Single nucleotide polymorphism–mediated translational suppression of endoplasmic reticulum mannosidase I modifies the onset of end-stage liver disease in alpha1-antitrypsin deficiency

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Inappropriate accumulation of the misfolded Z variant of alpha1-antitrypsin in the hepatocyte endoplasmic reticulum (ER) is a risk factor for the development of end-stage liver disease. However, the genetic and environmental factors that contribute to its etiology are poorly understood. ER mannosidase I (ERManI) is a quality control factor that plays a critical role in the sorting and targeting of misfolded glycoproteins for proteasome-mediated degradation. In this study, we tested whether genetic variations in the human ERManI gene influence the age at onset of end-stage liver disease in patients homozygous for the Z allele (ZZ). We sequenced all 13 exons in a group of unrelated Caucasian ZZ transplant recipients with different age at onset of the end-stage liver disease. Homozygosity for the minor A allele at 2484G/A (refSNP ID number rs4567) in the 3′-untranslated region was prevalent in the infant ZZ patients. Functional studies indicated that rs4567(A), but not rs4567(G), suppresses ERManI translation under ER stress conditions. Conclusion: These findings suggest that the identified single-nucleotide polymorphism can accelerate the onset of the end-stage liver disease associated with alpha1-antitrypsin deficiency and underscore the contribution of biosynthetic quality control as a modifier of genetic disease. (HEPATOLOGY 2009.)

Alpha1-antitrypsin (AAT) is the predominant circulating protease inhibitor (Pi) in human plasma and protects lung elastin fibers from excess proteolytic destruction.1, 2 The Z variant of AAT (Pi Z), which bears a single amino acid substitution (G342K),3 is defective in secretion from primary and transfected cells in culture due to its inappropriately folded conformation.4, 5 The decreased concentration of circulating protease inhibitor leads to the hydrolytic destruction of lung elastin as an etiologic factor for panlobular emphysema.6 The vast majority of retained Pi Z becomes a substrate for intracellular elimination7–9 by a process known as endoplasmic reticulum (ER)-associated degradation (ERAD).10 However, the polymerization-prone conformation of variant Pi Z leads to the accumulation of loop-sheet polymers in the hepatocyte ER.11 Liver injury, as the primary gain-of-toxic-function disorder, is associated with the intracellular accumulation of Pi Z polymers and can progress to end-stage pathology requiring liver transplantation.12, 13 In recent years, AAT deficiency has emerged as a paradigm to investigate the pathologic variability associated with conformational diseases, all of which are caused by cytotoxicity associated with the inappropriate accumulation of a structurally aberrant protein.14

Based on the most comprehensive study by Sveger and colleagues, who screened 200,000 newborns in Sweden in the 1970s and the follow-up studies in the 1990s, about 1 among 1700 mixed North American and European populations are homozygous for the Z allele (referred to as “PiZZ” individuals). However, only about 17% of the ZZ newborns exhibited clinically significant liver disease, and less than 3% of those progressed to life-threatening end-stage disease as infants.15, 16 The high variation in the severity of the liver disease in PiZZ individuals suggests that genetic modifiers or environmental factors might contribute to the development of end-stage pathology. Understanding the mechanistic principles that underlie the variability of liver disease progression among ZZ homozygotes is predicted to aid in the identification of biomarkers for improved prognosis, and have far-reaching consequences with regard to the development of effective treatment strategies.17

Accumulation of variant Pi Z in the hepatocyte ER lumen, rather than in the cytosol where proteasomal degradation takes place, implies that a defeat in an early step in the intracellular disposal process might exist. In support of this notion, a lag in degradation of transduced Pi Z in skin fibroblasts from a subgroup of “susceptible” ZZ homozygotes18 led to the realization that a subtle defect in the protein disposal machinery likely contribute to the etiology of the disorder. For misfolded asparagine-linked secretory glycoproteins like AAT, the removal of mannose units from N-linked glycans represents a crucial early event in the disposal system.19, 20 Studies from several groups including ours have demonstrated that ERManI, a putative ER mannosidase, plays a stochastic and rate-limiting role in both distinguishing and targeting misfolded glycoproteins for ERAD.21–23 In this study, a candidate gene approach was performed to determine whether variations in the corresponding gene (MAN1B1) might influence the age at onset of the end-stage liver disease.

