www.elsevier.com/locate/jpedsurg

Journal of Pediatric Surgery (2012) 47, 303–307

Identification of novel genes in Hirschsprung diseasepathway using whole genome expression studyAtif Saeed a,⁎, Lenka Barreto a, Sudhesna Guha Neogii b, Andrea Loos b,Ian Mcfarlane b, Adil Aslama

aDepartment of Paediatric Surgery, Box 267, Addenbrookes Hospital, Cambridge University Hospitals NHS Foundation Trust,Hills Road, Cambridge CB2 0QQ, UKbGenomics Core Lab, National Institute of Health Research, Biomedical Research Centre Metabolic Science,Cambridge University Hospitals NHS Foundation Trust, Hills Road, Cambridge CB2 0QQ, UK

Received 4 November 2011; accepted 10 November 2011

P

1

0d

Key words:Hirschsprung disease;Gene microarray;RELN

AbstractAim: This study aims to identify new genes not described previously that may be relevant in the etiologyor pathophysiology of patients with Hirschsprung disease (HD). This was done by identifyingdifferences in gene expression between normal and abnormal segments of bowel in HD patientscompared with controls.Methods: Full-thickness colonic tissue samples were taken from HD patients, both from the diseased(Ds) and normal segment of the colon (Nr), and from controls (Ct). Samples were further dissected intomucosa (MUC) and muscle (MUS). RNA was extracted and analyzed on Affymetrix Gene Chip HumanGene 1.0 ST arrays. Statistical analyses using ANOVA with a fold change cut off of 2 was applied todetect a number of differentially expressed genes. Selected genes were revalidated by quantitative real-time reverse transcriptase polymerase chain reaction.Results: Thirty-four samples (18 MUS and 16 MUC) were analyzed. MUC (1.64 ± 0.46 μg/mg) andMUS (0.83 ± 0.48 μg/mg) showed good RNA extraction yield and quality. Of the 24,987 filtered onexpression genes, MUS showed 220 genes with expression difference of 2-fold, out of which 120 geneswere significant with P ≤ .05. Similarly, MUC demonstrated 206 genes with 2-fold changes and 9 hadP ≤ .05. Some genes showing differential expression between groups and therefore subject to furtheranalysis were RELN, GAL, GAP43, NRSN1, and GABRG2.Conclusion: Analyzed data showed significant differences in expression of above sets of genes with up-and down-regulation, which has not been described before in HD and could have a role in pathogenesisof this condition.© 2012 Elsevier Inc. All rights reserved.

Presented at the 58th Annual Meeting of the British Association ofaediatric Surgeons, Belfast, Northern Ireland, July 20-22, 2011.⁎ Corresponding author. Tel.: +44 1223245151x56276; fax: +44

223596367.E-mail address: atif.saeed@addenbrookes.nhs.uk (A. Saeed).

022-3468/$ – see front matter © 2012 Elsevier Inc. All rights reserved.oi:10.1016/j.jpedsurg.2011.11.017

Hirschsprung disease (HD) is characterized by theabsence of the ganglion cells in the myenteric andsubmucosal plexuses of the gastrointestinal tract resultingin neonatal and childhood intestinal obstruction andconstipation. HD is considered to be multifactorial and at

304 A. Saeed et al.

least 10 different genes and 5 chromosomal loci have beenidentified as playing some role in its pathogenesis inanimal models and humans [1,2].

Genetic information contained in the DNA is copied baseby base into a strand of messenger RNA (mRNA), whichtranscribes the information into proteins in the ribosomes.Different genes are switched on or off or up- and down-regulated in different cells to produce the relevant proteinsand products via the mRNA. All mRNA expressed in anygiven cell is referred to as the transcriptome.

We characterized the transcriptome of colonic tissue inthis gene expression project using oligonucleotide micro-arrays, which allowed the simultaneous analysis of largenumbers of expression targets [3]. Some differentiallyexpressed genes between normal and abnormal segmentsof bowel in HD patients compared with controls werefurther analyzed.

1. Methods

The study was approved by National Research EthicsCommittee and the local research and developmentdepartment. Current data reflects patients recruited to thestudy from December 2007 to December 2010. Informedconsent was obtained from patients for use of tissueremoved at operation.

All patients diagnosed with HD who had a pull-throughprocedure over this period were included. All patients hadshort-segment HD, and there were no patients withsyndromes or long-segment HD in this cohort. Tissuesamples were taken from the resected bowel both fromganglionic and aganglionic segments. Controls werepatients having colonic tissue removed for reasons otherthan HD (eg, anorectal malformations). The criterion usedfor controls was the absence of disease process in thetissue removed; therefore, patients with underlying inflam-matory conditions that could affect the bowel such asinflammatory bowel disease or necrotizing enterocolitiswere excluded. These controls were as near to normalsubjects as possible.

