fax: 0049 241 8082512.
The Marburg I variant (G534E) of the factor VII–activating protease determines liver fibrosis in hepatitis C infection by reduced proteolysis of platelet-derived growth factor BB†
Article first published online: 24 OCT 2008
Copyright © 2008 American Association for the Study of Liver Diseases
Volume 49, Issue 3, pages 775–780, March 2009
How to Cite
Wasmuth, H. E., Tag, C. G., Van de Leur, E., Hellerbrand, C., Mueller, T., Berg, T., Puhl, G., Neuhaus, P., Samuel, D., Trautwein, C., Kanse, S. M. and Weiskirchen, R. (2009), The Marburg I variant (G534E) of the factor VII–activating protease determines liver fibrosis in hepatitis C infection by reduced proteolysis of platelet-derived growth factor BB. Hepatology, 49: 775–780. doi: 10.1002/hep.22707
Potential conflict of interest: Nothing to report.
- Issue published online: 24 FEB 2009
- Article first published online: 24 OCT 2008
- Accepted manuscript online: 24 OCT 2008 12:00AM EST
- Manuscript Accepted: 9 OCT 2008
- Manuscript Received: 3 JUL 2008
- Deutsche Forschungsgemeinschaft. Grant Numbers: WA 2557/1-1, WE2554/4-1, SFB/TRR57
- University Hospital in Aachen
- SFB 542
Genetic risk factors play an important role for the progression of liver fibrosis in chronic hepatitis C virus (HCV) infection, but functional data on specific alleles and their related proteins are limited. Platelet-derived growth factor BB (PDGF-BB) is one of the strongest mitogens for hepatic stellate cells and is considered as a critical soluble mediator of liver fibrosis in vitro and in vivo. The biological activity of PDGF-BB is dependent on its degradation by the factor VII–activating protease (FSAP). Here, we demonstrate that a coding polymorphism (G534E) in the gene for FSAP is significantly associated with severe HCV-induced liver fibrosis (odds ratio, 2.59; P = 0.017), which is independent of age, gender, and presence of diabetes in multivariate analysis. These genetic findings were replicated in a cohort of patients with liver transplantation due to HCV-induced cirrhosis (OR, 2.56; P = 0.011). Functional dissection of the association demonstrates that the single amino acid change encoded by G534E in the FSAP protein does not influence PDGFβ receptor or α-smooth muscle actin expression but completely abrogates FSAP-mediated inhibition of PDGF-BB–induced proliferation of primary stellate cells in vitro. Conclusion: The G534E variant of FSAP is a risk locus for HCV-induced liver fibrosis and cirrhosis by determining PDGF-BB–mediated hepatic stellate cell proliferation through a single amino acid substitution in FSAP. FSAP G534E might be useful for risk stratification in patients with HCV infection. (HEPATOLOGY 2009.)
Progressive liver damage in patients with chronic hepatitis C virus (HCV) infection is associated with increased morbidity and represents the main indication for liver transplantation in the United States and Europe.1 In large epidemiologic studies, it is apparent that only about half of all infected patients develop end-stage liver disease and related complications and are primary candidates for antiviral therapies.2 Based on these epidemiological data, genetic predisposition is considered to play a critical role in the evolution of progressive liver disease in chronic hepatitis C, which is further modulated by viral and host factors like age, gender, diabetes, and alcohol consumption,3, 4 However, the genetics of liver fibrosis are complex; studies include relatively small numbers of patients and have been difficult to reproduce. In recent years, genetic analyses have entered a new era by the application of genome-wide association studies which have already revealed chromosomal loci with significant associations with severe liver fibrosis in well-defined patients with chronic hepatitis C.5, 6 It is important to note that these large genome scans report only statistical associations and yet lack functional data on disease-associated gene polymorphisms. Therefore, conclusive studies on fibrosis-related genes are still warranted and the investigated polymorphisms need to be evaluated for their pathophysiological relevance in vitro and in vivo.7
The factor VII–activating protease (FSAP), also known as the hyaluronic acid–binding protein (HABP2), is a serine-protease which is present in human plasma. FSAP is mainly secreted by hepatocytes and shows high concentrations in the liver, where it is known to activate pro-urokinase and is intrinsically involved in fibrinolysis.8 FSAP is down-regulated during fibrogenesis in toxically-induced liver damage, suggesting a functional implication of the protease in disease progression.9 Interestingly, FSAP has been shown to inhibit PDGF-BB–mediated proliferation of vascular smooth muscle cells by its strong proteolytic activity toward PDGF-BB.10 PDGF is secreted during liver fibrogenesis by a variety of cell types as a response to injury, and many proinflammatory cytokines mediate their mitogenic effects via the autocrine release of PDGF.11, 12
Within the FSAP (systemic name HABP2) gene, two nonsynonymous amino acid polymorphisms have recently been characterized. Although the E393Q polymorphism (Marburg II) is not associated with altered enzyme activity, the G534E polymorphism (Marburg I) severely affects the proteolytic activity of FSAP.13 The E-allele (Marburg I-FSAP) of this polymorphism has recently been identified as a risk factor for carotid stenosis,14 cardiovascular diseases,15 and thromboembolic events,16, 17 underscoring the functional importance of this variant in vivo.
