Abstract
A glutamic acid at residue 69(Glu69) in the HLA-DPB1 gene (Glu69) is associated with chronic beryllium disease (CBD) and possibly beryllium sensitization (BeS). This study tested the hypothesis that MHC class II polymorphisms are important in susceptibility to BeS and CBD and that the Glu69 variant is related to markers of disease severity. Genomic DNA was obtained from BeS (n = 50), CBD (n = 104), and beryllium-exposed nondiseased (Be-nondiseased) (n = 125) subjects. HLA-DPB1, -DRB1, and -DQB1 genotypes were determined by (sequence-specific primers) PCR. Disease severity was assessed by pulmonary function and exercise testing. A higher frequency of the DPB1 Glu69 gene was found in CBD and BeS compared with the Be-nondiseased subjects, with odds ratios of 10.1 for CBD vs Be-nondiseased and 9.5 for BeS vs Be-nondiseased. The majority of BeS and CBD subjects displayed non-0201 Glu69 alleles. Glu69 homozygosity was higher in the CBD subjects, while BeS subjects were intermediate and Be-nondiseased lowest. DRB1*01 and DQB1*05 phenotypes were reduced in CBD vs Be-nondiseased subjects, while DRB1*13 and DQB1*06 were associated with CBD in the absence of Glu69. Markers of disease severity, including a lower forced vital capacity, diffusion capacity for carbon monoxide, PaO2 at rest, maximum workload on exercise testing, and a higher arterial-alveolar gradient at rest, were associated with Glu69 homozygosity. We conclude that DPB1 Glu69 is a marker of sensitization and not specific for disease. Glu69 homozygosity acts as a functional marker associated with markers of CBD severity.
Following exposure to beryllium (Be), 3 up to 20% of workers develop sensitization to Be (BeS) (1). Many of these individuals have the granulomatous lung disease chronic beryllium disease (CBD) at the time of first evaluation, while many others eventually develop CBD, progressing from BeS at a rate of ∼10% per year (2, 3). Individuals with BeS demonstrate a Be-specific immune response, as evidenced by the proliferation of PBMC in the presence of Be salts in the blood Be lymphocyte proliferation tests (BeLPT) (4, 5). These individuals do not have any evidence of pulmonary pathology on lung biopsy or physiologic abnormalities. In contrast, CBD individuals display granulomatous inflammation as evidenced by noncaseating pulmonary granulomas and mononuclear cell infiltrates on lung biopsy. Individuals with CBD may demonstrate both a peripheral blood and bronchoalveolar lavage (BAL) lymphocyte proliferative response on BeLPT (4, 5).
The immunopathogenesis of CBD and BeS hinges on the development of an Ag-specific, cell-mediated immune response. Be-specific CD4+ T cells probably recognize a form of Be as an Ag or as a hapten, acting in combination with MHC class II molecules on APCs (6). This forms the basis of the BeLPT. These CD4+ T cells express a biased repertoire of TCRs, confirming the specificity of the Be-stimulated response in CBD (7, 8). Evaluating these TCRs in the development of granulomas after Be skin patch testing of CBD subjects suggests that clonal T cells, similar to those found in the lung, are mobilized from the blood and infiltrate the affected tissue during granuloma formation (9). In early studies, Abs to MHC class II but not class I molecules blocked the ability of Be to induce CD4+ T cell proliferation (6). Current studies indicate that HLA-DPB1 is the predominant class II molecule presenting a Be Ag to T cells, thus inducing Ag-specific proliferation (10, 11). In these studies, proliferation of Be-specific CBD T cell clones was reconstituted by APCs expressing a glutamic acid residue at aa position 69 (Glu69) in DPB1and inhibited by a mAb directed against HLA-DPB1. Other studies have also shown that a mAb against HLA-DPB1 is able to block Be-stimulated IFN-γ and TNF-α production, suggesting that this gene is not only important in Be-stimulated proliferation, but also in the production of Be-stimulated cytokines (10, 12). These multiple studies support a functional role for the Glu69 polymorphism in Be Ag presentation and the ensuing inflammatory response in CBD and possibly BeS (13, 14, 15).
