MATERIALS AND METHODS

Patients and families

A large European consortium recruited 381 IBD families in France (n=246), Sweden (n=45), Belgium (n=34), Spain (n=26), Denmark (n=16), Italy (n=10), and Ireland (n=4). Of these 381 families, 306 contained several affected members and 75 were simplex families. Of the 306 multiplex families, 180 contained only CD patients, 45 contained only UC patients, and 81 were mixed families with UC, CD, and indeterminate colitis (IC) patients. In the 75 simplex families, 13 of the probands had UC, 55 had CD, and seven had IC.

Diagnostic criteria were based on clinical, radiological, endoscopic, and histological findings, as previously described.¹² A standardised questionnaire was completed for all patients, including date of birth, sex, family history, age at onset, and details of disease location (at onset and at its maximal severity), granuloma formation, stenosis, transmural involvement, extradigestive symptoms, and therapeutic management. Stenosis and transmural involvement were defined by the occurrence in the digestive tract of at least one stricture or by the presence of a fistula or abscess, respectively, as shown by radiological, endoscopic, or pathological examinations during the evolution of the disease. The study was approved by the relevant ethics committees and informed consent was obtained from all participants.

Genetic screening of CARD4 gene

A set of 63 unrelated index patients (including 54 CD and nine UC patients) was used for mutation screening. For each of the 63 index patients, the 11 exons and intron-exon boundaries of CARD4 (Genbank accession number NM_006092) were amplified from their DNA by polymerase chain reaction (PCR) using 15 couples of primers (table 1) and sequenced on an ABI 377 automated sequencer using a Dye Terminator Cycle Sequencing Ready reaction kit (Perkin-Elmer Applied Biosystems, Foster City, California, USA). Sequence data were then aligned using the Sequence Navigator analysis software version 1.0.1 (Perkin-Elmer Applied Biosystems) and compared with the previously reported CARD4 sequence.

Genotyping methods

Families were then genotyped for the E266K polymorphism. Genomic DNA was amplified by PCR on a thermal cycler (PTC-200; MJ Research, Waltham, Massachusetts, USA) in a 20 μl mixture containing 100 ng of genomic DNA, 1× Taq-Gold polymerase buffer II (Perkin Elmer), 1.5 mM MgCl₂, 0.2 mM dNTPs, 1 U Taq-Gold polymerase (Perkin Elmer), and 1 μM of forward and reverse primers (line 4, table 1). After denaturation (12 minutes at 95°C), the reaction consisted of 30 PCR cycles (30 seconds at 94°C, 30 seconds at 58°C, and 30 seconds at 72°C) followed by a final extension (seven minutes at 72°C). The E266K mutation was detected by taking advantage of the destruction of an AvaI restriction enzyme site by the single base pair substitution and 10 μl of the PCR products were digested with AvaI (NEB, Beverly, Massachusetts, USA) according to the manufacturer’s recommendations. Genotypes were deduced from the migration profile on a 2% agarose gel: wild-type DNA was visible as a double band (209 bp and 170 bp) while the mutated DNA was visible as a single 379 bp band.

Finally, the families were analysed for the three main CD associated mutations of CARD15/NOD2, as previously described.¹³

Linkage disequilibrium analysis

The transmission disequilibrium test (TDT) statistic¹⁴ was computed using the TDT option of the Genehunter 2.0 package. Phenotype-genotype relationships were analysed by χ² tests (qualitative variables) or ANOVA analyses (quantitative variables).

RESULTS

Seven nucleotide changes were observed in exon 3 and two in exon 5 (table 2). A few variants located in introns were also found (data not shown). However, because they were not expected to alter splicing sites, they were assumed to have no functional relevance and were not studied further. Five of the nine variations identified in the coding sequences change the peptide chain of either the NBD (E266K, D372N) or LRR domains (R705Q, T787M, T787K) (table 2). Among these five variations, four amino acid changes (D372N, R705Q, T787M, and T787K) were observed in only one CD patient and were considered private mutations (table 2).

Three genetic variations (E266K, D161D, and A574A) were frequent enough to perform association studies with reasonable power in our IBD families (table 2). These three polymorphisms were found to be in nearly complete linkage disequilibrium (data not shown). As the E266K variant was the only one to encode a changed protein, altering a glutamic acid residue (E) that is conserved in CARD15, we concentrated our study on this variant.

