling pathways activated by ML290 might play a role in its antifibrotic effects.

Recently, ML290 was shown to modulate the transcription of fibrosis-related genes in LX-2 cells (30). We used whole transcriptome analysis of activated LX-2 cells to identify additional gene enrichment pathways induced by the compound. Cotreatment with ML290 led to a significant down-regulation of TGF-β1–induced collagen genes, markers of myogenic differentiation, and an increase of ECM-degrading enzymes. We showed that in addition to these genes, a number of other genes critical for fibrosis were affected. This includes TNF, IL, cytokines, and platelet-derived growth factor signaling. As in primary human HSCs, the expression of PPARGC1A, a regulator of PPARG activity, was significantly increased in LX-2 cells, suggesting an involvement of ML290/RXFP1 in ligand-independent activation of PPARG in HSCs (38). Expression of α-SMA and RXFP1 were reduced, indicating the suppression of the myogenic differentiation of HSCs. Overall, the data suggest pleotropic effects of ML290/RXFP1 signaling on multiple biological pathways that support the antifibrotic action of the small molecule RXFP1 agonist.

While monolayer cell culture experiments are useful, the liver response to profibrotic stimuli involves a complex interaction of different cell types through cell–cell communications. To model these interactions, we first analyzed the antifibrotic properties of ML290 in fully functional bioengineered 3-dimensional multicellular human liver organoids (31, 32). Two treatment models were used. In a preventive treatment model, ML290 was added concurrently with fibrosis-inducing LPS. In a chronic fibrosis model, organoids were treated first with LPS and then with ML290. The collagen content in organoids treated with ML290 in both experiments was significantly reduced even at double-digit nanomolar concentrations.

The ML290 series of RXFP1 agonists do not activate rodent receptors (35). To test various properties of the small molecule agonists in vivo, we produced a mouse model in which we substituted the mouse gene with human RXFP1 (23). Similar to the previously described increase in RXFP1 expression in rat, mouse, and human cirrhotic livers (9, 37), hRXFP1 expression was dramatically up-regulated in CCl₄-treated livers. Pharmacokinetic analysis of ML290 after multiple intraperitoneal injections over 14 d demonstrated the remarkable stability of the compound in livers of treated animals for at least 72 h after the last treatment, even at a slightly lower dosage that was used in CCl₄-induced fibrosis (30 vs. 37 mg/kg). Because of the high stability shown by ML290 in liver, we injected the compound 3 d/wk for 2.5 wk. Collagen content in ML290-treated animals was significantly reduced, although it remained elevated compared with the control livers. Interestingly, the effects of ML290 treatment was not uniform. Areas around portal tracts demonstrated lower α-SMA staining coupled with reduced HSC proliferation rates. It is therefore possible that the activated HSCs and portal fibroblasts localized in the perisinusoidal space between hepatocytes and endothelial cells are the primary targets of ML290. These cells are believed to be the primary source of activated myofibroblasts that drive the fibrogenic process (2). Suppression of myofibroblast differentiation and proliferation might be the additional mechanisms of ML290/RXFP1 action.

Our data support previous reports that suggest that targeting the relaxin signaling pathway has strong beneficial effects in several models of fibrotic diseases. By overcoming problems associated with the peptide stability and delivery, small molecule agonists of RXFP1 provide an attractive therapeutic alternative. The beneficial effects of such compounds may have a wide application in a variety of fibrotic conditions.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.

Glossary

α-SMA

α-smooth muscle actin protein

ACTA2

α-smooth muscle actin gene

CAV1

caveolin 1

CCl₄

carbon tetrachloride

COL1A1

collagen type I-α 1

ECM

extracellular matrix

hRXFP1

humanized RXFP1

HSC

hepatic stellate cell

IHC

immunohistochemistry

LacZ

β-galactosidase

LPS

lipopolysaccharide

MMP

matrix metalloproteinase

PPARGC1A

peroxisome proliferator–activated receptor-γ coactivator 1-α

RNA-seq

RNA sequencing

RXFP1

relaxin family peptide receptor 1

TIMP

tissue inhibitor of metalloproteinase

TGF

transforming growth factor

AUTHOR CONTRIBUTIONS

E. M. Kaftanovskaya, J. J. Marugan, C. E. Bishop, I. U. Agoulnik, and A. I. Agoulnik conceptualized the study; E. M. Kaftanovskaya, H. H. Ng, C. E. Bishop, I. U. Agoulnik, and A. I. Agoulnik designed the study; E. M. Kaftanovskaya, H. H. Ng, M. Soula, B. Rivas, C. Myhr, B. A. Ho, B. A. Cervantes, T. D. Shupe, M. Devarasetty, X. Hu, X. Xu, S. Patnaik, K. J. Wilson, E. Barnaeva, M. Ferrer, N. T. Southall, C. E. Bishop, I. U. Agoulnik, and A. I. Agoulnik conducted the experiments; E. M. Kaftanovskaya, H. H. Ng, J. J. Marugan, C. E. Bishop, I. U. Agoulnik, and A. I. Agoulnik analyzed and interpreted the data; and E. M. Kaftanovskaya, H. H. Ng, C. Myhr, C. E. Bishop, and A. I. Agoulnik wrote the manuscript.