Studies in humans (16, 17) and animals (18, 19) exposed to Be support the importance of genetic susceptibility in CBD. In a landmark study, Richeldi et al. (13) reported an increased prevalence of HLA-DPB1 Glu69 in cases of CBD (97%) compared with Be-exposed nondiseased (Be-nondiseased) controls (30%). These results were later confirmed by additional studies, although the frequency of the Glu69 variant was slightly lower (14, 20, 21). Wang et al. (14) also noted that CBD subjects were more likely to be homozygous for the Glu69 gene than were Be-nondiseased controls. Furthermore, the predominant form of the Glu69 expressed in CBD subjects was not the more common HLA-DPB1 (*0201), but more likely to be rarer DPB1 non-*0201 variants. Subsequent studies have provided mixed results (20, 21). Initially it was thought that the Glu69 variant was a marker of disease susceptibility. Two recent studies indicate that Glu69 is a marker of the immune response to Be, found at the same frequency in BeS as CBD (20, 22), while another study found a lower rate of Glu69 in BeS than CBD (21), leaving this issue less clear. These differences are probably due to the small sample size and limited case characterization with potential for misclassification present across many of these studies. As only 80–85% of CBD subjects have a Glu69, other class II markers are likely to be involved in the Be-specific immune response and Be-stimulated proliferation, although this has been less well studied, to date, with conflicting results (20, 21).
Based on these studies, we hypothesized that Glu69 would be associated with BeS and not specific for CBD. We also hypothesized that Glu69, as a functional variant, would be associated with a higher Be-stimulated proliferative response and with more severe disease, which has not been evaluated to date. Specifically, we hypothesized that subjects with one or more copies of the Glu69 gene would be likely to demonstrate a stronger proliferative and inflammatory response to Be and thus develop more severe disease. Finally, it is likely that other HLA-DPB1, -DRB1, or -DQB1 genes are involved in BeS and CBD. To address these questions, we enrolled a large cohort of clinically evaluated, well-defined cases of BeS and CBD and controls in this study. We used sequence-specific primers PCR (SSP-PCR) to determine HLA-DPB1, -DRB1, and -DQB1 genotypes in BeS, CBD, Be-exposed subjects without BeS or disease and compared genotypes and markers of disease severity in CBD.
Materials and Methods
Study design
We conducted a case control study of subjects with CBD (n = 104) and BeS (n = 50) and controls with Be exposure without evidence of sensitization or disease (Be-nondiseased, n = 125) to evaluate HLA-DPB1, -DRB1, 3, 4, 5, and -DQB1 genotypes and disease. A case comparison study was undertaken to determine whether the Glu69 genotype was associated with markers of disease severity, including the blood and BAL BeLPT. The cases were enrolled from individuals evaluated in the outpatient Occupational and Environmental Medicine Clinic at National Jewish Medical and Research Center for abnormal respiratory symptoms or referral for an evaluation of an abnormal BeLPT detected during workplace medical surveillance. These subjects had been exposed in a number of industries across the United States, including nuclear weapons manufacture, Be manufacture and processing, ceramics manufacture, Be machining, and other industries. The Be-nondiseased controls were enrolled from a group of workers undergoing medical surveillance for CBD at a precision Be machining facility (23, 24, 25). All study subjects provided informed consent according to a protocol reviewed by the Human Subjects Institutional Review Board at National Jewish.
Study subjects
Subjects with BeS were enrolled if they had evidence of sensitization demonstrated by two abnormal BeLPTs or a positive Be skin patch test and no evidence of granulomatous inflammation on transbronchial lung biopsy. Cases of CBD were enrolled if they demonstrated sensitization with two abnormal BeLPTs or an abnormal BAL LPT or Be skin patch test and evidence of granulomatous inflammation on lung biopsy. Findings of noncaseating granulomas or mononuclear cell infiltrate on lung biopsy or BAL cell lymphocytosis (≥15%) along with an abnormal BAL LPT were used to define evidence of inflammation consistent with CBD. Be-nondiseased controls were defined as individuals employed in a Be industry who had at least one normal blood BeLPT without an abnormal BeLPT.
Cases of CBD and BeS underwent clinical evaluation consisting of venipuncture, a BeLPT, chest radiograph, pulmonary function testing, exercise testing, bronchoscopy with BAL, transbronchial biopsies, and BAL LPT at the time of diagnosis as previously described (26). Control subjects underwent venipuncture and BeLPT. All subjects completed a modified version of the American Thoracic Society questionnaire, providing demographic information (26).