Abbreviations

AAT, alpha1-antitrypsin; ERManI, endoplasmic reticulum mannosidase I; PCR, polymerase chain reaction; Pi, protease inhibitor; Pi Z, the Z variant of alpha1-antitrypsin; UTR, untranslated region; ZZ, individuals homozygous for the Z variant of AAT.

Patients and Methods

Study Subjects.

We obtained from the Liver Tissue Cell Distribution System (LTCDS, University of Minnesota and University of Pittsburgh, #N01-DK-7-0004 / HHSN267200700004C) all available excised livers (n = 30) from unrelated PiZZ individuals who had undergone orthotopic liver transplantation for end-stage liver disease. No additional confounding factors were identified. All ZZ patients were Caucasian, and included 10 females, and 20 males. The age at which each patient was listed for liver transplantation was used to indicate the onset of end-stage of liver disease.

We also acquired liver samples from 32 liver transplant donors (wild-type for AAT) from LTCDS. These are unrelated Caucasian individuals who had died from diseases other than liver disease. In addition, we obtained genomic DNA samples from 200 adult Caucasian ZZ patients (via the Alpha-1 Research Program at the University of Florida), none of whom ever exhibited a history of clinically diagnosed liver disease. The entire study was approved by members of the appropriate institutional review boards.

Genotyping.

Genomic DNA was extracted from each of the liver samples using the Easy DNA kit (Invitrogen, Carlsbad, CA) following the manufacturer's instructions.

The 13 exons, plus intron/exon junctions, of the ERManI gene (GenBank accession number NM_016219) were amplified by polymerase chain reaction (PCR). Amplification primers were designed to target regions incorporating 5′ ends complementary to forward and reverse sequence primers and are available upon request. PCR targets were amplified using the HotStarTaq Master Mix kit (Qiagen #203446) according to the manufacturer's instructions with some modifications. PCR products were generated using 10 ng of DNA for both the experimental and cell line reference samples, 2.8 μL of the 2× Hot Star Taq Master Mix, 3.2 pmol of each primer, and sterile H2O for a total reaction volume of 8 μL. Reactions were cycled following the Qiagen protocol with annealing and extension times of 45 seconds for each amplification cycle. Excess primers and deoxynucleotide triphosphates were removed from the PCR reaction by treatment with 5 μL of a 1:10 dilution of Exosap-IT (USB #78202). The PCR products were incubated with Exosap-IT at 37°C for 15 minutes and then inactivated by heating at 80°C for 15 minutes. Samples were then diluted with 22 μL of 1 × TE (10 mM Tris, pH 8.0; 0.2 mM ethylene diamine tetraacetic acid) to a concentration of approximately 20-40 ng/μL in preparation for cycle sequencing. Sanger reactions were generated using Applied Biosystems BigDye Terminator version 3.1 at 1/64 dilution, 4 pmol primer, 40 ng of PCR product, and standard cycling conditions. Reactions were purified by ethanol precipitation and dried under a vacuum. The reactions were resuspended in 20 μL of 0.1 mM ethylene diamine tetraacetic acid and sequenced with an Applied Biosystems 3730xl DNA Analyzer using the RapidSeq36 run module. Base calls were determined using the 3XX basecaller software provided by Applied Biosystems.

Statistical Analysis.

Genotype frequencies for single-nucleotide polymorphisms (SNPs) identified between the early-onset group and late-onset groups were compared to their distribution in the general Caucasian population using the χ2 test. P values < 0.01 were considered statistically significant.

Plasmid Construction.

The complementary DNA (cDNA) for ERManI containing the 3′-untranslated region (3′UTR) was amplified using I.M.A.G.E. Clone 3533651 purchased from American Type Culture Collection (Manassas, VA) as a template. The following primers were used: ataagcttgcctgggtggcgaattc and agcggccgcatagatgcctcgag, with HindIII and NotI restriction endonuclease sites incorporated into 5′ and 3′ ends, respectively. The amplified fragment was inserted into the pMH vector and confirmed by DNA sequencing.

Cell Culture, Transient Transfection, and Western Blotting.

HeLa cells were cultured in Dulbecco's Modified Eagle's Medium (Mediatech Inc.) supplemented with 10% fetal bovine serum (Gemeni Bio-Products) and 1% ampicillin/streptomycin (Invitrogen). The day prior to transfection, cells were plated into six-well dishes and allowed to reach 80% confluence by the time of transfection. Then, 4 μg of the plasmids were transfected into each well with Lipofectamine 2000 (Invitrogen) following the manufacturer's instructions. At 24 hours after transfection, cells were lysed and immunoblotted for ERManI following the protocol described previously.24

Metabolic Radiolabeling and Immunoprecipitation.

