Regulator of G-Protein Signaling-5 Is a Marker of Hepatic Stellate Cells and Expression Mediates Response to Liver Injury

Liver fibrosis is mediated by hepatic stellate cells (HSCs), which respond to a variety of cytokine and growth factors to moderate the response to injury and create extracellular matrix at the site of injury. G-protein coupled receptor (GPCR)-mediated signaling, via endothelin-1 (ET-1) and angiotensin II (AngII), increases HSC contraction, migration and fibrogenesis. Regulator of G-protein signaling-5 (RGS5), an inhibitor of vasoactive GPCR agonists, functions to control GPCR-mediated contraction and hypertrophy in pericytes and smooth muscle cells (SMCs). Therefore we hypothesized that RGS5 controls GPCR signaling in activated HSCs in the context of liver injury. In this study, we localize RGS5 to the HSCs and demonstrate that Rgs5 expression is regulated during carbon tetrachloride (CCl4)-induced acute and chronic liver injury in Rgs5LacZ/LacZ reporter mice. Furthermore, CCl4 treated RGS5-null mice develop increased hepatocyte damage and fibrosis in response to CCl4 and have increased expression of markers of HSC activation. Knockdown of Rgs5 enhances ET-1-mediated signaling in HSCs in vitro. Taken together, we demonstrate that RGS5 is a critical regulator of GPCR signaling in HSCs and regulates HSC activation and fibrogenesis in liver injury.


Introduction
Liver fibrosis and its sequelea of cirrhosis and hepatocellular carcinoma (HCC) are responsible for 29,000 deaths a year in the United States, making it the 12 th leading cause of death [1]. Injury to the liver results in a wave of cytokine mobilization [2], many of which are secreted by Kupffer cells, the liver's resident macrophages. These cytokines (e.g. tumor necrosis factor a (TNF-a) [3] and transforming growth factor-b1 (TGF-b) [4]) then activate HSCs, which deposit extracellular matrix (e.g. collagen) as part of the wound repair response. With chronic injury, ongoing inflammation and HSC activation results in accumulation of scar tissue and eventual decreased liver function [5].
We hypothesized that RGS5 controls liver injury via its ability to modulate GPCR-mediated signaling in activated HSCs. In this study, we localize expression of RGS5 to HSCs in the liver, and demonstrate that Rgs5 expression is regulated in both acute and chronic liver injury. Furthermore, mice lacking RGS5 expression develop increased hepatocyte damage and fibrosis in response to carbon tetrachloride (CCl 4 ). Rgs5 expression is regulated in cultured HSCs in response to fibrogenic agonists, and ET-1mediated signaling is potentiated in the absence of Rgs5 expression. Taken together, we demonstrate that RGS5 is a critical regulator of GPCR signaling in HSCs, and controls HSC activation and fibrogenesis in response to liver injury.

Animals
Rgs5 LacZ mice, which have a nuclear localized b-galactosidase reporter gene knocked into exon 2 of the Rgs5 locus and mice have been backcrossed to a C57BL/6 background, were purchased (Deltagen [45]). To induce acute liver injury, mice were injected (i.p.) with 10 ml/g body weight CCl 4 (Sigma-Aldrich) diluted 10% (v/v) in olive oil once. To induce chronic liver injury, mice were injected twice weekly for four weeks. Olive oil-injected animals served as controls for CCl 4 -injected mice. At the indicated time-points, mice were sacrificed using CO 2 inhalation. The Institutional Animal Care and Use Committee of the University of Washington, which is certified by the Association for Assessment and Accreditation of Laboratory Animal Care International, approved all experiments.

Measurement of relative hepatocyte cytoplasmic clearing
Approximately twenty 106 fields of H&E stained 5 mm paraffin embedded sections liver were imaged, each representing a 1.19x.89 mm area. Using the ImageJ software package, the relative cleared hepatocyte cytoplasmic area was measured. Images were automatically adjusted for contrast, converted to grey scale, and sharpened to enhance borders. The threshold tool isolated the tissue from slide background. Particle analysis included sizes 49 mm 2 to 324 mm 2 and circularity value between 0.3 and 1.0, yielding total particle area measured as a percentage of slide area. Data was processed by macro to remove potential bias.

