The relaxin family peptide receptor 1 (RXFP1): An emerging player in human health and disease

Abstract Background Relaxin/relaxin family peptide receptor 1 (RXFP1) signaling is important for both normal physiology and disease. Strong preclinical evidence supports relaxin as a potent antifibrotic molecule. However, relaxin‐based therapy failed in clinical trial in patients with systemic sclerosis. We and others have discovered that aberrant expression of RXFP1 may contribute to the abnormal relaxin/RXFP1 signaling in different diseases. Reduced RXFP1 expression and alternative splicing transcripts with potential functional consequences have been observed in fibrotic tissues. A relative decrease in RXFP1 expression in fibrotic tissues—specifically lung and skin—may explain a potential insensitivity to relaxin. In addition, receptor dimerization also plays important roles in relaxin/RXFP1 signaling. Methods This review describes the tissue specific expression, characteristics of the splicing variants, and homo/heterodimerization of RXFP1 in both normal physiological function and human diseases. We discuss the potential implications of these molecular features for developing therapeutics to restore relaxin/RXFP1 signaling and to harness relaxin's potential antifibrotic effects. Results Relaxin/RXFP1 signaling is important in both normal physiology and in human diseases. Reduced expression of RXFP1 in fibrotic lung and skin tissues surrenders both relaxin/RXFP1 signaling and their responsiveness to exogenous relaxin treatments. Alternative splicing and receptor dimerization are also important in regulating relaxin/RXFP1 signaling. Conclusions Understanding the molecular mechanisms that drive aberrant expression of RXFP1 in disease and the functional roles of alternative splicing and receptor dimerization will provide insight into therapeutic targets that may restore the relaxin responsiveness of fibrotic tissues.


| RELAXIN/RXFP1 SIGNALING
Relaxin is a heterodimeric peptide hormone with a twochain structure (Wilkinson et al., 2005). It was first identified by Frederick Hisaw in a guinea pig model of pregnancy and parturition (Fevold, Hisaw, & Meyer, 1930;Hisaw, 1926). Relaxin was observed to loosen pelvic ligaments to facilitate parturition by reducing the density of collagen bundles and relaxing the collagen fibers (Chihal & Espey, 1973;Hisaw, 1926;Wilkinson et al., 2005). Additional roles of relaxin/RXFP1 signaling axis were identified in many physiological processes including development of mammary nipples and vaginal epithelium in mice (Kaftanovskaya et al., 2017), cervix growth during pregnancy in rats and pigs (Burger & Sherwood, 1998;Huang, Li, & Anderson, 1997), growth of vagina and uterus in pregnant pigs (Min, Hartzog, Jennings, Winn, & Sherwood, 1997), new blood vessel formation and endometrial connective tissue maintenance in early pregnancy of rhesus monkeys (Goldsmith et al., 2004), and improvement of spermatozoan motility (Lessing et al., 1986).

| EC region
The EC region of RXFP1 consists of an N-terminus and three extracellular loops (ECL1-ECL3) (Venkatakrishnan et al., 2013). ECLs link TM segments and contribute to ligand binding, TM positioning, and activation of GPCRs (Palczewski et al., 2000;Wheatley et al., 2012). Three protein domains are identified in the EC region: an LDLa module, a linker domain, and a LRR domain (Hoare et al., 2019).
The LDLa was first described and characterized in LDL receptor and was subsequently identified in other proteins with diverse biological functions (Brown & Goldstein, 1986;Hopkins, Bathgate, & Gooley, 2005). It contains three disulfide bonds and requires a bound calcium ion for its correct folding and stabilization (Hopkins et al., 2005). Although relaxin does not bind to LDLa directly, the binding of relaxin to RXFP1 stabilizes the LDLa/linker structure that leads to the direct contact between the EC and TM region (Diepenhorst et al., 2014;Hoare et al., 2019;Sethi et al., 2016). Removing LDLa module from RXFP1 abolished the ligand-activated receptor signaling (Scott et al., 2006). Mutagenesis introduced in the LDLa module altered its native three-dimensional structure (Hopkins et al., 2005;Koduri & Blacklow, 2001;Varret et al., 1997) and fully disrupted receptor activity (Hopkins, Layfield, Ferraro, Bathgate, & Gooley, 2007).
There is a 32-residue linker hitching the N-terminus LDLa module and the LRR domain together in RXFP1 (Sethi et al., 2016). This helically shaped linker provides a binding site essential for the steady binding of the relaxin A-chain (Scott et al., 2006;Sethi et al., 2016). Although mutations in the linker residues have not been shown to affect receptor trafficking or G-protein coupling, profound effects on reducing relaxin binding and decreasing cAMP response were observed (Sethi et al., 2016).
Different from most of the class A GPCR ligands that bind to the TM region directly, relaxin binds to the RXFP1 through the primary ligand binding site in the LRR domain of the EC region (Hoare et al., 2019). A shallow curvature structure formed by the 10 LRRs is predicted to serve as the primary high-affinity relaxin binding site (Petrie, Lagaida, Sethi, Bathgate, & Gooley, 2015). The LRR domain potentially interacts with the linker after relaxin binding (Petrie et al., 2015;Scott et al., 2006). When relaxin binds to the LRR, it induces a conformational change of the receptor to position the LDLa module for interacting with the TM region (Hopkins et al., 2007).

