A Chromosome 8 Gene-Cluster Polymorphism with Low Human Beta-Defensin 2 Gene Copy Number Predisposes to Crohn Disease of the Colon

Patients

The patients from three separate cohorts received diagnoses, with use of the same standard criteria, in Cleveland, Stuttgart, and Vienna and were treated in specialized tertiary-care out- and inpatient centers. Patients gave their informed consent, and local ethics committees approved the study protocols. A small cohort—of patients with colonic CD (n=10) and healthy control individuals (n=10)—was recruited in Stuttgart. Their blood-derived genomic DNA was subjected to the microarray analysis described below. The exploratory cohort was a surgical group of patients who underwent surgical resection at the Cleveland Clinic for treatment of ileal or colonic Crohn disease (table 1). The confirmatory cohort were all whites with CD or ulcerative colitis (UC [MIM 191930]) who were treated at the Robert-Bosch-Hospital (Stuttgart) or the University Hospital (Vienna) (table 1). Diagnostic and (Vienna) classification criteria were the same at both European centers.¹⁷ In this classification, L1 is defined as ileal disease only, L2 as colonic disease only, and L3 as ileal as well as colonic disease. Finally, the Stuttgart control group was combined with blood donors (n=103), individuals unaffected by inflammatory bowel disease who underwent surveillance colonoscopy (n=46), and 20 control individuals unaffected by intestinal disease from Cleveland. All colonoscopies that included biopsies were performed on patients from the Stuttgart cohort.

Array CGH (Array-Based Comparative Genomic Hybridization) Analysis

DNA preparation, labeling, hybridization, and analysis procedures were performed as published elsewhere.^25,26 In brief, genomic DNA from fresh-frozen blood obtained from 10 patients with colonic CD and from 10 healthy control individuals was isolated using the Blood and Cell Culture Kit (Qiagen) following the instructions of the supplier. Sample DNA and reference DNA (pooled DNA from six healthy individuals) were labeled differentially with use of the Bioprime Labelling Kit (Invitrogen) and were hybridized on a DNA microarray consisting of ∼8,000 genomic fragments covering the human genome at a resolution of ∼0.5 Mb.²⁷ For the beta-defensin locus at 8p23.1, all additional genomic fragments that were available at the time were added to the microarray to enhance the resolution in the region of interest. The chromosomal mapping information was based on the Ensembl Genome Browser release 36.35i (December 2005). Arrays were scanned, and fluorescence intensities of all spots were filtered (intensity/local background >3; mean/median intensity <1.3; SD of genomic fragment log ratios <0.25) and were normalized blockwise. Chromosomal breakpoints delimiting regions of different copy number status were detected by GLAD (gain and loss analysis of DNA).²⁸

Determination of HBD-2 Gene Copy Number

A TaqMan real-time PCR assay, specifically for amplification of genomic HBD-2, was established by using a specific set of amplification primers (forward 5′-CACCTGTGGTCTCCCTGGAA-3′; reverse 5′-AGCTTCTTGGCCTCCTCATG-3′) and a probe (6-FAM-ATGCTGCAAAAAG-MGB). Quantitative HBD-2 amplification data were normalized to ALB (albumin [MIM 103600]) as an internal reference gene, which was coamplified simultaneously in a single-tube biplex assay. The primers and probe for HBD-2 were designed using Primer Express software, version 1.5 (Applied Biosystems). For albumin, we used the primers and probe that were published elsewhere.²⁹ Primers were purchased from MWG-Biotech, and probes were obtained from Applied Biosystems. Real-time PCR was performed using the ABI Prism 7700 sequence-detection system. Amplification reactions (25 μl each) were performed in triplicate with 20 ng of template DNA, 1× TaqMan Universal Master Mix buffer (Applied Biosystems), 300 nM of each primer, and 200 nM of each fluorogenic probe. Thermal cycling was initiated with a 2-min incubation at 50°C, followed by a first denaturation step of 10 min at 95°C and then by 40 cycles for 15 s at 95°C and for 1 min at 60°C. In each assay, a standard curve was recorded and a no-template control was included. To amplify HBD-2 and albumin in a one-tube biplex assay, limiting primer conditions were identified, to avoid competition of the two reactions. Quantification was performed by both the standard-curve method and the comparative C_T (threshold cycle) method, as described elsewhere.²⁹ The assay was validated with a selection of DNA samples, genotyped elsewhere (kindly provided by E. J. Hollox, Nottingham, United Kingdom), that contained 3, 4, 5, and 7 HBD-2 gene copies.²⁴