Quantification of collagen deposition
To label collagen deposition (indicative of fibrosis), 5 mm paraffin embedded tissue sections were stained with picrosirius red (365548 and P6744, Sigma) and 0.1% Fast Green (F7258, Sigma) at room temperature for 30 minutes. The relative quantification of fibrosis was measured using ImageJ. Ten 46 fields were imaged for each liver, representing a 2.9562.21 mm area. Using the color threshold tool, red staining was isolated. The image is then converted to grey scale, and threshold is used to subtract remaining background. Measurement of the remaining area yielded the area of the slide with red staining. This was normalized to the area of the tissue (vessel lumen subtracted) for each slide to yield the percent of tissue with picrosirius red staining. Data was processed by macro to remove potential bias.

Cell Culture
The LX-2 HSC cell line was generously provided by Scott L. Friedman [47]. LX-2 cells were passaged in DMEM high glucose (Gibco/Life Technologies) supplemented with 2% FBS and penicillin-streptomycin and used between passages 10 and 20.

HSC isolation
Mouse primary HSCs were prepared by perfusion with collagenase/pronase and density centrifugation using Optiprep (Sigma) as reported previously [48].

RNA isolation and quantitative RT-PCR (qPCR)
RNA was isolated from LX-2 cells and mouse livers using the E.Z.N.A. Total RNA Kit I (Omega Bio Tek). For tissue lysates, 27 mm 3 were homogenized in TRK lysis buffer using a statorrotor homogenizer, as per kit instructions. cDNA was prepared by reverse transcription using the High-Capacity cDNA Reverse Transcription Kit; (Applied Biosystems). 20 ng cDNA was used in each qPCR reaction, using PerfeCTa SYBR Green FastMix (Quanta biosciences). Gene expression was calculated by the DDCt method: Fold expression = 2 2DDCt . Gene expression was normalized to Gapdh expression within each sample then normalized to individual control treatment conditions within each dataset.

Statistics
Quantitative data were analyzed by unpaired t-test in excel. A p-value of less than 0.05 was considered significant. Where indicated, significance was analyzed using a non-parametric Mann-Whitney U test.

Ethics statement
Mice were housed in a specific pathogen-free environment overseen by the Department of Comparative Medicine at the University of Washington with IACUC approval under protocol 4253-01.

RGS5 is expressed in HSCs
Using Rgs5 LacZ reporter mice, we localized the expression of RGS5 in the liver by X-gal staining (Fig. 1A). b-gal + cells, and therefore RGS5 + cells, are observed adjacent to liver sinusoids. As expected, a subset of vascular SMCs of large vessels (arrows, Fig. 1A) express both RGS5 and smooth muscle a-actin (SMA) (arrows, Fig. 1B) by IF, consistent with published findings [35]. Glial fibrillary acidic protein (GFAP) is expressed in HSCs [50,51], and co-labeling with the GFAP and the b-gal antibody demonstrates co-localization of the RGS5 reporter and GFAP in HSCs adjacent to sinusoids (arrows, Fig. 1C). Another protein expressed in HSCs, cellular retinol-binding protein-1 (CRBP1 [52]), also colocalizes with b-gal, confirming that these cells are HSCs (arrows, Fig. 1D). Von Willebrand Factor (VWF) is expressed in endothelial cells, including those in the liver sinusoids (LSECs) [53]. IF for VWF and b-gal demonstrates that LSECs are distinct from b-gal + cells (Fig. 1E); LSECs are VWF + , whereas the b-gal + cells are sparse and evenly distributed, and do not overlap with VWF + cells. Similarly, F4/80 + Kupffer cells and b-gal + cells are also distinct, with no co-localization observed (Fig. 1F). Confocal imaging for GFAP and b-gal confirms that the nuclei of GFAP + HSC are bgal + (arrows, Fig. 1G). No co-localization is observed in confocal imaging for CD31 + , another endothelial cell marker, and b-gal + cells (Fig. 1H). Finally, to verify that RGS5 expression is restricted to HSCs, we measured Rgs5 expression in primary HSCs isolated from WT mice [48]. We found that primary HSCs express high levels of Rgs5 relative to other non-parenchymal cell markers (Fig.  S1A, B). In summary, our co-localization analyses determined that RGS5 is specifically expressed in HSCs, but not in hepatocytes, LSECs, or Kupffer cells.