SSP-PCR determination of the HLA-DPB1, -DRB1, 3, 4, 5, and -DQB1 loci
Genomic DNA was prepared by using the Wizard Genomic Purification kit (Promega, Madison, WI). HLA-DPB1, -DRB1, 3, 4, 5, and -DQB1 genotyping was performed with blinding to the subject’s disease status using SSP-PCR methodology as described by Bunce et al. (27) and Gilchrist et al. (28).
Statistical analysis
Continuous variables were compared using Student’s t test or Wilcoxon’s rank sum test as appropriate. For three-way comparisons, the Kruskal-Wallis test was used and Dunn’s method was used to control for each set of pairwise comparisons. Categorical variables were analyzed using χ2 and Fisher’s exact methods. To analyze the phenotypic frequencies of specific variants, the frequency of subjects with and without the polymorphism of interest was compared using either a three-way comparison of CBD, BeS, and Be-nondiseased controls or a two-way comparison, combining cases of CBD and BeS compared with Be-nondiseased controls. For the case comparison analysis, cases with the genotype of interest were compared with those without the genotype. A corrected p value (Pc) was determined for all multiple comparisons using the Bonferroni method. The correction factor used was the number of comparisons made: for HLA-DPB1–21; for the amino acid epitopes -15; for HLA-DRB1-13; and for HLA-DQB1–8. An odds ratio (OR) with 95% confidence intervals (95% CI) was determined. Statistical analysis was performed using Knowledge Studio (Angoss Software, Guildford, Surrey, U.K.) JMP-SAS or SAS software (SAS Institute, Cary, NC). All tests were two sided and a p < 0.05 was used to determine statistical significance.
Results
Demographics of cases and controls
The demographics of the cases and controls are shown in Table I⇓. The Be-nondiseased subjects were younger, had fewer years since first Be exposure, and were more likely to be male and non-Hispanic than were the BeS and CBD cases. Because of the difference in ethnicity noted between cases and controls, the genotypic analysis was conducted two ways: with all subjects and with the non-Hispanic Caucasian subjects to ensure no confounding due to ethnicity. The cases of BeS and CBD did not differ in any of the demographic variables.
Comparison of demographic characteristics among CBD, BeS, and Be-nondiseased subjectsa
HLA-DPB1 genotypes in CBD and BeS
Phenotype frequencies of HLA-DPB1.
We summarized the phenotype frequencies of HLA-DPB1 alleles with CBD, BeS, and Be-exposed controls in Table II⇓. DPB1*0601 allele was increased in both CBD (12.8%, p = 0.00008, Pc = 0.001) and BeS (14.6%, p = 0.00003, Pc = 0.0006) compared with Be-nondiseased subjects (0%). As found previously, DPB1*0201 alleles were found at higher frequency in cases of CBD (39.4%, p = 0.02, Pc = NS) and BeS (39.6%, p = NS) compared with the Be-nondiseased controls (24.3%). Similarly, DPB1*0901 and DPB1*1001 were also increased in CBD (9.6%, p = 0.004, Pc = NS for DPB1*0901; 16.0%, p = 0.005, Pc = NS for DPB1*1001) and BeS (6.3%, for DPB1*0901; 14.6%, p = 0.02, Pc = NS for DPB1*1001) compared with Be-exposed control (0.9% for DPB1*0901; 4.3% for DPB1*1001). Conversely, DPB1*0301 was identified at a lower frequency in cases of CBD (8.5%, p = 0.03, Pc = NS) and BeS (4.2%, p = 0.01, Pc = NS) compared with controls (19.1%), while DPB1*0401 was present at a lower frequency in CBD (35.1%) compared with BeS (58.3%, p = 0.008, Pc = NS) and Be-nondiseased controls (68.7%, p = 0.000001, Pc = 0.00002).
Comparison of HLA-DPB1 group by phenotypic frequency among CBD, BeS, and Be-nondiseased subjects
Glu69 in CBD and BeS.
