Potential conflict of interest: Dr. Friedman received grants from Celera. Drs. Kwok, Sninsky, and Christopherson own stocks in Celera.
The work was supported by Celera (to S.L.F.) and National Institutes of Health grant RO1 DK56621 (to S.L.F.), the Israel Science Foundation (to C.K.), and by a grant from the Deutsche Forschungsgemeinschaft (to P.K.). A.J.P. was supported by a Doris Duke Clinical Research Fellowship and an Alpha Omega Alpha Student Research Fellowship.
Among several single-nucleotide polymorphisms (SNPs) that correlate with fibrosis progression in chronic HCV, an SNP in the antizyme inhibitor (AzI) gene is most strongly associated with slow fibrosis progression. Our aim was to identify the mechanism(s) underlying this observation by exploring the impact of the AzI SNP on hepatic stellate cell (HSC) activity. Seven novel AZIN1 splice variants (“SV2-8”) were cloned by polymerase chain reaction from the LX2 human HSC line. Expression of a minigene in LX2 containing the AZIN1 slow-fibrosis SNP yielded a 1.67-fold increase in AZIN1 splice variant 2 (AZIN1 SV2) messenger RNA (mRNA) (P = 0.05). In healthy human leukocytes, the SNP variant also correlated with significantly increased SV2 mRNA. Cells (293T) transfected with short hairpin RNA (shRNA) complementary to the exonic splicing chaperone SRp40 expressed 30% less SRp40 (P = 0.044) and 43% more AzI SV2 (P = 0.021) than control shRNA-expressing cells, mimicking the effect of the sequence variant. LX2 cells transfected with AZIN1 full-length complementary DNA expressed 35% less collagen I mRNA (P = 0.09) and 18% less α-smooth muscle actin mRNA (P = 0.09). Transient transfection of AZIN1 SV2 complementary DNA into LX2 cells reduced collagen I gene expression by 64% (P = 0.001) and α-smooth muscle actin by 43% (P = 0.005) compared to vector-transfected controls, paralleling changes in protein expression. Both AZIN1 and AZIN-SV2 mRNAs are detectable in normal human liver and reduced in HCV cirrhotic livers. The AZIN1-SV2 acts via a polyamine-independent pathway, as it neither interacts with antizyme nor affects the ability of AZIN1 lacking this variant to neutralize antizyme. Conclusion: An SNP variant in the AZIN1 gene leads to enhanced generation of a novel alternative splice form that modifies the fibrogenic potential of HSCs. (HEPATOLOGY 2011)
Hepatic fibrosis, or accumulation of interstitial “scar” matrix, can occur in response to many hepatic injuries. Cirrhosis is the end-stage of fibrosis, and is characterized by septal scar encircling nodules of hepatocytes.1 “Activation” of HSCs into a myofibroblast-like cell is a hallmark of fibrotic liver disease, and is characterized by the induction of fibrogenic and proliferative cytokines, as well as enhanced deposition of interstitial or “scar” extracellular matrix (ECM), especially collagen I. The myofibroblast is the major source of fibrosis in injured liver.2 Sustained or chronic liver injury accelerates the accumulation of these α-smooth muscle actin (αSMA)-expressing myofibroblasts, whose deposition of ECM disrupts normal liver structure and function.3
Chronic hepatitis C (CHC) is the most common cause of cirrhosis and the leading indication for liver transplantation in the United States and Western Europe. The rate of progression from fibrosis to cirrhosis varies widely among patients with CHC, such that some patients harbor HCV for decades with minimal scarring, whereas others can progress in a decade or less to cirrhosis and liver failure.4 Although definable factors including alcohol, age, and obesity5 may contribute to some of the accelerated risk, host genetic factors are increasingly recognized to affect the rate of fibrosis progression. Recently, large-scale studies using whole genome analysis have sought the genetic risk factors involved in complex disease, as well as host factors involved in pathological consequences of viral infections by identifying single-nucleotide polymorphisms (SNPs) that correlate with specific disease or treatment outcomes. As expected for the hypothesis-independent nature of these genetic association studies, including several of the risk-associated SNPs of a reported cirrhosis risk score,1 the mechanisms underlying the role of each SNP remain unclear. Following resequencing of implicated genes and fine-mapping, the original risk score SNPs appear to be the most strongly associated variants,6 removing the ambiguity of possible linkage to genetic variants in nearby genes. Moreover, whereas the original cirrhosis risk score was derived from a cross-sectional study, the score has now been validated in two separate, prospective studies.7, 8
We sought to determine the functional involvement of a highly protective SNP variant in the AZIN1 gene.1 This SNP is a synonymous (tyrosine) coding substitution in AZIN1. This allelic SNP variant is found as the minor allele in Caucasians and is rarely present in African Americans and Asians.