| TM region
In addition to the high-affinity binding site in the EC region, there is a low-affinity relaxin binding site in the TM region of RXFP1 (Halls et al., 2005). The TM regions of GPCRs form the main structural core of the receptor with seven α-helices (TM1-TM7) folded together. Conformational changes of different TMs are important for transducing the ligand/receptor interaction to the IC region (Venkatakrishnan et al., 2013). Mutations in the TM region affected receptor conformational selectivity and ligand-binding affinity in vitro (Dore et al., 2011;Heitz et al., 1999). Two single amino acid changes in the TM6 resulted in dose-dependent increases of cAMP production (Hsu et al., 2000).

| IC region
The IC region of GPCR interfaces with cytosolic signaling proteins. It includes three intracellular loops (ICL1-ICL3), an intracellular amphipathic helix, and a unique C-terminal tail containing a phosphorylation site (Hsu et al., 2000;Scheerer et al., 2008). The C-terminal half of ICL3 plays an important role in linking the relaxin-activated RXFP1 receptor with G protein (Shpakov et al., 2007).
In summary, activation of RXFP1 by relaxin is a complex multistep process. Relaxin binding initiates RXFP1 signaling. However, the completion of RXFP1-dependent signal transduction requires the interactions between different receptor regions and correct conformation of the ligand/receptor complex to initiate downstream IC signaling (Sethi et al., 2016).

| Truncated N-terminus RXFP1 retaining both LDLa module and linker domain
In contrast to the splicing variants that retain the LDLa module but lack the linker domain, three of the known splicing variants encoded truncated RXFP1 proteins retain both the LDLa module and the linker domains (Figure 1c) (Kern et al., 2008;Muda et al., 2005). These variants were identified initially in the human fetal membrane and placental tissues and encode RXFP1 proteins lacking the majority of the LRR domain, the TM, and the IC regions (Kern et al., 2008;Muda et al., 2005). One of them (LGR7.1) has two novel exons after exon 6 (exon 6A) and exon 15 (15A) (Muda et al., 2005). Exon 15A contains an alternative poly-A signal while 6A has a premature stop codon (Muda et al., 2005).
LGR7.1 is translated into a RXFP1 protein containing the N-terminus region with only two LRRs and 10 nonhomologous amino acids at the C-terminus of the truncated protein (Muda et al., 2005). Kern and colleagues have also identified and characterized two splicing variants (LGR7-D and LGR7-F) that encode similarly truncated RXFP1 proteins as LGR7.1 (Kern & Bryant-Greenwood, 2009;Kern et al., 2008). The LGR7-D protein is encoded by a splicing variant lacking exon 6 through exon 15 generated by cryptic splice sites and contains the LDLa module, one LRR, and 25 nonhomologous amino acids (Kern et al., 2008). The LGR7-F is a result of an alternative splicing of exon 6 to exon 18 with cryptic splicing sequences (Kern et al., 2008). It encodes a RXFP1 protein containing the N-terminus part up to and including the first two LRRs and 10 nonhomologous amino acids at the end (Kern et al., 2008). The LGR7.1 is expressed in different tissues (Muda et al., 2005). Direct comparison of expression levels in the placenta demonstrated much lower levels of the LGR7-D and LGR7-F compared to the WT RXFP1 (Kern et al., 2008). Functional analysis revealed that the LGR7.1 and LGR7-F are predominantly retained within cells (Kern et al., 2008;Muda et al., 2005), while the LGR7-D is expressed intracellularly and on the cell surface (Kern et al., 2008). When co-expressed with WT in HEK-293 cells, both LGR7-D and LGR7-F colocalized with WT RXFP1 within the cells and reduce RXFP1-mediated cAMP accumulation (Kern et al., 2008). In addition, these two truncated variants have dominantly negative effects in WT RXFP1 maturation, homodimerization in the endoplasmic reticulum, and cell surface expression (Kern et al., 2008).

