*Corresponding author: Alexandre Serra, M.D., Ph.D., Department of Pediatric Surgery, Technische Universität Dresden, Fetscherstrasse 74, D-01304 Dresden, Germany. Tel: +49 351 458 4901; Fax: +49 351 458 5343. E-mail: Alexandre.email@example.com
Hirschsprung disease (HSCR) is transmitted in a complex pattern of inheritance and is mostly associated with variants in the RET proto-oncogene. However, RET mutations are only identified in 15–20% of sporadic HSCR cases and solely in 50% of the familial cases. Since genomic rearrangements in particularly sensitive areas of the RET proto-oncogene and/or associated genes may account for the HSCR phenotype in patients without other detectable RET variants, the aim of the present study was to identify rearrangements in the coding sequence of RET as well as in three HSCR-associated genes (ZEB2, EDN3 and GDNF) in HSCR patients by using Multiplex Ligation-dependent Probe Amplification (MLPA). We have screened 80 HSCR patients for genomic rearrangements in RET, ZEB2, EDN3 and GDNF and did not identify any deletion or amplification in these four genes in all patients. We conclude that genomic rearrangements in RET are rare and were not responsible for the HSCR phenotype in individuals without identifiable germline RET variants in our group of patients, yet this possibility cannot be excluded altogether because the confidence to identify variation in at least two percent of the individuals was only 95%.
Hirschsprung disease (HSCR; MIM 142623), also known as aganglionic megacolon, is a congenital disorder characterized by the absence of enteric ganglia along a variable length of the intestine of affected children. It is more common in males than in females (4:1 ratio) and affects ca. 1 in 5000 newborns (de Pontual et al., 2007). HSCR is transmitted in a complex pattern of inheritance, and the principal gene involved is the RET proto-oncogene located at 10q11.2 (MIM 164761) (Amiel et al., 1996; Edery et al., 1994; Fitze et al., 2002; Garcia-Barcelo et al., 2005; dePontual et al., 2006). Although RET mutations resulting in either RET-protein truncation or functional inactivation have been associated with the HSCR phenotype, these are only identified in 15–20% of sporadic HSCR cases, and in only 50% of the familial cases (Amiel et al., 2008; Chattopadhyay et al., 2003). Further studies have associated a single-nucleotide polymorphism (SNP) located in an enhancer sequence on intron 1 with a stronger predisposition to HSCR. This T>C SNP increases the susceptibility risk for HSCR by 20-fold by reducing the enhancer activity, (Chattopadhyay et al., 2003) and its frequency seems to vary among different ethnic populations proportionally to the HSCR prevalence (Amiel et al., 2008). Additionally, HSCR-phenotype modifying effects due to a within-gene interaction in patients harbouring RET germline mutations (such as RET c.135G>A) as well as a role for RET haplotypes containing the −5A promoter variant in the aetiology of HSCR have been described (Emison et al., 2005; Fitze et al., 2003).
Although HSCR is seen in most cases as an isolated phenotype associated with variants of RET, syndromic cases in association with variants of other genes have been reported. Namely, mutations in the ZEB2 (SIP-1) gene (located on chromosome 2q22.3) cause the Mowat-Wilson syndrome (or Hirschsprung disease-mental retardation syndrome, MIM235730), while mutations in the EDN3 gene (located on chromosome 20q13.32) can result in the Shah-Waardenburg syndrome (MIM277580) (Amiel et al., 2008). Further mutations in several genes such as EDNRB, GDNF, NRTN (NTN), SOX10 and ECE1 (either isolated or combined with a RET germline mutation) have been identified in up to 5% of HSCR cases, supporting a genetic heterogeneity for this disorder which may result from a cumulative effect of at least two mutations in different genes (Fitze et al., 2003).
Another interesting aspect in the attempt to elucidate the complex genetic background in HSCR is the role of microdeletions of RET, particularly those leading to allelic loss. Such allelic losses following microdeletions may result in loss-of-function due to protein shortening or inactivation, such as the recently described ZEB2 microdeletions in Mowat-Wilson syndrome (Engenheiro et al., 2008). The initial mapping studies on HSCR patients have coupled large deletions on chromosome 10 with the long-segment HSCR phenotype (Lyonnet et al., 1993) which was further narrowed to the RET gene through linkage analysis of a familial microdeletion in a large pedigree of HSCR cases (Luo et al., 1993). Accordingly, microdeletions or amplifications in sensitive areas of RET or in any of the other aforementioned associated genes may account for some of the HSCR cases (in familial, but also particularly important in sporadic cases) in which RET germline variants have not been identified. Therefore, the aim of the present study was to identify genomic rearrangements in the coding sequence of the RET proto-oncogene as well as in three HSCR-associated genes (ZEB2, EDN3 and GDNF) in 80 HSCR patients by using Multiplex Ligation-dependent Probe Amplification (MLPA), a method which primarily identifies genomic deletions and/or insertions in the gene.