Table III⇓ shows polymorphic amino acid residues of DPB1 alleles. DPB1*0201, *0202, *0601, *0801, *0901, *1001, *1301, *1601, and *1701 are all Glu69-containing alleles. As shown in Fig. 1⇓, Glu69-containing alleles were found expressed at a much higher frequency in CBD (86.2%, χ2 = 47.4, Pc < 0.0001, OR = 10.0, 95% CI, 5.0–20.2) and BeS (85.4%, χ2 = 28.3, p < 0.001, OR = 9.5, 95% CI, 3.9–22.9) than controls (38.3%). The DPB1 non-*0201Glu69 variants were expressed at a higher rate in CBD (62.8%, χ2 = 56.8, Pc < 0.0001, OR = 12.2, 95% CI, 6.1–24.4) and BeS (56.3%, χ2 = 32.6, Pc < 0.0001, OR = 9.3, 95% CI, 4.2–20.6) compared with controls (13.9%; Fig. 1⇓). Furthermore, cases of CBD (25.5%, χ2 = 24.7, Pc < 0.001,OR = 19.4, 95% CI, 4.4–84.5) and BeS (14.6%, χ2 = 8.4, p = 0.004, OR = 9.7, 95% CI, 1.9–48.3) were more likely to carry two copies of a Glu69-containing gene than the controls (1.7%). There was no statistically significant difference between CBD and BeS cases. However, when comparing Glu69 homozygous subjects to Glu69 heterozygotes, the OR for CBD was 8.8 (95% CI, 2.0–39.5, p < 0.02) and was not significant for BeS (OR = 4.3, 95% CI, 0.8–22.2, p = NS) compared with controls. We determined Glu69 allele frequency in 12 of our CBD cases who had progressed from BeS to CBD during follow-up at National Jewish. All but 1 of these subjects carried at least 1 Glu69 (91.7%), while 5 of the 12 (41.7%) were homozygous for Glu69.
The genotypic frequency of any Glu69 (▨), DPB1 non-*0201 Glu69 alleles (▦) and Glu69 homozygosity (▪) is shown for CBD, BeS, and Be-nondiseased subjects.
Polymorphic amino acid residues of HLA DPB1 alleles
Other HLA-DPB1 amino acid epitopes.
We evaluated whether other DPB1 amino acid epitopes besides the well-known Glu69 variant were associated with CBD or BeS by comparing the sequence of those alleles (Table III⇑). The DPB1 epitopes with a histidine at aa position 9 (H9), leucine at 11 (L11), valine at 36 (V36), and aspartic acid at 55 and glutamic acid at 56 (D55E56) were increased in cases of BeS and/or CBD compared with the Be-nondiseased controls (Table IV⇓). Various combinations of these epitopes were evaluated to determine whether they were expressed at higher frequencies in the cases than the controls. A higher frequency of subjects who had a number of the above epitopes, the V36D55E56Glu69 was found in cases of BeS and CBD than in the controls. The OR associated with the presence of all of these epitopes in CBD was lower than that found for Glu69 alone (OR = 7.5, 95% CI, 4.0–14.1), as it was for any of the amino acid epitopes alone. Most of these amino acid epitopes were also found in combination with the Glu69. Thus, none was found to be differentiating BeS or CBD from the controls better than Glu69.
Comparison of distribution of HLA-DPB1 amino acids by CBD, BeS, and Be-nondiseased subjects
HLA-DRB1, 3, 4, and 5 in CBD and BeS
DRB1*01 allele was found at a significantly lower frequency in CBD (7.9%) than in BeS (22.9%, χ2 = 5.3, p = 0.008, Pc = NS, OR = 0.29, 95% CI, 0.11–0.78) or Be-nondiseased subjects (28.9%, χ2 = 14.2, p = 0.00008, Pc = 0.001, OR = 0.21, 95% CI, 0.09–0.48). The frequency of DPB1 non-Glu69 phenotypes in the case and control subjects with and without DRB1*01 alleles did not differ (p ≥ 0.05), suggesting that this association was not due to the linkage disequilibrium with Glu69. Although not statistically significant, DRB1*03 was underrepresented in CBD (19.8%) compared with Be-nondiseased (27.1%) and DRB1*13 was found at a greater frequency in CBD (33.7%) than in BeS (20.8%) or Be-nondiseased subjects (23.1%; Table V⇓).
Comparison of DRB1 group among CBD, BeS, and Be-nondiseased subjects
The presence of HLA-DRB3, 4, and 5 alleles were determined for cases and controls. There were no significant differences in the presence of any of these alleles between cases and controls (data not shown).
To determine whether HLA-DRB1 Ag groups are associated with CBD or BeS in the absence of Glu69, we evaluated DRB1 groups in those CBD (n = 12), BeS (n = 7), and Be-nondiseased (n = 65) subjects without a Glu69. DRB1*13 was overrepresented in CBD cases (58.3%) compared with controls (18.5%, p = 0.003). Although the BeS subjects displayed a similar frequency of DRB1*13 (14.3%) compared with controls, this did not differ significantly from CBD. No other significant differences were noted in DRB groups.