| Truncated N-terminus RXFP1 retaining LDLa module, linker domain, and majority of LRRs
Another splicing variant (LRP7-C) misses the TM and IC regions but retains 8 of the 10 LRRs (Figure 1d) (Kern et al., 2008). The LGR7-C is a result of alternative splicing between exon 12 and exon 18 that creates a novel stop codon at the beginning of exon 18 (Kern et al., 2008). Although it contains 8 LRRs, the LGR7-C is mainly retained inside the cells and has a similar function as the three truncated N-terminus RXFP1 retaining only 1 or 2 LRRs (Figure 1c).

| RXFP1 variants with in-frame deletions
Two splicing variants of RXFP1 result from in-frame deletions and have been characterized (Figure 1e) (Hsu et al., 2000;Muda et al., 2005). One variant skips exon 3 [LGR7.10 based on Muda et al. (2005), LGR7(2) based on Hsu et al. (2000) or LGR7-Short based on Scott et al. (2006), Scott, Tregear, et al. (2005) and will be referred as LGR7.10 in this review] and the other skips both exon 12 and 13 (LGR7.2) which is different from the LGR7(2) mentioned above (Hsu et al., 2000) (Muda et al., 2005). The LGR7.10 is detected in the ovary, pituitary, placental, prostate, and uterus tissues and encodes a RXFP1 with an in-frame deletion of the linker region (Hsu et al., 2000;Muda et al., 2005). When LGR7.10 or LGR7.2 are overexpressed in HEK-293T cells, only the LGR7.10 was detected on the cell surface but at a very lower level compared to the WT (Muda et al., 2005). The LGR7.2 lost its responsiveness to relaxin and relaxin binding (Muda et al., 2005). Specific binding of relaxin to the LGR7.10 was not detected in a study reported by Muda et al., however, a later study demonstrated specific relaxin binding to this RXFP1 variant in HEK-293T cells (Muda et al., 2005;Scott et al., 2006).
In summary, all characterized RXFP1 splicing variants have been shown to lose their ability to activate relaxin-dependent cAMP accumulation. Seven of the nine splicing variants have been shown to interfere with cAMP accumulation mediated by WT RXFP1 signaling. Given the large size of the RXFP1 gene and the numbers of coding exons, we speculate that tissue-specific splicing variants will be discovered in the future. The differential tissue expression and antagonistic (dominant-negative) function of these splicing variant receptors suggest that complex posttranscriptional regulation of RXFP1 gene may play important roles in spatial and temporal expression and signaling of relaxin/RXFP1 (Halls, van der Westhuizen, Bathgate, & Summers, 2007).
Are splice variants associated with disease states? Overexpression of LDLa module of RXFP1 in prostate cancer cells resulted in a decrease in proliferation, soft agar colony formation, adhesion and invasion in vitro, and tumor growth in mouse model (Feng & Agoulnik, 2011). These findings suggest that alternative splicing variants retaining different domains of the RXFP1 protein may modulate relaxin function in cancer. In addition, the formation of both GPCR homodimer and heterodimer contributes to the complexity of GPCR signaling (Angers, Salahpour, & Bouvier, 2002). RXFP1 forms a homodimer when it is transported from the ER to the cell membrane and negative cooperativity occurs when it forms a heterodimer with RXFP2 Svendsen et al., 2009;Svendsen, Zalesko, et al., 2008). Therefore, dimerization with other receptors or RXFP1 splicing variants may play important roles in normal RXFP1 function and in diseases.