Based on these genetic and functional data, we hypothesized that the FSAP G534E polymorphism (Marburg I) is a primary risk locus for progressive liver fibrosis in HCV infection and that this association is mediated by modulation of PDBF-BB–mediated stellate cell proliferation by the mutated FSAP protein.
Patients and Methods
Overall, 433 Caucasian patients chronically infected with hepatitis C were included in the genetic study. The diagnosis was based on a positive anti-HCV test and a positive HCV RNA (TaqMan Assay with a lower limit of detection of 50 IU/mL). None of the included patients tested positive for hepatitis B surface antigen or antibody to human immunodeficiency virus. Other chronic liver diseases were ruled out by appropriate serological tests. Furthermore, none of the patients admitted alcohol consumption of more than 20 g/day. The patients were recruited from the outpatient departments of the universities of Aachen, Berlin, and Regensburg. The selection criteria and characteristics of these patients have been described before.18, 19 All included patients underwent percutaneous liver biopsies prior to antiviral therapy in which the histological grading and staging of liver disease was performed according the scoring system proposed by Desmet and Scheuer20 by pathologists not aware of the current study design.
To minimize false positive genetic associations due to sampling error in the biopsied cohort, we also recruited 171 patients who had undergone orthotopic liver transplantation due to end-stage HCV-induced liver cirrhosis. These subjects were recruited from the hospitals in Berlin and Villejuif.
We also included 192 control subjects without hepatitis C infection in the current study. These individuals were selected according to stringent epidemiological criteria for case-control studies.21 Specifically, these control subjects were matched to the HCV-infected patients with regards to age and gender distribution. In all of these patients, viral infections (HCV, hepatitis B virus and human immunodeficiency virus) and other liver diseases were excluded by appropriate serological tests and ultrasound. Importantly, all patients and controls were of Caucasian origin to exclude ethnicity-related spurious allelic associations.
The study protocol was approved by the local ethics committees, and patients gave informed consent for participation in the study.
Genotyping of the Marburg I Polymorphism and Genetic Analysis.
Genotyping was performed on DNA samples isolated from patients' peripheral blood that was treated with ethylene diamine tetraacetic acid; analysis was conducted by LightCycler PCR as described in detail.22 Sample DNA (50 ng) was amplified with the primers: forward 5′-CAG ATG TCT CTG GTT CAC G-3′; reverse 5′-GTT GTC TCT GCT TAG AGT AG-3′. The following probes were used for LightCycler analysis: sensor probe 5′-LC-Red640-TGG CCT CTT CCC ACA CTC C-Ph-3′; anchor1 probe 5′-AGG AAT TTG GTA ACT TGG GTG TAG ACC C-fluorescein-3′.
Stellate Cell Cultures and Stimulation with rhPDBF-BB.