At 24 hours after transfection, cells were starved in methionine-free and cysteine-free medium for 1 hour, and then subjected to metabolic pulse-radiolabeling with [35S]methionine for 20 minutes. In some experiments, the prestarvation step was eliminated and 10% fetal bovine serum (Invitrogen) was added during the pulse. For chase experiments, pulse medium was replaced by fresh medium containing unlabeled methionine, and cells were incubated at 37°C for the indicated time. At each time point, cells were then incubated on ice in lysis buffer containing 50 mM Tris HCl, 150 mM NaCl, 0.5% Nonidet P-40, 2 mM phenylmethylsulfonyl fluoride and protease inhibitors (Sigma). Following centrifugation at 4°C for 30 minutes, the supernatant of each sample was collected and mixed with polyclonal anti-ERManI antibody24 and protein G agarose beads (Calbiochem). The mixtures were rotated at 4°C overnight. After stringent washes with lysis buffer, and then lysis buffer containing 500 mM NaCl, the immunoprecipitates were eluted with sodium dodecyl sulfate (SDS) sample buffer and separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Radiolabeled proteins were detected by fluorography and quantified by either densitometric scanning using the National Institutes of Health (NIH) Image Program, or by PhosphoImager analysis.

Results

Identification of an SNP Associated with Early-Onset End-Stage Liver Disease in Patients with the ZZ Genotype.

Genotyping of the unrelated Caucasian ZZ transplant recipients led to the identification of six SNPs: one in exon 1 (176G/A), one in exon 4 (590C/T), and four in exon 13 (2046C/T; 2484G/A; 2542G/T; 2587C/CC) (Fig. 1A). Two of these, 176G/A and 2484G/A, have been reported by the dbSNP database as rs968733 and rs4567, respectively. The remaining four polymorphisms were novel. Both 176G/A and 590C/T represent nonsynonymous SNPs of S59N and P197L, respectively, while 2046C/T generates a synonymous change (D681D). Nucleotide polymorphisms 2484A/G, 2542G/T, and 2587C/CC were located in the 3′-UTR. Of all the identified SNPs, 2484G/A (rs4567) was the only one that exhibited a clustered distribution with age. Six of 10 patients who develop end-stage liver disease at age less than 2 years were homozygous for the A allele at rs4567 (Fig. 1B), but only three of 20 in the remaining patients displayed the same genotype (P = 0.0003). The distribution of rs4567, but not any of the other five identified SNPs, was significantly different between these two groups based on chi-squared analysis (Fig. 2A). The distribution of 2484A/G in infant ZZ transplant patients, which was 60% (AA), 40% (A/G), and 0% (G/G) (Fig. 2A), is also significantly different from the reported distribution in the Caucasian population with a P value < 0.01 (Fig. 2A).To further confirm this finding, we extracted genomic DNA samples from 32 unrelated Caucasians carrying wild-type AAT and determined their genotypes at rs4567. The results were: 18% (AA), 69% (A/G), and 13% (G/G), which is significantly different from that of the infant ZZ transplant patients (P < 0.0004). The prevalence of homozygosity for the A allele at rs4567 among the PiZZ infant group strongly suggests its contribution to an accelerated onset of the end-stage liver disease.

Figure 1.

Identified SNPs and distribution of rs4567. (A) Illustration of the positions of the six SNPs identified in the ERManI gene. (B) Genotypes of rs4567 among PiZZ individuals who developed end-stage liver disease, requiring transplantation at different ages. The age at which each patient was placed on the transplant list (age at listing) was used as an indicator for age of onset. TX, transplant.

Figure 2.

Percentage of identified SNPs in different groups. (A) Percentage of allelic genotypes of each SNP in the indicated groups. P values are calculated with the chi-squared statistic with 2 degrees of freedom. (B) Frequency of the rs4567 genotypes in the Caucasian population and in 200 ZZ individuals without liver disease.

Because a few PiZZ patients over 2 years of age are also homozygous for the rs4567(A) allele (Fig. 1B), it is unlikely that this genotype is solely responsible for the development of end-stage liver disease. To provide further evidence, we acquired (from the Alpha-1 Research Program at the University of Florida) genomic DNA samples from 200 adult Caucasian ZZ patients without a history of clinically diagnosed liver disease. Homozygosity for rs4567(A) among these individuals was nearly identical to the general Caucasian population (Fig. 2B), excluding the possibility that the SNP plays a causative role in end-stage liver disease. Rather, the data imply that rs4567(A) functions as a modifier of the disorder, accelerating the time at onset.