| Lung and skin fibrosis
SSc is a group of heterogeneous disorders characterized by varying degrees of fibrosis of the skin and internal organs (Haustein, 2002;Silman, 1997). Lung fibrosis is one of the most common manifestations and is a major cause of SSc-related mortality (Denton, Wells, & Coghlan, 2018). The protective role of relaxin signaling in lung fibrosis has been demonstrated in relaxin knockout mice Unemori et al., 1996). Similarly, the RXFP1-null mice develop early onset peribronchiolar and perivascular fibrosis compared to the relaxin-null mice (Kamat et al., 2004). Relaxin has been tested in SSc patients as early as 1958 with no beneficial effects (Casten & Boucek, 1958;Jefferis & Dixon, 1962). A smaller study with relaxin showed some efficacy in reducing skin fibrosis (Seibold et al., 2000) which was not validated in a large clinical trial with SSc patients (Khanna et al., 2009). RXFP1 protein expression in fibrotic lung and skin of SSc patients is dramatically reduced. RXFP1 is similarly downregulated in SSc lung and skin fibroblasts (Corallo et al., 2019;Giordano et al., 2012;Tan et al., 2016). Increased relaxin in peripheral blood was also reported in SSc patients (Giordano et al., 2005). However, the relative reduction of RXFP1 expression in fibrotic tissues may potentially render these tissues insensitive to relaxin. Interestingly, bulk RNA sequencing of SSc skin fibroblasts detected upregulation of 13 different mRNA isoforms without detectable expression of RXFP1 protein in these cells (Corallo et al., 2019). This study supports that the splicing variants of RXFP1 may be important regulators of RXFP1 expression in different fibrotic diseases.
Idiopathic pulmonary fibrosis (IPF) is a progressive disease with an average survival of 2.5 years (King, Pardo, & Selman, 2011). Patients with IPF or other forms of interstitial lung disease may have better pulmonary function if their lung-specific RXFP1 expression is higher (Tan et al., 2016). In the bleomycin lung fibrosis mouse model, treating with a relaxin-like agonist reduced bleomycin-induced collagen deposition in vivo (Pini et al., 2010;Tan et al., 2016). Most notable, RXFP1 expression is dramatically decreased in both lung tissues and lung fibroblasts of IPF patients (Tan et al., 2016). In vitro, silencing of RXFP1 expression was associated with insensitivity to exogenous relaxin, which could be reversed by enhancement of RXFP1 expression in IPF lung fibroblasts (Tan et al., 2016). The findings in both SSc and IPF support that the lack of or reduced expression of RXFP1 in fibrotic tissues of IPF and SSc contributes to the failed responses to relaxin for IPF lung fibroblasts in vitro and relaxin-based therapies in SSc clinical trials (Casten & Boucek, 1958;Jefferis & Dixon, 1962;Khanna et al., 2009;Tan et al., 2016).
What drives downregulation of RXFP1? Reduction of RXFP1 mRNA suggests that transcriptional mechanisms may account for this. TGFβ decreases expression of RXFP1 at the level of mRNA (Bahudhanapati et al., 2019;Corallo et al., 2019;Tan et al., 2016). Our group recently reported that mi-croRNA-144-3p (miR-144-3p) regulates RXFP1 in fibrotic lung fibroblasts (Bahudhanapati et al., 2019). MiR-144-3p is upregulated in IPF fibroblasts compared with control donor lung fibroblasts. Overexpression of a miR144-3p mimic and anti-miR144-3p in the donor lung fibroblasts resulted in the down-and upregulation RXFP1, respectively. Interestingly, Yong and colleagues have also demonstrated that knocking down RXFP1 gene by a synthetic microRNA resulted in a loss of relaxin responsiveness of human dermal fibroblasts (Yong, Callander, Bergin, Samuel, & Bathgate, 2013). In addition to the micro-RNA regulation, RXFP1 may be regulated by transcription factors important in fibrotic diseases. Therefore, abnormal regulation of RXFP1 expression in fibrotic lung and skin tissues is a therapeutic target for reversing tissue fibrosis.

| Liver fibrosis
The major cause of liver fibrosis is the over activation of hepatic stellate cells and their transformation into myofibroblast-like cells after liver damage (Williams et al., 2001). In the rat carbon tetrachloride model, relaxin increases intrahepatic NO level and reduces hepatic expression of profibrotic markers and portal pressure (Fallowfield et al., 2014). A phase II randomized open-label clinical study of serelaxin in patients with alcohol-related liver cirrhosis and portal hypertension was reported (Snowdon et al., 2017). The small molecule relaxin agonist, ML290, also shows antifibrotic effects in an in vitro liver organoid model and an in vivo liver fibrosis mouse model (Kaftanovskaya et al., 2019). Interestingly, unlike skin and lung fibrosis, dramatically increased hepatic expression of RXFP1 has been observed in a rat model of liver cirrhosis and-in contrast to lung and skin--higher expression of RXFP1 is correlated with increased liver fibrosis in human (Fallowfield et al., 2014;McBride et al., 2017;Nagorniewicz et al., 2019). However, whether the upregulation RXFP1 is related to the alternative splicing transcripts or protein variants is not known. The regulation of RXFP1 in liver fibrosis may be fundamentally different from that in lung, skin, and other fibrotic organs.

| CONCLUSIONS
Relaxin/RXFP1 signaling is important in both normal physiology and in human diseases. Reduced expression of RXFP1 in fibrotic lung and skin tissues surrenders both relaxin/ RXFP1 signaling and their responsiveness to exogenous relaxin treatments. Several questions remain. These include how splice variants of RXFP1 regulate expression and relaxin sensitivity. Understanding the molecular mechanisms that drive aberrant expression of RXFP1 in disease will provide insight into therapeutic targets that may restore the relaxin responsiveness of fibrotic tissues.