Material and Methods
Our population comprised 80 Caucasian HSCR patients from Germany. All patients fulfilled the histological and immunohistochemical criteria of HSCR: absence of neuronal ganglia on histological evaluation of the aganglionic tract and increased acetylcholinesterase histochemical staining in nerve fibres on suction-biopsies of the rectal submucosa. Only 5 investigated HSCR patients had a family history for the disease. In addition to HSCR, 2 patients had also Down syndrome and another had concomitant Fallot's tetralogy. In this cohort, 24 patients had long-segment aganglionosis and 56 had short-segment HSCR. All patients had undergone complete sequence analysis of the RET proto-oncogene, which revealed 22 different mutations in 21 patients, including 2 nonsense, 13 missense, 3 splice-site and 4 silent mutations. In addition, several polymorphisms located in the coding or the intronic region of RET have been found. Two HSCR patients harboured two mutations each – one missense and one silent mutation. Two non-consanguineous HSCR patients harboured the same missense mutation in c.2618 G>A (exon 15). (Fitze et al., 2003) All patients (or parents if the patient was a minor) gave written informed consent to participate in the study protocol, which was approved by the ethics committee of the University of Technology of Dresden.
Screening for rearrangements by multiplex ligation-dependent probe assay (MLPA)
MLPA was performed as described by Schouten et al., (2002) on genomic DNA isolated from peripheral blood leukocytes. The primary goal of the study was to assess genomic rearrangements in the RET proto-oncogene using the MLPA test kit P169 lot 0106 (MRC-Holland, Amsterdam, The Netherlands) performed according to the supplied protocol. Besides our primary interest in the RET proto-oncogene, other genes also associated to Hirschsprung disease (ZEB2, EDN3 and GDNF) were concomitantly studied since they were included in the MLPA test Kit for HSCR. The RET gene is located on chromosome 10q11.21 and spans 55 kb of genomic DNA. Probes were used for a total of 19 of 20 RET exons (probes for RET exon 4 are not yet available in this kit). The ZEB2 (SIP-1) gene is located on chromosome 2q22.3 and spans 135 kb of genomic DNA. Probes were used for 9 ZEB2 exons. The EDN3 gene is located on chromosome 20q13.32 and spans 25 kb of genomic DNA. Similarly, probes were used for 5 EDN3 exons. Finally, GDNF is located on chromosome 5p13.2 and spans 24 kb of genomic DNA, and probes were used for each of the 2 GDNF exons.
In summary, genomic DNA (20–500 ng) was denaturated in 5 μl Tris EDTA (TE) at 98°C for 5 min and incubated with the probe mix for 16h at 60°C. Following probe hybridization, ligation proceeded for 15 min at 54°C. Then, ligation products were amplified by polymerase chain reaction (PCR) using a FAM-labelled primer and an unlabeled primer. PCR products were analysed on ABI 3700 capillary sequencer using GeneMapper® software (Applied Biosystems, Foster City, CA, USA). Specific peaks corresponding to each exon were identified according to their migration in relation to size standards. Peak heights of each fragment were compared to those of control samples and deletions were suspected when peak height differed by more than 30%.[Shouten 2002] The probability for identifying a novel variant when 80 patients are screened was calculated using the approach of Gregorius (1980).
We have successfully screened 77 of 80 HSCR patients for genomic rearrangements in RET, ZEB2, EDN3 and GDNF, since in 3 patients DNA degradation precluded the analysis with MLPA. With this sample size, the confidence to identify deletions or amplifications in at least 2% of the patients was greater than 95% assuming Hardy-Weinberg equilibrium. We did not identify any deletion or amplification in these four genes in all patients, but in two cases we found a less than 50% reduction of the MLPA peak in exons 2 and 5 (Fig. 1). Both patients harboured a single base pair substitution in the heterozygous state, which is located in the binding region of the respective MLPA probe: c.159 C>T (exon 2) and c.1013 C>T (exon 5). All other mutations located in probe-binding regions such as c.1825 T>A (exon 10), c.1858 T>C (exon 10), c.2338 A>T (exon 13), c.2618 G>A (exon 15) in two patients, c.2734 G>A (exon 16) and the exon 13 c.2307 T>G variant did not show a detectable influence on peak height. Additionally, we found two new RET variants not previously described in two patients with long-segment sporadic HSCR, namely a polymorphism in intron 9 (IVS9 +1G>A) located at a splice site and a nonsense mutation in exon 2 (c.110 G>A).
More recent studies have tried to provide evidence for possible mechanisms which may account for this discrepancy in mutation frequency and the HSCR phenotype. Namely, SNPs in the coding region of RET either individually or combined (particularly in the enhancer region in intron 1) have been shown to affect the regulation of the transcription of the gene, eventually resulting in the HSCR phenotype (Chattopadhyay et al., 2003; Emison et al., 2005; Garcia-Barceló et al., 2005). Studies in inbred families suggest that the penetrance of RET gene mutations for HSCR is dependent on allele dosage and modifier-loci (Basel-Vanagaite et al., 2007). One of these proposed modifier-loci is the TTF-1 gene, now named NKX2–1, a transcription factor which increases RET promoter activity and whose binding site within the RET promoter has been shown to harbour SNPs in HSCR patients (Garcia-Barceló et al., 2005). Similarly, linkage studies in families with short-segment HSCR showed susceptibility loci at 3p21, 10q11 (RET) and 19p12 but mutations were found in only 40% of the linked families, suggesting that RET-dependent modifiers and non-coding variations might be equally important in an oligogenic model of transmission such as short-segment HSCR (Gabriel et al., 2002). Nonetheless, even these findings do not completely clarify the frequency of sporadic and syndromic HSCR patients without identifiable variants in the RET proto-oncogene.