To investigate the possible role of the E266K variant in IBD predisposition, 381 IBD families were genotyped using a PCR-restriction fragment length polymorphism procedure. The E and K alleles were found to be in Hardy-Weinberg equilibrium. The frequency of the K variant was 0.23 in IBD patients and 0.25 in the panel of alleles not transmitted to the IBD patients used as controls (NS). In families with only CD or UC segregating, the frequency of the K allele was, respectively, 0.27 in CD and 0.22 in UC patients. These frequencies were identical in familial and sporadic cases allowing pooling of the multiplex and simplex families for the genotype-phenotype analyses. TDT analysis failed to demonstrate preferential transmission of the E or K allele for any of the three tested phenotypes (IBD, CD, or UC) (table 3).

A subgroup of 235 unrelated CD patients was randomly selected for more detailed analyses from families with only CD segregating. These patients were subdivided into three groups according to the number of K variant alleles they carried: 0, 1, or 2. No difference was observed between groups for the following variables: sex, age at onset, family history, disease location (at onset and at its maximal severity), granuloma formation, extradigestive symptoms, and therapeutic management. The only observed difference was an excess of the stricturing phenotype in patients carrying one or more K allele(s) (p=0.02). However, this was not significant after applying the Bonferroni correction for multiple comparisons. Finally, we examined the relationship between CARD4 and CARD15 in CD patients to identify a putative gene interaction. The frequencies of the three CARD15 CD associated variants (R702W, G908R, and 1007fs) were similar in patients with the wild-type genotype or carrying the K allele of the E266K-CARD4 variation (data not shown).

DISCUSSION

The recent identification of CARD15/NOD2 as a CD susceptibility gene makes its homologous gene CARD4/NOD1 a potential candidate gene for predisposition to IBD.^3–5 The function of CARD4 has recently been clarified. The gene encodes for a cytosolic protein involved in the pathway by which intracellular LPS activates nuclear factor κB and JNK.¹⁰

Screening of all 11 exons and the flanking intronic sequences of CARD4 in 63 patients revealed nine nucleotide changes. Four are silent variations (D161D, A554A, A574A, and N686N) and have no expected functional effect. Five mutations encoding amino acid changes (E266K, D372N, R705Q, T787M, and T787K) were identified, including three mutations occurring in the LRR domain of the protein, as was observed for the CARD15 mutations¹³ in CD. Of these five variations, four occurred in only one patient (D372N R705Q, T787M, and T787K). In these patients, the medical records did not suggest a specific effect of these genetic variants. However, it is not possible to rule out an effect of these variations in the patient carriers without functional studies.

It is unlikely that additional mutations in the coding sequence of CARD4 with a large frequency in IBD patients would have been missed in our study and the E266K variation appears as the only frequent missense variant. However, the possible role of undetected sequence changes in the promoter or regulatory domains of the gene cannot be ruled out by this study and we only tested the hypothesis that, as for CARD15, mutations would be predominantly found in the coding sequence rather than in regulatory regions of the gene.

The G796A variation encodes a non-conservative change (E266K) in the NBD domain. The corresponding glutamic acid residue appears to be conserved in CARD15, suggesting a potential functional effect of the mutation. No transmission distortion of a particular allele of the E266K polymorphism was observed for any of the three tested phenotypes (IBD, CD, and UC). For the CD phenotype, this observation is unlikely to be due to lack of power considering the large number of families studied. In addition, when examining phenotypic subgroups, no phenotype-genotype relationship was evident in CD patients. For the UC phenotype, the power of this study was more limited but the data do not argue for a major role of CARD15 in UC. Finally, no CARD4/CARD15 interaction was identified in CD patients but the power of the statistical test was limited. Taken together, these results argue against a major role of CARD4 in IBD genetic susceptibility.

Considering the strong functional similarities between CARD4 and CARD15, which are both involved in the same pathway by which LPS activates nuclear factor κB and apoptosis, the lack of involvement of this gene could be seen as surprising. In fact, the main known difference between CARD4 and CARD15 is their tissue specific expression. While CARD4 is widely expressed in epithelial cells, CARD15 is mainly expressed in monocytes.¹⁵ Thus the present observations argue for a pivotal role of the monocyte lineage in the development of CD.