Advanced Review
Genome-wide association studies in biliary atresia Mylarappa Ningappa,1 Jun Min,2 Brandon W. Higgs,1 Chethan Ashokkumar,1 Sarangarajan Ranganathan1 and Rakesh Sindhi1∗ Biliary atresia (BA) is a model complex disease resulting from interactions between multiple susceptibility loci and environmental factors. This perception is based on a heterogeneous phenotype extending beyond an absent extrahepatic bile duct to include gut and cardiovascular anomalies, and the association of BA with viral infections. Refractory jaundice and progression to cirrhosis shortly after birth can be fatal without surgical correction, and further suggests a pathogenesis during liver and bile duct development. Conclusive proof for a developmental origin would require documentation of disease progression in the perinatal or fetal liver, an impossible task for obvious reasons. We review three different sets of genome-wide association studies (GWAS) from three different cohorts of BA patients by three different groups of investigators, which address this knowledge gap. Knockdown of each susceptibility gene identified by GWAS in zebrafish embryos impairs excretion of bile from the liver, duplicating the characteristic diagnostic finding seen in affected children. This finding is associated with impaired intrahepatic biliary network formation in zebrafish morphants. Although distinct, these susceptibility genes share several functions including roles in mechanisms for organogenesis (glypican 1 or GPC1, and adenosine diphosphate ribosylation factor 6, or ARF6) or a greater expression in fetal liver than in adult liver (adducin 3 or ADD3). Together, these studies emphasize the importance of the human evidence, and present opportunities to map novel pathways which explain the phenotypic heterogeneity of BA. © 2015 Wiley Periodicals, Inc. How to cite this article:
WIREs Syst Biol Med 2015, 7:267–273. doi: 10.1002/wsbm.1303
BILIARY ATRESIA: A COMPLEX DISEASE TRAIT
B
y any standard, the heterogeneous phenotype of biliary atresia (BA) defies a unified mechanistic explanation, is a likely consequence of multiple susceptibility loci interacting with environmental factors, and is, therefore, a model complex disease.1,2 Characterized by impaired extrahepatic bile drainage ∗ Correspondence
to: [email protected]
1 Department
of Surgery, Hillman Center for Pediatric Transplantation, Children’s Hospital of Pittsburgh of University of Pittsburgh Medical Center (UPMC), Pittsburgh, PA, USA 2 Department of Bioengineering, University of California San Diego, La Jolla, CA, USA Conflict of interest: The authors have declared no conflicts of interest for this article.
Volume 7, September/October 2015
and refractory jaundice in newborns, the disease is fatal if not corrected surgically.3 BA can also be associated with a variety of structural gastrointestinal and cardiovascular anomalies, some of which are also fatal without surgical correction.4,5 Most cases are ‘isolated’ or ‘perinatal’ with obstructive fibrosis of the extrahepatic biliary system.6 The presumed etiology is a perinatal viral cholangitis with exaggerated immune responses in the setting of familial predisposition to autoimmunity. The evidence for this hypothesis is circumstantial because of the impossibility of demonstrating progressive viral invasion and inflammation in sequential fetal biliary tissue samples. This evidence includes (1) viral particles of or viral sensitization toward reovirus, rotavirus, or cytomegalovirus in affected tissue in some human
© 2015 Wiley Periodicals, Inc.
267
wires.wiley.com/sysbio
Advanced Review
studies,7–9 (2) the induction of an analogous phenotype with these viruses in experimental models,10 and (3) seasonal peaks in BA cases from many parts of the world.11–13 The less common ‘syndromic’ or ‘embryonic’ BA consists of those cases which present earlier than isolated BA and also manifest major structural anomalies including laterality defects such as asplenia, malrotation, and heterotaxy in varying combinations or to varying degrees.4,5 These phenotypic categories are linked by an important fact: a discernible proportion of children with isolated BA also manifest minor anomalies in the same extrahepatic systems as those with syndromic BA.4 Recent genome-wide association studies (GWAS) of BA patients reflect a collective quest for novel unbiased hypotheses from a multidisciplinary constituency of caregivers comprising basic scientists, physicians, and surgeons, who are committed to alleviating the societal burden of BA. With an incidence of 1:5000 live births in Asian populations, and 1:18,000 live births in western Caucasian populations, BA accounts for at least one-third of all liver transplant procedures performed in children worldwide.14–17
GWAS AS A DISCOVERY METHOD FOR A COMPLEX TRAIT Toward hypothesis-generation, trait mapping with GWAS is based on the premise that a marker allele which ‘tags’ or represents a haplotype, and which shows a strong association with a disease trait, is likely to be associated with a novel, potentially causal variant(s) within that haplotype because of linkage disequilibrium.18–20 The causal variant(s) may be identified in subsequent studies, or may remain simply associated with the disease trait. In practice, this approach relies on arrays which characterize hundreds of thousands or a few million genome-wide ‘tag’ single-nucleotide polymorphisms (SNPs) or copy number variants (CNVs). These SNP arrays are used to genotype-unrelated cases, or subjects expressing the phenotype and unrelated controls, or healthy subjects not expressing the phenotype, all with a similar genetic background. Candidate SNPs are those with significantly different minor allele frequencies in the cases population compared with the control population. Replication in several independent disease and control subject cohorts, and across different geographic or ethnic groups strengthens the validity of an association. If this population heterogeneity is appropriately accounted for at this stage of the association testing, unrecognized disease heterogeneity among cases may be one reason for failure to replicate an association in a case–control design. This limitation is pertinent to 268
BA, where multiple loci with small biological effects acting in various combinations may contribute to the constellation of ‘minor’ or ‘major’ malformations associated with BA. Another compounding factor is the fact that the exact extent of associated extrahepatic anomalies, and, therefore, of the complexity of BA, is unknown. This is illustrated by the more subtle correlates of BA such as a ‘hypoplastic’ portal vein.4,5 Despite controlled data accrual and comprehensive analysis methods conducted by well-organized national consortia, recognition of a hypoplastic portal vein in a child with BA may occur in retrospect in the form of asymptomatic portal venous thrombosis on surveillance abdominal scans several years after liver transplantation. Virtually all children with this rare surgical complication have received liver transplantation for BA (Rakesh Sindhi, personal communication, December 2014). Also pertinent to rare diseases like BA is the validity and biological plausibility of any association. Dense genome-wide SNP scans mandate correction for testing of multiple hypotheses necessitating P-values less than or approaching 10−7 for 500,000 SNPs or 10−8 for 1–2 million SNPs.18 This objective requires large numbers of cases to achieve adequate power to detect an association. An acceptable alternative is demonstrating genome-wide significance in combined discovery and replication cohorts, though in BA, even combined cohorts may not provide sufficient sample sizes. Identification of putative associated SNP allele(s) is only the first step in GWAS analysis. Post-GWAS efforts should include discovery of novel causal variants with resequencing of the disease loci tagged by the primary SNP association, then functional studies, because the causal determinant (risk or protective) is assumed to be the direct link to the expressed phenotype. This objective can be challenging, however, because such loci often occupy sequences with potentially unknown function, at great distances from the coding region of the nearest gene, thus raising important questions around the length and depth of coverage for the resequencing effort.21,22 The ultimate goal of any association study is to identify the functional relevance of the disease-specific variant to disease mechanism. The GWAS reports summarized below illustrate several approaches, which may help to map disease pathways for a rare and complex disease like BA. Each of these studies presents differential (protein) expression of the candidate gene(s) in diseased human tissue as evidence of disease-specificity. Additional validation efforts range from classical discovery set-replication set testing, to mapping signaling
© 2015 Wiley Periodicals, Inc.
Volume 7, September/October 2015
WIREs Systems Biology and Medicine
GWAS in biliary atresia
pathways from the larger list of significant SNPs as supportive evidence for the mechanism suggested by the primary association. Modeling the developmental effects of the candidate(s) with knockdown in zebrafish embryos adds to the biological relevance of these studies.
