DISTRIBUTION OF HLA-DPBl, -DQBI-DQAl ALLELES AMONG SARDINIAN CELIAC PATIENTS

The Sardinian population in many aspects differs from other Caucasoid populations, particularly for its degree of homogeneity. Forthis reason we have studied 50 adult Sardinian patients with celiac disease (CD) and 50 control healthy Sardinian individuals by RFLP analysis and by extensive oligotyping for 17 HLA-DPB I, 8-DQB I and 9-DQA I alleles, and established their -DPB I alleles and -DQB I -DQA I genotypes. The heterodimer HLA-DQB I *020 I I-DQA I *050 I, present in 96% of our patients, is strongly associated with CD susceptibility, confirming published reports. On the other hand we found in I I of 50 probands (22%) the presence of the allele -DQB I *05021 DQAI*0102. This genotype is extremely rare in other Caucasian populations and appears to confer limited protection in CD Sardinian patients.


INTRODUCTION
Human leukocyte antigens (HLA) are associated with genetic susceptibility to a large number of diseases. Some of the associations are related to specific aminoacids or epitopes ofHLA class II molecules. The major histocompatibility complex (MHC), class II molecules, are cell surface heterodimers found on antigen presenting cells and B lymphocytes.
The HLA-D region of the short arm of chromosome 6, contains 12-15 non allelic class II genes and has been subdivided into three subregions, HLA DR, DQ and DP, each containing at least one functional A and B locus.
Celiac disease (CD) is an autoimmune disorder characterized by small intestinal mucosal injury and malabsorption, caused by abnormal sensitivity to gliadins, components of wheat gluten and related proteins of others grains. The condition has been reported to be MHC, class II, associated in a number of studies (Bugawan et al., 1989).
CD was initially associated with HLA class I antigens Al and B8 (Falchuck et al., 1972), later with the HLA class II, DR3 and DR7 (De Marchi et al., 1983) antigens, later still with HLA DQW2 (Tosi et ai., 1983); finally a significant association with DPBI alleles emerged. Moreover, different reports indicate significant associations of CD, with HLA DQB I -DQA 1 (Roepetai. , 1988). and DPB I alleles defined by restriction fragment length polymorphism (RFLP) analysis (Howell et al. , 1988;Niven et aI. , 1987) or by probing polymerase chain reaction amplified genomic DNA, with sequence specific oligonucleotides (Bugawan et ai., 1989;Roep et al., 1988;Howell et aI., 1988;Niven et aI. , 1987;Rosemberg et aI. , 1989). Data on the DP associations are nevertheless questionable not only among different ethnic groups, but also within various regions populated by the same ethnic group. On the other hand, several studies confirm that the heterodimer-alphalbeta, encoded by the DQAl *0501fDQB 1 *0201 genes, is the HLA genetic determinant most closely associated with CD (Sollid et aI., 1989, Sollid andThrosby, 1990;Spurkland et aI., 1990).
We have studied a group of Sardinian patients, with the late onset form of CD and have focused on these genes, hoping to establish more clearly whether alleles of the DQ and DP regions are always implicated in CD susceptibility.

Patients and normal controls
The patient group comprised 50 unrelated adults (13 males, 37 females), with the late onset form of celiac disease. Family studies have been performed and complete HLA haplotypes were determined. Fifty ethnically matched controls, with no family history of autoimmune disorders, were typed.

Serological HLA typing
HLA class I and II antigens were determined using the standard microlymphocytotoxicity assay.
RFLP analysis (28 patients and 22 controls with DQw2-DR3 genotype were investigated using traditional methods) High molecular weight DNA was extracted and then digested with the following restriction enzymes: Bam HI, Eco RI, Hind III, Taq I, Rsa I, Pvu 1, Msp I, Pst I, Xba I, Bgi II. It was subjected to electrophoresis in 0.8 % agarose gel and transferred onto nylon membranes (Z probe Biorad), by Southern' s method. Hybridization was performed with the following eDNA probes: betaDR, betaDQ and alphaDQ, betaDP and alphaDP.

Oligonucleotides
Oligonucleotides were synthesized using a DNA synthesizer (Beckman system I plus) and purified by the applied biosystem OPC (oligonucleotide purification cartridge) method.
The sequence of the ASO probers and primers used to amplify the second exon of the HLA-DQB 1, -DQA I and -BPB I genes have already been reported (Bugawan et aI., 1989;Spurkland et aI. , 1990;Khalil et al., 1990) .

peR amplification
Amplification by PCR was performed on approximately 1 /lg human genomic DNA, using 50 pmol of each primer. Two units ofTaq polymerase (Perkin-Elmer-Cetus) were added to a 100ml vol. of polymerase buffer containing 50 mM KCL, 10 mM TRIS pH 8.4,2.5 mM MgCI, and 0.001 % gelatin.
Each of the four deoxynucleotide triphosphates was present at 2 mM (8 mM total dNTP) . The reactions were amplified for 30 cycles using a DNA thermal cycler (Perkin-Elmer-Cetus) with a two step temperature cycle (denaturation: 96° for 30"; annealing and extension: 65° for 45").