Increased RGS5 expression is associated with liver tumor and liver fibrosis
Multiple studies have demonstrated RGS5 expression in liver tumors [41,42]. Because these tumors are associated with HSC activation, we examined RGS5 expression in two mouse models of HCC, mice with hepatocyte-specific deletion of either tuberous sclerosis complex 1 (Tsc1) or phosphatase and tensin homolog (Pten). The loss of either of these genes results in disruption of the (PI3K)/AKT/mTORC1 pathway, leading to HCC or cholangiocarcinoma [54]. First we performed trichrome staining on Tsc1 fl/fl ; Alb Cre and Pten fl/fl ;Alb Cre liver sections spanning HCCs and adjacent non-tumor liver. Tsc1 fl/fl ;Alb Cre mice had collagen deposition in tumors (Fig. 2C), but not in adjacent non-tumor liver ( Fig. 2A), while Pten fl/fl ;Alb Cre mice had significant trichrome staining in both tumor and non-tumor liver (Fig. 2B, D). When we macro-dissected tumors from these mice and performed qPCR for Rgs5 expression (Fig. 2E, F), we found that Rgs5 expression mirrored collagen deposition: 1) Rgs5 is elevated in fibrotic Tsc1 fl/fl ; Alb Cre tumors but not adjacent non-fibrotic liver; 2) Rgs5 is elevated in both tumor and surrounding liver (both fibrotic) in Pten fl/fl ;Alb Cre mice. The up-regulation of Rgs5 expression in fibrotic liver tissue suggests that Rgs5 expression may be associated with HSC activation, which occurs both in liver injury and in HCC [55].

RGS5 expression is up-regulated in acute CCl 4 injury
Acute injection of CCl 4 leads to hepatocyte death and a subsequent injury response [56][57][58]. We found that acute CCl 4induced injury induces a 4-fold up-regulation of Rgs5 mRNA expression 48 hours after injection (Fig. 3A). Elevated Rgs5 expression correlates with the increased expression of additional markers of HSC activation and fibrosis in the murine liver, including Sma (Fig. 3B), Desmin (Fig. 3C), and Pdgfrb ( Fig. 3D) [15,59], which also peak at 48 hours post injury. Pdgfra ( Fig 3E) and Col1a (Fig 3F) are up-regulated at 48 hours, but peak at 72 hours post injury. Taken together, these data support the hypothesis that RGS5 is associated with fibrosis and HSC activation in both tumor and non-tumor associated fibrosis, and that the observed up-regulation of RGS5 expression in fibrotic liver tumors [41,42] may be due to tumor associated activated HSCs.