SUMMARY OF GWAS STUDIES AND SUSCEPTIBILITY LOCI FOR BILIARY ATRESIA Glypican 1 and BA To evaluate the association of BA with germline CNV, Leyva-Vega et al.23 compared genotypes at 550,000 SNP loci for 35 children and 2026 controls using a previously described CNV calling algorithm.23,24 Heterozygous deletions were identified in two of 35 unrelated BA patients in a common 280-SNP 1.76 MB region comprising 30 genes on chr 2q37.3. The deletions were confirmed with fluorescent in situ hybridization (FISH). Both children received liver transplantation after a failed portoenterostomy by ages 10 and 11 months, respectively. In the first of two patients, the deletion was also identified in the father, who did not experience liver disease. A venous hemangioma was identified in the right tentorium at age 17 years in this first child. In the second child, the 1.76 MB deletion was contained within a larger 1346 SNP, 5.87 Mb heterozygous deletion on chromosome 2q37.3, and accompanied by dysmorphic facies, developmental delay, hypothyroidism, asplenia, and the 47 XXY karyotype consistent with Klinefelter’s syndrome. Of the 30 genes in the common 1.76 MB deletion, the authors sequenced the alanine-glyoxylate aminotransferase (AGXT) gene to identify additional causal variants. AGXT is only expressed in the liver. Several common SNPs were identified. These findings suggested that, although the CNV was inherited in the first child, the deletion alone was insufficient to produce BA because the father with the deletion did not manifest liver disease. Therefore, extraneous factor(s) such as toxins, viral infections, and so on, could have been potential participating event(s).25 In a subsequent study from the same group, Cui et al.26 evaluated the association of BA with germline CNVs by comparing 61 BA cases and 5088 controls at ≥550,000 SNP loci using the Penn CNV calling algorithm.27 A CNV consisting of a heterozygous deletion was localized to the same 2q37.3 region, but to a single gene, glypican 1 (GPC1). GPC 1 is a proteoglycan, which regulates hedgehog signaling and inflammation.28–30 In zebrafish larvae, gpc1 knockdown with morpholino antisense oligonucleotides Volume 7, September/October 2015
(MOs) resulted in poor biliary network formation, a small gall bladder, and poor bile excretion from the liver 5 days post-fertilization (dpf). These changes were associated with upregulation of hedgehog target genes, gli2a, ptch1, foxl1, znf697, and ccnd1, and partly rescued by injecting zebrafish larvae with gpc1 messenger RNA and the hedgehog antagonist, cyclopamine. The authors concluded that the decreased availability of the glypican1 ligand acted as a ‘sink’, resulting in upregulated hedgehog signaling genes. Immunostaining of normal human liver tissue demonstrated apical distribution of GPC1. This type of staining was poorly seen or absent in the liver tissues from the three patients with BA, none of whom carried deletions in the GPC1 gene. These human findings also suggested that abnormal GPC1 expression may be an independent correlate of BA.
Adducin-3 and BA Garcia-Barcelo et al.31 compared 181 Chinese BA patients and 481 healthy ethnically matched controls at >289,000 SNP loci using an Affymetrix genome-wide SNP array. In this discovery phase, several SNPs including rs12571674 and rs17095355 in a 129-kb intergenic locus on chr 10q24.2 were strongly associated with BA. In 124 independent Chinese BA patients and 90 controls, the association between rs17095355 and BA was replicated. This intergenic SNP is flanked by the X-prolyl aminopeptidase P (XPNPEP1) and adducin 3 (ADD3) genes. ADD3 is expressed in the liver and biliary epithelial cells (BECs), more so in fetal livers, and is necessary for remodeling of membrane cytoskeleton at points of cell-cell contact. XPNPEP1 is also found in liver epithelial cells and may be important in bile acid excretion. No functional role has been assigned to the rs17095355 SNP locus in databases such as ENCODE. However, the size of the BA cohort and replication in an independent cohort strongly support a role for this locus including its flanking genes in BA. To ‘fine’ map the disease locus further, Cheng et al.32 from the same investigator group genotyped 107 SNPs in this candidate 10q24.2 locus in 339 Han Chinese BA patients and 401 matched controls. The SNP rs17095355 achieved genome-wide significance with a p-value of 10−10 . Furthermore, a risk haplotype comprising 5 tag SNPs was identified with odds ratio 2.38 for BA. The risk haplotype was significantly correlated (P = 0.003) with ADD3 gene expression in RNA samples from 37 of 339 genotyped cases. These RNA tissue samples were obtained at portenterostomy in 26 cases and at liver transplantation in 11 cases. Sanger sequencing did not reveal a novel variant
© 2015 Wiley Periodicals, Inc.
269
wires.wiley.com/sysbio
Advanced Review
within this risk haplotype. The prevalence of the risk haplotype mirrored the population incidence of BA among Chinese populations. Because several SNPs that were associated with BA were located in the ADD3 promoter region or putative enhancer or transcription factor binding sites, the contribution of this ADD3 locus to BA is felt to be regulatory. The association of the 10q24.2 locus with BA in a large cohort of Chinese BA patients led Tsai et al.33 to test whether this association was also present in a North American BA cohort. Genotypes at 550,000 SNP loci were evaluated in 171 BA cases and 1630 controls. From these data, 333 SNPs within a 2 MB region around SNP rs17095355 and encompassing the XPNPEP1 and ADD3 genes were selected for comparison after imputation. The SNP rs17095355 itself was not present in the Illumina 550 k or 660 k SNP arrays used for genome-wide genotyping of this North American cohort. The imputed SNP with the most significant association with BA was rs7906904, which was located in the first intron of ADD3. To establish disease-specificity, liver tissue was immunostained for XPNPEP1 and ADD3 proteins in BA and normal control groups. Significant differences in ADD3 but not XNPEP1 immunostaining were observed between normal and BA liver tissue. In a personal communication, one of the co-authors of this study also reported that add3 knockdown in zebrafish resulted in the same defects in biliary network which were observed with gpc1 knockdown. This association suggested that, although ADD3 was a viable susceptibility gene for BA, the markers reflecting this genetic association were different in the Chinese and the North American populations. Interestingly, rs17095355, the SNP locus associated with BA in Chinese populations, was also significantly associated with BA in a Thai cohort of 124 BA patients compared with 114 controls.34 As in the North American study, these replication results were arrived at by evaluating the candidate locus in isolation, without simultaneous consideration of genome-wide SNP loci outside of this locus.
Adenosine Diphosphate Ribosylation Factor 6 and BA To identify potential susceptibility loci, Ningappa et al.35 compared genotypes at >550,000 SNP loci in 80 Caucasian BA cases and 2818 Caucasian controls. BA cases were accrued at two centers, the Children’s Hospital of Pittsburgh (CHP) of University of Pittsburgh Medical Center and the Center for Applied Genomics at the Children’s Hospital of Philadelphia (CAG). Controls were recruited at CAG. Population substructure was further evaluated with principal components analysis (PCA) of genotyping data. 270
Thirty-nine of 50 CHP cases, 24 of 30 CAG cases, and 1914 normal controls, which clustered together on this PCA plot, were carried forward for association analysis. To boost confidence in the association, SNPs were further ranked based on proximity to neighboring SNP(s) in a 10 kb window from among the top-ranked 1000 SNPs in the CHP cohort. The SNPs rs3126184 and rs10140366 in a 3′ flanking region of the ARF6 gene on chr. 14q21.3 demonstrated higher minor allele frequencies (MAF) in each cohort, and 63 combined cases when compared with controls (0.286 vs 0.131, P = 5.94 × 10−7 , 0.286 vs 0.13, P = 5.57 × 10−7 ). Fourteen additional BA cases were genotyped at rs3126184 only using commercially available probes. When these data were combined with those for the 63 cases, the association between rs3126184 and BA remained significant in 77 combined cases (P = 4.2 × 10−8 , OR = 2.66). Both SNPs occupied a putative enhancer region in the 3′ flanking region of the ARF6 gene. ARF6 gene expression is negatively correlated with the minor alleles of rs1040366 (r = −0.1346, P = 0.023) and rs3126184 (r = −0.1128, P = 0.061) in the combined HapMap populations (SNPexp v1.2). ARF6 regulates liver development.36 ARF6 is activated when the signaling molecule GEP100 ligates to the EGF-EGFR complex.37 Activated ARF6 exerts several downstream effects. Therefore, a systems biology approach was used to search for additional loci among the 1000 top-ranked SNPs, which might also contribute to ARF6 signaling. This analysis was based on the presumption that multiple genes, each with small effects, together accounted for the disease phenotype. Pathway analysis of these top-ranked SNPs revealed enrichment of genes for EGF regulation, ERK/MAPK and CREB KEGG canonical pathways and functional networks for cellular development and proliferation. Activation of ARF6 is known to promote cell proliferation and development via sequential activation of the ERK/MAPK and CREB signaling pathways.38–41 To establish disease-specificity, normal control, disease control, and BA liver tissues from 29 of 39 CHP cases were immunostained with antibody to ARF6. No differences were seen between 27 BA liver explants which had diffuse cirrhosis and normal liver tissue. Poor ARF6 immunostaining was seen in the remaining two liver explants with bile duct paucity and minimal fibrosis, compared with normal liver tissue. Together these findings suggested that decreased ARF6 expression might contribute to intrahepatic bile duct paucity in a subset of BA patients. This hypothesis was pursued further by evaluating the effects of
© 2015 Wiley Periodicals, Inc.
