INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS REFERENCES

Diffuse panbronchiolitis (DPB), first described in 1969, is a distinctive chronic obstructive pulmonary disease of unknown etiology (1, 2). Patients present with chronic cough, purulent sputum, and progressive dyspnea on exertion. Chronic inflammatory lesions around respiratory bronchioles are characteristic of the disease and are seen as diffuse small nodular shadows on a plain chest X-ray film and further delineated as centrilobular micronodules on computed tomographic (CT) images. Titer of cold hemagglutinin, an autoantibody against red blood cells, is consistently high for years in most cases without evidence of mycoplasma infection. Patients commonly have a history of chronic sinusitis. Typical DPB is seldom misdiagnosed because of the characteristic radiographic and pathologic findings. Of the other chronic inflammatory lung diseases, cystic fibrosis is extremely rare in the Japanese population and mutations of the cystic fibrosis transmembrane conductance regulator gene have not been found in patients with DPB (3). Recently, prognosis of this disease has been improved dramatically by low-dose erythromycin therapy, the effect of which is attributed to an anti-inflammatory action of macrolides (4).

DPB and sporadic familial cases (7) were initially found in the Japanese population. As it was recognized internationally, cases of DPB have been noted in other East Asians such as Chinese and Koreans (8). In non-Asian populations, however, only a limited number of cases have been reported (11, 12). Neither environmental factors nor infectious agents specific to the disease have been demonstrated so far (2). Considering these observations, we hypothesized that unique alleles of genes involved in the immune reaction or inflammatory response in DPB are shared by East Asians.

The human leukocyte antigen (HLA) system is essential for the appropriate immune response mediated by T-cell receptors and associations between HLA types and diseases, especially those with a presumed immune etiology have been extensively studied (13). In 1990, Sugiyama and coworkers serologically typed HLA-A, -B, and -C antigens of 38 patients with DPB and showed that 63% of patients possessed Bw54 antigen, compared with 11% of control subjects (14). Although the number of patients in their study was relatively small, it was noteworthy because HLA-Bw54 is a serotype that has been found predominantly in East Asians (15). In the present study, we newly collected 76 patients with DPB to analyze HLA genes. We directly identified individual HLA alleles at the nucleotide sequence level by using polymerase chain reaction (PCR)-based techniques and evaluated the contribution of each allele to genetic predisposition in DPB.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS REFERENCES

Subjects

Blood samples were obtained from 76 Japanese patients with DPB diagnosed in Toranomon Hospital, Tenri Hospital, or Nippon Medical School Hospital. The diagnostic criteria for DPB proposed in 1995 by a working group of the Ministry of Health and Welfare of Japan were: (1) persistent cough, sputum, and exertional dyspnea; (2) past or present history of chronic sinusitis; (3) bilateral diffuse small nodular shadows on a plain chest X-ray film or centrilobular micronodules on chest CT images; (4) coarse crackles; (5) FEV₁/FVC less than 70% and Pa_O2 lower than 80 mm Hg; and (6) titer of cold hemagglutinin equal to or higher than 64. All patients recruited for this study fulfilled criteria 1, 2, 3 and at least two of criteria 4, 5, and 6. These cases did not involve other HLA-Bw54-related diseases such as autoimmune hepatitis and silicosis. We also excluded cases in which the diagnosis of other chronic lung diseases was made. Patient characteristics are summarized in Table 1. The normal control subjects were 110 healthy volunteers from the same area of Japan.