Functional Studies.

Previous studies have indicated that the ERManI concentration is primarily regulated at translational and/or post-translational levels.24, 25 Therefore, we asked whether the rs4567 genotype might influence the protein expression level for ERManI. To test this hypothesis, we generated mammalian expression constructs containing either nucleotide A or G at rs4567 in the 3′UTR of the human ERManI cDNA, which were designated ERManI(A) and ERManI(G), respectively. Equal amounts of the cDNA expression constructs were separately transfected into the human cervical carcinoma cell line HeLa, and the synthesis of the recombinant ERManI was monitored by metabolic radiolabeling. Based on our sequence data, HeLa cells are heterozygous (A/G) at rs4567. Therefore, overexpression of ERManI(A) or ERManI(G) should have similar effects on the A and G alleles, respectively. No differences in the translational efficiency of the encoded products were detected under basal conditions. However, when cells were cotransfected with the Z variant of AAT, ERManI(A) exhibited a 50% decrease in synthesis as compared to rs4567(G) (Fig. 3A). Consistently, the steady-state concentrations of the corresponding proteins were diminished to a similar extent, providing additional support for the validity of the observed phenomenon (Fig. 3B). The doublet bands detected during steady-state conditions were not unexpected, because the post-translational modification of the protein has been previously reported.23 Also, these distinctions did not reflect differences at either the transcriptional or post-translational levels because the respective messenger RNA levels were similar by northern blotting (Fig. 4A), as were the intracellular stabilities of the translated proteins (pulse-chase radiolabeling) (Fig. 4B), indicating that rs4567(A) influences ERManI at the translational level. Moreover, the pre-starvation step used in our routine metabolic method does not seem to influence the effect of rs4567 (A) on the translational suppression of ERManI, because an identical result was obtained in the absence of serum starvation. (Fig. 4C).

Figure 3.

rs4567(A) suppresses the translation of transfected ERManI in HeLa cells. (A) The translational efficiency of transfected ERManI (during a 20-minute pulse with [35S]methionine) in HeLa cells, with or without cotransfected Pi Z, were evaluated by a combination of immunoprecipitation, SDS-PAGE, autoradiography, and quantified by PhosphoImager analysis. The error bars represent standard error of the mean. (B) The concentration of transfected ERManI (from above) under steady-state conditions was evaluated by western blotting of cell lysates. The relative ERManI concentrations, as compared with control, were quantified in five separate experiments. The error bars represent a standard error of the mean.

Figure 4.

rs4567 does not affect transcription or turnover rate of ERManI and its expression is not affected by serum. (A) HeLa cells were either not transfected [C], or cotransfected with ERManI(A) or ERManI(G) plus empty vector or Pi Z. At 24 hours after transfection, cells were lysed and total messenger RNA was extracted and separated on agarose gels, and then subjected to northern blotting with ERManI-specific or actin-specific probes. The bands were quantified using NIH Image J, and the ratio between ERManI and actin was calculated. (B) HeLa cells were cotransfected with ERManI(A) or ERManI(G) plus empty vector or Pi Z. At 48 hours after transfection, cells were metabolically labeled with [35S]methionine and chased for 0.5, 1, or 2 hours. At each time point, cells were lysed and subjected to immunoprecipitation of ERManI. The immunoprecipitates were resolved by SDS-PAGE and detected by autoradiography. After quantification by PhosphoImager analysis, the percentage of ERManI left at each chase time point was calculated and plotted. Error bars represent standard deviation from three independent experiments. (C) HeLa cells were cotransfected with ERManI(A) plus empty vector or Pi Z, and then metabolically labeled with [35S]methionine in the presence of 10% fetal bovine serum for 20 minutes. Cells were then lysed followed by the immunoprecipitation of ERManI. The immunoprecipitates were resolved by SDS-PAGE and detected by autoradiography. Quantification was performed by PhosphoImager analysis. The percentage of ERManI level in the presence of Z versus in the absence of Z was calculated with error bars representing the standard error of the mean.