Another possible explanation would be a technical failure of the hitherto employed analytical methods, which may oversee very small (sometimes even single-pair) variants in key areas of the encoding region of RET. Such variants, which possibly could lead to allelic losses, may result in loss-of-function due to protein shortening or inactivation, such as the recently described ZEB2 microdeletions in Mowat-Wilson syndrome (Engenheiro et al., 2008). Similar findings were described in the initial mapping studies on HSCR patients, where large deletions on chromosome 10 were associated with the long-segment HSCR phenotype (Lyonnet et al., 1993) and further emphasized through linkage analysis of a familial microdeletion in a large pedigree of HSCR cases (Luo et al., 1993). Consequently, RET rearrangements may account for some of the HSCR cases (both sporadic and familial) in which RET germline mutations have not been identified. MLPA techniques have been useful in the detection of quantitative changes in genomic DNA (deletions and duplications), with several advantages over more traditional techniques such as fluorescent in-situ hybridization (FISH), Southern blots and quantitative PCR (den Dunnen & White, 2006). The use of MLPA for the assessment of microdeletions or amplifications in the 4 genes in 80 HSCR patients should allow the detection of variants previously overlooked by other conventional DNA sequencing methods. However, the analysis of such rearrangements using MLPA in 80 sporadic HSCR patients has not shown any genomic rearrangements, only a small decrease in the peak whenever the probes were bound in areas of previously known variants.
As a tyrosine-kinase receptor, RET is dependent for activation on 4 ligands [glial-derived neurotrophic factor (GDNF), artemin (ARTN), persephin (PSPN) and neurturin (NRTN)], whose variants could theoretically also contribute to the development of HSCR (Martucciello et al., 1998). Notably, very few cases of mutations in the GDNF gene have been associated with HSCR, and in most of these cases there were either concomitant RET variants present or trisomy 21 (Salomon et al., 1996). Accordingly, we have not detected any GDNF microdeletions or amplifications in our 80 HSCR patients.
Similarly, mutations involving genes of the Endothelin signalling pathway, such as EDNRB and EDN3 have been found in HSCR patients. Notably, a G>T missense mutation in EDNRB exon 4 (W276C) was associated with a risk of 74% (homozygotes) and 21% (heterozygotes) for HSCR development in a Mennonite population (Puffenberger et al., 1994). Other studies further demonstrated this association between HSCR and EDNRB mutations by findings of an interstitial deletion of 13q22 in several patients with HSCR (Shanske et al., 2001) and by a significant LOD-score at 13q22 in a community with several HSCR cases (Van Camp et al., 1995). Furthermore, EDNRB mutations such as S390R, P383L and C109R have been found in up to 5% of HSCR patients in some series (Amiel et al., 2008; Fuchs et al., 2001; Tanaka et al., 1998) and the mechanism for HSCR in EDNRB variants seems to be haploinsufficiency (Amiel et al., 1996, 2008; Kusafuka et al., 1996). Since EDN3 is considered the major ligand for EDNRB, it is feasible to assume that variants of the EDN3 gene might have been associated with the HSCR phenotype. However, mutations in the EDN3 gene have only seldom been seen in HSCR patients, and in our cohort of 80 patients we have not detected any microdeletions or amplifications in this gene.
In this study mostly children with sporadic HSCR were analyzed. Studies in syndromic HSCR patients have shown that RET may act as a modifier gene inducing the HSCR phenotype in patients with congenital central hypoventilation syndrome, Bardet-Biedl Syndrome and Down syndrome, but not in patients with Mowat-Wilson Syndrome (MWS) or Shah-Waardenburg-Syndrome (WS) (dePontual et al., 2007). In these latter syndromes, other genes such as ZEB2 (SIP-1) for MWS and EDN3 for WS have been suggested as associated with the HSCR phenotype (Amiel et al., 2008; Engenheiro et al., 2008). Since our cohort was composed mostly of sporadic HSCR patients, we did not expect the MLPA analysis of ZEB2 to show variants; fittingly, none was found.
In conclusion, MLPA assessment of rearrangements in the RET proto-oncogene and in 3 other associated genes did not show any variants in 80 sporadic HSCR patients. We conclude that genomic rearrangements in RET are rare and were not responsible for the HSCR phenotype in individuals without identifiable germline RET variants in our group of patients, yet this possibility cannot be excluded altogether because the confidence to identify variation in at least two percent of the individuals was only 95%.