HLA-DQB1 in CBD and BeS
A statistically significant difference in the phenotype frequency of DQB1*05 was noted between the cases of CBD (19.8%) compared with the Be-nondiseased controls (37.2%, χ2 = 7.2, p = 0.005, Pc = 0.04, OR = 0.42, 95% CI, 0.23–0.77, Table VI⇓). The BeS subjects displayed an intermediate frequency of DQB1*05 (27.1%), which did not differ from the CBD cases or the controls. As noted above for DRB1*01, the frequency of DPB1 non-Glu69 phenotypes did not vary by DQB1*05 phenotype in the cases or controls (p ≥ 0.05). DQB1*06 alleles were present at a higher frequency in CBD (51.5%) compared with the BeS (41.7%) and Be-nondiseased (38.8%), although there was no significant difference between them. To determine whether HLA-DQB1 Ag groups are associated with CBD or BeS in the absence of Glu69, we evaluated DQB1 groups in those CBD (n = 12), BeS (n = 7), and Be-nondiseased (n = 65) subjects without a Glu69. DQB1*06 was more prevalent in CBD (75.0%) compared with controls (38.5%, p = 0.03, Pc = NS). BeS subjects displayed an intermediated frequency (57.1%), which did not differ significantly from either group.
Comparison of DQB1 group among CBD, BeS, and Be-nondiseased subjects
Disease severity and DPB1 Glu69
Since the Glu69 variant is a functional polymorphism that has been associated with Be-stimulated proliferation and cytokine production in CBD, we hypothesized that it would be associated with more severe disease in CBD and a higher proliferative response in cases of BeS and CBD. Using the results from CBD subjects’ initial evaluation, Glu69 subjects displayed evidence of abnormal gas exchange with a lower diffusion capacity for carbon monoxide (DLCO) (median, 95.5) than the non-Glu69 subjects (median 110, p = 0.05). Of note, the DLCO of those Glu69 subjects with homozygosity was significantly lower (p = 0.05, Table VII⇓) than non-Glu69 subjects, while the Glu69 heterozygotes displayed an intermediate DLCO (p > 0.05). Other evidence of abnormal gas exchange in the Glu69 homozygotes included a lower partial pressure for oxygen (PaO2) at rest and higher arterial-alveolar (A-a) gradient at rest than the Glu69 heterozygous and non-Glu69 subjects (p ≤ 0.04, Table VII⇓). There was evidence of reduced exercise capacity with a reduction in maximum workload in the Glu69 homozygous subjects (p = 0.03). The Glu69 homozygous CBD subjects also displayed a lower FVC percent predicted than those with a non-Glu69 gene (p = 0.05). No significant differences were noted in forced expiratory volume in 1 s (FEV1), total lung capacity, or BAL fluid cell white blood cell count and lymphocyte percent by Glu69 genotypes (Table VII⇓).
Pulmonary and exercise physiology in CBD by HLA-DPB1 with a glutamic acid at amino acid position 69 (Glu69) copy number;a
When comparing the peak stimulation index (SI) from CBD and BeS cases’ first BeLPT, we found no statistically significant association with the Glu69 variant, regardless of whether the subjects were compared by Glu69 vs non-Glu69 or Glu69 homozygosity vs Glu69 heterozygosity vs non-Glu69 (data not shown). Evaluating the CBD cases alone, those with any Glu69 tended to have a higher BAL LPT peak SI than those without, although this was not statistically significant (p = 0.14). Those with Glu69 heterozygosity had a significantly higher peak BAL LPT SI than those with no Glu69 (Table VII⇑). Although not statistically significant, the Glu69 homozygotes tended to demonstrate a lower BAL LPT peak SI than did the Glu69 heterozygotes (p = 0.08).
Discussion
This is the largest study to date evaluating HLA class II genes as a risk factor in CBD and BeS compared with subjects who have been exposed to Be but have no evidence of sensitization or disease. We found an increased risk of both CBD and BeS associated with the HLA-DPB1 gene with a glutamic acid at aa position 69, confirming this as a marker of the Be-specific immune response and not specific for disease per se. Glu69 homozygosity was associated with CBD (OR 19.4) and BeS (OR = 9.7). Of those subjects with a Glu69, prevalence of less frequently expressed HLA-DPB1 non-*0201 Glu69 alleles was comparable in BeS and CBD and increased compared with the Be-nondiseased controls, with an OR similar to that for any Glu69 allele.
