Relaxin modulates human and rat hepatic myofibroblast function and ameliorates portal hypertension in vivo


  • See Editorial on Page 1223

    Potential conflict of interest: Nothing to report.

  • Jonathan Fallowfield and Damian Mole are supported by Academy of Medical Sciences / The Health Foundation Clinician Scientist Fellowships. Annette Hayden was funded jointly by Ferring Research Ltd as part of an industrial collaborative studentship with the Medical Research Council (MRC) UK and the University of Southampton. Victoria Snowdon and Ben Stutchfield are funded by Scottish Translational Medicine and Therapeutics / Wellcome Trust Clinical Research Fellowships. John Iredale gratefully acknowledges the support of the Medical Research Council and the Wellcome Trust.


Active myofibroblast (MF) contraction contributes significantly to the increased intrahepatic vascular resistance that is the primary cause of portal hypertension (PHT) in cirrhosis. We sought proof of concept for direct therapeutic targeting of the dynamic component of PHT and markers of MF activation using short-term administration of the peptide hormone relaxin (RLN). We defined the portal hypotensive effect in rat models of sinusoidal PHT and the expression, activity, and function of the RLN-receptor signaling axis in human liver MFs. The effects of RLN were studied after 8 and 16 weeks carbon tetrachloride intoxication, following bile duct ligation, and in tissue culture models. Hemodynamic changes were analyzed by direct cannulation, perivascular flowprobe, indocyanine green imaging, and functional magnetic resonance imaging. Serum and hepatic nitric oxide (NO) levels were determined by immunoassay. Hepatic inflammation was assessed by histology and serum markers and fibrosis by collagen proportionate area. Gene expression was analyzed by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) and western blotting and hepatic stellate cell (HSC)-MF contractility by gel contraction assay. Increased expression of RLN receptor (RXFP1) was shown in HSC-MFs and fibrotic liver diseases in both rats and humans. RLN induced a selective and significant reduction in portal pressure in pathologically distinct PHT models, through augmentation of intrahepatic NO signaling and a dramatic reduction in contractile filament expression in HSC-MFs. Critical for translation, RLN did not induce systemic hypotension even in advanced cirrhosis models. Portal blood flow and hepatic oxygenation were increased by RLN in early cirrhosis. Treatment of human HSC-MFs with RLN inhibited contractility and induced an antifibrogenic phenotype in an RXFP1-dependent manner. Conclusion: We identified RXFP1 as a potential new therapeutic target for PHT and MF activation status. (Hepatology 2014;59:1492-1504)


alpha-smooth muscle actin


alanine aminotransferase


bile duct ligation


blood oxygen dependent-magnetic resonance imaging


carbon tetrachloride


collagen proportionate area


(endothelial) nitric oxide synthase






(human) hepatic stellate cell


indocyanine green


intrahepatic vascular resistance


L-NG-Nitroarginine methyl ester


mean arterial pressure




nitric oxide


portal hypertension


relaxin receptor family peptide.

The dynamic component of portal hypertension (PHT), mediated by active contraction of myofibroblasts (MFs) predominantly derived from activated hepatic stellate cells (HSCs),[1, 2] accounts for about 30% of the increased intrahepatic vascular resistance (IHVR) in cirrhosis and is potentially reversible.[3, 4] The increase in sinusoidal tone reflects a functional disturbance of the liver circulation in response to excessive production of contractile agonists (e.g., endothelin-1 [ET-1]), reduced bioavailability of relaxant factors (mainly nitric oxide [NO]), and impaired responses to NO.[5, 6] The structural component of IHVR may be decreased by effective etiological treatment,[7] but this is not achievable in the majority of patients with PHT. Current treatments for PHT (vasopressin and somatostatin analogs, nonselective beta-blockers) decrease portal pressure by splanchnic vasoconstriction, but this may further compromise hepatic perfusion in cirrhosis and impair organ function.[8]

