Chronic beryllium disease (CBD) is associated with the allelic substitution of a Glu69 in the HLA-DPB1 gene. Although up to 97% of CBD patients may have the Glu69 marker, about 30–45% of beryllium-exposed, unaffected individuals carry the same marker. Because CBD occurs in only 1–6% of exposed workers, the presence of Glu69 does not appear to be the sole genetic factor underlying the disease development. Using two rounds of direct automated DNA sequencing to precisely assign HLA-DPB1 haplotypes, we have discovered highly significant Glu69-containing allele frequency differences between the CBD patients and a beryllium-exposed, nondiseased control group. Individuals with DPB1 Glu69 in both alleles were almost exclusively found in the CBD group (6/20) vs the control group (1/75). Whereas most Glu69 carriers from the control group had a DPB1 allele *0201 (68%), most Glu69 carriers from the CBD group had a non-*0201 DPB1 Glu69-carrying allele (84%). The DPB1 allele *0201 was almost exclusively (29/30) associated with DPA1 *01 alleles, while the non-*0201 Glu69-containing DPB1 alleles were closely associated with DPA1 *02 alleles (26/29). Relatively rare Glu69-containing alleles *1701, *0901, and *1001 had extremely high frequencies in the CBD group (50%), as compared with the control group (6.7%). Therefore, the most common Glu69-containing DPB1 allele, *0201, does not seem to be a major disease allele. The results suggest that it is not the mere presence of Glu69, per se, but specific Glu69-containing alleles and their copy number (homozygous or heterozygous) that confer the greatest susceptibility to CBD in exposed individuals.

Beryllium is used in a large number of modern industries because of its unique chemical and physical properties. Inhalation of beryllium dust has long been recognized as a cause of interstitial lung disease (1). Although acute beryllium disease was reported among the industrialized countries in the 1940s and 1950s, with improvements in environmental conditions in the workplace, chronic beryllium disease (CBD)3 is now the major illness caused by beryllium exposure. CBD is a systemic granulomatous disorder that predominantly affects the lungs (2). Although the incidence of CBD in exposed workers averages from 1 to 6% (3), attack rates as high as 16% have been reported for some occupations (4). Effective disease prevention strategies in the workplace have been difficult to implement for two main reasons. First, the latency between the first beryllium exposure and the first symptom of CBD ranges from a few months to up to 30 yr (5). Hence, individuals can have subclinical disease while they unknowingly continue to work and be exposed. Second, the incidence rate of CBD correlates inconsistently with exposure levels (6), in that cases of CBD occur among some workers having minimal beryllium exposure (7). These results suggest that an individual susceptibility factor may predispose certain individuals to development of CBD even at low levels of beryllium exposure.

CBD is characterized by hypersensitivity to beryllium measured by the lymphocyte proliferation test (LPT) (8, 9). Lung T cells from CBD patients proliferate in vitro specifically in response to beryllium, with an increase in total T cells, in CD4+ T cells, and in CD4/CD8 ratio (10). Because CD4+ T cells, i.e., helper cells, predominate in the response to beryllium, it was hypothesized that the TCRs, with the help of CD4 molecules, recognize hapten-tagged peptides (peptides with a chemical or metal group attached to amino acid residues) presented by the MHC HLA class II. This interaction on the surface of APCs triggers T cell activation and proliferation. This hypothesis was supported by the finding that the proliferation of beryllium T cell lines was blocked only by the anti-HLA class II Ab, but not by anti-HLA class I Ab (10) and that a much higher rate of T cell subset expansions occurred in the bronchoalveolar lavage from CBD patients compared with the control group (11). Thus, not surprisingly, an association was found between CBD cases and an HLA class II gene, specifically in the HLA-DPB1 locus (12, 13). Although over 95% of the CBD patients were found to have a glutamic acid residue encoded at the 69th position of the mature HLA-DPB1 protein, 30–45% of the control group (exposed but not affected) also expressed the same amino acid. With the low prevalence of CBD, the majority of beryllium-exposed Glu69-carrying individuals do not develop the disease, which suggests that the presence or absence of Glu69 is not the sole susceptibility factor for disease development. In this study, we sought to extend these findings to widen the difference between CBD individuals from controls and find a marker or markers that may eventually be useful for determining individual susceptibility to development of CBD. We report here that other positions on the HLA-DPB1 gene, in addition to the Glu69 marker, also appear to play an important role in the disease. In addition, the HLA-DPA1 gene, which encodes for the α-chain of the HLA-DP heterodimer complex, appears also to be involved.