RGS5 expression is regulated in response to inflammatory and profibrotic stimuli in vitro
HSCs are activated in response to the growth factor and cytokine milieu released during the response to hepatocyte injury, including TNFa [60], TGFb [61], PDGF-BB [62], and ET-1 [63]. To determine whether RGS5 expression is affected by any of these factors, we stimulated LX-2 cells and assayed the expression of Rgs5 by qPCR. Rgs5 and endothelin receptor B (ET B ) expression were up-regulated by TNFa stimulation but inhibited by TGFb stimulation (Fig. 4) ET B is a marker of HSC activation [64], and activated receptors are rapidly internalized, serving as a sink for ET-1 agonists [65]. The simultaneous co-regulation of RGS5 and ET B expression makes sense, as ET B sequestering ET-1 agonist and RGS5 inhibiting Ga q -mediated signaling both result in the blockade of ET-1-mediated signal transduction. Regulation of RGS5 in response to cytokines released after hepatocyte injury could enable tunable control of ET-1-mediated signaling in HSCs.
RGS5 + HSCs participate in the response to hepatic injury As stated above, RGS5 expression is up-regulated in both genetically-induced HCC (Fig. 2) and acutely in response to CCl 4induced liver injury (Fig. 3). To investigate the role of RGS5 in mediating HSC activation and liver fibrosis, we induced liver injury (CCl 4 model, as above) in Rgs5 LacZ/LacZ mice to assess the changes in RGS5 cellular expression over time during injury. To identify RGS5 + HSCs during the injury response, we co-localized expression of GFAP and b-gal by IF. In uninjured liver tissue, HSCs are sparsely distributed throughout the liver of both Rgs5 +/+ and Rgs5 LacZ/LacZ mice (Fig. 5A,D). However, 48 hr post CCl 4 injury, HSCs are present in the necrotic foci (Fig. 5B). b-gal + HSCs are clustered in the necrotic foci, and are scarce in the uninjured parenchyma (Fig. 5E). At 96 hours, HSCs are tightly clustered at the foci of injury and are rare in the parenchyma (Fig. 5C, F). Co-localization analysis (Fig. S2) shows 75% of b-gal + cells are GFAP + , and that this ratio does not significantly change over the course of injury. The expression of RGS5 remains localized to GFAP + HSCs before and during injury (Fig. 5G-I), and its up-regulation correlates with expression of HSC activation markers (desmin and SMA) [15,59], peaking at 48 hr post injury (Fig. 3). Since RGS5 functions to inhibit GPCR signaling [25], the  up-regulation of RGS5 mRNA in HSCs during the liver injury response suggests a role in controlling GPCR mediated HSC activation.

RGS5 deficient mice have increased hepatocyte damage in response to acute liver injury
Given the induction of RGS5 expression after CCl 4 injection, we next used Rgs5 LacZ/LacZ mice to assess the functional consequences of loss of RGS5 expression. Uninjured liver appears histologically normal in Rgs5 LacZ/LacZ mice (Fig. 6A,D). At 48 hr post injury, foci of necrosis are visible around the central veins (Fig. 6B,E). At 96 hr post injury, necrotic hepatocytes undergo clearance and infiltrating cells remain at the site of injury (denoted by arrow in Fig. 6C&F). In Rgs5 LacZ/LacZ mice, necrotic foci are similar to wild-type mice. At 96 hr post injury, the foci of necrosis are diminished; however, hepatocytes throughout the liver have cleared cytoplasm and have a ballooned appearance (Fig. 6F). To determine if Rgs5 LacZ/LacZ mice are more susceptible to CCl 4induced liver damage than wild-type mice, we performed careful morphological analysis of H&E stained sections of liver tissue 96 hr after injury. Rgs5 +/+ mice have comparatively normal hepatocytes (Fig. 7A, B). In contrast, Rgs5 LacZ/LacZ mice demonstrated widespread hepatocyte ballooning (Fig. 7C, D). Furthermore, in Rgs5 LacZ/LacZ mice, the hepatocytes are characterized by cleared cytoplasm and central nuclei (Fig. 7D). Quantification of cleared hepatocyte area, using ImageJ, confirms a significant increase in injured hepatocyte area in Rgs5 LacZ/LacZ mice (Fig. 7E). Further, by analyzing the liver/body weight ratio in Rgs5 +/+ compared to Rgs5 LacZ/LacZ mice, we observe injured Rgs5 LacZ/LacZ livers are larger than uninjured livers, while Rgs5 +/+ liver/body weight ratios do not show an injury-induced increase in size (Fig. S3). Oil red-O staining shows that the cleared hepatocytes do not contain lipids (Fig. S4), nor accumulated glycogen (Fig. S5). TUNEL staining shows no difference in apoptosis between the genotypes, and no association with cleared hepatocytes (Fig. S6).
We propose that in the absence of RGS5 expression in HSCs, HSCs are increasingly sensitive to GPCR agonists, such as ET-1 and AngII, which contribute to HSC activation, portal hypertension, and the severity of fibrosis [14][15][16][17][18][19]. Excessive ET-1 or AngII signaling in Rgs5 LacZ/LacZ mice may induce inappropriate or excessive activation of HSCs, causing increased stress on hepatocytes during injury. Therefore, we compared the mRNA expression of HSC activation markers in Rgs5 +/+ and Rgs5 LacZ/LacZ mice in acutely injured liver tissue. Expression of desmin, endothelin-1 receptor B (ET B ), and PDGFRb [15,59] is increased in Rgs5 LacZ/LacZ mouse liver tissue relative to Rgs5 +/+ littermates at 48 hr post CCl 4 injury (Fig. 7F). The observed increase in expression of HSC activation markers may be attributed to an increase in HSC proliferation or an increase in HSC activation throughout the liver of Rgs5 LacZ/LacZ mice. Taken together these data suggest RGS5 controls HSC activation in acute liver injury.