Relaxin (RLN) is a naturally occurring peptide hormone with a two-chain structure similar to insulin.[9] Although initially identified as a hormone of pregnancy, gene knockout studies and observations in rodents and humans treated with exogenous RLN have revealed diverse effects in nonreproductive tissues in both males and females.[9] The product of the human RLN2 gene, human gene 2-relaxin (H2-RLN), is the major stored and circulating form. H2-RLN circulates in women at low concentrations during the luteal phase of the menstrual cycle (∼50 pg/mL) and at increased levels (∼1 ng/mL) throughout pregnancy. In men, H2-RLN is expressed locally in the prostate and may also be present at very low levels (1-5 pg/mL) in the circulation. The primary receptor for H2-RLN, relaxin receptor family peptide-1 (RXFP1), is a member of the leucine-rich repeat family of G-protein coupled receptors[10] and is widely distributed in many tissues in both sexes. RLN is thought to mediate its physiological actions in an autocrine or paracrine manner, whereby activation of RXFP1 results in the stimulation of multiple signal transduction pathways including 3′-5′-cyclic adenosine monophosphate (cAMP) (influenced by a variety of Gα isoforms), extracellular signal-regulated kinases, tyrosine kinases, and NO signaling.[10, 11] The effects of RLN on connective tissue remodeling in reproductive and nonreproductive tissues are well documented.[9, 12] Although circulating RLN has been detected in patients with chronic liver disease, serum concentrations were either stable or rose very slowly over 15 years of follow-up and did not correlate with the histological stage of fibrosis (M. Mayo, pers. commun.).

A number of humoral agents that elicit either HSC contraction or relaxation have been identified.[1] In human cirrhosis, circulating and tissue levels of the HSC contractile agonist ET-1 are elevated[13] and HSC sensitivity to ET-1 is also enhanced. Exogenous NO has been shown to abolish the contractile effects of ET-1 in isolated perfused liver and in cultured HSCs.[14] Through binding to RXFP1, RLN may induce activation of the NO pathway in target cells. Furthermore, RLN has also been shown to inhibit the vasoconstrictive properties of ET-1 and angiotensin-II in rodent models.[15, 16]

Here we show that RLN can modulate the dynamic component of PHT, through effects on intrahepatic NO and MF contractility; thereby identifying the RLN-RXFP1 axis as a novel vasoactive target in cirrhosis.

Materials and Methods

Tissue was used in accordance with the local ethical review committees and the Human Tissue Act 2004. Procedures involving animals were conducted with ethical approval and in accordance with the Use of Animals in Scientific Procedures Act 1986.

Isolation and Culture of Cells

Primary HSCs and hepatic sinusoidal endothelial cells (HSECs) were isolated and cultured as described in the online Supporting Information.

Quantitation of HSC Apoptosis

Apoptosis of human HSCs was assessed using acridine orange as described in the online Supporting Information.

Quantitative Polymerase Chain Reaction (qPCR)

Total RNA was extracted using the RNeasy kit (Qiagen, Manchester, UK) and qPCR performed. Full details are provided in the online Supporting Information and primer sequences listed in Supporting Table 1.

cAMP Assay

Intracellular cAMP measurement in HSCs was performed using the HitHunter cAMP II kit (DiscoveRx, Birmingham, UK) following the manufacturer's instructions. Cells were resuspended in antibody solution at 10,000 cells/well with 0.5 mM isobutylmethylxanthine (IBMX).

Cyclic Guanosine Monophosphate (cGMP) Immunoassay

cGMP in whole liver homogenates and HSC lysates was measured by enzyme immunoassay (Cayman Chemical, Cambridge, UK). Full details are provided in the online Supporting Information.

Gel Contraction Assay

Activated passage-1 human HSCs (hHSCs) suspended in Dulbecco's Modified Eagle Medium (DMEM; Gibco, Paisley, UK) / 10% fetal bovine serum (FBS; Gibco) were layered on top of type-1 collagen lattices at a density of 100,000 cells/well in 24-well flat-bottom tissue culture plates and contractility assessed in response to 1 μM H2-RLN ± RXFP1 small interfering RNA (siRNA). Full details are provided in the online Supporting Information.

NO Measurement

Rat serum was centrifuged at 13,000g for 2 minutes before filtration using a 10,000 Da molecular weight cutoff filter (Millipore, Watford, UK). NO/NO3-/NO2- determination was performed using a Parameter Kit (R&D Systems, Abingdon, UK).

Histology, Immunohistochemistry, and Confocal Microscopy

Full details are provided in the online Supporting Information.

Western Blotting

Full details are provided in the online Supporting Information and antibody dilutions listed in Supporting Table 3.