Blood samples were collected from CBD patients (n = 20, predominantly U.S. Caucasian) and from a beryllium-exposed, unaffected control group (n = 75, predominantly U.S. Caucasian) with their informed consents. CBD diagnosis was based on 1) history of beryllium exposure, 2) noncaseating granulomas on lung biopsy, and 3) abnormal bronchoalveolar lavage beryllium LPT. The individuals in the control group (beryllium-exposed but nondiseased) were randomly picked among healthy beryllium workers. None of the control individuals showed symptoms of CBD or other respiratory diseases. All but two of the control individuals were negative in the blood LPT. The two individuals with positive LPT results were determined by clinical evaluation not to have CBD and so were kept in the control group. Genomic DNAs were isolated using QIAamp Blood Kits (Qiagen, Chatsworth, CA). The exon 2 of HLA-DPB1 was amplified from the genomic DNA on a GeneAmp PCR System 2400 (Perkin-Elmer, Norwalk, CT) using generic primers UG19 (GCTGCAGGAGAGTGGCGCCTCCGCTCAT) and UG21 (CGGATCCGGCCCAAAGCCCTCACTC) (14) for all DPB1 alleles. The purified double-stranded PCR products were directly subjected to an automatic DNA sequencing on an ABI PRISM 310 Genetic Analyzer using dRhodamine Terminator Cycle Sequencing Kit (PE Applied Biosystems, Foster City, CA).

The DNA samples carrying Glu69 (either heterozygous or homozygous) in the DPB1 gene were further analyzed for precise assignment of both alleles from each individual to reveal the possible differences in the allelic distribution between the disease group and the control group. The designation of alleles of each individual was obtained by a comparison of all polymorphic positions in the determined sequence with all known allele sequences retained in the existing HLA-DP database along with their heterozygous combinations (15). Ambiguous DPB1 allele combinations were further resolved by a second round of direct sequencing of selectively amplified alleles using group-specific primers. To specifically amplify one allele of the individuals who have an ambiguous allele combination, either primers complementary to position 69 or to positions 84–87 were used in combination with one generic 5′ primer (UG19). The choice of using group-specific primers complementary to which position (69 or 84–87) depended on the information obtained from the first-round DNA sequencing in which generic intron primers (UG19 and UG21) were used at both 5′ and 3′ ends. Three position 84–87-specific primers were designed: Gly84R (GCTGCAGGGTCATGGGCCCGC), Asp84R (GCTGCAGGGTCACGGCCTCGT), and Val84R (GCTGCAGGGTCATGGGCCCGA). Two of these three primers were chosen for each DNA sample according to the information from the first-round DNA sequencing. Three position 69-specific primers were designed: Glu69R (CTGTCCGGCACTGCCCGCTC), Lys69R (CTGTCCGGCACTGCCCGCTT), and Arg69R (CTGTCCGGCACTGCCCGCC). Two of these three primers were chosen for each DNA sample according to the information from the first-round DNA sequencing. Because exon 2 of the HLA-DPA1 gene has less variation and fewer allelic types than exon 2 of the DPB1 gene, one round of sequencing of PCR products of the DPA1 exon 2 amplified by two generic primers (DPA1Fil GCTTTGACCACTTGCATATTCAAACTGA and DPA1Ri2 CCTTCCAGTTGGGCTACAGA) for all alleles was sufficient for precise allele assignment of every DNA sample.

Table I shows the frequencies of individuals carrying at least one copy of GAG (Glu) at position 69 of their DPB1 genes from the disease group (95%) and the control group (45%). These data are consistent with published data on the frequency of the Glu69 marker in the CBD and unaffected individuals (12).

Table I.

Frequencies of variations in the 69th position of the HLA-DPB1 chain alleles in the CBD group and in the beryllium-exposed, unaffected control groupa

Be-Exposed, Unaffected (n = 75)CBD (n = 20)
SequenceRatio%SequenceRatio%
AAG /AAG 39 /75 52 AAG /AAG 1 /20 
GAG /GAG 1 /75 1.3 GAG /GAG 6 /20 30 
GAG /AAG 33 /75 44 GAG /AAG 12 /20 60 
AAG /AGG 2 /75 2.7 GAG /AGG 1 /20 
Total GAG 34 /75 45 Total GAG 19 /20 95 
Be-Exposed, Unaffected (n = 75)CBD (n = 20)
SequenceRatio%SequenceRatio%
AAG /AAG 39 /75 52 AAG /AAG 1 /20 
GAG /GAG 1 /75 1.3 GAG /GAG 6 /20 30 
GAG /AAG 33 /75 44 GAG /AAG 12 /20 60 
AAG /AGG 2 /75 2.7 GAG /AGG 1 /20 
Total GAG 34 /75 45 Total GAG 19 /20 95 
a

The p value was corrected for the total number of six different highly variable regions compared in the DPB1 gene, indicated in the parentheses.