Determination of Alleles of HLA Class I and Class II Genes

Genomic DNA was extracted from whole blood samples by the method described elsewhere (16). Class I HLA-A, -B, and -C antigens were screened serologically by the conventional National Institutes of Health (NIH) standard method. Class I HLA-B22 group alleles including B54 were genotyped by PCR-single-strand conformation polymorphism (SSCP) analysis developed recently (17). Briefly, exons 2 and 3 of HLA-B gene encoding polymorphism of B antigens were amplified separately by the seminested PCR method and then PCR products mixed with deionized formamide were electrophoresed at 30° C on a 10% nondenaturing polyacrylamide gel without glycerol. Single-stranded DNA fragments on the gel were visualized by silver staining (Daiichi Pure Chemicals, Tokyo, Japan). Each allele was identified by comparison with the electrophoretic patterns of standard samples, B*5401, B*5502, B*5504, B*5601, and B*5603, of which nucleotide sequences were determined by the direct sequencing method (ABI PRISM dRhodamine Terminator Cycle Sequencing Ready Reaction Kit; PE Applied Biosystems, Foster City, CA). Alleles of class II HLA-DRB1 gene were identified by the PCR-microtiter plate hybridization method reported previously (18). Briefly, exon 2 of HLA-DRB1 gene encoding polymorphism of DR antigens was amplified by the PCR method using biotin-labeled primers. The labeled PCR products were hybridized with a panel of sequence-specific oligonucleotide probes immobilized on the bottom of microtiter wells (Wakunaga Pharmaceutical Co. Ltd., Hiroshima, Japan). After unbound DNA was washed off, the wells were incubated with streptavidin-horseradish peroxidase followed by the substrate 2,2'-azino-di(3-ethylbenzthiazoline-6-sulfonic acid) and then the developed color indicating hybridized DNA was detected at 415 nm by a microtiter-plate reader.

Statistical Analysis

Disease association was assessed by chi-square (²) test. When any expected number in the 2 × 2 contingency table is less than 5, the p value was directly calculated by Fisher's exact test. In this study, these standard p values less than 0.05 were generally considered significant. Corrected p values (p_c), p values multiplied by the number of comparisons, were used to show a stronger association. Odds ratio (OR) was defined as the cross-product ratio of the numbers shown in the 2 × 2 table. Comparison of clinical data between B*5401-positive and -negative groups was made by using an unpaired t test or the Mann-Whitney U test for nonparametric data. The p values less than 0.05 were considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS REFERENCES

Association between HLA Class I Antigens and the Disease

Distribution of HLA-A, -B, and -C antigens in 76 DPB patients and the 110 normal subjects is shown in Table 2. A11, B54, and Cw1 presented positive association with the disease (p < 0.05). A33 and B44 exhibited negative association with the disease (p < 0.02). Of these, B54 antigen showed the strongest association; OR = 3.4, ² = 12.4, and p = 0.0004 (p_c < 0.01). HLA-B54, B55, and B56 are serologic subdivisions of HLA-B22. Thus, association between the whole B22 family and the disease was also calculated and more positive association was observed (49% versus 22%, OR = 3.4, ² = 14.7, and p = 0.0001). In addition, when a related B7 antigen was added to B22 antigen, the association was even more robust (table not shown; 63% versus 32%, OR = 3.7, ² = 17.9, and p < 0.0001).

Determination of Alleles of the HLA-B22 Family Including B54 by SSCP Analysis

To examine HLA-B54 molecule associated with the disease at the nucleotide sequence level, we performed PCR-based SSCP analysis and detected alleles encoding HLA-B22 group antigens which are B54, B55, and B56 (Table 3). All B54 alleles were exclusively identified as B*5401 and all alleles for non-B54 antigens were genotyped as non-B*5401. Therefore, 37% of DPB patients and 15% of the healthy subjects possessed the B*5401 allele. In B55-positive patients, B*5504, a rare allele identified recently (17), was predominant, while only B*5502 was observed in normal subjects. B56 alleles were mostly genotyped as B*5601, which was possessed by 8% of the patients and 3% of the normal control subjects. No other alleles for B22 group antigens were detected in this study.

Association between HLA Class II Alleles and the Disease

Alleles of HLA-DRB1 gene were determined for all the DPB patients and normal subjects by the microtiter plate hybridization method (Table 4). No single allele showed positive association, whereas DRB1*1302 exhibited significant negative association with the disease (OR = 0.18, p < 0.01). The presence of any DRB1*0101, *0404, *0405, and *0410 alleles which share amino acid residues 70 to 74 was only weakly associated with the disease (table not shown, OR = 1.8, p = 0.04).