Rat Carbon Tetrachloride Models

Cirrhosis and PHT was induced in age-matched male Sprague-Dawley rats by 8 or 16 weeks administration of carbon tetrachloride (CCl4; Sigma). Littermates (∼250-300 g body weight) were randomly allocated to treatment with either twice-weekly injections of intraperitoneal CCl4 (0.1 mL/100g CCl4 for 2 weeks then 0.05 mL CCl4/100g for 6-14 weeks) or sterile vehicle (olive oil; Sigma). 72h RLN Model: Twenty-four hours after the final injection of CCl4 or olive oil (“peak fibrosis”), rats either underwent portal pressure measurement (n = 4-6/group) or insertion of subcutaneous osmotic minipumps (2ML2 Alzet; Durect, Cupertino, CA) under isoflurane anesthesia. Minipumps were preloaded with recombinant human H2-RLN (0.5 mg/kg/day; Corthera, San Carlos, CA) or vehicle (sodium acetate pH 5) for 72h continuous infusion (n = 10). Subgroups of rats (n = 5) were also coadministered L-NG-Nitroarginine methyl ester (L-NAME; Sigma) 250 mg/mL in drinking water starting 24h before minipump insertion and continuing for the 72h infusion period. Serum RLN levels were measured using the Quantikine Human Relaxin-2 Immunoassay (R&D Systems). Blood was collected from the inferior vena cava (systemic blood) or portal vein (splanchnic blood), tissues were fixed in 10% buffered formalin or snap-frozen in liquid nitrogen. Acute RLN Model: Groups of CCl4 or olive oil treated anesthetized rats were randomly allocated to treatment with intravenous RLN (4 μg/mL over 2 minutes) or equivalent volume of vehicle by way of an internal jugular vein catheter (PE-50; Durect), with continuous hemodynamic monitoring for 180 minutes. Animals were warmed using a heat pad and received intraoperative 0.9% NaCl 10 mL/kg/h.

Rat Bile Duct Ligation Model

Liver fibrosis and PHT was induced in age-matched male Sprague-Dawley rats (∼250-300 g) by bile duct ligation (BDL) under isoflurane anesthesia. Briefly, the common bile duct was exposed after laparotomy, ligated with 7-0 silk and cut. Intraoperatively, animals were warmed and received 0.9% NaCl 10 mL/kg/h. Eighteen days post-BDL, animals were randomly allocated to 72h subcutaneous treatment with RLN (0.5 mg/kg/day) or vehicle by minipump (n = 5).

Hemodynamic Measurements

Hemodynamic data were recorded from the femoral artery by way of a PE-50 catheter and portal vein by way of a 24G cannula or indwelling PE-50 catheter using Powerlab 4/35 with LabChart 7 Pro software (ADInstruments, Oxford, UK). Portal blood flow was measured using a perivascular flowprobe (Transonic MA1.5PRB; ADInstruments).

Dynamic Indocyanine Green (ICG) Imaging

Noninvasive in vivo imaging was performed using the IVIS Spectrum Imaging System and Living Image v. 3.2 software (PerkinElmer, Waltham, MA). Hepatic uptake and peripheral disappearance rate of 2.5 mg/kg ICG (Cardiogreen; Sigma) was assessed by dynamic fluorescence imaging in anesthetized 8-week CCl4 rats after pretreatment with 4 μg/mL intravenous H2-RLN or vehicle. Full details are provided in the online Supporting Information.

Blood Oxygen Level-Dependent Magnetic Resonance Imaging (BOLD-MRI)

BOLD-MRI uses the paramagnetic properties of deoxyhemoglobin to acquire images sensitive to local tissue oxygen concentration.[17] MRI data were acquired using a 7-Tesla preclinical MRI scanner (Agilent Technologies, Edinburgh, UK) in anesthetized 8-week CCl4 rats, 30 and 60 minutes after administration of 4 μg/mL intravenous H2-RLN or vehicle. Full details are provided in the online Supporting Information.


GraphPad Prism 5 was used for statistical analysis. Data are presented as mean ± standard error of the mean or 95% confidence intervals. Parametric data were compared by unpaired t test or one-way analysis of variance (ANOVA) with post-hoc Tukey test for multiple groups, and nonparametric data by equivalent methods.


RXFP1 Is Expressed in Rat and Human Cirrhotic Liver and MFs Are the Major Cellular Source

We studied hepatic expression of the primary receptor for RLN, RXFP1, in progressive rat CCl4 fibrosis. Rxfp1 messenger RNA (mRNA) was barely detectable in normal rat liver but transcripts were increased 38-fold and 44-fold, respectively, after 8 weeks and 16 weeks of CCl4 (Fig. 1A). Quiescent primary rat HSCs did not express Rxfp1 mRNA (data not shown) but RXFP1 and alpha-smooth muscle actin (α-SMA) colocalized in culture-activated rat HSCs (Fig. 1B). Neither hepatocytes nor hepatic macrophages expressed RXFP1 (data not shown). Expression of RXFP1 protein was shown in a range of mechanistically distinct rat liver fibrosis models (Fig. 1C) in areas of fibrosis and within the sinusoids. No RXFP1 protein was detected in olive oil-treated (Fig. 1C) or normal rat liver (data not shown). Quiescent human HSCs (hHSCs) did not express RXFP1 or other HSC activation markers, but culture-activated hHSCs expressed RXFP1 mRNA (Fig. 1D) and protein (Fig. 1E). We have previously demonstrated that the activated HSC is not a committed phenotype and HSCs can effectively become “deactivated” in vitro after replating onto a synthetic basement membrane-type matrix.[18] Culture of activated hHSCs on Matrigel abrogated expression of RXFP1 and COL1A2 mRNA. (Fig. 1D). In addition, human sinusoidal endothelial cells (HSECs) isolated from fibrotic liver also expressed RXFP1 (Supporting Fig. 2D). Finally, we showed that RXFP1 was undetectable in normal human liver but abundant in areas of fibrous scarring within explanted cirrhotic livers (Fig. 1F). Taken together, these data indicate that expression of RXFP1 is restricted to activated HSC-MFs and HSECs in fibrotic liver.