RGS5 suppresses ET-1 signaling in HSCs
As stated above, the increased hepatocyte injury and fibrosis observed in the Rgs5 LacZ/LacZ mouse may be due to excessive signaling through GPCRs, since RGS5 inhibits Ga q signal transduction [25]. To elucidate the role of RGS5 in HSCs during activation and fibrosis, we utilized a human HSC cell line (LX-2 HSCs [47]) to investigate RGS5-mediated ET-1 signaling in vitro. LX-2 cells are transfected with RGS5 siRNA or non-specific siRNA, stimulated with ET-1, and the activation of MAPK signaling was determined. As shown in Fig. 8A and 8B, knockdown of RGS5 expression significantly increases ERK1/2 phosphorylation in response to ET-1 in LX-2 HSCs. Therefore, in the context of increased ET-1-mediated signaling during liver injury, the absence of RGS5 expression could further exacerbate fibrogenic signaling.

RGS5 expression is up-regulated in chronic liver injury
While acute injury models provide insight to the HSC response to damage, HSCs produce detectable fibrosis and scarring during chronic liver injury. Chronic exposure to CCl 4 in mice recapitulates the fibrosis observed in human chronic liver disease [58]. Rgs5 +/+ and Rgs5 LacZ/LacZ mice were repeatedly injected with CCl 4 over a period of 4 weeks. In damaged Rgs5 LacZ/LacZ liver tissue, SMA expression localizes activate HSCs in fibrotic septa (Fig. 9A). HSCs (GFAP + , bGAL + ) were associated with fibrotic septa bridging the portal veins, near the sites of collagen deposition (Fig. 9B). Interestingly, the b-gal reporter is limited to these GFAP expressing cells in the chronically injured liver. RGS5 mRNA expression is increased 8-fold in the chronic CCl 4 injured liver (Fig. 9C), while HSC activation markers desmin, SMA, PDGFRa, and PDGFRb are elevated, yet below the levels observed in acute CCl 4 injury (Fig. 3). Elevated expression of RGS5 in HSCs during the chronic CCl 4 injury indicates an ongoing role during fibrosis, while classic markers of HSC activation have subsided, suggesting RGS5's function may extend beyond controlling HSC activation.

RGS5 deficient mice have increased liver fibrosis during chronic injury
In the acute liver injury model, Rgs5 LacZ/LacZ mice were characterized by an increase in ballooning of hepatocytes (Fig. 7E) and increased liver/body weight ratio (Fig. S3), as compared to Rgs5 +/+ mice. To determine whether this phenotype persists in the chronic liver injury setting, hepatic fibrosis was visualized and quantified using picrosirius red. Fibular collagen is stained red, and the relative degree of fibrosis is quantitated by ImageJ (Fig. 10  A-D). Chronic CCl 4 -treated Rgs5 +/+ mice develop scar tissue bridging between portal veins (Fig. 10A), while oil-treated control mice are histologically normal, with minimal picrosirius red staining in Rgs5 +/+ (Fig. 10C) and Rgs5 LacZ/LacZ (Fig. 10D). In contrast, chronic CCl 4 -treated Rgs5 LacZ/LacZ mice develop severe fibrosis, as demonstrated by fibrotic septa bridging between portal veins and encircling hepatic lobules (Fig. 10B). Quantification of picrosirius red staining reveals significantly increased fibrosis in Rgs5 LacZ/LacZ mice relative to Rgs5 +/+ mice (Fig. 10E). Taken together with the finding that RGS5 is up-regulated in chronic injury (Fig. 9A), the increase in fibrosis in Rgs5 LacZ/LacZ mice suggests a RGS5-dependent role in controlling HSC-mediated collagen deposition.