Real-Time Quantitative PCR

Frozen biopsies were disrupted in 1 ml of Trizol (Gibco BRL) until complete fragmentation occurred. Total RNA was extracted according to the supplier’s protocol. RNA quality was determined by electrophoresis and was quantified by photometry. Subsequently, 2 μg of RNA was reverse transcribed with oligo-dT primers and 200 U Superscript (Gibco BRL), according to the routine procedure.

cDNA samples were subjected to real-time PCR as outlined elsewhere.²¹ In brief, an aliquot corresponding to 50 ng of RNA was set up in a 20-μl reaction mixture containing 4 mM MgCl₂, 0.5 μM of each primer (for HBD-2, forward 5′-ATCAGCCATGAGGGTCTTGT-3′ and reverse 5′-GAGACCACAGGTGCCAATTT-3′ [annealing temperature 60°C; product 172 bp]; for HBD-3, forward 5′-TGAAGCCTAGCAGCTATGAGGATC-3′ and reverse 5′- CCGCCTCTGACTCTGCAATAA-3′ [annealing temperature 62°C; product 128 bp]; for interleukin 8 [IL-8 (MIM 146930)], forward 5′-ATGACTTCCAAGCTGGCCGTGGC-3′ and reverse 5′-TCTCAGCCCTCTTCAAAAACTTC-3′ [annealing temperature touch down protocol 66°C–60°C; product 292 bp] [Sigma]) and 1× LightCycler-FastStart DNA Master SYBR Green I Mix (Roche Molecular Biochemicals) and was loaded in capillary columns. PCR was performed for 45 cycles in a LightCycler (Roche Molecular Biochemicals). After each cycle, fluorescence emission readings reflecting the increase in PCR products were monitored and analyzed using LightCycler software (Roche Molecular Biochemicals).

DNA Sequencing

Exons 1 and 2 of the HBD2 gene and ∼50 bp of adjacent noncoding regions were PCR amplified from genomic DNA and were sequenced on an Applied Biosystems 3100 capillary sequencer with use of Big-Dye chemistry. The sequences of the amplification and nested sequencing primers are available on request.

Statistics

Statistical comparisons of copy numbers of (1) ileal and colonic subsets in the Cleveland cohort as well as (2) patients with L1, L2, and L3 UC and (3) control individuals from the European collective were performed with the (two-sided) Mann-Whitney test. Additionally, Kruskal-Wallis analysis of variance of ranks, including post hoc assessment by Dunn’s test, was performed to correct for multiple testing. Differences in copy number distribution among the clinical cohorts were assessed by the Pearson χ² test, with continuity corrections. Again, the Mann-Whitney test was used to assess differences in HBD-2 mRNA expression with increasing HBD-2 gene copy numbers.

Microarray Analysis

Using an array CGH with ∼8,000 genomic fragments covering the human genome, with an average resolution of ∼0.5 Mb, we screened 10 patients with colonic CD and 10 healthy control individuals for DNA copy number variations. No gross chromosomal aberrations were present in either group (data not shown), but we detected several known regions of CNP, such as the IGHG1 gene cluster at 14q32.33 (MIM 147100) (data not shown) or the amylase gene cluster at 1p21.1 (MIM 104700) (fig. 1A). Except for the beta-defensin gene cluster at 8p23.1 (fig. 1B), however, no region showed a bias toward copy number loss or gain in patients versus controls. In the Ensembl Genome Browser (release 36.35i), the beta-defensin cluster is shown to be organized as a pair of inverted repeats separated by a sequence gap (fig. 1C). This probably represents the minimal size of the locus in humans. In the patient and control samples that showed copy number variation compared with the control DNA, the size of the variable region was always the same, covering ∼900 kb (megabase 7.1–8.0), including BAC clones RP11-278P18, RP11-1005B13, and RP11-52B19 (Clone Registry). This region contains both copies of the beta-defensin cluster but not the alpha-defensin cluster, as is shown for patient X1604 in figure 1C. When comparing the copy number profiles of the patient and control groups, it became evident that patients with CD, on average, seemed to have fewer copies than did healthy individuals. Relative to a DNA pool from 6 healthy control individuals, 8 of 10 patients with CD displayed small copy number losses, and none displayed copy number gains (fig. 1D). In control individuals, no such bias was observed (fig. 1E). The neighboring alpha-defensin gene cluster (fig. 1C), which mapped to megabase 6.76–6.90, and other beta-defensin–like loci—identified on chromosomal bands 6p12, 20q11.1, and 20p13 by sequence similarity search³⁰—did not show any copy number variation in patients or control individuals (data not shown).