AZIN1 is expressed in liver9 and promotes cell growth and proliferation by increasing intracellular polyamine levels and stabilizing cyclin D1.10, 11 Polyamines are organic cations that regulate fundamental cellular processes, especially cellular proliferation.12 Ornithine decarboxylase (ODC) is the first and rate-limiting enzyme in the biosynthesis of polyamines. ODC is a short-lived, highly regulated, enzyme subject to autoregulation by polyamines.13, 14 In the center of this autoregulatory circuit is a small polyamine-induced protein named antizyme (Az), which promotes ubiquitin-independent degradation of ODC and inhibits cellular uptake of polyamines. Az itself is regulated by the ODC-related protein antizyme inhibitor (AZIN1), which titrates intracellular polyamine levels by antagonizing Az function, allowing ODC to escape Az-mediated degradation and stimulate cellular polyamine uptake.
The proteomic complexity derived from a relatively small number of genes has been ascribed in large measure to pre-mRNA alternative splicing, which allows a single gene to produce multiple proteins.15 Coding-region SNPs may exert phenotypic variability by modifying pre-mRNA splicing accuracy or efficiency.16 Exonic cis-elements contain RNA sequences that bind exonic splicing enhancers (ESE), proteins that promote exon definition by direct recruitment of the splicing machinery via their serine/arginine-rich (SR) regions or by antagonizing nearby silencing elements.17 SNPs can create or ablate known ESE binding regions, thereby affecting the outcome of pre-mRNA processing.
The recognized primary transcript of AZIN1 encodes a functional 448–amino-acid protein. Given the location of the SNP relative to the 5′ intron/exon border of the 12th exon, we speculated that it might affect AZIN1 pre-mRNA splicing. SRp40 is an ESE that binds to specific exonic cis-element and chaperones the formation of the spliceosomal subunits.18 Its expression is up-regulated during liver regeneration.19 Therefore, we hypothesized that the silent SNP associated clinically with slower rates of hepatic fibrosis alters the splicing pattern of AZIN1 pre-mRNA to favor the expression of a variant protein that would suppress the fibrogenicity of HSCs.
Semiquantitative PCR and Sequencing to Identify Splice Variants.
Polymerase chain reaction (PCR) was performed using AZIN1 forward 5′-ACAGTGAAGTGCAACTCTGCTCCA-3′ and AZIN1 reverse 5′-TGGGCTTCCATCTCCACTAAGTCA-3′ with LX2 complementary DNA (cDNA) as a template. The reaction products were separated using gel electrophoresis. Bands were extracted using the QIAquick Gel Extraction Kit (Qiagen, Chatsworth, CA). The DNA fragments were then cloned into the pcDNA3.1/V5-His TOPO TA Expression Kit (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Inserts were sequenced using the T7 forward and BGH reverse priming sites.
AZIN1 Minigene and Site-Directed Mutagenesis.