Primary stellate cells (HSCs) were isolated as described in detail.23 Wild-type and mutated Marburg I (MI)-FSAP protein were isolated from human plasma over an anti-FSAP antibody column as described previously.13 HSCs were plated in Dulbecco's modified Eagle medium (DMEM; 10% vol/vol) fetal bovine serum (FBS). On the third day in culture, the serum was reduced to 0.5% FBS and cells were starved for 24 hours. Thereafter, they were stimulated in DMEM containing 0.5% FBS and 1 mg/mL bovine serum albumin, and increasing concentration of PDGF-BB that was preincubated for 1 hour at 37°C with heparin (10 μg/μL) in buffer alone or in combination with FSAP or mutated MI-FSAP (12 μg/mL). [3H]Thymidine incorporation into newly synthesized DNA was determined after 48 hours and is represented as fold increase compared to unstimulated control cells.24
Furthermore, the influence of wild-type and MI-FSAP on PDGF receptor type β (PDGFRβ) and α-smooth muscle actin (α-SMA) was assessed in vitro. For this purpose, stellate cells were starved in DMEM with 1 mg/mL bovine serum albumin (BSA) and 0.5% FBS for 24 hours at day 4 of primary culture. Thereafter, the cells were washed (DMEM containing 1 mg/mL BSA) and incubated in DMEM supplemented with 1 mg/mL BSA and addition of wild-type FSAP or MI-FSAP (each 120 or 240 ng/mL) for 24 hours. FSAP and MI-FSAP were activated in a citrate-buffered solution containing heparin (100 or 200 ng/mL) before addition. Whole-cell lysates were prepared and equal amounts of proteins (30 μg) were resolved in NuPAGE Bis-Tris gels (Novex Invitrogen, Karlsruhe, Germany) and electroblotted onto a Protran membrane (Schleicher & Schuell, Dassel, Germany). The blot was probed with an antibody specific for PDGFRβ (sc-432; Santa Cruz Biotechnology, Santa Cruz, CA) and α-SMA (clone asm-1 obtained from Cymbus Biotechnology, Chandlers Ford, UK) and subsequently with an antibody specific for β-actin (A5441; Sigma, Taufkirchen, Germany). The primary antibodies were visualized using horseradish peroxidase–conjugated anti-mouse or anti-rabbit immunoglobulin G (Santa Cruz Biotechnology) and the SuperSignal chemiluminescent substrate (Pierce, Rockford, IL).
Demographic data are given as mean ± standard deviation (SD). Analysis of genetic data was performed by Armitage's trend test and chi-squared tests for allele frequency differences as provided on http://ihg.gsf.de. For this analysis patients were divided according to their histological fibrosis stage in a control cohort (n = 192), a group of individuals with mild to moderate fibrosis (F0-F2, n = 314), a group of subjects with severe fibrosis (F3 and F4, n = 119) and a cohort of 171 patients who had undergone liver transplantation due to HCV-induced liver cirrhosis. In all these cohorts, accordance of genotype distribution with Hardy-Weinberg equilibrium was tested with an exact test (http://ihg.gsf.de). Fibrosis progression (stage of fibrosis divided by known duration of infection) was available in a subgroup of patients (n = 256) and was analyzed independently with the Student t-test comparing the mean progression rate in patients with wild-type versus variant FSAP. Additionally, a multivariate analysis was performed to correct the genetic association for known fibrosis risk factors of age, gender, and the presence of diabetes. For this analysis, severity of fibrosis was the dependent variable and age, gender, diabetes, and the Marburg I genotype were included as covariates. For all analyses, a P value of less than 0.05 was considered as significant. The statistical analyses were performed with SPSS version 14.0 software (SPSS Inc., Chicago, IL).
The E-Allele of FSAP Is Associated with Severe Liver Fibrosis in HCV Infection.
The demographic data of the included patients and controls are shown in Table 1. Our patient population represents a typical cohort of HCV-infected individuals at tertiary referral centers with respect to age, gender distribution, and degree of liver damage.5, 6 Overall, the FSAP G543E genotype distributions were in Hardy-Weinberg equilibrium in all our patient and control cohorts.