HBD-2 Gene Copy Numbers in Inflammatory Bowel Diseases

Our array CGH analyses clearly indicated that low DNA copy number at the main beta-defensin locus might be connected to CD; however, the size of this initial sample was much too small for reliable correlation analysis. Furthermore, whereas array CGH is extremely useful for whole-genome scans, its quantitative performance is suboptimal at loci that are rich in low copy number DNA repeats (LCRs), such as the beta-defensin cluster. At such loci, the reported copy number ratios can be severely affected by cross-hybridization of the LCRs.²⁵

Therefore, we decided, as an alternative, to use quantitative PCR analysis of the HBD-2 gene to estimate the DNA copy number of the beta-defensin cluster. This HBD2 gene–specific approach was applied to a control population and to two patient cohorts, one from the United States and one from Europe (table 1). In the control population (n=169) the copy numbers had a range of 2–10 per genome, with a median number of 4 copies (fig. 2A). The numbers of control individuals who carry the median (4), below-median (<4), or above-median (>4) number of copies were about equal. Details of the copy number frequencies of all cohorts and subgroups are given in table 2.

The U.S. patient cohort from the Cleveland Clinic consisted of 85 surgical patients with CD who had indications for ileal versus colonic resection. In patients with ileal resections (fig. 2B), the median number of copies was identical to that of the control group (4 copies); also, the frequency distribution of the three subgroups (with <4, with 4, and with >4 gene copy numbers) was not significantly different from that of the control group. In contrast, the majority (72%) of patients with colonic resections had a copy number <4, with a median of 3 copies (fig. 2C). The difference in copy numbers was highly significant (P=.008 [Mann-Whitney]) between the ileal and colonic subgroups with CD. This was maintained when tested by Kruskal-Wallis analysis of variance (P<.01 for both post hoc Dunn test ileal vs. colonic and colonic vs. control). Similarly, the proportion of the three copy number groups in the two subgroups with CD was significantly different (P=.018 [Pearson χ²]).

An independent second cohort of European patients with CD (n=165) from Stuttgart and Vienna was classified according to the Vienna classification of location into those with ileal disease only (L1), with colonic disease only (L2), or with ileal plus colonic disease (L3). Again, ileal CD (L1) exhibited a copy number distribution similar to controls (P>.05), with a majority in the group with 4 gene copies (fig. 3). The median was 4 copies in L1 and controls compared with 3 in L2 (P=.032 and P=.001, respectively [Mann-Whitney]). Analysis of variance and post hoc test identified a significant difference between L2 and controls only (P<.05). Colonic CD only (L2) was characterized by a shift to lower copy numbers, with the majority (52%) carrying <4 copies of HBD-2 (P=.002 vs. controls; P=.037 vs. L1 [Pearson χ²]). Individuals with ⩽3 copies have a significantly higher risk of developing colonic CD than do individuals with ⩾4 copies (odds ratio 3.06; 95% CI 1.46–6.45). Patients with L3 showed an intermediate distribution pattern between controls and L2, and, although the median was the normal 4 copies, the shift in distribution was significant (P=.034 [Mann-Whitney]). In UC, differences with controls were not significant. Detailed noncategorized copy number data are given in table 2.

HBD-2 Expression Related to HBD-2 Gene Copy Numbers in Inflamed Mucosa

Mucosal biopsies were analyzed for HBD-2, HBD-3, and IL-8 expression in patients from the Stuttgart cohort. Since HBD-2 expression is negligible in normal mucosa but is enhanced during inflammation, the biopsy samples were taken from a subgroup of 44 patients with inflammation due to CD (n=17) or UC (n=27). HBD-2 and HBD-3 expression was highly correlated (r=0.84 in CD; r=0.86 in UC). When the mucosal mRNA expression of HBD-2 was related to the HBD-2 gene copy number in the same patients (fig. 4), HBD-2 mRNA expression was significantly diminished in the group with <4 compared with 4 copies (P=.023 [Mann-Whitney]) or ⩾4 (P=.033), whereas IL-8 expression in these copy number groups was not significantly different (data not shown).

HBD-2 Gene Sequencing

Finally, to investigate the possibility that HBD-2 gene expression might be affected by the presence of gene copies that are inactivated by point mutations, we sequenced exon 1 and exon 2 of the HBD-2 gene in eight patients with colonic CD and eight control individuals. We did not find any nonsense or missense mutations in the coding region. The only sequence variations present were one synonymous SNP in exon 2 (rs2740090) and four SNPs in noncoding sequences (rs2740086, rs2740091, rs2737912, and rs2737913). These SNPs occurred without predominance in either patients or control individuals (data not shown; SNPs were retrieved from dbSNP).