Modeling of the HLA-DP β-chain predicts that Glu69 is located on the α helix with side groups pointing into the Ag peptide-binding cleft (Fig. 2⇓) (10). This observation suggests that the Glu69 is involved in Be Ag binding (29). There are other variable regions in the HLA-DP β-chain, that might also influence peptide binding. For example, the H residue at position 9, L at position 11, V at position 36, and D and E at position 55–65 are all predicted to be located in the peptide-binding cleft (Fig. 2⇓). Thus, like the Glu69 residue, these amino acids also determine the ability of certain peptides to bind. Interestingly these amino acids are usually found in conjunction with Glu69.
A proposed structure for HLA-DPB1 is presented based on the DRB1*01 molecule. A top view of the Ag-binding groove is shown, with hemagglutinin peptide as the purple stick, the α-chain in green, and β-chain in blue. The amino acids of interest are numbered and shown as red balls.
This is the first study to demonstrate an association between the Glu69 variant and clinical manifestations of disease. Glu69 homozygosity was associated with worse gas exchange, as indicated by a decreased DLCO, PaO2 at rest, and increased A-a gradient at rest, worse exercise capacity, with a reduced workload, and a lower FVC indicating more severe disease in CBD. Although we hypothesized that, as a functional gene, Glu69 would be associated with the BeLPT response in CBD and BeS, no association was noted between the blood BeLPT peak SI and any Glu69 variant. Surprisingly, we did find a decreased proliferative response in the BAL in CBD subjects homozygous for Glu69.
Our data suggest that in addition to the DPB1 allele association with BeS, CBD, and disease severity that some HLA class II genes may be protective. Specifically, we found that HLA-DRB1*01 and -DQB1*05 alleles were underrepresented in CBD compared with BeS and Be-nondiseased controls, suggesting a protective role for these alleles.
HLA-DPB1 in CBD and BeS
Studies show that HLA- DPB1 Glu69 is associated with CBD (13, 14, 20, 21, 30). The relationship between Glu69 and BeS has been less clear, as two studies have found an association (20, 21, 22) while one has not (21). These differences likely occurred for a number of reasons. First of all, most of the previous studies contained relatively small case populations, varying from a n = 6–25 cases of CBD (13, 14, 20, 21, 30) and n = 23–30 cases of BeS (20, 21, 22). Second, the clinical phenotype of the cases of BeS and CBD have varied. Some subjects with BeS had not undergone definitive clinical evaluation, including a bronchoscopy, to exclude a diagnosis of CBD (20, 22). Our study contains the largest number of CBD and BeS cases of any to date. We also used a stringent case definition of CBD and BeS, excluding individuals who had not undergone bronchoscopy, to avoid misclassification between BeS and CBD. As a result, it also contained subjects with a spectrum of clinical disease manifestations, including some with normal chest radiography and pulmonary physiology. We also enrolled a well-defined control group, who had undergone serial BeLPT testing every other year since 1995. The results of the most recent BeLPT were used to confirm that these subjects had no evidence of BeS.
This control group had fewer years of Be exposure, which may have provided them with less time to develop BeS or CBD. However, if anything this should only result in misclassification of individuals who will develop BeS or CBD as controls and acceptance of the null hypothesis, which did not occur. Our controls had fewer years of exposure, and their quantitative or qualitative exposures may have differed from the cases. However, the one study that examined genetic-exposure interactions suggested that HLA-DPB1 Glu69 and exposure are independent risk factors for BeS and CBD (30, 31). Thus, the difference in our cases and controls should not limit our conclusions. Furthermore, the frequency of Glu69 in CBD in our study was similar to that found in previous studies (13, 14, 20, 21). Our study supports the previous studies by Rossman et al. (20) and Wang et al. (22) confirming that the Glu69 variant is a marker of an immune response to Be and not specific to CBD.