|Parameter||Patients with HCV (n = 433)||Controls (n = 192)|
|Age (years)*||44 ± 17||48 ± 21|
|Inflammation (grade)*||2.0 ± 1.2||n.a.|
|Fibrosis (stage)*||1.9 ± 1.4||n.a.|
|Steatosis (grade)*||1.2 ± 0.8||n.a.|
|HCV genotype (% genotype1)||82.2||n.a.|
Within the cohort of 433 HCV-infected patients with liver biopsy, presence of the E-allele of the FSAP Marburg I polymorphism was associated with severe liver fibrosis (odds ratio [OR], 2.51, P = 0.019; Table 2). None of the genotyped individuals was homozygous for the FSAP 534E-allele. However, 4.1% of individuals (13 of 314) with mild to moderate fibrosis were heterozygous for the FSAP E-allele (Fig. 1), whereas 10.1% of patients (12 of 119) with severe liver fibrosis were heterozygotes (OR, 2.59; P = 0.017). The increased rate of heterozygosity for the FSAP E-allele was replicated in an independent cohort of patients with liver transplantation due to end-stage HCV liver cirrhosis (OR, 2.48; P = 0.011 compared to subjects with mild to moderate fibrosis). Importantly, subjects with severe fibrosis or liver transplantation also had an increased frequency of heterozygosity for the Marburg I polymorphism compared to our carefully selected control group (OR, 2.58 and OR, 2.27; both P < 0.05, respectively; Fig. 1) and to published control cohorts of Caucasian descent (2.3%-4.4%).14, 16 Therefore, the FSAP 534E-allele (Marburg I) represents a true risk allele for severe HCV-induced liver fibrosis and cirrhosis in our study. The genetic association of the FSAP 534E-allele with severe fibrosis was independent of the known risk factors of age, gender, and diabetes in multivariate analysis (P = 0.020). Furthermore, progression of fibrosis was higher in patients carrying the E-allele (0.38 fibrosis stages/year) compared to patients homozygous for the FSAP wild-type allele (0.25 fibrosis stages/year) in subgroup analysis, although this difference did not reach statistical significance due to the low frequency of the 534E-allele and the limited number of patients for whom reliable data on duration of disease was known.
|Tagging Polymorphism||Number (Frequency) of Alleles/Genotypes||Tests for Association*|
|FSAP Marburg I||Mild Fibrosis (2n = 628)||Severe Fibrosis (2n = 238)||OR||CI||P|
|534 G||615 (97.9)||226 (94.7)||2.51||1.12–5.58||0.019|
|534 E||13 (2.1)||12 (5.3)|
|534 GG||301 (95.9)||107 (91.9)||2.59||0.017|
|534 GE||13 (4.1)||12 (8.1)|
|534 EE||0 (0)||0 (0)|
The Mutated FSAP Protein Does Not Inhibit PDGF-BB–Induced Stellate Cell Proliferation.
Based on the genetic association of the Marburg I allele with hepatic fibrosis and the known proteolytic activity of FSAP toward PDGF-BB, we next functionally assessed whether the mutated FSAP protein has a different impact on PDGF-BB–induced stellate cell proliferation compared to the wild-type protein. PDBF-BB induces a strong proliferation of primary hepatic stellate cells as determined by [3H]thymidine incorporation (Fig. 2). Addition of wild-type FSAP to the cell culture completely abolishes PDBF-BB–induced stellate cell proliferation, underscoring the potent proteolytic activity of FSAP and supporting recent findings with vascular smooth muscle cells.10 In great contrast, addition of mutated MI-FSAP protein to the cell culture had virtually no inhibitory effect on PDBF-BB–induced stellate cell proliferation. These results strongly support our main genetic finding that the mutated E-allele is associated with severe liver fibrosis.
Because the difference between wild-type and mutated FSAP protein might have also been mediated by differences in PDGF-BB receptor expression or stellate cell activation, we next evaluated PDGFRβ and α-SMA expression after treatment with FSAP by western blot. Neither wild-type FSAP nor mutated FSAP had an effect on PDGFRβ or α-SMA protein expression (Fig. 3). Therefore, the observed differences of wild-type versus mutated FSAP seem to be mainly mediated through its direct proteolytic activity toward PDGF-BB–induced proliferation in vitro.
The main findings of our study are that the E-allele of the Marburg I polymorphism of the serine-protease FSAP is a genetic risk factor for severe liver disease in patients with chronic hepatitis C infection and that this correlation is functionally mediated by complete lack of inhibition of PDBF-BB–mediated stellate cell proliferation by mutated MI-FSAP compared to wild-type FSAP. To our knowledge, this is the first genetic study which directly demonstrates a plausible fibrogenic effect of a mutated protein with alteration in only a single amino acid in vitro, thereby providing direct evidence for the observed genotype-phenotype association.