Figure 1.

(A) Hepatic expression of Rxfp1 mRNA in progressive rat CCl4 injury (**P < 0.01, ***P < 0.001; NS, not significant; n = 4). (B) Dual immunofluorescence in activated rat HSCs (α-SMA, red; RXFP1, green; DAPI (diamidino-2-phenylindole) nuclear stain, blue). (C) Hepatic RXFP1 immunostaining in rat fibrosis models (TAA, thioacetamide). Positive staining indicated in fibrotic scars and sinusoids (arrows). (D) Representative gels showing gene expression by RT-PCR in hHSCs activated on plastic (P) and deactivated on Matrigel (M) (COL1A2, procollagen-1; GAPDH, glyceraldehyde 3-phosphate dehydrogenase). (E) RXFP1 immunofluorescence in activated hHSCs. (F) Immunostaining for RXFP1 in normal and cirrhotic (explanted) human liver, with localization to areas of fibrotic scarring (arrows). ALD, alcohol-related liver disease; AI, autoimmune.

RLN Treatment Reduces Portal Pressure in 8-Week CCl4 Rats

Cirrhosis and preascitic PHT was induced by 8 weeks CCl4 (Supporting Fig. 1A-D). After 72h subcutaneous RLN the mean serum concentration was 62 ± 10 ng/mL, whereas in controls it was undetectable. Rats receiving RLN had lower portal pressures (Fig. 2A; 9.7 ± 0.28 mmHg versus 11.7 ± 0.29 mmHg) and a reduction in splenomegaly (Fig. 2B; spleen:body weight ratio 0.20 ± 0.01 versus 0.27 ± 0.03). Livers were assessed for hepatocellular injury and fibrosis (Supporting Fig. 1E,F). There was no difference in serum alanine aminotransferase (ALT) level (Fig. 2C), hepatic necro-inflammatory score (Fig. 2D), or collagen proportionate area (CPA; Fig. 2E) between RLN-treated rats and controls. Acknowledging that matrix remodeling was unlikely within the short time-frame chosen for our hemodynamic study, we quantified hepatic expression of profibrogenic marker genes (Fig. 2F). Relative to control, transcripts for all profibrotic genes were reduced by RLN.

Figure 2.

Effect of 72h RLN treatment on portal pressure, hepatic inflammation, and fibrosis in 8-week CCl4 rats. (A) Portal pressure (***P = 0.0003) and (B) spleen size (**P = 0.01) were compared after 72h subcutaneous RLN (0.5 mg/kg/day) or vehicle (veh; sodium acetate pH 5) (n = 8-9). (C) Serum ALT, (D) hepatic histological inflammation score (0 (absent), 1 (<2 foci), 2 (2-4 foci), 3 (>4 foci), and (E) collagen proportionate area (CPA) (n = 5-10). (F) Relative hepatic expression of profibrotic genes by qRT-PCR (mean ± 95% confidence intervals; *P < 0.05 for Col1a2 (procollagen-1), Mmp3 (matrix metalloproteinase-3) and Acta2 (α-smooth muscle actin). Col3a1, procollagen-3; Lox, lysyl oxidase; Timp1, tissue inhibitor of metalloproteinases-1; n = 6.