Clinical Features of B*5401-positive and -negative Patients

To examine whether B*5401-positive patients compose a clinically separated subgroup, we compared sex ratio, necessity of CT scanning to detect nodular lesion, effectiveness of erythromycin, FEV₁/FVC, Pa_O2, titer of cold hemagglutinin, age at the time of diagnosis, and duration of the disease, between B*5401-positive group and -negative group. However, no items were significantly different between the two groups (data not shown).

DISCUSSION

In the present study, we demonstrated that HLA-B*5401 allele encoding B54 antigen is definitely associated with DPB. HLA-A11 and Cw1 were weakly associated with the disease. The latter appears to be secondary association due to linkage disequilibrium because A24/11-B54-Cw1-DR4 (DRB1*0405) is a typical haplotype observed in 4% of the Japanese population (19, 20). In contrast, HLA-A33, B44, and DRB1*1302 showed negative association. B44 is in strong linkage disequilibrium with A33 and DRB1*1302, and A33-B44-C blank-DR13 (DRB1*1302) is another representative haplotype in the Japanese population (19, 20). These findings indicate that HLA-B gene or a closely linked gene in the HLA region is involved in the susceptibility of Japanese to DPB.

By analyzing a larger number of cases than that reported previously (14), we confirmed positive association between HLA-Bw54 and the disease. Furthermore, we directly identified HLA alleles in DPB patients at the nucleotide sequence level for the first time. This enabled us to deduce the primary structure of the HLA molecules involved.

B*5401 allele is mainly restricted to East Asians including Japanese, Koreans, and Chinese, not found in Africans or Native Americans and rare in Caucasians (15, 17). The nucleotide and amino acid sequence of the B*5401 molecule closely resembles B*5501-04 which encode B55 antigen and B*5601-03 encoding B56 and these are grouped into the HLA-B22 family (17, 21). B*5502, B*5504, B*5601, and B*5603 described in this study are also oriental alleles and these appear to have diverged from the ancestral form in Asia (17).

Although we have not had any direct evidence that B*5401 itself is involved in the pathogenesis of DPB, recent progress in our understanding of HLA class I molecules suggests this possibility. The biologic function of HLA class I molecules is to bind antigenic peptides of 9 or 10 amino acids in length and to present them to specific T cells for recognition. Polymorphisms within HLA class I genes play an important role in determining the repertoire of presented peptides (13, 22). HLA-B*5401 bears a characteristic amino acid sequence in the ₁-helix of the molecule, a part of which, residues 57 to 84, is shared by other members of the B22 family, B7 and B42 alleles (21). Of these, residues 67 to 71, tyrosine 67-lysine 68-alanine 69-glutamine 70-alanine 71, are a particularly unique motif. These residues create a shared epitope and also cause a unique binding specificity for antigenic peptides carrying proline at position 2, whereas B44 and its related alleles preferentially bind distinct peptides bearing acidic residues such as glutamic acid at the same position (21, 22). When we postulate that HLA-B gene itself contributes to the disease susceptibility, we should focus on the structural feature discussed previously because our results suggest that B22 or B22 + B7 might form a group of susceptibility alleles and B44 is considered to be a resistant allele for DPB. B42, related to the B22 family, is not found in Japanese. Furthermore, it will be interesting if B*5401 and B*5504 share a more restricted peptide binding motif than other alleles of B22 or B7, because a rare B*5504 allele was predominant in B55-positive DPB patients.