PCR was performed using AZIN1 minigene forward 5′-TGCTCCTCCTGCTCTCTGATGTTT-3′ and AZIN1 minigene reverse 5′-CCAAGGGCAATTTGATGGCAATGG-3′ with genomic DNA from LX2 cells as a template. The products of the reaction were separated using gel electrophoresis. Bands were excised and extracted using the QIAquick Gel Extraction Kit (Qiagen) according to the manufacturer's protocol. The DNA was then cloned into the pcDNA3.1/V5-His TOPO TA Expression Kit (Invitrogen,) according to the manufacturer's protocol. The SNP was introduced using the QuickChange II XL Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA) according to the manufacturer's protocol using the following primers: AZIN1 SNP forward 5′-cCAGAGGTTCACAAGGCAAAAATATAAGGAAGATGAGC-3′ and AZIN1 SNP reverse 5′-GCTCATCTTCCTTATATTTTTGCCTTGTGAACCTCTGG-3′.
AZIN1 WT and SV2 Expression Vectors and SRp40 shRNA.
PCR was performed using the following primers: AZIN1 WT/SV2 forward 5′-GGCTTTGTGGAATACGGCTGAGATG-3′, AZIN1 WT reverse 5′-GATCTAAAGAAGCGTTAATGCCTG-3′ and AZIN1 SV2 reverse 5′-CTTCAGCGGAAAAGCTCATCACAG-3′ with cDNA from LX2 cells as a template. The products were subjected to gel electrophoresis with a 2% agarose gel. Bands were excised and extracted using the QIAquick Gel Extraction Kit (Qiagen) according to the manufacturer's protocol. The DNA was then cloned into the pcDNA3.1/V5-His TOPO TA Expression Kit (Invitrogen) according to the manufacturer's protocol. Four clones with shRNA targeting SRp40 and their control (#TG309487; Origene, Rockville, MD,) were tested for their ability to reduce expression of SRp40 mRNA. These studies are based on the shSRp40 sequence 5′-GTTGAGAATTTATCCTCAAGAGTCAGCTG-3′. Transient transfection of these clones and gene assessment was performed as described below.
RNA and Quantitative Real-Time PCR Analysis.
RNA was extracted from cells overexpressing either LacZ, AZIN1 WT, AZIN1 SV2, shSRp40, or shControl using the RNeasy Mini Kit (Qiagen). All RNA was treated with deoxyribonuclease (Roche). A total of 1 μg RNA per condition was reverse-transcribed with first-strand cDNA synthesis with random primers using Sprint RT Complete-Double PrePrimed tubes (Clontech, Mountain View, CA). Quantitative real-time PCR (RT-PCR) was performed using the following primers on a LightCycler 480 (Roche): AZIN1 SV2 forward 5′-CGGAAGTGATGAACCAGCCTTCATGT-3′ and AZIN1 SV2 reverse 5′-GCTTCAGCGGAAAAGCTCAT-3′; AZIN1 WT F 5′-GAAAGCTGTCTTCTTCCTGAGCTG-3′ and AZIN1 WT R 5′-GGTTCATGGAAAGAATCTGCTCCCA-3′ AZIN1 SV8 forward 5′-CAGAACCCGGAAGCTACTATGTGT-3′ and AZIN1 SV8 reverse 5′-TGTAATAAATGGCTGGCCTCTTGTG-3′ αSMA forward 5′- AGGCACCCCTGAACCCCAA-3′ and αSMA reverse 5′-CAGCACCGCCTGGATAGCC-3′; collagen I forward 5′-GGCTTCCCTGGTCTTCCTGG-3′ and collagen I reverse 5′-CCAGGGGGTCCAGCCAAT-3′; GAPDH forward 5′-CAATGACCCCTTCATTGACC-3′ and GAPDH reverse 5′-GATCTCGCTCCTGGAAGATG-3′. All experiments were done in triplicate and normalized to GAPDH. Cycle numbers (CT) for AZIN1 isoform mRNA detection were: AZIN1 = 24.1; SV1 = 32.6; SV2 = 28.3; SV7 = 27.7.