Volume 7, September/October 2015
WIREs Systems Biology and Medicine
GWAS in biliary atresia
arf6 knockdown in 2–5 days post-fertilization (dpf) zebrafish larvae. arf6 knockdown in zebrafish was associated with (1) a sparse intrahepatic biliary network and poor bile excretion from the liver into the extrahepatic biliary tree, despite an anatomically normal extrahepatic biliary tree, (2) reduced biliary canalicular length and reduced BEC mobility leading to enhanced clustering of BEC during bile duct morphogenesis, (3) upregulation of EGFR pathway genes, egfra, gep100, arf6,ddef1, and the hedgehog signaling gene, ptch1, but not gli1 and gli2a, the other genes in the canonical hedgehog signaling pathway, in zebrafish liver tissue, and (4) poor development of other endodermal organs manifested by a small pancreas and intestinal bulb. These defects were rescued with arf6 mRNA and reproduced with the EGFR blocking agent AG1478 in 2–5 dpf zebrafish larvae. These findings led to the conclusion that a chromosome 14q21.3 region encompassing the ARF6 gene was a susceptibility locus for BA. That this locus likely contributes to poor bile duct development in BA via inhibitory effects on ERK-MAPK and CREB signaling pathways awaits additional studies.
WHAT DO THESE GWAS STUDIES TELL US ABOUT BA? The abovementioned studies lead us to the following inferences: • BA is associated with genetic variants, which affect multiple susceptibility genes. By themselves, these variants may be insufficient to cause the development of BA and may require help from other yet to be determined susceptibility genes, or extraneous triggers, e.g., toxins.42,43 This is exemplified best by an inherited 1.76 MB heterozygous deletion in the GPC1 gene in a child with BA, but not in the father who is an asymptomatic carrier. • With the exception of the Chinese BA cohort which consisted of nonsyndromic, presumably ‘isolated’ BA, the disease cohorts in these studies
appear inclusive of the various subtypes of BA, of which the isolated or perinatal form is four times more common than the embryonic form of BA. Therefore, the biology suggested by GWAS findings is applicable to all types especially the isolated variety. This type of an inclusive test cohort is unavoidable, because some subsets of patients including, those with post-infectious BA, cannot be identified with certainty. Therefore, susceptibility genes that participate in bile duct development may contribute to pathogenesis of both forms of BA. This is important to note when considering the clinical relevance of either infectious or developmental models of BA, or in predicting the outcome of therapeutic interventions based exclusively on either model. • Although distinct, GPC1, ADD3, and ARF6 share several functional attributes including roles in the development of the liver and bile ducts. All are expressed in the liver and the bile ducts. ADD3 and ARF6 regulate membrane function, cell growth and cell development. ADD3 is also expressed to a greater extent in fetal liver than in the adult liver.44 Knockdown of each of these genes impairs biliary network formation in zebrafish larvae. GPC1 and ARF6 participate in or are targeted by FGF and EGFR signaling,45,38–41 respectively. Both of these signaling pathways can facilitate organogenesis via branching morphogenesis of epithelial cells. Further, the majority of the identified susceptibility loci including ADD3 and ARF6 may contribute to BA via regulatory effect. Finally, the contribution of each locus may be modest in causing the development of BA. • Systems biology re-analyses of GWAS results can strengthen the validity of an association further by identifying other susceptibility genes in pathways related to the primary association. Subsequent validation of these multiple genes in established models like the zebrafish knockout model may help to map unique pathways of BA by generating supportive evidence.
ACKNOWLEDGMENTS This work was supported by the Hillman Foundation of Pittsburgh, Nancy Foster and Team Transplant, and the Herridge and Giventer-Braff families. We thank Ms. Dale Zecca for this manuscript preparation.
Volume 7, September/October 2015
© 2015 Wiley Periodicals, Inc.
271
wires.wiley.com/sysbio
Advanced Review
REFERENCES 1. Perlmutter DH, Shepherd RW. Extrahepatic biliary atresia: a disease or a phenotype? Hepatology 2002, 35:1297–1304.
15. Shim WK, Kasai M, Spence MA. Racial influence on the incidence of biliary atresia. Prog Pediatr Surg 1974, 6:53–62.
2. Weeks DE, Lathrop GM. Polygenic disease: methods for mapping complex disease traits. Trends Genet 1995, 11:513–519.
16. Petersen C, Harder D, Abola Z, Alberti D, Becker T, Chardot C, Davenport M, Deutschmann A, Khelif K, Kobayashi H, et al. European biliary atresia registries: summary of a symposium. Eur J Pediatr Surg 2008, 18:111–116. doi:10.1055/s-2008-1038479.
3. Karrer FM, Price MR, Bensard DD, Sokol RJ, Narkewicz MR, Smith DJ, Lilly JR. Long-term results with the Kasai operation for biliary atresia. Arch Surg 1996, 131:493–496. 4. Schwarz KB, Haber BH, Rosenthal P, Mack CL, Moore J, Bove K, Bezerra JA, Karpen SJ, Kerkar N, Shneider BL, et al. Extrahepatic anomalies in infants with biliary atresia: results of a large prospective North American multicenter study. Hepatology 2013, 58:1724–1731. doi:10.1002/hep.26512. 5. Guttman OR, Roberts EA, Schreiber RA, Barker CC, Ng VL, Canadian Pediatric Hepatology Research Group. Biliary atresia with associated structural malformations in Canadian infants. Liver Int 2011, 31: 1485–1493. doi:10.1111/j.1478-3231.2011.02578.x. 6. Mack CL, Sokol RJ. Unraveling the pathogenesis and etiology of biliary atresia. Pediatr Res 2005, 57:87R–94R. 7. Tyler KL, Sokol RJ, Oberhaus SM, Le M, Karrer FM, Narkewicz MR, Tyson RW, Murphy JR, Low R, Brown WR. Detection of reovirus RNA in hepatobiliary tissues from patients with extrahepatic biliary atresia and choledochal cysts. Hepatology 1998, 27:1475–1482. 8. Riepenhoff-Talty M, Gouvea V, Evans MJ, Svensson L, Hoffenberg E, Sokol RJ, Uhnoo I, Greenberg SJ, Schäkel K, Zhaori G. Detection of group C rotavirus in infants with extrahepatic biliary atresia. J Infect Dis 1996, 174:8–15. 9. Tarr PI, Haas JE, Christie DL. Biliary atresia, cytomegalovirus, and age at referral. Pediatrics 1996, 97:828–831. 10. Petersen C, Biermanns D, Kuske M, Schäkel K, MeyerJunghänel L, Mildenberger H. New aspects in a murine model for extrahepatic biliary atresia. J Pediatr Surg 1997, 32:1190–1195. 11. Yoon PW, Bresee JS, Olney RS, James LM, Khoury MJ. Epidemiology of biliary atresia: a population-based study. Pediatrics 1997, 99:376–382. 12. Strickland AD, Shannon K. Studies in the etiology of extrahepatic biliary atresia: time-space clustering. J Pediatr 1982, 100:749–753. 13. Caton AR, Druschel CM, McNutt LA. The epidemiology of extrahepatic biliary atresia in New York State, 1983–98. Paediatr Perinat Epidemiol 2004, 18:97–105. 14. Nakamura K, Tanoue A. Etiology of biliary atresia as a developmental anomaly: recent advances. J Hepatobiliary Pancreat Sci 2013, 20:459–464. doi:10.1007/s00534-013-0604-4.