χ2 = 15.791. Corrected (6) p < 0.0006. Odds ratio = 22.9 (confidence interval = 2.9–180.1).

After first-round automated DNA sequencing of exon 2 of the HLA-DPB1 gene using the generic intron primers UG19 and UG21, the non-Glu69-carrying individuals were removed from subsequent analysis. The DNA samples that carried the Glu69 marker were further analyzed for precise allele assignment of Glu69-carrying alleles as described in Materials and Methods. Fig. 1, A and B, is an example of how the ambiguous DPB1 allele combinations were resolved by a second round of DNA sequencing in which the exon 2 of HLA-DPB1 genes was selectively amplified by group-specific primers. Different alleles from three DNA samples (two homozygous samples, sample A and B, and one heterozygous sample, sample C) were specifically PCR amplified only by the group-specific 3′ primers (Gly84R or Asp84R) that were perfectly complementary to the corresponding alleles shown in Fig. 1,A, and therefore a complete separation of all the polymorphic positions of a heterozygous DNA sample (sample C shown in Fig. 1,A) was obtained during the second-round automated DNA sequencing shown in Fig. 1 B.

FIGURE 1.

A, Group-specific amplification of exon 2 of the HLA-DPB1 gene from three samples carrying different DNA sequences at the aa positions 84–87. Sample A (lanes 1 and 2) carries GGC GGG CCC ATG at positions 84–87. Sample B (lanes 3 and 4) carries GAC GAG GCC GTG. Sample C (lanes 5 and 6) carries both. A generic 5′ primer, UG19, was used for all lanes. Two group-specific 3′ primer (13 ) Gly84R (complementary to sample A) and Asp84R (complementary to sample B) were used for odd lanes and even lanes, respectively. Lane 7 is 1-kb ladder DNA size marker (Promega, Madison, WI). B, Automated DNA sequencing of PCR products amplified separately with three different 3′ primers from a DNA sample (sample C shown in A) carrying heterozygous DNA sequences at positions 84–87. A generic 5′ primer, UG19, was used for PCR in combination with a generic 3′ primer UG21 (A), a group-specific 3′ primer Asp84R (B), and another group-specific 3′ primer Gly84R (C). All sequencing reactions were performed in the presence of the generic 5′ primer UG19. All the polymorphic sites in A, which is a sequencing result of PCR products amplified by two generic primers at both 5′ (UG19) and 3′ (UG21) ends, were completely separated by group-specific PCR amplification shown in B and C.

FIGURE 1.

A, Group-specific amplification of exon 2 of the HLA-DPB1 gene from three samples carrying different DNA sequences at the aa positions 84–87. Sample A (lanes 1 and 2) carries GGC GGG CCC ATG at positions 84–87. Sample B (lanes 3 and 4) carries GAC GAG GCC GTG. Sample C (lanes 5 and 6) carries both. A generic 5′ primer, UG19, was used for all lanes. Two group-specific 3′ primer (13 ) Gly84R (complementary to sample A) and Asp84R (complementary to sample B) were used for odd lanes and even lanes, respectively. Lane 7 is 1-kb ladder DNA size marker (Promega, Madison, WI). B, Automated DNA sequencing of PCR products amplified separately with three different 3′ primers from a DNA sample (sample C shown in A) carrying heterozygous DNA sequences at positions 84–87. A generic 5′ primer, UG19, was used for PCR in combination with a generic 3′ primer UG21 (A), a group-specific 3′ primer Asp84R (B), and another group-specific 3′ primer Gly84R (C). All sequencing reactions were performed in the presence of the generic 5′ primer UG19. All the polymorphic sites in A, which is a sequencing result of PCR products amplified by two generic primers at both 5′ (UG19) and 3′ (UG21) ends, were completely separated by group-specific PCR amplification shown in B and C.