Blood specimens from normal donors at Blood Centers of the Pacific were collected in EDTA and Paxgene tubes. Genomic DNA was isolated from the EDTA tubes using the QIAmp Extraction kit (Qiagen) and RNA was isolated from the Paxgene tubes using the PAXgene Blood RNA kit (Qiagen). All samples were stripped of identifying information and unlinked from the donor. Institutional review board approval was obtained from the UCSF Committee on Human Research.
Human Liver Tissue Samples.
A total of 10 human liver samples obtained either from liver resection or transplantation were analyzed. All samples were obtained from the HCC Genomic Consortium: Mount Sinai School of Medicine, New York, NY (US), Hospital Clinic, Barcelona (Spain), and Istituto Nazionale dei Tumori, Milan (Italy). Laboratory techniques were performed in the laboratories of the Division of Liver Disease at the Mount Sinai School of Medicine. The research protocol was approved by the institutional review boards, and informed consent was obtained in all cases. Five samples were obtained from a non–tumor carrying liver lobe of patients undergoing liver resection or transplantation for HCC on the background of HCV cirrhosis. Results were compared with samples of normal tissue obtained from five patients undergoing liver resection for nonmalignant lesions (hepatic hemangioma [two patients], focal nodular hyperplasia [two patients], complicated cyst [one patient]). A detailed description of tissue collection and RNA extraction from liver samples is detailed elsewhere.20, 21
The AZIN1 hCV25635059 polymorphism was genotyped by PCR using allele-specific primers (5′-AACAGAGGCTCATCTTCCTTA-3′ or 5′-ACAGAGGCTCATCTTCCTTG-3′ and 5′-TGAGCACATTGGATCTGAGACAG-3′). Approximately 3 ng of DNA was amplified with 5 units AmpliTaq Gold in a buffer containing 15 mM Tris (pH 8.0), 2.5 mM MgCl2, 50 mM KCl, 200 μM dAGC and 400 μM dUTP, 2 units of uracil-N-glycosylase (UNG), 1× ROX, and 0.2× SYBR green. The reactions were amplified on the ABI-Prism 7500 as follows: 50°C for 2 minutes, 95°C for 10 minutes, followed by 45 cycles of 15 seconds at 95°C and 1 minute at 60°C.
AZIN1 Splice Variant Expression Analysis.
Total RNA was reverse-transcribed using the High Capacity cDNA Archive Kit (Applied Biosystems, Norwalk, CT) with random hexamers. The transcription levels of total AZIN1 and the AZIN1 SV2 were quantified by real-time quantitative PCR using a primer set that amplifies all variants of AZIN1: total forward 5′-GCTTGCAAAGAATCTCAAGTATA-3′ and AZIN1 total reverse 5′-GTTCCCGTGAATCCTCCACCA-3′ and a primer set that is specific for SV2 (see above). A total of 5 ng cDNA was used for quantification of total AZIN1 and 20 ng for SV2. A dilution series of LX2-LACZ cDNA (2.5 pg to 25 ng) was amplified with each primer set and the cycle thresholds (Ct) used to generate a standard curve. The abundance of total AZIN1 WT and AZIN1 SV2 in each sample was determined by extrapolating the Ct values of the respective standard curves. The amount of AZIN1 SV2 relative to total AZIN1 was determined by measuring the ratio of AZIN1 SV2 to total AZIN1.
Cell Culture and Transient Transfection.