272
17. Hsiao CH, Chang MH, Chen HL, Lee HC, Wu TC, Lin CC, Yang YJ, Chen AC, Tiao MM, Lau BH, et al. Universal screening for biliary atresia using an infant stool color card in Taiwan. Hepatology 2008, 47:1233–1240. doi:10.1002/hep.22182. 18. McCarthy MI, Abecasis GR, Cardon LR, Goldstein DB, Little J, Ioannidis JP, Hirschhorn JN. Genome-wide association studies for complex traits: consensus, uncertainty and challenges. Nat Rev Genet 2008, 9:356–369. doi:10.1038/nrg2344. 19. Stranger BE, Stahl EA, Raj T. Progress and promise of genome-wide association studies for human complex trait genetics. Genetics 2011, 187:367–383. doi:10.1534/genetics.110.120907. 20. Cantor RM, Lange K, Sinsheimer JS. Prioritizing GWAS results: a review of statistical methods and recommendations for their application. Am J Hum Genet 2010, 86:6–22. doi:10.1016/j.ajhg.2009.11.017. 21. Zeggini E, Weedon MN, Lindgren CM, Frayling TM, Elliott KS, Lango H, Timpson NJ, Perry JR, Rayner NW, Freathy RM, et al. Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes. Science 2007, 316:1336–1341. 22. Gudmundsson J, Sulem P, Manolescu A, Amundadottir LT, Gudbjartsson D, Helgason A, Rafnar T, Bergthorsson JT, Agnarsson BA, Baker A, et al. Genome-wide association study identifies a second prostate cancer susceptibility variant at 8q24. Nat Genet 2007, 39:631–637. 23. Leyva-Vega M, Gerfen J, Thiel BD, Jurkiewicz D, Rand EB, Pawlowska J, Kaminska D, Russo P, Gai X, Krantz ID, et al. Genomic alterations in biliary atresia suggest region of potential disease susceptibility in 2q37.3. Am J Med Genet A 2010, 152A:886–895. doi:10.1002/ajmg.a.33332. 24. Shaikh TH, Gai X, Perin JC, Glessner JT, Xie H, Murphy K, O’Hara R, Casalunovo T, Conlin LK, D’Arcy M, et al. High-resolution mapping and analysis of copy number variations in the human genome: a data resource for clinical and research applications. Genome Res 2009, 19:1682–1690. doi:10.1101/gr.083501.108. 25. Schreiber RA, Kleinman RE. Genetics, immunology, and biliary atresia: an opening or a diversion? J Pediatr Gastroenterol Nutr 1993, 16:111–113. 26. Cui S, Leyva-Vega M, Tsai EA, EauClaire SF, Glessner JT, Hakonarson H, Devoto M, Haber BA, Spinner NB, Matthews RP. Evidence from human and
© 2015 Wiley Periodicals, Inc.
Volume 7, September/October 2015
WIREs Systems Biology and Medicine
GWAS in biliary atresia
zebrafish that GPC1 is a biliary atresia susceptibility gene. Gastroenterology 2013, 144:1107e3–1115e3. doi:10.1053/j.gastro.2013.01.022. 27. Glessner JT, Wang K, Cai G, Korvatska O, Kim CE, Wood S, Zhang E, Estes A, Brune CW, Bradfield JP, et al. Autism genome-wide copy number variation reveals ubiquitin and neuronal genes. Nature 2009, 459:569–573. 28. Grisaru S, Cano-Gauci D, Tee J, Filmus J, Rosenblum ND. Glypican-3 modulates BMP and FGF-mediated effects during renal branching morphogenesis. Dev Biol 2001, 231:31–46. 29. Yan D, Lin X. Drosophila glypican Dally-like acts in FGF-receiving cells to modulate FGF signaling during tracheal morphogenesis. Dev Biol 2007, 312:203–216. 30. Yan D, Lin X. Opposing roles for glypicans in Hedgehog signalling. Nat Cell Biol 2008, 10:761–763. 31. Garcia-Barcelo MM, Yeung MY, Miao XP, Tang CS, Cheng G, So MT, Ngan ES, Lui VC, Chen Y, Liu XL, et al. Genome-wide association study identifies a susceptibility locus for biliary atresia on 10q24.2. Hum Mol Genet 2010, 19:2917–2925. doi:10.1093/hmg/ddq196.
36. Suzuki T, Kanai Y, Hara T, Sasaki J, Sasaki T, Kohara M, Maehama T, Taya C, Shitara H, Yonekawa H, et al. Crucial role of the small GTPase ARF6 in hepatic cord formation during liver development. Mol Cell Biol 2006, 26:6149–6156. 37. Morishige M, Hashimoto S, Ogawa E, Toda Y, Kotani H, Hirose M, Wei S, Hashimoto A, Yamada A, Yano H, et al. GEP100 links epidermal growth factor receptor signalling to ARF6 activation to induce breast cancer invasion. Nat Cell Biol 2008, 10:85–92. 38. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res 2000, 28:27–30, KEGG ID (hsa04010). 39. Iguchi H, Mitsui T, Ishida M, Kanba S, Arita J. cAMP response element-binding protein (CREB) is required for epidermal growth factor (EGF)-induced cell proliferation and serum response element activation in neural stem cells isolated from the forebrain subventricular zone of adult mice. Endocr J 2011, 58:747–759. 40. Sarró E, Tornavaca O, Plana M, Meseguer A, Itarte E. Phosphoinositide 3-kinase inhibitors protect mouse kidney PCT3 cells from cyclosporine- induced cell death. Kidney Int 2008, 73:77–85.
32. Cheng G, Tang CS, Wong EH, Cheng WW, So MT, Miao X, Zhang R, Cui L, Liu X, Ngan ES, et al. Common genetic variants regulating ADD3 gene expression alter biliary atresia risk. J Hepatol 2013, 59:1285–1291. doi:10.1016/j.jhep.2013.07.021.
41. Hu Z, Du J, Yang L, Zhu Y, Yang Y, Zheng D, Someya A, Gu L, Lu X. GEP100/Arf6 is required for epidermal growth factor-induced ERK/Rac1 signaling and cell migration in human hepatoma HepG2 cells. PLoS One 2012, 7:e38777.
33. Tsai EA, Grochowski CM, Loomes KM, Bessho K, Hakonarson H, Bezerra JA, Russo PA, Haber BA, Spinner NB, Devoto M. Replication of a GWAS signal in a Caucasian population implicates ADD3 in susceptibility to biliary atresia. Hum Genet 2014, 133:235–243. doi:10.1007/s00439-013-1368-2.
42. Harper P, Plant JW, Unger DB. Congenital biliary atresia and jaundice in lambs and calves. Aust Vet J 1990, 67:18–22.
34. Kaewkiattiyot S, Honsawek S, Vejchapipat P, Chongsrisawat V, Poovorawan Y. Association of Xprolyl aminopeptidase 1 rs17095355 polymorphism with biliary atresia in Thai children. Hepatol Res 2011, 41:1249–1252. doi:10.1111/j.1872-034X.2011. 00870.x. 35. Ningappa M, So J, Glessner J, Ashokkumar C, Ranganathan S, Min J, Higgs BW, Sun Q, Haberman K, Schmitt L, et al. The role of ARF6 in biliary atresia. PLoS One. In press.
Volume 7, September/October 2015
43. Lorent K, Koo KA, Gong W, Whittaker S, Porter JR, Wells R. Michael pack isolation and identification of a plant toxin that induces biliary atresia in livestock and zebrafish. In: The 63rd Annual Meeting of American Association for the Study of Liver Diseases, Boston, MA, 2012. 44. Ku NO, Zhou X, Toivola DM, Omary MB. The cytoskeleton of digestive epithelia in health and disease. Am J Physiol 1999, 277:G1108–G1137. 45. Jen YH, Musacchio M, Lander AD. Glypican-1 controls brain size through regulation of fibroblast growth factor signaling in early neurogenesis. Neural Dev 2009, 4:33.
© 2015 Wiley Periodicals, Inc.