Samples were further separated into mucosa (MUC) andmuscularis (MUS) using microdissection. Adequate separa-tion of samples was further validated by histology and alsoby act-B and MUC-2 gene assay on quantitative real-timereverse transcriptase polymerase chain reaction (qRT-PCR),which is specific for MUS and MUC, respectively. Thesamples were kept at −70°C until processing to ensureminimal degradation of RNA. Samples were then fraction-ated to obtain RNA and protein subfractions by MicroarrayCore Laboratory protocols using Qiagen RNeasy mini kit.The protein fractions were stored for later use, and the RNAfractions were further analyzed. Complementary DNA(cDNA) were generated from the RNA extract and labeledusing standard Affymetrix reagents and protocols. Validated

protocols were used for the generation of cDNA from as littleas 10 ng of total RNA. Fragmented cDNA was hybridized toAffymetrix Gene Chip Human Gene 1.0 ST arrays andscanned using the high-resolution scanner. This generatedAffymetrix Cell Intensity (CEL) files, which contain thesingle representative intensity value for each feature presentin the scanned images.

Quality control parameters were used at each step toensure acceptable standards. Selected genes were revalidatedby qRT-PCR. Clinical data regarding the preoperative andpostoperative course of patients was analyzed for anyunusual factors.

1.1. Data and statistical analyses

Genes with differential expression were identified byusing a fold-change cut off of 2. Further analysis was done todetermine significant gene networks and canonical pathwaysusing Ingenuity Pathway Knowledgebase (IPA) version 8.6(Ingenuity Systems, www.ingenuity.com) for the genes thatshowed a change in expression level between the groups.

Microarray gene expression analyses were performedusing Agilent GeneSpring GX 11.0.2 software and PartekGenomic Suite 6.5. Robust Multichip Average (RMA) (5)with Quantile normalization was used for backgroundcorrection and normalization to preprocess CEL files. AnANOVA test with multiple comparisons (Benjamini-Hotchberg correction) was applied to detect genes withconsistent differential expressions between groups. Datawere quoted as mean (SD). A P value ≤.05 wasconsidered significant.

2. Results

Twelve (9 boys and 2 girls) patients and 6 (2 boys and 4girls) controls were recruited for the duration of the study.

Median age of HD cohort was 2.85 months comparedwith 7.9 months for controls. RNA was extracted from 48samples, 36 diseased and 12 controls. Both MUC (1.64 ±0.46 μg/mg) and MUS (0.83 ± 0.48 μg/mg) showed goodRNA extraction yield. Forty-four samples showed goodquality and integrity of RNA on Agilent and qRT-PCR(Figs. 1 and 2). Thirty-four sample gene chips wereconsidered suitable for further data analyses.

Analyses were performed in 2 main groupings:

1. tissue type (ie, MUS and MUC)2. disease state (ie, disease segment, normal segment,

or control.

The distribution of different groups is shown in Table 1.Of the 24,987 filtered on expression genes, MUS samples

showed 220 genes with expression difference of 2-fold, ofwhich 120 genes were significant with P ≤ .05. Similarly,

Fig. 1 Agilent capillary electrophoresis showed good integrity and no degradation of RNA samples. RNA showed good banding at 28S and18S, which is comparable to a ladder (reference RNA).

305Novel genes in Hirschsprung disease pathway

MUC demonstrated 206 genes with 2-fold changes and 9 hadP ≤ .05. A few genes were selected from the above forfurther analysis (Table 2). These genes were cross validatedwith qRT-PCR.

RELN showed significant up-regulation in diseasedsegment (N 2 fold difference) compared with normal andcontrol segment. GAL, GAP43, NRSN1, and GABRG2were significantly down-regulated in the disease segment(Table 2).

3. Discussion

HD is the commonest congenital gut motility disorder andis caused by the absence of ganglion cells in the bowel wall.HD is thought to be multifactorial, and multiple genes are

Fig. 2 qRT-PCR showed good reproducibility of RNA sampleslinear to dilution in assay. This proves viability of RNA samples forcDNA synthesis.

thought to play a role in the pathogenesis of the condition [4].Although progress has been made in identifying some of thegenetic basis of HD many aspects are still unexplained.Recent advances in bioinformatics and high-throughputtechnologies such as microarray analysis are bringingabout a revolution in the understanding of the molecularmechanisms underlying normal and abnormal biologicalprocesses. Gene expression profiling or microarray analysishas enabled the identification and analysis of thousands ofgenes in a single RNA sample [5]. This technique has notbeen used widely for HD before though there are few studiesin animal models [6]. Microarray analysis has been used toidentify markers and candidate genes in the development ofthe enteric nervous system of RET-mutant mice [7].Recently, a genome-wide association study in a Chinesepopulation was used to identify additional genetic loci in HDand found NRGN1 as susceptibility locus analyzing single-nucleotide polymorphism [8]. This is the only significanthuman study to date.