LX2 and 293FT cells were plated in six-well dishes and serum-starved for 4 hours prior to transfection. Cells were transiently transfected using the Lipofectamine 2000 (Invitrogen) in DMEM containing 0.2% bovine serum albumin. Human embryo kidney cells (293 HEK) were transiently transfected by the calcium phosphate method.22
LX2 cells grown in six-well dishes were transfected with 2 μg of DNA/well of LacZ, AZIN1-WT or AZIN1-SV2 encoding cDNAs. Cell extracts for Western blotting were harvested in the Complete-Lysis M buffer (Roche) 48 hours after transfection. Equal amounts of protein (25 μg), as determined by the Bio-Rad DC protein quantification assay, were loaded and separated by PAGE and transferred to PVDF membranes. Western blotting was performed using rabbit polyclonal antibodies to αSMA (Abcam) and collagen I (Rockland). Other immunoblots were carried out as follows: cellular extracts were prepared by lysis of cells in lysis buffer (50 mM Tris-Hcl, pH 8, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM dithiothreitol, and 1 mg/mL each of leupeptin, aprotinin, and pepstatin). Insoluble material was removed by centrifugation. Equal amounts of protein were mixed with Laemmli sample buffer (4% SDS, 20% glycerol, 10% 2-mercaptoethanol, and 0.123 M Tris-HCl [pH 6.8]), heated for 5 minutes at 95°C, and fractionated by electrophoresis in polyacrylamide-SDS gel. The resolved proteins were electroblotted onto a nitrocellulose membrane. The membrane was incubated with the indicated antibodies followed by horseradish peroxidase–conjugated anti IgG antibodies. The antibodies used were mouse monoclonal anti-FLAG (Sigma) and rabbit polyclonal anti-Az.23 Signals were developed using EZ-ECL (Biological Industries) and the membranes were exposed to x-ray film and developed.
Coimmunoprecipitation experiments were carried out with extracts from cells transiently cotransfected with Flag-tagged human AZIN1 or AZIN1 SV2 and Az. The reactions were incubated overnight with anti-Flag beads (Sigma). The immunoprecipitated material was resolved by electrophoresis in an sodium dodecyl sulfate (SDS)-polyacrylamide gel and the proteins were visualized using anti-Flag or anti-Az antibodies, following electroblotting onto nitrocellulose.
Protein synthesis was inhibited by the addition of cycloheximide (20 μg/mL) to the growth medium of transiently transfected cells, either alone or together with the proteasome inhibitor MG132, and harvested at the indicated times. Cells were lysed, and aliquots containing equal amounts of protein were fractionated by SDS-PAGE and examined by western blot.
ODC Activity Assay.
ODC activity assay was performed as previously described.24 100 μg aliquots of cellular extracts were mixed with 100 μL of ODC buffer (25 mM Tris-HCl [pH 7.5], 2.5 mM DTT, 0.1 mM EDTA, 0.2 mM pyridoxal phosphate, and 0.33 mM L-ornithine), containing 0.5 μCi L-[14C]ornithine. The reaction mix was incubated at 37°C in a 96-well plate for 4 hours. The liberated [14C]CO2 was trapped in a covering 3 mm paper soaked with saturated barium hydroxide solution. The paper was washed with acetone, dried and the results quantified using the Fuji Bas2500 phosphoimager.
AZIN1 mRNA Is Alternatively Spliced.
The location of the AZIN1 SNP, six nucleotides from the 5′ splice-site of the 12th exon, suggested that the SNP could affect splicing of AZIN1 pre-mRNA. Furthermore, analysis of the AZIN1 genomic sequence, using the ESEFinder 2.0 software,25 predicted decreased binding of the exonic splicing enhancer, SRp40, to the SNP containing sequence by affecting regional splice-site recognition. Review of AZIN1 transcripts in the RefSeq database26 (www.genome.ucsc.edu) revealed one AZIN1 splice variant that reportedly alters the transcript in the 5′-untranslated region. A shotgun analysis, performed as part of the sequencing of the human genome,27 had previously revealed one variant, termed here as “AZIN1 SV1,” that includes intron 11. To determine if there are additional AZIN1 splice variants, we performed PCR on cDNA from a cultured human stellate cell line (LX2) using primers targeting AZIN1's 5′ and 3′ noncoding regions. The resulting amplicons were separated by agarose gel electrophoresis (Fig. 1A), extracted, and sequenced. In total, seven novel AZIN1 splice variants, termed AZIN1 SV2-8, were detected (Fig. 1B).