Glu69 homozygosity in CBD and BeS
Glu69 homozygosity was the greatest risk factor for CBD (OR, 19.4 vs 10.0 for any Glu69). In fact, comparing homozygous to heterozygous subjects showed an increased risk for CBD compared with controls but not for BeS. This may indicate that by increasing the number of Glu69-containing HLA-Ag complexes on APCs, activation of Be-specific TCRs and an inflammatory response may be increased. However CBD and BeS subjects did not differ significantly with respect to rate of homozygosity probably due to the small number of BeS cases. The prevalence of Glu69 homozygosity was intermediate in BeS, suggesting that this group is a mixed population of individuals, some of whom may progress to CBD and some who may not. To determine whether Glu69 homozygosity is a risk factor for progression from BeS to CBD, a longitudinal study of BeS subjects would need to be conducted, evaluating progression over time based on Glu69 status.
Non-0201 Glu69 alleles and implications for Ag binding
Previous studies have demonstrated that the Glu69 gene is functional, as Abs to HLA-DP can block the proliferative response to Be in T cell clones and cytokine production (10, 11, 12, 32). Our studies and others (14, 20, 22) indicate that specific Glu69 alleles, such as the DPB1 non-*0201 alleles, are more prevalent than the more common DPB1 *0201 Glu69-containing allele. Although we found an increased frequency of the DPB1*0201 variant associated with both BeS and CBD, this was only modest (OR = 2.0) compared with the presence of the DPB1 non-*0201 Glu69 variants (OR = 12.2 for CBD and 9.3 for BeS). This would suggest that DPB1 non-*0201 variants might affect the putative Ag or TCR binding and thus affect the immune response to Be (22).
Our study and that of Rossman et al. (20) found an L at aa position 11 on HLA-DPB1 and a DE at positions 55–56 were increased in CBD compared with Be-nondiseased controls, suggesting that these other amino acids may affect putative Ag binding. We also found an increased prevalence of an H at position 9 and a V at position 36, the latter which has also been noted in sarcoidosis (33). Although the majority of these changes in amino acid positions are relatively conservative, subtle changes in amino acid structure or side chain composition may still affect Ag binding. We have modeled a proposed structure of the HLA-DPB1 molecule based on the known structure of HLA-DRB1*01 with an influenza hemagglutinin peptide to indicate the position of these various amino acids (Fig. 2⇑). These amino acids, including the Glu69 and DE at 55–56, may affect Ag binding either by affecting binding directly to Be itself (29) or affecting a peptide bound in the groove (34). For example, the change from lysine (K) to E at position 69 significantly alters the charge of the amino acid and likely peptide binding, as well as the conformation of the binding structure, as would the change in amino acid at position 55–56. Additionally, the H at aa position 9 provides a neutral side chain, which could provide an additional hydrogen binding site and thus could have an impact on the charge in this region of the protein. The lack of the asymmetric glycine (G) and the presence of the more symmetric L at position 11 would likely affect the Ag-binding structure or binding to other amino acids in the pocket or nearby. However, since the ORs for each of these amino acid epitopes were less than that for Glu69, it is likely that these are less important than the Glu69. Interestingly, although Scott et al. (29) found amino acids at positions 84–85 important in modeled Be Ag binding, we found no association between these variants and either CBD or Bes.
HLA-DRB1 and DQB1 in CBD and BeS
An alternative route for Be-induced proliferation must exist, since 15% of CBD and BeS subjects do not have a Glu69. Despite small numbers, when we evaluated non-Glu69 subjects, an increased frequency of DRB1*13 and, to a lesser degree, DQB1*06 was noted in CBD cases compared with controls. These two variants are in linkage disequilibrium, partially explaining these findings. In addition, another study demonstrated that Be-stimulated proliferation of T cell clones from one CBD subject could be reconstituted with a DR-matched APC, DRB1* 1501 (10), indicating that DRB1 may also be important in the Be-stimulated immune response.
When all cases and controls were evaluated, we did not find any significantly increased DRB1 or DQB1 variants associated with CBD or BeS. On the contrary, reduced frequencies of DRB1*01 and DQB1*05 were associated with CBD, but not BeS, compared with Be-nondiseased controls. Thus, these HLA variants, which are in linkage disequilibrium, are “protective” for CBD, associated with a reduced risk of disease. This does not appear to be due to linkage disequilibrium with DPB1 non-Glu69 alleles, as we found no difference in non-Glu69 phenotypes in subjects with or without DRB1*01 or DQB1*05. Interestingly, HLA-DRB1*01 is also associated with reduced risk of sarcoidosis (35, 36). In addition to affecting binding of a putative Ag and thus affecting Ag-specific T cells with appropriate T cell receptors, HLA class II alleles may have other immunoregulatory effects. In some autoimmune disease, there is evidence to suggest that protective or resistant class II molecules are associated with a shift in the Th1 and Th2 cytokine balance, supporting a more Th2-predominant milieu (32, 37, 38, 39). In sarcoidosis, a granulomatous lung disease clinically indistinguishable from CBD, a specific class II allele is associated with a reduced Th1 cytokine response, better prognosis, and disease remission (40). In animal models of autoimmunity, protective class II molecules may influence the immune response by deleting a potentially reactive subset of T cells or by inducing T cell tolerance (39, 41, 42, 43). Whether one or all of these mechanisms is relevant to CBD is beyond the scope of this article, but warrants future study from a mechanistic standpoint.