RLN Increases Intrahepatic But Not Systemic NO Levels in 8-Week CCl4 Rats

To define the role of NO in mediating the portal hypotensive effect of RLN, we measured serum NO metabolites and intrahepatic cGMP as a surrogate index of NO bioavailability. We found no difference in total nitrite levels in systemic venous (Fig. 3A) and portal venous blood (data not shown), but RLN-treated rats had increased intrahepatic cGMP levels compared to controls (Fig. 3B). Hepatic expression of phosphorylated Akt (Ser473) (Supporting Fig. 2A) and phosphorylated eNOS(Ser1179) protein (Fig. 3C) were also increased. The eNOS regulatory protein caveolin-1 is abundant in cirrhotic liver but, relative to control, RLN reduced caveolin-1 mRNA and protein (Fig. 3D). These data indicate that RLN can stimulate intrinsic NO generation in fibrotic liver by activating the Akt/eNOS/cGMP pathway. The pivotal role of NO was confirmed by showing that NOS inhibition with L-NAME inhibited hepatic eNOS expression (Supporting Fig. 2B) and abrogated the effect of RLN on portal pressure (Fig. 3E) and splenomegaly (Supporting Fig. 2C). We went on to demonstrate that primary HSECs cultured from fibrotic human liver expressed RXFP1 (Supporting Fig. 2D) and increased NO production when treated with RLN (Supporting Fig. 2E).

Figure 3.

Effect of 72h RLN treatment on NO bioavailability and signaling in 8-week CCl4 rats. (A) Serum total nitrite (P = 0.9372) and (B) intrahepatic acetylated cGMP (**P = 0.0018) levels were compared after 72h subcutaneous RLN (0.5 mg/kg/day) or vehicle (veh; sodium acetate pH 5) (n = 6). (C) Relative p-eNOS protein by western blotting and densitometry (*P = 0.0125; n = 6). (D) Hepatic caveolin-1 (Cav1) mRNA by qRT-PCR (**P = 0.0059) and representative photomicrographs showing protein by immunohistochemistry with quantitation by morphometry (*P = 0.0253) (n = 6). (E) Effect of cotreatment with the NOS inhibitor L-NAME on portal pressure (*P < 0.05; n = 5-8).

RLN Treatment Down-Regulates Expression of Cytoskeletal Filament Proteins in 8-Week CCl4 Rats

Liver fibrosis is characterized by increased contractile filament expression and increased cytoskeletal tension. After 72h RLN, we observed a profound reduction in hepatic α-SMA protein by immunostaining (Fig. 4A) and western blotting (Fig. 4C). Desmin filaments run between anchorage points of the contractile apparatus in cells (the foci of mechanical stress) and are also a marker of the activated HSC-MF phenotype.[19] We also observed a marked reduction in desmin protein by immunostaining (Fig. 4B) and western blotting (Fig. 4D) following RLN. In contrast, glial fibrillary acidic protein (GFAP), predominantly a marker of quiescent HSCs, did not decrease significantly (Supporting Fig. 3A). Additionally, treatment of activated rat HSCs with a range of RLN doses in vitro had no effect on cell numbers (Fig. 4E). Indeed, we observed no difference in apoptotic markers in liver (TUNEL staining [Fig. 4F] and cleaved caspase-3 levels [Supporting Fig. 3B]), indicating that the observed decrease in hepatic α-SMA and desmin by RLN involved a specific reduction of fibrotic markers in these cells. We found no evidence for inhibition of hepatic Rho/Rho-kinase by RLN in our study (Supporting Fig. 4A,B).

Figure 4.

Effect of 72h RLN treatment on hepatic expression of cytoskeletal proteins in 8-week CCl4 rats. Representative photomicrographs of (A) α-SMA (*P = 0.026) and (B) desmin (P = 0.0422) immunostaining in liver after 72h subcutaneous RLN (0.5 mg/kg/day) or vehicle (veh; sodium acetate pH 5) with quantitation by morphometry (n = 6). Representative western blots showing (C) α-SMA and (D) desmin protein expression in whole liver extracts. (E) Effect of RLN on numbers of activated rat HSCs by MTS assay (P < 0.001; n = 4). (F) Representative image showing TUNEL (green) immunofluorescence from a vehicle-treated CCl4 rat liver with DAPI (blue) nuclear counterstain. Enlarged panels show TUNEL-positive cells within a fibrotic band (arrow) and in liver parenchyma (arrowhead) (n = 6).

The major profibrogenic cytokine responsible for initiation and perpetuation of HSC activation is transforming growth factor beta-1 (TGFβ1). After 72h subcutaneous RLN we showed reduced hepatic expression of TGFβ1 mRNA and protein and reduced transcription of other genes associated with TGFβ activity and signaling (Supporting Fig. 3C-E).

Acute Intravenous RLN Treatment Decreases Portal But Not Systemic Pressure in 8-Week CCl4 Rats

To exclude the impact of systemic hypotension on the apparent portal hypotensive effect of RLN, we measured the hemodynamic effects of intravenous RLN in 8 week CCl4 rats (Fig. 5A). Dose-response studies had previously been undertaken by our group (data not shown) and others.[15, 20] Rats were monitored for 180 minutes after injection of RLN (4 μg/mL over 2 minutes) or vehicle. RLN induced a rapid (<60 minutes) and robust reduction in portal pressure (mean change from baseline 28 ± 6% versus −6.2 ± 1.5%), an increase in heart rate (mean change from baseline 16 ± 3% versus −1 ± 2.5%), and modest increase in mean arterial pressure (MAP; mean change from baseline 20 ± 6.4 versus −6 ± 9.5%). Vehicle had no significant hemodynamic effects (Fig. 5A). There was no difference is spleen size between RLN and vehicle-treated rats in this acute model.