The Fibrosis-Protective SNP in AZIN1 Increases mRNA Alternative Splicing In Vitro and In Vivo.
The presence of the protective SNP variant predicts an increased likelihood of AZIN1 mRNA alternative splicing based on analysis using the ESE finder software.25 Specifically, the G-to-A polymorphism, located six nucleotides from an intron/exon boundary, abolished a high scoring (4.3949) SRp40 binding motif (Fig. 2A). To test whether this variant altered the propensity for the native AZIN1 mRNA to undergo alternative splicing, we generated two ∼4 kilobase pair (kbp) mini-genes spanning the middle of the 10th exon to the middle of the 13th exon (corresponds to the genomic sequence in Fig. 1B). One minigene contained the major allele nucleotide, at base pair 1345, the other contained the protective minor allele, T. Relative production of AZIN1 splice forms was then quantified by real-time quantitative PCR. Minigenes, rather than the full genomic locus, were used because the latter is 37.8 kbp, which cannot be reliably transfected and expressed in LX2 cells. Indeed, LX2 cells transfected with the SNP-containing minigene expressed 1.67-fold (P = 0.05) more AZIN1 SV2 mRNA than cells transfected with a minigene containing the major allelic sequence, a biologically meaningful increase given that this increased expression reflects of a germline sequence change rather than mutation (Fig. 1C). AZIN1 SV2 encodes a truncated, 362–amino acid (AA) protein that involves the loss of 87 and gain of two amino acids at the AZIN1 C-terminus.
We next examined whether the presence of the variant altered the relative expression of AZIN1 splice forms in humans by comparing splice form expression using real-time PCR in leukocytes from healthy donors with homozygous expression of the major allele sequence, to those with heterozygous expression. We were only able to detect meaningful expression of AZIN1 SV1, SV2, and SV7 (Supporting Table 1). In agreement with the mini-gene experiments in cultured cells, leukocytes from heterozygous patients (Table 1) with one copy of the protective SNP variant-containing sequence had a significantly higher ratio of AZIN1 SV2 to AZIN1 (17.9220), than patients homozygous for the major allele sequence, 13.6782 (P = 0.0022) (Fig. 1D). In contrast, there were no significant differences in expression of other splice variants (Supporting Table 1). Even though there were more males in the homozygous group, we did not detect a correlation between relative AZIN1 SV2 expression and sex (Supporting Table 2).
Table 1. PBMC Donor Demographics
Summary of the donor demographics for patients who are homozygous for the WT sequence (C/C) or contain one copy of the SNP (C/T).
Loss of SRp40 Binding Increases Expression of AZIN1 SV2.
Computer analysis suggested that the introduction of the SNP variant to the AZIN1 sequence would significantly decrease the binding of SRp40 (Fig. 2A), an exonic binding enhancer, whose expression is increased during liver regeneration. To mimic the loss of SRp40 binding, we compared mRNA expression of AZIN1 and AZIN1 SV2 in cells expressing shRNA against SRp40 to cells expressing a control shRNA (shC). Thirty-six hours after transient transfection, we observed a 30% (P = 0.04) decrease in SRp40 expression and a 43% (P = 0.02) increase in AZIN1 SV2 expression in 293FT cells transfected with shSRp40 relative to cells transfected with shC (Fig. 2B). This suggests that the increased AZIN1 SV2 expression observed in patients harboring the SNP variant is mediated by loss of SRp40 binding.
AZIN1 SV2 Inhibits Expression of Fibrogenic Genes.