We use microarray analysis to look at the differentialexpression of genes in diseased and normal colonic tissues ofHD patients and compared it with controls. The data showedsignificant differential expression in hundreds of genesbetween the groups. The genes identified for further analyses

Table 1 Distribution of genome chip samples for analysis

Tissue type Disease state

DIS NORM CONT

MUS 6 6 6MUC 5 5 6

Abbreviations: CONT, control sample; DIS, disease sample; MUC,mucosa samples; MUS, muscle samples; NORM, normal sample.

Table 2 Some significant genes identified between the groups

Gene Role Fold changes betweendifferent groups

Nr vs Ct Ds vs Nr Ds vs Ct

RELN Neuronal migrationand selection

1.31 ↑ 2.80⁎ ↑ 2.13⁎ ↑

GAL Muscle contraction. 1.19 ↓ 3.60⁎ ↓ 3.02⁎ ↓GAP43 Nerve growth 1.14 ↓ 5.09⁎ ↓ 5.81⁎ ↓NRSN1 Nerve signal 1.01 ↓ 2.39⁎ ↓ 2.36⁎ ↓GABRG2 Neurotransmitter 1.23 ↑ 3.47⁎ ↓ 4.29⁎ ↓⁎P ≤ .05.Abbreviations: GAL, galanin prepropeptide; GAP43, growth associatedprotein 43; RELN, reelin; NRSN1, neuresin 1; ↑, up-regulated gene; ↓,down-regulated.

306 A. Saeed et al.

were selected using previously described features ofneuronal and muscle development, function, or regenerationas areas of interest and therefore having a possible predictedrole in the pathogenesis of HD. We identified a number ofgenes not described in HD before. These findings alsoindicate that cell-specific gene expression profiling is anefficient means of identifying candidate genes for humandisease susceptibility loci.

RELN is mainly implicated in brain and CNS develop-ment. This gene encodes a large secreted extracellular matrixprotein Reelin, thought to control cell-cell interactionscritical for cell positioning and neuronal migration [9,10].It acts on migrating neuronal precursors and thus controlscorrect cell positioning in the cortex and other brainstructures. It affects migration of sympathetic preganglionicneurons in the spinal cord, where it seems to act as a barrierto neuronal migration [11,12]. The proposed role is one of adissociation signal for neuronal groups, allowing them toseparate and go from tangential chain migration to radialindividual migration [13]. It binds to 2 surface receptors,apolipoprotein E receptor 2 (ApoER2) and very low–densitylipoprotein receptor (VLDLR) [14,15]. This induces tyrosinephosporylation of Dab1 and modulates Tau phosporylation.This is thought to trigger a cytoplasmic signaling cascadethat is essential for the correct positioning of neurons [16].RELN has also got signaling pathways components Dab1,ApoER2, and VLDLR in rat small intestine, which havebeen found recently [17]. It is also found to be producedby myofibroblasts.

We have not only shown the expression of RELN inhuman colonic tissue for the first time but it is also up-regulated in the diseased segment of HD compared with thenormal segment and controls. This raises the possibility of arole in the pathogenesis of HD. The observation suggestsRELN may have a role in migration of the neural tissueand ganglion cells, and there is possibility that it acts as aninhibitory factor for neural migration, which is thought to bea major factor in the pathogenesis of HD. The role of RELNin the positioning of autonomic neurons in Reeler micespinal cord has been observed, and it has been suggested that

RELN may act as a stop signal to terminate the migration ofsympathetic preganglionic neurons on radial glia fibers [12].

An extensive literature search has not shown any previousstudies suggesting a role for RELN in HD or related humangut motility disorders. Many genes seem to play role in thepathogenesis of HD [4], and it is difficult at present toexplain if the impact of RELN is isolated or linked to theother genes implicated in the pathology of HD. What weknow through pathway analysis is that it has a role in tyrosinekinase phosphorylation. We suggest the role of RELN is vitaland has the potential to explain the rejection or inhibition ofenteric ganglia phenomenon in the pathogenesis of HD.

Alongside RELN, GAL, GAP43, NRSN1, and GABAR2were also studied. They play various roles like smoothmuscle contraction, nerve growth, and neurotransmitterreceptors. These were all down-regulated in the diseasedsegment of HD.