Figure 5.

Acute hemodynamic effects of RLN treatment in 8-week CCl4 rats. (A) Mean change in portal pressure (PP; **P = 0.0035), heart rate (HR; *P = 0.0324), and mean arterial pressure (MAP; P = 0.0869) from baseline after intravenous RLN (4 μg/mL) or vehicle (veh; sodium acetate pH 5) (n = 4). (B) Effect of RLN on portal venous blood flow by perivascular flowprobe (*P = 0.0352 baseline versus 60 minutes; n = 5). (C) Representative whole-animal fluorescence images in an RLN-treated CCl4 rat pre-ICG, 30 seconds and 20 minutes after tail vein ICG injection (2.5 mg/kg) with regions of interest indicated over liver (*), abdominal wall (background; #), right paw (red box). (D) Hepatic ICG uptake rate after pretreatment with 100 μL intravenous RLN (4 μg) or equivalent volume of vehicle (*P = 0.0303; n = 5). (E) BOLD-MRI showing change in R2* (1/T2*) after 1 mL intravenous vehicle (P = 0.3343 baseline versus 60 minutes; n = 5) and (F) after intravenous RLN (4 μg/mL) (*P = 0.0346 baseline versus 60 minutes; n = 4).

Acute Intravenous RLN Treatment Increases Portal Venous Blood Flow and Hepatic Oxygenation in 8-Week CCl4 Rats

Given its effects on intrahepatic NO, we went on to study portal blood flow and hepatic oxygenation in response to RLN. By perivascular flowprobe we demonstrated an increase in portal venous flow in response to RLN in 4 of 5 rats studied (Fig. 5B; baseline 12.8 ± 0.42 mL/min versus 60 minutes 15.0 ± 1.1 mL/min), but no significant change was observed in controls (data not shown). Injection of ICG with dynamic fluorescence imaging[21] was used to measure blood flow noninvasively (Fig. 5C,D; Supporting Fig. 5A). Hepatic uptake of ICG at 20 minutes was increased in rats pretreated with RLN compared to vehicle (Fig. 5D), as was peripheral disappearance rate, although the difference was not statistically significant (Supporting Fig. 5B). Assessment of tissue oxygenation by BOLD-MRI (Fig. 5E,F) showed that RLN decreased R2* (1/T2* = deoxyhemoglobin) after 60 minutes, whereas vehicle had no effect.

RLN Reduces Portal But Not Systemic Pressure and Improves Liver Function in Advanced Cirrhosis

We then confirmed that RLN mediated a portal hypotensive effect in advanced CCl4 cirrhosis and also in a mechanistically distinct model of fibrosis (BDL). Rats injured for 16 weeks with CCl4 had PHT but no significant ascites. Treatment with subcutaneous RLN for 72h reduced portal pressure (Fig. 6A; 6.1 ± 1.1 mmHg versus 9.12 ± 0.9 mmHg), but not spleen size (Supporting Fig. 6A), heart rate (Supporting Fig. 6B), or MAP (Fig. 6B). Serum albumin levels were increased following RLN (Supporting Fig. 6C; 27.4 ± 0.54 g/L versus 24.7 ± 0.67 g/L), but there was no difference in liver fibrosis (Fig. 6C) by CPA (Supporting Fig. 6D), or hepatocellular injury by serum ALT (Fig. 6A) and histology (data not shown).

Figure 6.

Effect of 72h RLN treatment in 16-week CCl4-treated and bile duct ligated rats. (A) Portal pressure (*P = 0.0414), (B) MAP (P = 0.8574), (C) Sirius red staining and serum ALT levels (P = 0.8808) were compared in 16-week CCl4 rats after 72h subcutaneous RLN (0.5 mg/kg/day) or vehicle (veh; sodium acetate pH 5) (n = 5). (D) Portal pressure (**P = 0.0067), (E) MAP (P = 0.3072), (F) Sirius red staining and serum alkaline phosphatase (ALP) levels (P = 0.8713) were compared in 3-week BDL rats after 72h subcutaneous RLN (0.5 mg/kg/day) or vehicle (veh; sodium acetate pH 5) (n = 5).