Discussion
In this study, we have identified RGS5 as a marker of HSCs and a regulator of GPCR-mediated signaling in the liver. Upregulation of Rgs5 expression correlates with HSC activation during liver injury, and Rgs5 is highly expressed in the fibrotic liver. RGS5 deficient mice develop more severe liver injury following acute CCl 4 exposure, and increased fibrosis after chronic CCl 4 administration. In vitro, profibrotic and pro-inflammatory mediators regulate Rgs5 expression in HSCs, and RGS5 knockdown in HSCs resulted in increased ERK1/2 signaling in response to ET-1, a possible mechanism for its effects in the liver. Taken together, these data suggest that the regulation of RGS5 in HSCs allows for tunable sensitivity to ET-1 signaling following hepatic injury. Loss of RGS5 disrupts this fine level of control, leading to increased HSC activation, hepatic injury, and fibrosis.
The Rgs5 LacZ reporter mouse provides a robust and sensitive method to specifically localize RGS5 expression in vivo. Colocalization of b-galactosidase activity and traditional HSC markers (GFAP [50] and CRBP1 [66]) in these mice establishes RGS5 as a marker of HSCs in the liver. RGS proteins have been implicated in the progression of HCC [42,67], and an earlier study localized hepatic Rgs5 expression to endothelial cells in HCC [41] using in situ hybridization (ISH). However, the nature of ISH makes precise localization and discrimination between adjacent cells difficult. IF labeling of endothelial cells with an antibody to detect VWF confirms that RGS5 + HSCs are distinct from endothelial cells. A recent publication correlates RGS5 expression with increased vascular invasion, tumor recurrence, and decreased survival in patients with HCC [42]. However, HSCs are the major stromal cell in the tumor microenvironment, and promote HCC proliferation and invasion [68][69][70]. The correlation of increased RGS5 expression with decreased survival may reflect the level of HSC activation within the tumor, therefore predicting tumor metastasis and proliferation. High RGS5 expression has also been found in select patient-derived HCC cell lines, suggesting expression by transformed hepatocytes. These data should be interpreted with caution, however, as high passage number in culture conditions containing serum and associated growth factors may select for a phenotype that does not reflect in vivo expression. We therefore propose that the specific co-localization of b-gal expression (and therefore RGS5 expression) with markers of HSCs, and not with markers of other non-parenchymal cells, establishes RGS5 as a marker of HSCs.
We observed that RGS5 deficient mice had widespread hepatocyte clearing after a single CCl 4 injection, while this phenotype was not observed in Rgs5 +/+ mice (Fig. 7). The hepatocyte cytoplasmic clearing observed after acute CCl 4 treated Rgs5 LacZ/LacZ mice is similar to ballooning observed in human cases of non-alcoholic fatty liver disease (NAFLD). Conflicting evidence suggests RGS5 plays a role in maintaining body weight and steatosis, with one group reporting that Rgs5 2/2 mice exhibit spontaneous hepatic steatosis and obesity [71] while another study demonstrates RGS5 2/2 mice have low body weight [34]. Although RGS proteins have been implicated in the control of hepatic fatty acid oxidation [72] and homeostasis [73], we did not observe any differences in baseline mouse body weight or oil red-o staining (Fig. S3, S4) after CCl 4 injection, indicating that the ballooning is not due to lipid accumulation. The differences in our findings may be due to differences in mouse strains or housing conditions used in the different institutions, and the overall effects of RGS5 expression on body weight are unclear. Glycogen storage can also induce ballooning in hepatocytes, but periodic acid-schiff staining showed no difference between Rgs5 LacZ/LacZ mice and wild type littermates (Fig. S5). Hepatocyte swelling reflects reversible cell injury [74] in Rgs5 LacZ/LacZ mice, and it has been suggested that ballooning is protective in hepatotoxicity [59,75]. To explore this possibility, TUNEL staining was conducted to determine whether ballooned hepatocytes were undergoing apoptosis. However, there was minimal staining in hepatocytes 96 hrs after CCl 4 injection, and no difference was observed between Rgs5 LacZ/LacZ mice and wild type littermates (Fig. S6). We similarly did not appreciate a difference in liver necrosis between the genotypes, indicating that the centrilobular zonal expression of cytochrome p450, and thus CCl 4 metabolism, is unchanged in Rgs5 LacZ/LacZ mice. Further, Rgs5 LacZ/LacZ mice survive repeated doses of CCl 4 in the chronic liver injury model, suggesting that the hepatocyte swelling is reversible.