GAL is a gene that encodes the expression of galanin.Galanin is small neuropeptide that functions as a cellularmessenger within the central and peripheral nervoussystems, modulating diverse physiologic function. Con-tracts smooth muscle of the gastrointestinal and genitouri-nary tract regulates growth hormone release, modulatesinsulin release, and may be involved in the control ofadrenal secretion [18,19]. GAP43 gene controls neuronregeneration and nerve growth [20]. It is expressed at highlevels in neuronal growth cones during development andaxonal regeneration. NRSN1 may play an important role inneuronal organelle transport and in conduction of nervesignals. It is particularly localized in actively extendingneurites and may have role in axon regeneration anddevelopment [21]. GABRG2 encodes γ2 subunit of γ-aminobutyric acid (GABA) receptor. GABA is animportant inhibitory neurotransmitter in mammalian brain.It binds to GABA-A receptor, which is a chloride channeland mediates neuronal inhibition [22]. Its role in entericnerve system is not established yet.

The analyzed data showed significant differences inexpression of above sets of genes in diseased and normalsegments of HD compared with controls, which has notdescribed before and which may have a have a role inpathogenesis of HD. These genes require further study todefine their exact role.

References

[1] Kapur RJ. Practical pathology and genetics of Hirschsprung's disease.Semin Pediatr Surg 2009;18:212-23.

[2] Tam PKH, Garcia-Barcelo M. Genetic basis of Hirschsprung's disease.Pediatr Surg Int 2009;25:543-58.

[3] Roland BS. Application of DNA microarrays in biology. Annu RevBiochem 2005;74:53-82.

[4] Kenny SE, Tam PK, Garcia-Barcelo M. Hirschsprung's disease. SeminPediatr Surg 2010;19:194-200.

[5] Raychaudhuri S, Sutphin PD, Chang JT, et al. Basic microarrayanalysis: grouping and feature reduction. Trends Biotechnol 2001;19:189-93.

307Novel genes in Hirschsprung disease pathway

[6] Vohra PS, Tsuji K, Nagashimada M, et al. Differential gene expressionand functional analysis implicate novel mechanism in enteric nervoussystem precursor migration and neuritogenesis. Developmentalbiology 2006;298:259-71.

[7] HeunneT, Pachnis V. Expression profiling of the developingmammalianenteric nervous system identifies marker and candidate Hirschsprung'sdisease genes. Proc Natl Acad Sci U S A 2006;103:6919-24.

[8] Garcia–Barcelo M, Tang CS, Ngan ES, et al. Genome-wideassociation study identifies NRG 1 as a susceptibility locus forHirschsprung's disease. PNAS 2009;106:2694-9.

[9] Weeber EJ, Beffert U, Jones C, et al. Reelin and ApoE receptorscooperate to enhance hippocampal synaptic plasticity and learning.J Biol Chem 2002;277:39944-52.

[10] D'Arcangelo G. Apoer2: a reelin receptor to remember. Neuron2005;47:471-3.

[11] Yip JW, Yip YPL, Nakajima K, et al. Reelin controls position ofautonomic neurons in the spinal cord. PNAS 2000;97:8612-6.

[12] Yip YP, Metha N, Megdaleno M, et al. Ectopic expression of reelinalters migration of sympathetic preganglionic neurons in the spinalcord. J Comp Neurol 2009;515:260-8.

[13] Hack I, Bancila M, Loulier K, et al. Reelin is a detachment signal intangential chain-migration during postnatal neurogenesis. Nat Neu-rosci 2002;5:939-45.

[14] Hiesberger T, Trommsdorff M, Howell BW, et al. Direct binding ofReelin to VLDL receptor and ApoE receptor 2 induces tyrosine

phosphorylation of disabled-1 and modulates tau phosphorylation.Neuron 1999;24:481-9.

[15] Trommsdorff M, Gotthardt M, Hiesberger T, et al. Reeler/Disabled-like disruption of neuronal migration in knockout mice lacking theVLDL receptor and ApoE receptor 2. Cell 1999;97:689-701.

[16] Rice DS, Curran T. Role of the reelin signaling pathway in centralnervous system development. Annu Rev Neurosci 2001;24:1005-39.

[17] García-Miranda P, Peral MJ, Ilundain AA. Rat small intestineexpresses the reelin–Disabled-1 signalling pathway. Exp Physiol2010;95:498-507.

[18] Evans H, Baumgartner M, Shine J, et al. Genomic organization andlocalization of the gene encoding human preprogalanin. Genomics1993;18:473-7.

[19] Mechenthaler I. Galanin and the neuroendocrine axes. Cell Mol LifeSci 2008;65:1826-35.

[20] Kosik KS, Orecchio LD, Bruns GA, et al. Human GAP-43: its deducedamino acid sequence and chromosomal localization in mouse andhuman. Neuron 1988;1:127-32.

[21] Suzuki H, Tohyyama K, Taketani S, et al. Regulatory expression ofneurosin-1 in the spinal motor neurons after mouse sciatic nerve injury.Neurosci lett 2007;42:152-7.

[22] Nymann-Anderson J, Wang H, Chen L, et al. Subunit specificity andinteraction domain between GABA (A) receptor-associated protein(GABARAP) and GABA (A) receptors. J Neurochem 2002;80:815-23.