The livers of BDL rats were intensely scarred and inflamed and PHT was marked with most rats developing ascites. Treatment with 72h subcutaneous RLN reduced portal pressure (Fig. 6D; 8.8 ± 0.73 mmHg versus 14.13 ± 1.28 mmHg), although there was no significant difference in spleen:liver weight ratio (Supporting Fig. 7A), heart rate (Supporting Fig. 7B), or MAP (Fig. 6E). There were no significant differences in fibrosis (Fig. 6F; Supporting Fig. 7C), hepatocellular injury (Supporting Fig. 7D,E), serum alkaline phosphatase (ALP; Fig. 6F), or albumin (Supporting Fig. 7F).

RXFP1 Activation and Signaling in Human HSCs In Vitro

Functional activity of RXFP1 was investigated by cAMP accumulation assay following treatment with exogenous RLN. hHSCs exhibited a full cAMP dose-response curve with the preferential RXFP1 ligand H2-RLN (Fig. 7A). Transfection of hHSCs (and LX-2 cells, not shown) with RXFP1 siRNA inhibited the cAMP response to RLN (Fig. 7B). No significant effects were observed following treatment with RXFP1 siRNA in H2-RLN unstimulated forskolin-spiked assays (data not shown).

Figure 7.

Effects of RLN on human HSCs in vitro. (A) cAMP accumulation in hHSCs after stimulation with RLN for 90 minutes (cps = counts per second; n = 6). (B) Effect of RXFP1 siRNA pretreatment on RLN-induced cAMP accumulation (n = 3). (C) Effect of RLN (1 μM) on 10% serum-induced hHSC contraction by collagen gel contraction assay (*P < 0.05; n = 4) and (D) after pretreatment with RXFP1 siRNA (n = 2). (E) Representative photomicrographs showing hematoxylin and eosin (H&E)-stained hHSCs in collagen gel lattices after RLN and vehicle treatment. Arrows indicate long cellular processes. (F) Intracellular cGMP levels after 30 minutes treatment with 100 ng/mL RLN or vehicle (*P < 0.05 versus hHSCs incubated with vehicle/untreated hHSCs/DMSO control; n = 3).

Modulation of hHSC Contraction by RLN In Vitro Supports a Role for a Portal Hypotensive Effect in Humans

In collagen gel contraction assays, RLN reduced basal (10% serum-induced) hHSC contractility by ∼40% after 72h measured by final gel area (Fig. 7C). Visualization of hHSCs seeded into collagen lattices revealed morphological differences associated with RLN treatment, including a reduction in the number of cytoplasmic protrusions (Fig. 7E). Knockdown of RXFP1 with siRNA completely inhibited the cell relaxation effect of RLN, indicating that this was mediated by way of RXFP1 (Fig. 7D). Treatment with RLN also increased intracellular cGMP in hHSCs (Fig. 7F).

Exogenous RLN Induces an Antifibrotic Phenotype in Activated Human HSCs

Treatment of activated hHSCs with RLN for 72h decreased expression of profibrotic genes and increased expression of MMPs in an RXFP1-dependent manner (Supporting Fig. 8A,B). Additionally, RLN reduced TGFβ1 mRNA and protein expression (Supporting Fig. 8C), and inhibited basal and TGFβ-stimulated α-SMA protein expression in hHSCs (Supporting Fig. 8D), but had no effect on hHSC apoptosis (Supporting Fig. 8E).


In this study we show comprehensively that RLN induced a selective and significant portal hypotensive effect in the most relevant rodent model of sinusoidal PHT (CCl4),[22] through augmentation of intrahepatic NO signaling and down-regulation of contractile filament expression. Critically, reduction in portal pressure was also reproduced in advanced cirrhosis and a second mechanistically distinct model of fibrosis (BDL). Finally, we showed that activated hHSCs expressed functional RLN receptors in vitro and that stimulation with exogenous RLN decreased cellular contractility and induced an antifibrogenic phenotype. Therapeutic deployment of RLN in diseases such as cirrhosis that are characterized by fibrosis and pathological vasoconstriction is therefore an appealing approach. Increased IHVR, resulting from structural changes related to scarring and increased vascular tone mediated by dynamic contraction of hepatic MFs, is the primary cause of PHT in cirrhosis. As such, it is the ideal target for treatment.