273
Genome-wide association studies in biliary atresia Mylarappa Ningappa,1 Jun Min,2 Brandon W. Higgs,1 Chethan Ashokkumar,1 Sarangarajan Ranganathan1 and Rakesh Sindhi1∗ Biliary atresia (BA) is a model complex disease resulting from interactions between multiple susceptibility loci and environmental factors. This perception is based on a heterogeneous phenotype extending beyond an absent extrahepatic bile duct to include gut and cardiovascular anomalies, and the association of BA with viral infections. Refractory jaundice and progression to cirrhosis shortly after birth can be fatal without surgical correction, and further suggests a pathogenesis during liver and bile duct development. Conclusive proof for a developmental origin would require documentation of disease progression in the perinatal or fetal liver, an impossible task for obvious reasons. We review three different sets of genome-wide association studies (GWAS) from three different cohorts of BA patients by three different groups of investigators, which address this knowledge gap. Knockdown of each susceptibility gene identified by GWAS in zebrafish embryos impairs excretion of bile from the liver, duplicating the characteristic diagnostic finding seen in affected children. This finding is associated with impaired intrahepatic biliary network formation in zebrafish morphants. Although distinct, these susceptibility genes share several functions including roles in mechanisms for organogenesis (glypican 1 or GPC1, and adenosine diphosphate ribosylation factor 6, or ARF6) or a greater expression in fetal liver than in adult liver (adducin 3 or ADD3). Together, these studies emphasize the importance of the human evidence, and present opportunities to map novel pathways which explain the phenotypic heterogeneity of BA. © 2015 Wiley Periodicals, Inc. How to cite this article:
WIREs Syst Biol Med 2015, 7:267–273. doi: 10.1002/wsbm.1303
BILIARY ATRESIA: A COMPLEX DISEASE TRAIT
B
y any standard, the heterogeneous phenotype of biliary atresia (BA) defies a unified mechanistic explanation, is a likely consequence of multiple susceptibility loci interacting with environmental factors, and is, therefore, a model complex disease.1,2 Characterized by impaired extrahepatic bile drainage ∗ Correspondence
to: [email protected]
1 Department
of Surgery, Hillman Center for Pediatric Transplantation, Children’s Hospital of Pittsburgh of University of Pittsburgh Medical Center (UPMC), Pittsburgh, PA, USA 2 Department of Bioengineering, University of California San Diego, La Jolla, CA, USA Conflict of interest: The authors have declared no conflicts of interest for this article.
Volume 7, September/October 2015
and refractory jaundice in newborns, the disease is fatal if not corrected surgically.3 BA can also be associated with a variety of structural gastrointestinal and cardiovascular anomalies, some of which are also fatal without surgical correction.4,5 Most cases are ‘isolated’ or ‘perinatal’ with obstructive fibrosis of the extrahepatic biliary system.6 The presumed etiology is a perinatal viral cholangitis with exaggerated immune responses in the setting of familial predisposition to autoimmunity. The evidence for this hypothesis is circumstantial because of the impossibility of demonstrating progressive viral invasion and inflammation in sequential fetal biliary tissue samples. This evidence includes (1) viral particles of or viral sensitization toward reovirus, rotavirus, or cytomegalovirus in affected tissue in some human
© 2015 Wiley Periodicals, Inc.
267
wires.wiley.com/sysbio
Advanced Review
studies,7–9 (2) the induction of an analogous phenotype with these viruses in experimental models,10 and (3) seasonal peaks in BA cases from many parts of the world.11–13 The less common ‘syndromic’ or ‘embryonic’ BA consists of those cases which present earlier than isolated BA and also manifest major structural anomalies including laterality defects such as asplenia, malrotation, and heterotaxy in varying combinations or to varying degrees.4,5 These phenotypic categories are linked by an important fact: a discernible proportion of children with isolated BA also manifest minor anomalies in the same extrahepatic systems as those with syndromic BA.4 Recent genome-wide association studies (GWAS) of BA patients reflect a collective quest for novel unbiased hypotheses from a multidisciplinary constituency of caregivers comprising basic scientists, physicians, and surgeons, who are committed to alleviating the societal burden of BA. With an incidence of 1:5000 live births in Asian populations, and 1:18,000 live births in western Caucasian populations, BA accounts for at least one-third of all liver transplant procedures performed in children worldwide.14–17
GWAS AS A DISCOVERY METHOD FOR A COMPLEX TRAIT Toward hypothesis-generation, trait mapping with GWAS is based on the premise that a marker allele which ‘tags’ or represents a haplotype, and which shows a strong association with a disease trait, is likely to be associated with a novel, potentially causal variant(s) within that haplotype because of linkage disequilibrium.18–20 The causal variant(s) may be identified in subsequent studies, or may remain simply associated with the disease trait. In practice, this approach relies on arrays which characterize hundreds of thousands or a few million genome-wide ‘tag’ single-nucleotide polymorphisms (SNPs) or copy number variants (CNVs). These SNP arrays are used to genotype-unrelated cases, or subjects expressing the phenotype and unrelated controls, or healthy subjects not expressing the phenotype, all with a similar genetic background. Candidate SNPs are those with significantly different minor allele frequencies in the cases population compared with the control population. Replication in several independent disease and control subject cohorts, and across different geographic or ethnic groups strengthens the validity of an association. If this population heterogeneity is appropriately accounted for at this stage of the association testing, unrecognized disease heterogeneity among cases may be one reason for failure to replicate an association in a case–control design. This limitation is pertinent to 268
BA, where multiple loci with small biological effects acting in various combinations may contribute to the constellation of ‘minor’ or ‘major’ malformations associated with BA. Another compounding factor is the fact that the exact extent of associated extrahepatic anomalies, and, therefore, of the complexity of BA, is unknown. This is illustrated by the more subtle correlates of BA such as a ‘hypoplastic’ portal vein.4,5 Despite controlled data accrual and comprehensive analysis methods conducted by well-organized national consortia, recognition of a hypoplastic portal vein in a child with BA may occur in retrospect in the form of asymptomatic portal venous thrombosis on surveillance abdominal scans several years after liver transplantation. Virtually all children with this rare surgical complication have received liver transplantation for BA (Rakesh Sindhi, personal communication, December 2014). Also pertinent to rare diseases like BA is the validity and biological plausibility of any association. Dense genome-wide SNP scans mandate correction for testing of multiple hypotheses necessitating P-values less than or approaching 10−7 for 500,000 SNPs or 10−8 for 1–2 million SNPs.18 This objective requires large numbers of cases to achieve adequate power to detect an association. An acceptable alternative is demonstrating genome-wide significance in combined discovery and replication cohorts, though in BA, even combined cohorts may not provide sufficient sample sizes. Identification of putative associated SNP allele(s) is only the first step in GWAS analysis. Post-GWAS efforts should include discovery of novel causal variants with resequencing of the disease loci tagged by the primary SNP association, then functional studies, because the causal determinant (risk or protective) is assumed to be the direct link to the expressed phenotype. This objective can be challenging, however, because such loci often occupy sequences with potentially unknown function, at great distances from the coding region of the nearest gene, thus raising important questions around the length and depth of coverage for the resequencing effort.21,22 The ultimate goal of any association study is to identify the functional relevance of the disease-specific variant to disease mechanism. The GWAS reports summarized below illustrate several approaches, which may help to map disease pathways for a rare and complex disease like BA. Each of these studies presents differential (protein) expression of the candidate gene(s) in diseased human tissue as evidence of disease-specificity. Additional validation efforts range from classical discovery set-replication set testing, to mapping signaling
© 2015 Wiley Periodicals, Inc.
Volume 7, September/October 2015
WIREs Systems Biology and Medicine
GWAS in biliary atresia
pathways from the larger list of significant SNPs as supportive evidence for the mechanism suggested by the primary association. Modeling the developmental effects of the candidate(s) with knockdown in zebrafish embryos adds to the biological relevance of these studies.