We examined if the increased generation of AZIN1 SV2 conferred by the fibrosis-protective SNP inhibited fibrogenic gene expression by stellate cells. To do so, LX2 cells were transiently transfected with either LacZ, AZIN1 WT, or AZIN1 SV2 cDNAs and expression of collagen I and αSMA mRNAs was assessed using real-time PCR. Expression of AZIN1 WT reduced collagen I and αSMA mRNAs by 35.5% and 18.2%, respectively (P = 0.089 for both genes), relative to LacZ-transfected cells. AZIN1 SV2 more significantly suppressed gene expression of collagen I and αSMA mRNAs by an additional 64.1% (P = 0.001) and 42.6% (P = 0.005), respectively (Fig. 3A,B). Furthermore, LX2 cells expressing AZIN1 SV2 expressed 41.1% less collagen I (P = 0.0036) and 29.9% less αSMA (P = 0.017) and LX2 cells expressing AZIN1 WT (Fig. 3A,B). These mRNA findings were correlated with parallel changes in protein expression by western blot (Fig. 3C). To confirm that the AZIN isoforms are expressed in human liver, we performed real-time PCR of AZIN1 WT and SV2 in both normal and cirrhotic whole-liver mRNA, documenting their expression (Fig. 3D). Interestingly, expression of both isoforms decreased significantly in cirrhotic tissues, the significance of which is unclear, because the cellular sources of these mRNAs are not defined.
AZIN1 SV2 Functions Via a Polyamine-Independent Pathway.
Because the main function of AZIN1 is to regulate the degradation of ODC and the uptake of polyamines by neutralizing antizyme, we examined whether AZIN1 SV2 affects the cellular polyamine metabolism. Coimmunoprecipitation using transfected FLAG-tagged AZIN1 and SV2 in 293T cells was performed to examine the relative affinity of AZIN1 and AZIN1 SV2 for antizyme. As shown in Fig. 4A, whereas wild-type AZIN1 without this variant interacts efficiently with antizyme, there was no interaction between AZIN1 SV2 and antizyme. Cycloheximide chase demonstrated that the half-life of AZIN1 SV2 is markedly shorter than that of the wild-type AZIN1 protein lacking this variant (Fig. 4B), reflecting the lack of interaction with Az that stabilizes the protein.28 In contrast to AZIN1, and in agreement with its inability to bind Az, AZIN1 SV2 failed to stimulate ODC activity in transfected cells (Fig. 4C). Moreover, when coexpressed, AZIN1-SV2 did not interfere with the ability of AZIN1 to stimulate ODC activity (Fig. 4D).
Patients with chronic hepatitis C (CHC), the leading cause of cirrhosis in the United States and Western Europe, accumulate fibrosis at highly variable rates. A gene-centric functional genome SNP scan that sought to characterize the genetic risk factors of fibrosis progression identified a SNP in the antizyme inhibitor (AZIN1) gene. The SNP in this gene had the most robust cross-validation score in the first step in the development of the multi-SNP score, as well as the highest odds ratio among the predictive SNPs correlating with delayed fibrosis progression.1 However, similar to several of the risk-associated SNPs that emerged from this and other genetic association studies, the mechanism underlying this gene's involvement and the specific protective effect of the minor allelic SNP variant remain to be clarified. Recently, studies from our laboratory have established mechanisms underlying effects of additional SNP variants of the TLR4,29 another gene included in the cirrhosis risk score, and DDX5,30 a separately reported gene on fibrogenic activities of HSCs, both of which were identified from the same genetic scan.
The AZIN 1 SNP encodes a silent polymorphism that does not alter the primary protein sequence. However, the SNP's proximity to an intron–exon boundary led us to hypothesize that the SNP altered the splicing of AZIN1 pre-mRNA, and computational analysis of the exonic sequence surrounding the SNP predicted decreased binding of an exonic splicing enhancer. Indeed, we have identified eight AZIN1 splice variants that are proximal to C-terminus of the AZIN1 protein, suggesting that the SNP resides in a region of alternative splicing.