Glu69 disease severity and Be proliferation
Studies from our group show that HLA-DP functions not only to mediate Be-stimulated proliferation, but also Be-stimulated IFN-γ and TNF-α production, as an Ab to HLA-DP will abrogate Be-stimulated CBD BAL cell IFN-γ and TNF-α production (32, 39). Our study findings of an association between Glu69 and worse gas exchange, exercise capacity and FVC suggests that the HLA-DPB1 Glu69 copy number may affect the inflammatory response within the lung that results in more severe granulomatous inflammation and more severe disease.
Contrary to our hypothesis, we found no association between Be-induced blood cell proliferation and Glu69 copy number. CBD subjects with one copy of the Glu69 gene had a higher BAL LPT response than did those with no Glu69 genes, but contrary to our hypothesis, homozygotes tended to have a lower median response compared with the heterozygotes. Previous work by our group indicates that proliferation is not directly related to the number of Ag-specific cells, even though an HLA-DP Ab was able to block most if not all Be-stimulated proliferation (32). Previous studies indicate that CBD BAL CD4+ cells are previously activated memory T cells (32, 44), the majority of which are terminally differentiated and release cytokine upon Ag stimulation (32). Although it is unknown whether disease severity is associated with the number of Be-specific T cells, it is possible that with advancing disease the lung becomes infiltrated with memory T cells that can secrete cytokines, but which do not proliferate. This speculation would help explain our findings of the diminished proliferative capacity of BAL T cells in our Glu69 homozygous CBD subjects with more severe disease.
Conclusion
HLA-DPB1 Glu69 is a genetic marker associated with development of BeS and is not specific for CBD. Glu69 homozygosity, while associated with both BeS and CBD, confers greater risk of developing CBD than does heterozygosity and as such might be important in the progression from BeS to CBD. Glu69 copy number appears to be important in the development of more severe disease, possibly due to its effect on cytokine production. The class II genes DRB1*13 and/or DQB1*06 may also be associated with CBD in the absence of Glu69, while DRB1*01 and/or DQB1*05 appear to be protective for CBD. Future studies will be needed to address gene-environment and gene-gene interactions in BeS, progression to CBD, and development of more severe CBD.
Acknowledgments
We thank Lynn Jui, MPH, Margaret Mroz, MSPH, and Barbara Bardenheier, MSPH, for technical assistance and Sharon Warren for expert administrative support. We also thank the clinical support staff at National Jewish for assistance in patient care and acknowledge the ongoing support from beryllium patients and workforces that make this and other beryllium-related research possible.
Footnotes
-
↵1 This work was supported by Grants K08 HL03887, PO1 ES011810, GCRC M01 RR00051, and R01 ES06538 from the National Institutes of Health and Grant CDC/ NIOSH CCU 812221.
-
↵2 Address correspondence and reprint requests to Dr. Lisa A. Maier, National Jewish Medical and Research Center, 1400 Jackson Street, Denver, CO 80206. E-mail address: maierL{at}njc.org
-
↵3 Abbreviations used in this paper: Be, beryllium; BeS, Be sensitization; CBD, chronic Be disease; BeLPT, Be lymphocyte proliferation test; LPT, lymphocyte proliferation test; Be-nondiseased, Be-exposed nondiseased; SSP, sequence-specific primer; Pc, corrected p; OR, odds ratio; CI, confidence interval; FEV1, forced expiratory volume in 1 s; SI, stimulation index; A-a, arterial-alveolar gradient; FVC, forced expiratory volume; DLCO, diffusion capacity for carbon monoxide.
- Received June 18, 2003.
- Accepted October 7, 2003.
- Copyright © 2003 by The American Association of Immunologists
References
In this issue