Feathery hepatocyte degeneration is commonly observed in cholestasis [76] and is characterized by enlarged periportal hepatocytes with a flocculent cytoplasm. Intrahepatic cholestasis due to endotoxemia or drug toxicity can induce feathery degeneration; however, we did not observe pigmented hepatocytes in Rgs5 LacZ/LacZ mice, and the hepatocyte swelling was equally distributed throughout the liver lobule. Whether hepatocyte swelling enhances hepatocyte survival or is simply a surrogate for more severe injury, Rgs5 LacZ/LacZ mice show extensive hepatocyte swelling after a single injection of CCl 4 , while Rgs5 +/+ mice have normal histology.
Potential mechanisms for the increased injury observed in Rgs5 LacZ/LacZ mice include portal hypertension or increased activation of HSCs. ET-1 and AngII-mediated sinusoidal constriction by HSCs induces portal hypertension [19,77,78]; loss of RGS5 expression enhances GPCR signaling [31] in response to these agonists, and may thus result in sinusoidal constriction. AngII infusion exacerbates fibrosis in rodent models of fibrosis [14] and ET-1 infusion increases HSC contraction and portal hypertension [79]. These effects are similar to what is observed in the Rgs5 LacZ/LacZ mice during CCl 4 injury. Loss of VEGF induces sinusoidal capillarization, portal hypertension and HSC activation [80] in mice. However no parenchymal injury was observed, invalidating portal hypertension as a potential mechanism.
The increased hepatocyte injury in Rgs5 LacZ/LacZ mice may be due to enhanced HSC activation via GPCR signaling. ET-1mediated control of HSC activation has been demonstrated in previous studies, in which pharmaceutical inhibition of AngII and ET-1 reduces HSC activation in vitro [13,17] and in rat models of cirrhosis [15,19]. Our findings of increased expression of HSC markers (desmin, ET B , and PDGFRb) in Rgs5 LacZ/LacZ mice are consistent with this hypothesis, as removal of ET-1 inhibition would enhance expression of markers of HSC activation. ET-1 induced TGFb secretion [81,82] would be expected to increase in the absence of RGS5 expression, further activating HSCs and contributing to enhanced fibrosis after chronic CCl 4 administration. This hypothesis could be further confirmed by rescue of the RGS5-null phenotype through ET-1 antagonism. For instance, the ET A antagonist BQ-123 and the ET B antagonist BQ-788 could be administered to Rgs5 LacZ/LacZ mice during CCl 4 injury, and would be expected to mitigate the hepatocyte swelling and increased fibrosis that we found in these mice.
RGS5-deficient mice exhibit enhanced arterial hypertrophy and perivascular fibrosis in a hypertension-induced vascular injury model [83]. This pathogenic remodeling was attributed to enhanced MEK/ERK and Rho kinase (ROCK) signaling via increased AngII-induced Ga q signaling in RGS5-null mice. Enhanced ERK activity due to RGS5 knock-down has been previously reported in aortic SMCs [31], and we observed enhanced ERK signaling in LX-2 cells following RGS5 siRNA treatment (Fig. 8A,B). ET-1 induced Rho activation in HSCs has been shown to enhance migration in vitro [84], and inhibition of ROCK improves fibrosis in choline deficient diet fed rats [85]. Loss of RGS5-mediated inhibition of ROCK could be one mechanism behind the enhanced HSC activation and fibrosis observed in Rgs5 LacZ/LacZ mice. RGS5 inhibition of ET-1 signaling has potential in the treatment of liver fibrosis and cirrhosis. While animal models of cirrhosis benefit from ET-1 antagonism, patients risk liver failure [86] due the hepatotoxic effects these drugs. RGS5 inhibition of ET-1 signaling specifically in HSCs provides a novel avenue for future therapies. Enhancing the expression of RGS5 in HSCs could reduce their activation and subsequent development of fibrosis and cirrhosis. The anti-fibrotic effects of RGS5 overexpression could be validated with studies of chronic CCl 4 administration to mice in which RGS5 expression is conditionally overexpressed in HSCs. High RGS5 expression would be expected to block ET-1 mediated signaling in HSCs, potentially reducing HSC activation, contraction, migration, and ultimately fibrosis.