HLA-B54 has been reported to be associated with inflammatory disorders such as silicosis (23) and autoimmune hepatitis (24) in Japan, but recent DNA typing has often revealed that the primary associations are represented by class II genes and that associations with class I genes are simply due to linkage disequilibrium with those class II genes (24). In fact, DPB has been reported to be weakly associated with MC1 antigen indicating class II DR1 and a subgroup of DR4 (14). To determine which locus shows primary association, we performed genotyping of DRB1 class II gene encoding DR antigens in the patients with DPB. As a result, no single allele of DRB1 gene was significantly associated with the disease. Although the frequency of DRB1*1302 was decreased in patients with DPB, we could not determine whether this negative association is primarily significant, because DRB1*1302 is in strong linkage disequilibrium with B44 in Japanese (19, 20). DRB1*0101, *0404, *0405, and *0410 alleles share residues 70 to 74 and the shared epitope has been reported to confer a primary risk for adult rheumatoid arthritis (25). In the case of DPB, however, the presence of the sequence in these alleles did not show strong association with the disease. The slight increase in the frequencies of DRB1*0101 and DRB1*0405 appears simply due to linkage disequilibrium with B7 and B54, respectively, because these haplotypes are frequently observed in Japan (19, 20). These results suggest that the class I region is more important as a locus for susceptibility to DPB than the class II region.

Although it is unknown why ubiquitous class I molecules might be involved in chronic inflammation limited to upper and lower airways, concepts of disease mechanism regarding ankylosing spondylitis might provide a good model to consider. In the case of ankylosing spondylitis, HLA-B27 is known to be strongly associated with the disease (26). One possible mechanism is that cytotoxic T cells are sensitized to a particular pathogen-derived peptide presented by a subgroup of HLA-B molecules with a unique peptide binding specificity. Subsequently these sensitized T cells could crossreact with a tissue-specific peptide with a sequence motif similar to the exogenous peptide and induce chronic inflammation at the sites of disease. The idea that a certain microorganism modulates host immune response locally and is implicated in the pathogenesis of DPB is attractive, although we have not had any definite evidence that a particular group of pathogens interacts with the alleles of the class I molecules that we identified in patients with DPB.

In this study, the phenotypic frequency of B54 antigen in the patient group was relatively lower than the 63% reported previously (14). Because our serologic data were verified by DNA typing as described previously, mistyping is unlikely. Therefore the high frequency of B54 antigen in the previous study might be explained by the random deviation in their small sample size. In fact, the frequency of B54 antigen in the large population of normal volunteers was not different between the two studies.

Because the association between particular alleles of HLA-B gene and the disease was not 100%, we should also take into account the contribution of unidentified genetic and environmental factors. It is also possible that the pathogenesis of DPB is heterogenous, because multiple disease processes may give rise to the same clinical entity. To test this issue in our patients, we compared B*5401-positive and -negative cases in detail, but no clinical characteristics discriminated between the two groups. Especially when DPB cases in different countries are compared, possible heterogeneity should be carefully investigated.

A possibility that B*5401 is merely a marker for a closely linked "DPB-related" gene cannot be ruled out. If this were the case, a specific allele of the candidate gene should be in linkage disequilibrium with B*5401 and B54-related alleles, although it is plausible that there is more than one genetic factor that determine susceptibility to DPB. As mentioned earlier, similarities in nucleotide sequences strongly suggest that Japanese HLA-B22 group alleles including B*5401 have been derived from an ancestral HLA-B allele in Asia (17). Therefore it is also conceivable that a founder mutation in a "DPB-related" gene close to the HLA-B locus has occurred on a chromosome bearing the ancestral allele of B*5401 and that the disease allele thus generated has expanded with evolution of B22 alleles in East Asia including Japan. This ancestral mutation hypothesis could explain why DPB is observed predominantly in East Asia. To determine which hypothesis is more likely, comparison of results from HLA studies of DPB in other ethnic populations is essential. Linkage studies using familial cases, epidemiologic surveys to compare the incidence of DPB among Asian people in North America, and an extensive search for candidate genes in the HLA region should also be conducted.

In conclusion, we demonstrated that DPB is a lung disease associated with an HLA class I gene. Our results suggest that the distinctive molecular structure of HLA-B alleles or a closely linked gene in the HLA region contributes to genetic predisposition in DPB.