To assess the impact of the SNP variant on alternative splicing, we generated partial AZIN1 minigenes containing the two allelic SNP nucleotides. Quantitative RT-PCR demonstrated significant increases in AZIN1 SV2 and AZIN1 SV8 (Supporting Fig. 1). The increase in SV2 and SV8 was unexpected because although the SNP is located six nucleotides from the 5′ end of exon 12, it appeared to have the greatest influence on 3′ splice site recognition.
We have established the in vivo relevance of the HSC culture findings by demonstrating that the fibrosis-protective variant is correlated with the increased expression of the SV2 mRNA splice form in leukocytes from healthy patients whose genomic DNA contains at least one copy of this SNP variant. Although evidence of increased splicing in liver-derived cells rather than leukocytes would be more informative, there were insufficient numbers of liver samples containing the minor allelic variant to perform such an analysis. Nonetheless, we have been able to establish that both isoforms are expressed in normal human liver. However, because HSCs, the fibrogenic cell in liver, represent only a small percentage of resident liver cells, our analysis of whole-liver mRNA likely reflects expression primarily in hepatocytes rather than in cells where the SNP regulates fibrogenic gene expression.
Prior studies have demonstrated that polyamines, whose synthesis is promoted by AZIN1, play an essential role in hepatocellular regeneration following partial hepatectomy or toxic liver injury.31-33 Our study did not detect an AZIN1 splice form perturbation of polyamine content. Interestingly, AZIN1 knockout mice, which die at birth, have nonspecific liver abnormalities but otherwise normal tissues, suggesting that AZIN1 is essential for normal liver development, architecture, or function.9 It is unknown whether these liver abnormalities are independent of polyamine content, but the question merits reassessment in light of the results described here.
Because the AZIN1 SNP variant correlates with a slower rate of fibrosis and leads to increased AZIN1 SV2, we characterized the effects of the two allelic variants of AZIN1 in the hepatic stellate cell, which is the cellular effector of fibrosis. Moreover, AZIN1 has not previously been characterized in any fibrogenic cell type. Overexpression of AZIN1 SV2 significantly inhibited expression of αSMA and collagen I mRNAs and proteins. Although AZIN1 moderately reduced fibrogenic gene expression in stellate cells following transfection, the inhibition was markedly increased in the presence of AZIN1 SV2, suggesting that AZIN1 SV2 exerts a unique affect on stellate cell responses.
The specific inhibition of genes characteristic of activated HSCs by AZIN1 SV2 was polyamine-independent, as described above. AZIN1 SV2 does not interact with Az based on coimmunoprecipitation, and it does not interfere with the ability of AZIN1 to neutralize Az activity, excluding the possibility that AZIN1 SV2 functions as a dominant-negative protein. Accordingly, ODC activity was unaffected by overexpression of AZIN1 SV2. This finding was surprising, as the AZIN1 SV2 protein conserves the known Az binding site, suggesting that the interaction between AZIN1 and Az may require inclusion of sequences from the C-terminal part of the AZIN1 protein for proper tertiary structure and interaction with Az.
Polyamine biology and biosynthetic pathways have proven a target-rich arena for therapeutic intervention with cancer as a primary and hyperproliferative disorders as a secondary disease focus.34 Even though clinical success for these strategies has proven evasive, recruitment of the developed antagonists merit study for liver fibrosis given the implicated role of the AZIN1.
In summary, the minor allelic SNP variant in the 12th exon of AZIN1 associated with slower rates of fibrosis progression favors the expression of a novel splice form, AZIN1 SV2, that inhibits the expression of fibrogenic genes in hepatic stellate cells via a novel, polyamine-independent pathway.
We gratefully acknowledge the assistance of Michael Busch, M.D., of Blood Centers of the Pacific for acquisition of patient blood samples, and Johnny Loke for technical assistance with real-time PCR and shRNA studies.