SUMMARY OF GWAS STUDIES AND SUSCEPTIBILITY LOCI FOR BILIARY ATRESIA Glypican 1 and BA To evaluate the association of BA with germline CNV, Leyva-Vega et al.23 compared genotypes at 550,000 SNP loci for 35 children and 2026 controls using a previously described CNV calling algorithm.23,24 Heterozygous deletions were identified in two of 35 unrelated BA patients in a common 280-SNP 1.76 MB region comprising 30 genes on chr 2q37.3. The deletions were confirmed with fluorescent in situ hybridization (FISH). Both children received liver transplantation after a failed portoenterostomy by ages 10 and 11 months, respectively. In the first of two patients, the deletion was also identified in the father, who did not experience liver disease. A venous hemangioma was identified in the right tentorium at age 17 years in this first child. In the second child, the 1.76 MB deletion was contained within a larger 1346 SNP, 5.87 Mb heterozygous deletion on chromosome 2q37.3, and accompanied by dysmorphic facies, developmental delay, hypothyroidism, asplenia, and the 47 XXY karyotype consistent with Klinefelter’s syndrome. Of the 30 genes in the common 1.76 MB deletion, the authors sequenced the alanine-glyoxylate aminotransferase (AGXT) gene to identify additional causal variants. AGXT is only expressed in the liver. Several common SNPs were identified. These findings suggested that, although the CNV was inherited in the first child, the deletion alone was insufficient to produce BA because the father with the deletion did not manifest liver disease. Therefore, extraneous factor(s) such as toxins, viral infections, and so on, could have been potential participating event(s).25 In a subsequent study from the same group, Cui et al.26 evaluated the association of BA with germline CNVs by comparing 61 BA cases and 5088 controls at ≥550,000 SNP loci using the Penn CNV calling algorithm.27 A CNV consisting of a heterozygous deletion was localized to the same 2q37.3 region, but to a single gene, glypican 1 (GPC1). GPC 1 is a proteoglycan, which regulates hedgehog signaling and inflammation.28–30 In zebrafish larvae, gpc1 knockdown with morpholino antisense oligonucleotides Volume 7, September/October 2015
(MOs) resulted in poor biliary network formation, a small gall bladder, and poor bile excretion from the liver 5 days post-fertilization (dpf). These changes were associated with upregulation of hedgehog target genes, gli2a, ptch1, foxl1, znf697, and ccnd1, and partly rescued by injecting zebrafish larvae with gpc1 messenger RNA and the hedgehog antagonist, cyclopamine. The authors concluded that the decreased availability of the glypican1 ligand acted as a ‘sink’, resulting in upregulated hedgehog signaling genes. Immunostaining of normal human liver tissue demonstrated apical distribution of GPC1. This type of staining was poorly seen or absent in the liver tissues from the three patients with BA, none of whom carried deletions in the GPC1 gene. These human findings also suggested that abnormal GPC1 expression may be an independent correlate of BA.
Adducin-3 and BA Garcia-Barcelo et al.31 compared 181 Chinese BA patients and 481 healthy ethnically matched controls at >289,000 SNP loci using an Affymetrix genome-wide SNP array. In this discovery phase, several SNPs including rs12571674 and rs17095355 in a 129-kb intergenic locus on chr 10q24.2 were strongly associated with BA. In 124 independent Chinese BA patients and 90 controls, the association between rs17095355 and BA was replicated. This intergenic SNP is flanked by the X-prolyl aminopeptidase P (XPNPEP1) and adducin 3 (ADD3) genes. ADD3 is expressed in the liver and biliary epithelial cells (BECs), more so in fetal livers, and is necessary for remodeling of membrane cytoskeleton at points of cell-cell contact. XPNPEP1 is also found in liver epithelial cells and may be important in bile acid excretion. No functional role has been assigned to the rs17095355 SNP locus in databases such as ENCODE. However, the size of the BA cohort and replication in an independent cohort strongly support a role for this locus including its flanking genes in BA. To ‘fine’ map the disease locus further, Cheng et al.32 from the same investigator group genotyped 107 SNPs in this candidate 10q24.2 locus in 339 Han Chinese BA patients and 401 matched controls. The SNP rs17095355 achieved genome-wide significance with a p-value of 10−10 . Furthermore, a risk haplotype comprising 5 tag SNPs was identified with odds ratio 2.38 for BA. The risk haplotype was significantly correlated (P = 0.003) with ADD3 gene expression in RNA samples from 37 of 339 genotyped cases. These RNA tissue samples were obtained at portenterostomy in 26 cases and at liver transplantation in 11 cases. Sanger sequencing did not reveal a novel variant
© 2015 Wiley Periodicals, Inc.
269
wires.wiley.com/sysbio
Advanced Review
within this risk haplotype. The prevalence of the risk haplotype mirrored the population incidence of BA among Chinese populations. Because several SNPs that were associated with BA were located in the ADD3 promoter region or putative enhancer or transcription factor binding sites, the contribution of this ADD3 locus to BA is felt to be regulatory. The association of the 10q24.2 locus with BA in a large cohort of Chinese BA patients led Tsai et al.33 to test whether this association was also present in a North American BA cohort. Genotypes at 550,000 SNP loci were evaluated in 171 BA cases and 1630 controls. From these data, 333 SNPs within a 2 MB region around SNP rs17095355 and encompassing the XPNPEP1 and ADD3 genes were selected for comparison after imputation. The SNP rs17095355 itself was not present in the Illumina 550 k or 660 k SNP arrays used for genome-wide genotyping of this North American cohort. The imputed SNP with the most significant association with BA was rs7906904, which was located in the first intron of ADD3. To establish disease-specificity, liver tissue was immunostained for XPNPEP1 and ADD3 proteins in BA and normal control groups. Significant differences in ADD3 but not XNPEP1 immunostaining were observed between normal and BA liver tissue. In a personal communication, one of the co-authors of this study also reported that add3 knockdown in zebrafish resulted in the same defects in biliary network which were observed with gpc1 knockdown. This association suggested that, although ADD3 was a viable susceptibility gene for BA, the markers reflecting this genetic association were different in the Chinese and the North American populations. Interestingly, rs17095355, the SNP locus associated with BA in Chinese populations, was also significantly associated with BA in a Thai cohort of 124 BA patients compared with 114 controls.34 As in the North American study, these replication results were arrived at by evaluating the candidate locus in isolation, without simultaneous consideration of genome-wide SNP loci outside of this locus.
Adenosine Diphosphate Ribosylation Factor 6 and BA To identify potential susceptibility loci, Ningappa et al.35 compared genotypes at >550,000 SNP loci in 80 Caucasian BA cases and 2818 Caucasian controls. BA cases were accrued at two centers, the Children’s Hospital of Pittsburgh (CHP) of University of Pittsburgh Medical Center and the Center for Applied Genomics at the Children’s Hospital of Philadelphia (CAG). Controls were recruited at CAG. Population substructure was further evaluated with principal components analysis (PCA) of genotyping data. 270
Thirty-nine of 50 CHP cases, 24 of 30 CAG cases, and 1914 normal controls, which clustered together on this PCA plot, were carried forward for association analysis. To boost confidence in the association, SNPs were further ranked based on proximity to neighboring SNP(s) in a 10 kb window from among the top-ranked 1000 SNPs in the CHP cohort. The SNPs rs3126184 and rs10140366 in a 3′ flanking region of the ARF6 gene on chr. 14q21.3 demonstrated higher minor allele frequencies (MAF) in each cohort, and 63 combined cases when compared with controls (0.286 vs 0.131, P = 5.94 × 10−7 , 0.286 vs 0.13, P = 5.57 × 10−7 ). Fourteen additional BA cases were genotyped at rs3126184 only using commercially available probes. When these data were combined with those for the 63 cases, the association between rs3126184 and BA remained significant in 77 combined cases (P = 4.2 × 10−8 , OR = 2.66). Both SNPs occupied a putative enhancer region in the 3′ flanking region of the ARF6 gene. ARF6 gene expression is negatively correlated with the minor alleles of rs1040366 (r = −0.1346, P = 0.023) and rs3126184 (r = −0.1128, P = 0.061) in the combined HapMap populations (SNPexp v1.2). ARF6 regulates liver development.36 ARF6 is activated when the signaling molecule GEP100 ligates to the EGF-EGFR complex.37 Activated ARF6 exerts several downstream effects. Therefore, a systems biology approach was used to search for additional loci among the 1000 top-ranked SNPs, which might also contribute to ARF6 signaling. This analysis was based on the presumption that multiple genes, each with small effects, together accounted for the disease phenotype. Pathway analysis of these top-ranked SNPs revealed enrichment of genes for EGF regulation, ERK/MAPK and CREB KEGG canonical pathways and functional networks for cellular development and proliferation. Activation of ARF6 is known to promote cell proliferation and development via sequential activation of the ERK/MAPK and CREB signaling pathways.38–41 To establish disease-specificity, normal control, disease control, and BA liver tissues from 29 of 39 CHP cases were immunostained with antibody to ARF6. No differences were seen between 27 BA liver explants which had diffuse cirrhosis and normal liver tissue. Poor ARF6 immunostaining was seen in the remaining two liver explants with bile duct paucity and minimal fibrosis, compared with normal liver tissue. Together these findings suggested that decreased ARF6 expression might contribute to intrahepatic bile duct paucity in a subset of BA patients. This hypothesis was pursued further by evaluating the effects of
© 2015 Wiley Periodicals, Inc.