Conclusions
RGS5 is a marker of HSCs that is up-regulated as HSCs respond to injury. RGS5 controls HSC activation via inhibition of ET-1 signaling and reduction of ERK activation. The critical role of RGS5 in liver injury is demonstrated by enhanced HSC activation, hepatocyte ballooning, and fibrosis in CCl 4 treated RGS5-deficient mice. RGS5-mediated control of GPCR signaling in the liver is a novel mechanism by which HSC activation can be controlled, and a potential target of therapeutic intervention for liver fibrosis. Figure S1 RGS5 is expressed in freshly isolated primary HSCs. 1u HSC have astrocyte-like morphology. A. Isolated 1u HSCs have an astrocyte-like phenotype and store vitamin A in lipid droplets. B. Isolated 1u HSCs are characterized by high RGS5 expression and low Tie1 and CD68 expression, indicating there are few contaminating cells. Low PDGFRb expression demonstrates the 1u HSCs are in a quiescent state. (TIF) Figure S2 Co-localization of b-gal + nuclei and GFAP staining does not change during injury. Frozen sections of CCl 4 injected mouse liver were immunofluorescently labeled with antibodies for b-gal and GFAP. Co-localization analysis using Figure 10. Rgs5 LacZ/LacZ mice have increased fibrosis during chronic liver injury. Mice were injected with CCl 4 (or oil), twice weekly for 4 weeks. Formalin fixed, paraffin embedded liver sections were stained with picrosirius red to assess fibrosis. A. Rgs5 +/+ mice show periportal fibrosis and bridging fibrosis between portal veins. B. Rgs5 LacZ/LacZ mice show severe periportal fibrosis bridging between portal veins and surrounding lobules. C. Rgs5 +/+ and D. Rgs5 LacZ/LacZ oil injected mice have minimal fibrosis. Scale bars are 500 mm. E. ImageJ quantitation of total picrosirius red staining in chronic CCl 4 injury shows significantly more fibrosis in Rgs5 LacZ/LacZ mice, compared to Rgs5 +/+ mice. n = 6, * = p,0.05 by Mann-Whitney U test. doi:10.1371/journal.pone.0108505.g010

Supporting Information
ImagePro showed that the fraction of b-gal + cells that are b-gal + and GFAP + does not significantly change during the course of injury. n = 3 mice per group. (TIF) Figure S3 Liver weight to body weight ratio is moderately elevated in Rgs5 lacZ/lacZ mice at 96hr post CCl 4 injury. Body and liver weights of CCl 4 injected mice were measured at time of sacrifice. Rgs5 LacZ/LacZ mice had elevated liver to body weight ratio 96 h post injury. Rgs5 +/+ liver to body weight ratio was not significantly different from untreated mice. * p = .001, { p = .06, n = 4-6.