FSAP is predominantly secreted as an inactive zymogene from the liver and circulates in human plasma in the same concentration as used in our cell culture.8 Initially, FSAP has been shown to activate pro-urokinase and is believed to be a member of fibrinolysis pathways.25 However, potentially more relevant with regards to liver fibrosis, FSAP has been shown to be down-regulated in toxically-induced liver fibrosis9 and that it specifically cleaves PDGF-BB, thereby leading to its inactivation.13 Importantly, recent work from our groups has demonstrated that FSAP does not cleave transforming growth factor-β, a master cytokine in liver fibrosis, or other growth factors such as insulin-like growth factor, thrombin, sphingosine-1-phosphatase, basic fibroblast growth factor, or hepatocyte growth factor, some of which have also been involved in liver fibrogenesis.13
PDGF is known to be among the most potent stellate cell mitogens described thus far, and stellate cells display increased PDGF production as well as up-regulation of PDGF receptors during liver injury.26, 27 Furthermore, transgenic overexpression of PDGF-BB promotes fibrogenesis,28 and its neutralization by antisense technologies inhibits liver scarring in vivo.29 Therefore, interference with PDGF, such as cleavage by FSAP, is likely to modulate the natural history of liver fibrosis in HCV infection.
For interpretation of our data, it is also important to note that the effects of the Marburg I polymorphism are not only evident in vitro, but individuals carrying the E-allele indeed have strongly reduced FSAP activity in vivo.14 We used the same concentrations of FSAP in cell culture as the protease is present in human serum.8 Thus, we hypothesize that the effects determined in vitro might also operate in vivo. Furthermore, we show here that the Marburg I variant of FSAP exhibits extremely reduced proteolysis of PDGF-BB, but does not modulate PDGFβ receptor expression or stellate cell activation as determined by α-SMA expression. This suggests that the proteolytic activity toward PDGF-BB is the main mode of action which determines the in vitro effects observed in this study.
Despite advances in recent years, there is still a need for better prediction of progressive liver disease in patients with hepatitis C and other chronic liver diseases. Such risk prediction is of great clinical importance given the costs, adverse effects, and modest response rates to current antiviral treatment regimens.30 Although a genetic basis for liver damage has been widely accepted and genetic risk profiles for progressive disease have been postulated, the genetic variants that drive fibrogenesis have not yet been fully elucidated.31 Recently, genetic analysis has been improved by the advent of genome-wide association studies in well-characterized patients, and risk scores based on genetic data have been developed.6 However, mechanistic links between the associated polymorphisms and the biology of liver fibrosis are yet to be defined.7 It is therefore important to note that we were for the first time able to directly investigate a mutated protein encoded by a fibrosis-associated polymorphism and show a plausible profibrogenic effect on stellate cells in vitro.
Due to the low frequency of the E-allele in Caucasian populations, the Marburg I polymorphism is only a risk factor for a fraction of all patients infected with hepatitis C. We therefore reproduced our main genetic finding in a cohort of patients who have undergone liver transplantation due to HCV-induced cirrhosis. With such replication, we also strongly reduced the likelihood of a spurious allelic association due to sampling error in the biopsied cohort. Furthermore, we were able to show the independence of the FSAP genotype from the known risk factors of age, gender, and diabetes in multivariate analysis. Other polymorphisms of clinically relevant proteins with low minor allele frequencies have also been found to be strongly associated with progressive liver disease, such as the factor V Leiden mutation32 and the odds ratio of our genetic association is entirely in line with a complex disease and a rare genetic variant.33
In summary, we show here that the Marburg I polymorphism is associated with severe liver fibrosis in HCV and that this association is functionally mediated through reduced PDGF cleavage by the mutated FSAP protein in vitro. This combined genetic and functional analyses of the Marburg I polymorphism might significantly add new information to the statistical and biological risk profile driving progressive liver damage in individuals with chronic hepatitis C.
We thank all patients for participation in the study and Mrs. Michelle Gigou for excellent technical assistance.