Volume 7, September/October 2015
WIREs Systems Biology and Medicine
GWAS in biliary atresia
arf6 knockdown in 2–5 days post-fertilization (dpf) zebrafish larvae. arf6 knockdown in zebrafish was associated with (1) a sparse intrahepatic biliary network and poor bile excretion from the liver into the extrahepatic biliary tree, despite an anatomically normal extrahepatic biliary tree, (2) reduced biliary canalicular length and reduced BEC mobility leading to enhanced clustering of BEC during bile duct morphogenesis, (3) upregulation of EGFR pathway genes, egfra, gep100, arf6,ddef1, and the hedgehog signaling gene, ptch1, but not gli1 and gli2a, the other genes in the canonical hedgehog signaling pathway, in zebrafish liver tissue, and (4) poor development of other endodermal organs manifested by a small pancreas and intestinal bulb. These defects were rescued with arf6 mRNA and reproduced with the EGFR blocking agent AG1478 in 2–5 dpf zebrafish larvae. These findings led to the conclusion that a chromosome 14q21.3 region encompassing the ARF6 gene was a susceptibility locus for BA. That this locus likely contributes to poor bile duct development in BA via inhibitory effects on ERK-MAPK and CREB signaling pathways awaits additional studies.
WHAT DO THESE GWAS STUDIES TELL US ABOUT BA? The abovementioned studies lead us to the following inferences: • BA is associated with genetic variants, which affect multiple susceptibility genes. By themselves, these variants may be insufficient to cause the development of BA and may require help from other yet to be determined susceptibility genes, or extraneous triggers, e.g., toxins.42,43 This is exemplified best by an inherited 1.76 MB heterozygous deletion in the GPC1 gene in a child with BA, but not in the father who is an asymptomatic carrier. • With the exception of the Chinese BA cohort which consisted of nonsyndromic, presumably ‘isolated’ BA, the disease cohorts in these studies
appear inclusive of the various subtypes of BA, of which the isolated or perinatal form is four times more common than the embryonic form of BA. Therefore, the biology suggested by GWAS findings is applicable to all types especially the isolated variety. This type of an inclusive test cohort is unavoidable, because some subsets of patients including, those with post-infectious BA, cannot be identified with certainty. Therefore, susceptibility genes that participate in bile duct development may contribute to pathogenesis of both forms of BA. This is important to note when considering the clinical relevance of either infectious or developmental models of BA, or in predicting the outcome of therapeutic interventions based exclusively on either model. • Although distinct, GPC1, ADD3, and ARF6 share several functional attributes including roles in the development of the liver and bile ducts. All are expressed in the liver and the bile ducts. ADD3 and ARF6 regulate membrane function, cell growth and cell development. ADD3 is also expressed to a greater extent in fetal liver than in the adult liver.44 Knockdown of each of these genes impairs biliary network formation in zebrafish larvae. GPC1 and ARF6 participate in or are targeted by FGF and EGFR signaling,45,38–41 respectively. Both of these signaling pathways can facilitate organogenesis via branching morphogenesis of epithelial cells. Further, the majority of the identified susceptibility loci including ADD3 and ARF6 may contribute to BA via regulatory effect. Finally, the contribution of each locus may be modest in causing the development of BA. • Systems biology re-analyses of GWAS results can strengthen the validity of an association further by identifying other susceptibility genes in pathways related to the primary association. Subsequent validation of these multiple genes in established models like the zebrafish knockout model may help to map unique pathways of BA by generating supportive evidence.
ACKNOWLEDGMENTS This work was supported by the Hillman Foundation of Pittsburgh, Nancy Foster and Team Transplant, and the Herridge and Giventer-Braff families. We thank Ms. Dale Zecca for this manuscript preparation.
Volume 7, September/October 2015
© 2015 Wiley Periodicals, Inc.
271
wires.wiley.com/sysbio
Advanced Review
REFERENCES 1. Perlmutter DH, Shepherd RW. Extrahepatic biliary atresia: a disease or a phenotype? Hepatology 2002, 35:1297–1304.
15. Shim WK, Kasai M, Spence MA. Racial influence on the incidence of biliary atresia. Prog Pediatr Surg 1974, 6:53–62.
2. Weeks DE, Lathrop GM. Polygenic disease: methods for mapping complex disease traits. Trends Genet 1995, 11:513–519.
16. Petersen C, Harder D, Abola Z, Alberti D, Becker T, Chardot C, Davenport M, Deutschmann A, Khelif K, Kobayashi H, et al. European biliary atresia registries: summary of a symposium. Eur J Pediatr Surg 2008, 18:111–116. doi:10.1055/s-2008-1038479.
3. Karrer FM, Price MR, Bensard DD, Sokol RJ, Narkewicz MR, Smith DJ, Lilly JR. Long-term results with the Kasai operation for biliary atresia. Arch Surg 1996, 131:493–496. 4. Schwarz KB, Haber BH, Rosenthal P, Mack CL, Moore J, Bove K, Bezerra JA, Karpen SJ, Kerkar N, Shneider BL, et al. Extrahepatic anomalies in infants with biliary atresia: results of a large prospective North American multicenter study. Hepatology 2013, 58:1724–1731. doi:10.1002/hep.26512. 5. Guttman OR, Roberts EA, Schreiber RA, Barker CC, Ng VL, Canadian Pediatric Hepatology Research Group. Biliary atresia with associated structural malformations in Canadian infants. Liver Int 2011, 31: 1485–1493. doi:10.1111/j.1478-3231.2011.02578.x. 6. Mack CL, Sokol RJ. Unraveling the pathogenesis and etiology of biliary atresia. Pediatr Res 2005, 57:87R–94R. 7. Tyler KL, Sokol RJ, Oberhaus SM, Le M, Karrer FM, Narkewicz MR, Tyson RW, Murphy JR, Low R, Brown WR. Detection of reovirus RNA in hepatobiliary tissues from patients with extrahepatic biliary atresia and choledochal cysts. Hepatology 1998, 27:1475–1482. 8. Riepenhoff-Talty M, Gouvea V, Evans MJ, Svensson L, Hoffenberg E, Sokol RJ, Uhnoo I, Greenberg SJ, Schäkel K, Zhaori G. Detection of group C rotavirus in infants with extrahepatic biliary atresia. J Infect Dis 1996, 174:8–15. 9. Tarr PI, Haas JE, Christie DL. Biliary atresia, cytomegalovirus, and age at referral. Pediatrics 1996, 97:828–831. 10. Petersen C, Biermanns D, Kuske M, Schäkel K, MeyerJunghänel L, Mildenberger H. New aspects in a murine model for extrahepatic biliary atresia. J Pediatr Surg 1997, 32:1190–1195. 11. Yoon PW, Bresee JS, Olney RS, James LM, Khoury MJ. Epidemiology of biliary atresia: a population-based study. Pediatrics 1997, 99:376–382. 12. Strickland AD, Shannon K. Studies in the etiology of extrahepatic biliary atresia: time-space clustering. J Pediatr 1982, 100:749–753. 13. Caton AR, Druschel CM, McNutt LA. The epidemiology of extrahepatic biliary atresia in New York State, 1983–98. Paediatr Perinat Epidemiol 2004, 18:97–105. 14. Nakamura K, Tanoue A. Etiology of biliary atresia as a developmental anomaly: recent advances. J Hepatobiliary Pancreat Sci 2013, 20:459–464. doi:10.1007/s00534-013-0604-4.
