METHODS

Patients and controls

For this study 115 Italian coeliac patients with typical HLA haplotypes (either DQ2 or DQ8), 37 Italian coeliac patients with atypical HLA haplotypes, and 180 pan-ethnic controls (healthy blood donors with no history of either gastrointestinal or autoimmune diseases) were enrolled. Diagnosis of coeliac disease was performed following the ESPGHAN indications.27 The study was approved by the ethics committee at the Burlo Garofolo Children’s Hospital, Trieste. The importance of a positive result was explained to all participants, and verbal consent was obtained from the patients.

CD14 genotyping

DNA was extracted from peripheral whole blood using standard laboratory protocols.28

The detection of the bi-allelic polymorphism at position −159 was performed following the PCR-RFLP technique already described by Baldini et al.29

The detection of the bi-allelic polymorphism at position −1359 was performed with a modified version of the PCR-RFLP technique described by Holla et al.25 Briefly, PCR was performed with the forward primer 5′-AGCACTTTGAGAGGCCAAGG-3′ and the reverse primer 5′-CCACATCCCTAGACCTCTGGG-3′, both designed on the basis of the published CD14 human sequence with the software Primer Express 1.5 (Applied Biosystems, Foster City, CA, USA). After an initial denaturation step at 95°C for 10 min, 35 cycles with a denaturation step at 95°C for 30 s, an annealing step at 62°C for 30 s, and an elongation step at 72°C for 30 s were carried out. The PCR reaction mix was 1.5 mmol/l MgCl₂, 800 µmol/l dNTPs, 200 µmol/l for each primer and contained 1 U of AmpliTaq Gold (Applied Biosystems). PCR products were digested for 2 h at 37°C with the endonuclease FokI (New England Biolabs, Beverly, MA, USA) and subsequently separated on a 2% agarose gel. The G allele showed one fragment of 303 bp, while the T allele showed two fragments of 215 and 88 bp.

Statistical analysis

Gene frequencies were calculated from the observed number of genotypes. The significance of differences in allelic frequencies was calculated by Fisher’s exact test. To test the significance of differences in genotype frequencies a χ² test was used. Holm’s correction for multiple tests was performed and only corrected p values (p_corr) < 0.05 were considered to be significant. All the statistical analyses were carried out with SPSS software (Version 6.1.3; SPSS Inc., Chicago, IL, USA).

RESULTS

Allelic frequencies and genotype frequencies for the position −159 in the three populations studied are shown in Table 1. The frequency of the (−159)C allele was 53%, 46%, and 50% in typical HLA coeliac patients, atypical HLA coeliac patients, and healthy controls, respectively. No differences were observed in genotype frequencies between the three populations.

Table 2 shows the allelic and genotype frequencies for the G to T transversion at position −1359. The frequency of the (−1359)G allele did not differ in the three populations; it was 80% in typical HLA coeliac patients, 77% in atypical HLA coeliac patients, and 84% in healthy controls (Fisher’s exact test, p=0.631 for typical compared with atypical HLA coeliac patients; p=0.251 for typical HLA coeliac patients compared with healthy controls, and p=0.172 for atypical HLA coeliac patients compared with healthy controls). No differences were observed in genotype frequencies when healthy controls and typical HLA coeliac patients were compared (χ²=4.19, p=0.123). However an increased frequency of (−1359)TT homozygotes were observed in the latter population (odds ratio TT v GT and GG genotypes: 3.82; 95% confidence interval (CI) 0.97 to 15.10). In fact the frequencies of (−1359)GG, (−1359)GT, and (−1359)TT genotypes were 70%, 28%, and 2% in controls, and 67%, 27%, and 6% in typical HLA coeliac patients, respectively. The genotype frequencies in the atypical HLA coeliac patients were 67%, 19%, and 14% for the genotype (−1359)GG, (−1359)GT, and (−1359)TT, respectively. This distribution did not differ when compared with that of typical HLA coeliac patients (χ²=2.680, p=0.262), but it was significantly different from that of the healthy controls (χ²=12.640, p=0.0018, p_corr<0.05), mainly due to the difference in the frequency of (−1359)TT homozygous individuals. Odds ratio for TT genotype in atypical HLA coeliac patients was 9.22 (95% CI 2.10 to 40.50).