Close modal

After allele assignment for every individual who had at least one DPB1 allele carrying the Glu69 marker, the non-Glu69-carrying alleles were removed from subsequent analysis. All other five variable sites (including aa positions 8–11, 35–36, 55–57, 76, and 84–87) in Glu69-containing alleles were carefully analyzed to determine whether other polymorphic positions were also related to the disease. Among these five loci, two regions, position 84–87 and position 8–11, showed a close association with the disease. As shown in Table II, most people in the control group have “type A” amino acid and nucleic acid sequences (Gly Gly Pro Met; GGC GGG CCC ATG) at the position 84–87 of their Glu69-containing alleles (23/34 if both heterozygous and homozygous Glu69-carrying individuals are included or 22/33 if only heterozygous Glu69-carriers are included). However, most people in the CBD group have a different sequence denoted “type B” sequences (Asp Glu Ala Val; GAC GAG GCC GTG, 16/19 both heterozygous and homozygous Glu69 individuals or 11/14 heterozygous Glu69 individuals only). Furthermore, as shown in Table III, most individuals in the control group have type A sequences (Leu Phe Gly; CTT TTC GGA) at positions 8, 9, and 11 of their Glu69-containing alleles (24/34 both homozygous and heterozygous Glu69 carriers or 23/33 heterozygous Glu69 carries only), while most people in the CBD group have type B sequences (Val His Leu, GTG CAC TTA; or Val Tyr Leu, GTG TAC TTA; 15/19 both homozygous and heterozygous Glu69 carriers or 10/14 heterozygous Glu69 carries only). All of the individuals from both groups who have type A sequences at the position 84–87 also have type A sequences at positions 8, 9, and 11. Representatives of such alleles are almost exclusively DPB1 allele *0201 in these two groups with only one exception, which is allele *0202, a sister allele with a very similar protein structure to *0201 (aa 35 Phe in *0201 was substituted by Leu in *0202, and aa 55 Asp in *0201 was substituted by Glu in *0202). Most people (10/12 in the control group and 15/16 in the CBD group) with type B amino acid sequences at the position 84–87 also have type B sequences at positions 8, 9, and 11. Representatives of such alleles are DPB1 allele *1701, *1001, *0901, *1301, and *0601. Table IV is the DPB1 allele distribution of Glu69-carrying individuals from the disease and the control groups. Because the non-*0201 Glu69-carrying alleles are relatively rare alleles in the whole population, and, because of the small sample number of these alleles in this study, we did not compare any individual rare alleles between the disease the control groups. Instead, the rare Glu69-carrying alleles were grouped into one group, called the “non-*0201 Glu69-carrying alleles” group. This group was compared with a group of the most common Glu69-carrying allele, *0201, in phenotypic frequency between the disease and the control groups as shown in Table IV. This comparison was done to be more meaningful statistically. It is also more reasonable biologically since the current hypothesis about the susceptibility to CBD is based on the HLA-DP structural genes and these rare Glu69-carrying DPB1 alleles have common protein/DNA sequences at the aa positions 8–11 and 84–87, which are different from those of allele *0201. As shown in Table IV, the phenotypic frequency of non-*0201 Glu69-carrying alleles in Glu69 carriers showed a dramatic increase in the disease group (84.21%) compared with the control group (35.29%), while the phenotypic frequency of *0201 alleles in Glu69 carriers showed a substantial decrease in the CBD group (42.11%) compared with the control group (67.65%). The major differences between these two types of alleles (*0201, commonly seen in the control group, and non-*0201, commonly seen in the CBD group) involve a change of 7 amino acids (positions 8, 9, 11, and 84–87). Three of these 7 amino acid changes (positions 9, 84, and 85) are changes between different amino acid groups, i.e., from hydrophobic to hydrophilic amino acids. In addition, a change from Ala to Pro at position 86 may also be important. These differences could alter the folding property of the β-chain of the DPB1 gene and also affect the binding property to the α-chain coded by the DPA1 gene. As for allele *1901, which has the same sequence at positions 84–87 with non-*0201 Glu69-carrying alleles but the same sequence at position 8–11 with allele *0201, we grouped it into the more susceptible category, non-*0201, which was more conservative for the risk estimation. In this study, it will not affect the statistics significantly to group allele *1901 into either group (*0201 or non-*0201) because allele *1901 has a nearly identical phenotypic frequency in the disease (5.26%) and the control (5.88%) groups shown in Table IV.

Table II.

Protein and nucleic acid sequence distributions of the DPB1 gene at the amino acid positions 84–87 on the Glu69-carrying alleles in the disease and control groupsa

Be-Exposed, Unaffected (n = 34)CBD (n = 19)
A typeB typeA + BA typeB typeA + B
GGC GGG CCC ATG GAC GAG GCC GTG  GGC GGG CCC ATG GAC GAG GCC GTG  
Gly Gly Pro Met Asp Glu Ala Val  Gly Gly Pro Met Asp Glu Ala Val  
22 11 11 
64.7% 32.4% 2.9% 15.8% 57.9% 26.3% 
Total B type 12/34 (35.3%)   Total B type 16/19 (84.2%)   
Be-Exposed, Unaffected (n = 34)CBD (n = 19)
A typeB typeA + BA typeB typeA + B
GGC GGG CCC ATG GAC GAG GCC GTG  GGC GGG CCC ATG GAC GAG GCC GTG  
Gly Gly Pro Met Asp Glu Ala Val  Gly Gly Pro Met Asp Glu Ala Val  
22 11 11 
64.7% 32.4% 2.9% 15.8% 57.9% 26.3% 
Total B type 12/34 (35.3%)   Total B type 16/19 (84.2%)   
a

Individuals carrying type A only (predominantly heterozygous Glu69), type B only (heterozygous Glu69), or both type A and B (homozygous Glu69) sequences are marked with A type, B type, or A + B in the table, respectively.