We next asked whether rs4567(A)-mediated translational suppression of ERManI occurs only in the presence of Pi Z. To address this question, the ERManI constructs were cotransfected with wild-type AAT, and the expression levels of ERManI(A) and ERManI(G) were examined by metabolic radiolabeling. A minor, but significant difference was also observed under this condition (Fig. 5). This result suggests that the rs4567(A)-mediated translational suppression of ERManI is associated with certain ER stress conditions.

Figure 5.

Overexpression of mutant or wild-type AAT induces rs4567(A)-mediated translational suppression of ERManI. The translational efficiency of ERManI(A) or ERManI(G) in the presence of cotransfected Z or wild-type AAT were evaluated by metabolic radiolabeling and quantified by PhosphoImager analysis. The P value was calculated by the Student t-test.

Discussion

Due to its pathological variation and rare occurrence, the development of end-stage liver disease is difficult to predict in ZZ individuals. Although SNPs were recently implicated in the etiology of the Pi Z–associated lung and liver diseases,26, 27 findings from the present study represent a functionally-defined genetic modifier of the end-stage pathology. Based on the results of this study, we predict that lower levels of ERManI in individuals homozygous for the A allele at rs4567 generate a conditional hypomorphic allele for ERManI that impairs the liver's capacity to deal with the accumulation of misfolded AAT, likely accelerating the rate at which a tolerable threshold is surpassed, resulting in the liver failure. However, it is worth noting that the rs4567(A)-mediated translational suppression of ERManI itself could not be the only modifier, because several older ZZ individuals homozygous for rs4567(A) did not exhibit an accelerated onset of end-stage liver disease. Moreover, not all the affected ZZ infants exhibited this genotype. Importantly, our conclusions need confirmation with carefully collected liver tissue from the analysis of a large group of affected ZZ patients before the role of ERManI can be substantiated.

The suppression of ERManI occurs when Pi Z or AAT are overexpressed, suggesting that cellular stress is likely required. Previous studies have shown that overexpression of Pi Z does not activate the unfolded response, but rather augments the nuclear factor-κB and autophagy pathways.28–30 It is possible that certain components of either pathway are actually responsible for the suppression of ERManI translation. The exact mechanism for the above suppression is unknown. However, the in silico analysis (MicroInspector web server [http://www.imbb.forth.gr/microinspector]) of sequences flanking rs4567 has identified alternative microRNA binding sites with constitution of A or G allele (Fig. 6). The possibility that certain microRNAs are responsible for the rs4567(A)-mediated translational suppression of ERManI is currently under investigation.

Figure 6.

Predicted binding of microRNA-205 (miR-205) to rs4567 in ERManI. The predicted binding pattern between messenger RNA sequences of ERManI is shown. The nucleotide at rs4567 (red) and flanking sequences (black) are shown, as is the sequence for mature human miR-205 (blue). The seed region and predicted 3′ compensatory sequences that contribute to hybridization are underlined.

The frequency of homozygous A allele at rs4567 is ∼ 28% in the general Caucasian population. Therefore, under stress conditions similar to Pi Z accumulation, these individuals may not be able to cope with ER stress as efficiently as those homozygous for the G allele, due to the suppression of ERManI translation. This may eventually lead to other pathological conditions, and it will therefore be interesting to determine whether the genotype is associated with other conformational diseases.

Although present within the 3′UTR of ERManI, we cannot dismiss the possibility that rs4567 might also reside in a regulatory region of another gene involved in glycoprotein quality control. It will be interesting to eventually determine whether the genes upstream or downstream of ERManI are also affected by the nucleotide at rs4567.

This study demonstrates the utility of functional studies to validate the contribution of an SNP in disease pathogenesis, and introduces a novel paradigm in which a subtle defect in the multilevel regulation of gene expression can modify a classical gain-of-toxic-function disorder. Whether a similar modality might undermine pathologies associated with other conformational disorders is still unknown. However, the present findings provide an exciting prospect for early prognosis and hold considerable promise for tapping a new avenue for therapeutic intervention.

Acknowledgements

This study was supported by NIH grant DK064232 (R.N.S.), Fernandez Liver Initiative research grant FO3-12 from the Alpha-1 Foundation (R.N.S.), and a postdoctoral research grant (S.P.) from the Alpha-1 Foundation. Also, we acknowledge the Liver Tissue Cell Distribution System (LTCDS) (#N01-DK-7-0004 / HHSN267200700004C) and the Alpha1-Foundation–University of Florida DNA and Tissue Bank. Finally, we thank Marion Namenwirth's (LTCDS) contribution toward the collection of data, and Sandra McGill for scientific editing.

Ancillary