Dot spot
A lO ~ aliquot of each amplified DNA was mixed with 8 ~ of 10 N NaOH and lO ~ of 0.50 mM EDT A.
Samples were kept at room temperature for lO min., then applied to a nylon membrane (Biorad Z probe) using a DOT SPOT apparatus (Schleicher and Schuell, Keene, NH).
The filters were exposed to Kodak X-AR 5, at -80°C with intensifying screen for 2 hours and/or overnight.

Statistical evaluation
The Chi-Square test was used as appropriate.
Applying the Bonferroni criterium, P Values were corrected for the number of comparisons made.
The relative risk (according to Wolf's test with Haldane's correction) and confidence limits were calculated.

RESULTS
The frequencies of polymorphic restriction fragments observed in our patients and controls, versus those described by other authors, are shown in Table 1.
No significant differences between patients and controls DQw2, DR3 were found. In particular the alphaDP 3.5 Kb, BgI II fragment associated with CD, was found in 50% of our patients versus 56% of matched controls.
Results of HLA-DPB 1 genes analysis revealed a slight increase in the frequency of DPB I *0301 (31 %) and DPB I *0402 (21 %) alleles.

DISCUSSION
The findings in Sardinian CD patients differ somewhat from analysis of DQ and DP polymorphism as described for other Caucasian populations. CD patients in the USA and England showed a very high incidence of certain alphaDP and betaDP chain alleles (Howell et al., 1988;Niven et aI., 1987); the most common allele in the USA was DPB I *030 I followed by DPB I *0402 and by DPB I *0 IO 1 (Kagnoff et al., 1989), while in a Caucasian English population an increase of DPB 1 *0 10 I was observed (Rosemberg et aI., 1989). The same RFLPs did not reveal any significant DP association in an Argentinian population (Herrera et al., 1989). A low frequency of the alphaDP 3.5 Kb, BgI II fragment present in 50% of our patients was noted, and the other RFLPs examined showed a moderate increase when compared with controls carrying the DQw2-DR3 haplotype.
The hypothesis that can be invoked to explain the above findings is that, in the Sardinian population, these alleles are preferentially associated with the DQw2 genotype independently from the disease. Previous research studies of the DPB I alleles have been undertaken in different Italian region s: a group of CD pediatric patients from the Naples area in southern Italy, showed a significant association with DPB I *0402 and DPB I *0301 (Bugawan et al., 1989), while in the Bologna area in northern Italy, only the DPB 1 *030 I showed an association with CD (Colonna et aI., 1990). Our results, regarding DPB I alleles, do not confirm those found in other Italian regIons .
Moreover, results determined by oligotyping for HLA-DPB I alleles revealed in pediatric patients from the Naples area that the DPB I *0402 allele (52 %) prevails. Whilst in our patients the frequencies of DPB I *030 I and DPB I ':'0402, respectively 31 % and 21 %, almost overlap those of adults checked in the Bologna area.
Regarding the DQB I alleles, we found the most significant association in CD was homozygosity for the DQB I *0201 allele, identified in 38% of patients.
An interesting finding of this study is the relatively high frequency of DQB 1 *05021 -DQA I *0 I 02 genotype determined in II of SO probands (22 %) that confirms the limited general protection provided by these alleles in the Sardinian population).
The presence of the DQB I *0502 allele in 7.8 % of patients has been previously reported in a study among Sardinian IDDM (Carcassi et aI., 1991), in fact this allele, extremely rare in other Caucasian populations, was found in the Sardinian general population, with a frequency of 19.6% (Carcassi et al., 1991).
The DQB I *0502 has a serine in position 57 of the HLA-DQB I chain, the absence of aspartic acid in this position seems generally to confer susceptibility to IDDM and not resistence.
In keeping with the above we could hypothesize that the DQB I *05021DQA I *0 102 alleles are also involved in CD susceptibility in the Sardinian population.
At present it remains uncertain whether the DQA 1 ': '020 I and DQA I *0 I 02 alleles may have an additional role in susceptibility to CD.
Finally the heterodimer HLA DQB 1 *020 IIDQA I *050 I, present in 96% of our patients, confirms this as the genetic marker most closely associated with CD in Sardinian as in other patient groups.
It is possible to affirm therefore that while this association is constant in each population described, other associations may vary according to the particular linkage disequilibrium of each ethnic group.