272
17. Hsiao CH, Chang MH, Chen HL, Lee HC, Wu TC, Lin CC, Yang YJ, Chen AC, Tiao MM, Lau BH, et al. Universal screening for biliary atresia using an infant stool color card in Taiwan. Hepatology 2008, 47:1233–1240. doi:10.1002/hep.22182. 18. McCarthy MI, Abecasis GR, Cardon LR, Goldstein DB, Little J, Ioannidis JP, Hirschhorn JN. Genome-wide association studies for complex traits: consensus, uncertainty and challenges. Nat Rev Genet 2008, 9:356–369. doi:10.1038/nrg2344. 19. Stranger BE, Stahl EA, Raj T. Progress and promise of genome-wide association studies for human complex trait genetics. Genetics 2011, 187:367–383. doi:10.1534/genetics.110.120907. 20. Cantor RM, Lange K, Sinsheimer JS. Prioritizing GWAS results: a review of statistical methods and recommendations for their application. Am J Hum Genet 2010, 86:6–22. doi:10.1016/j.ajhg.2009.11.017. 21. Zeggini E, Weedon MN, Lindgren CM, Frayling TM, Elliott KS, Lango H, Timpson NJ, Perry JR, Rayner NW, Freathy RM, et al. Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes. Science 2007, 316:1336–1341. 22. Gudmundsson J, Sulem P, Manolescu A, Amundadottir LT, Gudbjartsson D, Helgason A, Rafnar T, Bergthorsson JT, Agnarsson BA, Baker A, et al. Genome-wide association study identifies a second prostate cancer susceptibility variant at 8q24. Nat Genet 2007, 39:631–637. 23. Leyva-Vega M, Gerfen J, Thiel BD, Jurkiewicz D, Rand EB, Pawlowska J, Kaminska D, Russo P, Gai X, Krantz ID, et al. Genomic alterations in biliary atresia suggest region of potential disease susceptibility in 2q37.3. Am J Med Genet A 2010, 152A:886–895. doi:10.1002/ajmg.a.33332. 24. Shaikh TH, Gai X, Perin JC, Glessner JT, Xie H, Murphy K, O’Hara R, Casalunovo T, Conlin LK, D’Arcy M, et al. High-resolution mapping and analysis of copy number variations in the human genome: a data resource for clinical and research applications. Genome Res 2009, 19:1682–1690. doi:10.1101/gr.083501.108. 25. Schreiber RA, Kleinman RE. Genetics, immunology, and biliary atresia: an opening or a diversion? J Pediatr Gastroenterol Nutr 1993, 16:111–113. 26. Cui S, Leyva-Vega M, Tsai EA, EauClaire SF, Glessner JT, Hakonarson H, Devoto M, Haber BA, Spinner NB, Matthews RP. Evidence from human and
© 2015 Wiley Periodicals, Inc.
Volume 7, September/October 2015
WIREs Systems Biology and Medicine
GWAS in biliary atresia
zebrafish that GPC1 is a biliary atresia susceptibility gene. Gastroenterology 2013, 144:1107e3–1115e3. doi:10.1053/j.gastro.2013.01.022. 27. Glessner JT, Wang K, Cai G, Korvatska O, Kim CE, Wood S, Zhang E, Estes A, Brune CW, Bradfield JP, et al. Autism genome-wide copy number variation reveals ubiquitin and neuronal genes. Nature 2009, 459:569–573. 28. Grisaru S, Cano-Gauci D, Tee J, Filmus J, Rosenblum ND. Glypican-3 modulates BMP and FGF-mediated effects during renal branching morphogenesis. Dev Biol 2001, 231:31–46. 29. Yan D, Lin X. Drosophila glypican Dally-like acts in FGF-receiving cells to modulate FGF signaling during tracheal morphogenesis. Dev Biol 2007, 312:203–216. 30. Yan D, Lin X. Opposing roles for glypicans in Hedgehog signalling. Nat Cell Biol 2008, 10:761–763. 31. Garcia-Barcelo MM, Yeung MY, Miao XP, Tang CS, Cheng G, So MT, Ngan ES, Lui VC, Chen Y, Liu XL, et al. Genome-wide association study identifies a susceptibility locus for biliary atresia on 10q24.2. Hum Mol Genet 2010, 19:2917–2925. doi:10.1093/hmg/ddq196.
36. Suzuki T, Kanai Y, Hara T, Sasaki J, Sasaki T, Kohara M, Maehama T, Taya C, Shitara H, Yonekawa H, et al. Crucial role of the small GTPase ARF6 in hepatic cord formation during liver development. Mol Cell Biol 2006, 26:6149–6156. 37. Morishige M, Hashimoto S, Ogawa E, Toda Y, Kotani H, Hirose M, Wei S, Hashimoto A, Yamada A, Yano H, et al. GEP100 links epidermal growth factor receptor signalling to ARF6 activation to induce breast cancer invasion. Nat Cell Biol 2008, 10:85–92. 38. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res 2000, 28:27–30, KEGG ID (hsa04010). 39. Iguchi H, Mitsui T, Ishida M, Kanba S, Arita J. cAMP response element-binding protein (CREB) is required for epidermal growth factor (EGF)-induced cell proliferation and serum response element activation in neural stem cells isolated from the forebrain subventricular zone of adult mice. Endocr J 2011, 58:747–759. 40. Sarró E, Tornavaca O, Plana M, Meseguer A, Itarte E. Phosphoinositide 3-kinase inhibitors protect mouse kidney PCT3 cells from cyclosporine- induced cell death. Kidney Int 2008, 73:77–85.
32. Cheng G, Tang CS, Wong EH, Cheng WW, So MT, Miao X, Zhang R, Cui L, Liu X, Ngan ES, et al. Common genetic variants regulating ADD3 gene expression alter biliary atresia risk. J Hepatol 2013, 59:1285–1291. doi:10.1016/j.jhep.2013.07.021.
41. Hu Z, Du J, Yang L, Zhu Y, Yang Y, Zheng D, Someya A, Gu L, Lu X. GEP100/Arf6 is required for epidermal growth factor-induced ERK/Rac1 signaling and cell migration in human hepatoma HepG2 cells. PLoS One 2012, 7:e38777.
33. Tsai EA, Grochowski CM, Loomes KM, Bessho K, Hakonarson H, Bezerra JA, Russo PA, Haber BA, Spinner NB, Devoto M. Replication of a GWAS signal in a Caucasian population implicates ADD3 in susceptibility to biliary atresia. Hum Genet 2014, 133:235–243. doi:10.1007/s00439-013-1368-2.
42. Harper P, Plant JW, Unger DB. Congenital biliary atresia and jaundice in lambs and calves. Aust Vet J 1990, 67:18–22.
34. Kaewkiattiyot S, Honsawek S, Vejchapipat P, Chongsrisawat V, Poovorawan Y. Association of Xprolyl aminopeptidase 1 rs17095355 polymorphism with biliary atresia in Thai children. Hepatol Res 2011, 41:1249–1252. doi:10.1111/j.1872-034X.2011. 00870.x. 35. Ningappa M, So J, Glessner J, Ashokkumar C, Ranganathan S, Min J, Higgs BW, Sun Q, Haberman K, Schmitt L, et al. The role of ARF6 in biliary atresia. PLoS One. In press.
Volume 7, September/October 2015
43. Lorent K, Koo KA, Gong W, Whittaker S, Porter JR, Wells R. Michael pack isolation and identification of a plant toxin that induces biliary atresia in livestock and zebrafish. In: The 63rd Annual Meeting of American Association for the Study of Liver Diseases, Boston, MA, 2012. 44. Ku NO, Zhou X, Toivola DM, Omary MB. The cytoskeleton of digestive epithelia in health and disease. Am J Physiol 1999, 277:G1108–G1137. 45. Jen YH, Musacchio M, Lander AD. Glypican-1 controls brain size through regulation of fibroblast growth factor signaling in early neurogenesis. Neural Dev 2009, 4:33.
© 2015 Wiley Periodicals, Inc.
273