The primary (H2-)RLN receptor, RXFP1, was undetectable in normal rat liver and quiescent HSCs but highly expressed in fibrotic liver and HSC-MFs. By subcutaneous infusion in cirrhotic rats we generated steady-state serum RLN concentrations comparable to midterm pregnancy levels, when gestational renal vasodilation is maximal,[23] and induced a portal hypotensive effect that would be clinically meaningful if translated in humans. Reduction in portal pressure was not due to regression of fibrosis or inflammation (both of which contribute to the mechanical component of PHT) or loss of HSC-MFs by apoptosis. Recent studies have demonstrated hepatic MF plasticity even in an advanced activated phenotype.[24] We conclude that RLN altered the fibrogenic state of MFs rather than influencing cell numbers, consistent with studies in renal and cardiac models.[25]

In experimental cirrhosis and human disease, impaired NO production by eNOS contributes to increased IHVR. Decreased expression of eNOS protein, decreased phosphorylation of eNOS by the serine-threonine kinase Akt, and inhibition of eNOS by enhanced expression of caveolin-1[26] underlies this endothelial dysfunction. Replenishing NO in cirrhotic liver by NOS gene transfer or drug therapy in rats and humans[27] has been shown to ameliorate PHT. In treating cirrhotic rats with RLN, we increased hepatic eNOS, Akt-dependent eNOS phosphorylation, and cGMP levels, and down-regulated caveolin-1 expression. Inhibition by L-NAME demonstrated the importance of this axis in mediating the portal hypotensive effect of RLN in vivo. Clinical application of nonselective NO donors such as nitrates in cirrhosis is complicated by impairment of systemic hemodynamics and subsequent risk of kidney dysfunction.[28] However, we showed that 72h subcutaneous RLN did not significantly increase systemic or portal venous nitrite levels and did not reduce MAP, even in advanced cirrhosis, indicating that RLN may represent a new liver-selective NO donor. Interestingly, RLN has been shown to preferentially dilate preconstricted vessels and vasodilatory effects were observed in small renal arteries of rats and mice but not in mesenteric or coronary septal arteries.[29]

We demonstrated a rapid and robust reduction in portal pressure following acute intravenous RLN with no effect on MAP. This is in agreement with rodent[30] and human[31] studies, where RLN elicited hemodynamic effects consistent with vasodilatation without inducing hypotension. The goal of therapy in PHT is to reduce portal pressure below critical thresholds without further deteriorating liver function. We showed that RLN infusion in early cirrhosis increased portal venous flow and tissue oxygenation while decreasing portal pressure, indicating a reduction in IHVR. Furthermore, in advanced CCl4 cirrhosis serum albumin levels were significantly increased.

We went on to underscore the potential for RLN in human disease by showing RXFP1 expression in explanted human cirrhosis specimens and activated hHSC-MFs, and confirmed H2-RLN as the cognate ligand for RXFP1 in these cells. Treatment with exogenous RLN in vitro inhibited HSC contractility and induced an antifibrogenic gene expression profile in an RXFP1-dependent manner. Compared to contraction, relatively little is known about the signaling pathways which regulate HSC or MF relaxation.[1] Both an increase in intracellular cAMP[32] and cGMP[6] have been shown to reduce HSC contraction. We showed that RLN increased intracellular cAMP and cGMP in hHSC-MFs. Interestingly, the observation that α-SMA expression in rat HSCs is regulated by NO and cGMP production[33] may provide a link between RLN induced intrahepatic NO generation and down-regulation of α-SMA protein in HSC-MFs in vivo. Indeed, we also showed that human HSECs isolated from fibrotic liver expressed RXFP1 in vitro and increased NO production in response to RLN treatment. Further work will define the HSEC response to RLN and crosstalk between HSECs and HSCs.

Our data demonstrate that short-term RLN treatment effectively down-regulates HSC-MF contractile filament expression and contractile function and exerts a significant portal hypotensive effect in vivo even in advanced cirrhosis models (Supporting Fig. 9). Although initial application in humans would likely involve short-term hemodynamic modulation (e.g., for variceal bleeding), it is tempting to speculate that longer-term targeting of RXFP1 might have dual effects on PHT through relaxation and deactivation of HSC-MFs, particularly as RXFP1 desensitization has not been reported to occur.[11] We have provided proof of concept for the use of RLN in cirrhotic PHT and other conditions associated with MF activation and pathological vasoconstriction and, given its excellent safety profile in clinical trials, support its evaluation in selected patients in early-phase studies.


We thank Will Mungall, Ross Lennen, Maurits Jansen, Cathy Payne and Deborah Mauchline (all University of Edinburgh) and Shishir Shetty (University of Birmingham, UK) for assistance; Hector Garcia (Hepatic Vascular Biology Lab, University of Barcelona, Spain), and Don Rockey (University of Texas Southwestern, USA) for technical advice; and Professor David Adams (University of Birmingham, UK) for access to human liver material. We thank Dr. Dennis Stewart (Corthera, USA) for supply of clinical grade recombinant human H2-RLN and placebo (vehicle).