DISCUSSION

Coeliac disease is one of the most common multifactorial autoimmune disorders in the white population (1 in 200 individuals). The principal genetic (HLA-linked genes) and environmental (wheat gluten) factors responsible for CD are well known, but few data are available for non-HLA-linked genes.

In this case–control study we have assessed the frequency of two polymorphisms in the promoter region of the CD14 gene, which regulate the expression of the protein in three selected populations.

LeVan et al.30 demonstrated that (−159)T allele promotes CD14 gene transcription and that homozygous subjects for this allele have increased CD14 serum levels.29 In our study the allelic and genotype frequencies of C(−159)T did not differ betwen the 115 coeliac patients carrying the typical HLA haplotype (DQ2 or DQ8), the 37 coeliac patients with an atypical HLA haplotype, and the 180 selected healthy controls of the same ethnic origin. Interestingly, there was a significant skewing in coeliac patients with atypical HLA for the genotype frequencies at position −1359 compared with healthy controls. This difference is mainly due to the higher frequency of (−1359)TT subjects in coeliac patients carrying the atypical HLA haplotype. The frequency of (−1359)TT patients was also greater in coeliac patients with typical HLA, even though our results do not reach statistical significance. As demonstrated by Vercelli et al31 (−1359)TT subjects have lower serum levels of CD14. The odds ratio of developing coeliac disease for (−1359)TT atypical HLA coeliac patients is 9.22 (95% CI 2.10 to 40.50).

Several hypotheses could explain the involvement of CD14 in coeliac disease. Firstly, individuals with low levels of sCD14, and thus mCD14, might be more prone to Gram-negative bacterial infections. These infections may disrupt the permeability of intestinal epithelia, leading to an increased concentration of gluten in the lamina propria, where most of the gluten-reactive lymphocytes are localised. Such individuals may be more susceptible to developing coeliac disease. Unfortunately, no studies linking the G(−1359)T polymorphism to a risk of infection have been carried out. Moreover, a recent study by Agnese et al32 showed a significant correlation between TLR4 polymorphisms and an augmented risk of gram negative bacterial infection: no association was found between CD14 C(−159)T polymorphism and the incidence of infection or patients’ outcome.

Another possibility, and in our opinion the most interesting, may derive from the role of CD14 in apoptosis regulation. The defective clearance and removal of apoptotic cells has been closely linked to autoimmune and persistent inflammatory diseases.33 Recently Boniotto et al26 found a significant association between allelic variants within the first exon of the mannose-binding lectin 2 (MBL2) gene and an increased risk of CD, and discussed the importance of mannose-binding lectin in the clearance and removal of apoptotic cells. Several studies demonstrate that CD14 is able to bind several apoptotic-cell-associated molecular patterns (ACAMPs),19,34,35 such as ICAM-3 on apoptotic leukocytes,36 and may either act as a phagocyte pattern-recognition receptor or bind directly to ACAMPs to initiate engulfment of apoptotic cells. Moreover, it is known that MBL binds to apoptotic cells and initiates their uptake by interacting with phagocyte calreticulin receptors on the macrophage cell surfaces.37 In the same way, CD14 is expressed mainly on the surface of these “professional” phagocytes that, unlike dendritic cells, are also able to present apoptotic-cell-derived antigens to T cells in a tolerogenic manner, providing functional inactivation or deletion of potentially autoreactive T cells.38,39 Furthermore, sCD14 is able to bind and switch off activated T cells.40 Taken together, these data explain why subjects with lower levels of CD14 (both mCD14 and sCD14) are more prone to develop autoimmune diseases, such as CD.

At the moment, linkage disequilibrium of the G(−1359)T allele with the real causative gene for CD in the genomic region 5q31–33 cannot be excluded.

In conclusion, a significant association between (−1359)TT genotype and an increased risk for CD was shown in coeliac patients with atypical HLA haplotype. There was a positive trend for a higher frequency of (−1359)TT individuals in typical HLA coeliac patients, although it did not reach statistical significance. To our knowledge, this is the first study that has found a significant association between CD and a polymorphism in a gene located in the candidate genomic region 5q31–33.

Finally, our findings demonstrate the importance of apoptosis regulation. They indicate that the genes encoding proteins, which play a key role in this process, are potential candidates for CD. A larger cohort of typical HLA coeliac patients should probably be studied, and functional studies carried out, to clarify the role of CD14 in coeliac disease.