χ2 = 11.704. Corrected (6) p = 0.0036. Odds ratio = 9.8 (confidence interval = 2.3–40.1).

Table III.

Protein and nucleic acid sequence distribution of the DPB1 gene at the amino acid positions 8, 9, and 11 on the Glu69-carrying alleles in the disease and control groupsa

Be-Exposed, Unaffected (n = 34)CBD (n = 19)
A typeB typeA + BA typeB typeA + B
CTT TTC GGA GTG CAC TTA  CTT TTC GGA GTG CAC TTA  
 (TAC)   (TAC)  
Leu Phe Gly Val His Leu  Leu Phe Gly Val His Leu  
 (Tyr)   (Tyr)  
24 10 
70.6% 26.5% 2.9% 21.1% 52.6% 26.3% 
Total B type 10/34 (29.4%)   Total B type 15/19 (78.9%)   
Be-Exposed, Unaffected (n = 34)CBD (n = 19)
A typeB typeA + BA typeB typeA + B
CTT TTC GGA GTG CAC TTA  CTT TTC GGA GTG CAC TTA  
 (TAC)   (TAC)  
Leu Phe Gly Val His Leu  Leu Phe Gly Val His Leu  
 (Tyr)   (Tyr)  
24 10 
70.6% 26.5% 2.9% 21.1% 52.6% 26.3% 
Total B type 10/34 (29.4%)   Total B type 15/19 (78.9%)   
a

Individuals carrying type A only (predominantly heterozygous Glu69), type B only (heterozygous Glu69) or both type A and type B (homozygous Glu69) sequences are marked with A type, B type or A + B in the table, respectively.

χ2 = 12.002. Corrected (6) p = 0.0030. Odds ratio = 9.0 (confidence interval = 2.4–34).

Table IV.

HLA-DPB1 allele distributions of the Glu69 carriers in the disease and the control groupsa

DPB1 AlleleBe-Exposed, Unaffected (n = 34)CBD (n = 19)
Total no. of alleles (n = 68)Allele frequency (%)Phenotypic Frequency (%)Total no. of alleles (n = 38)Allele frequency (%)Phenotypic frequency (%)
Non-Glu69 alleles 33 48.53 97.06 13 34.21 68.42 
0201 22 32.35 64.71 23.68 42.11 
0202 1.47 2.94 
0601 1.47 2.94 5.26 10.53 
0901 2.94 5.88 5.26 10.53 
1001 2.94 5.88 5.26 10.53 
1301 5.88 11.76 7.89 15.79 
1701 1.47 2.94 15.79 31.58 
1901 2.94 5.88 2.63 5.26 
       
Total 0201 (0201 and 0202) 23 33.82 67.65 23.68 42.11 
Total non-0201 Glu69 (0601, 0901, 1001, 1301, 1701, 1901) 12 17.65 35.29 16 42.11 84.21 
DPB1 AlleleBe-Exposed, Unaffected (n = 34)CBD (n = 19)
Total no. of alleles (n = 68)Allele frequency (%)Phenotypic Frequency (%)Total no. of alleles (n = 38)Allele frequency (%)Phenotypic frequency (%)
Non-Glu69 alleles 33 48.53 97.06 13 34.21 68.42 
0201 22 32.35 64.71 23.68 42.11 
0202 1.47 2.94 
0601 1.47 2.94 5.26 10.53 
0901 2.94 5.88 5.26 10.53 
1001 2.94 5.88 5.26 10.53 
1301 5.88 11.76 7.89 15.79 
1701 1.47 2.94 15.79 31.58 
1901 2.94 5.88 2.63 5.26 
       
Total 0201 (0201 and 0202) 23 33.82 67.65 23.68 42.11 
Total non-0201 Glu69 (0601, 0901, 1001, 1301, 1701, 1901) 12 17.65 35.29 16 42.11 84.21 
a

The non-Glu69 alleles are predominantly *0401, followed by *0402 and *0301 regardless of which group, the CBD or control, the *0201 or non-*0201.

Among non-*0201 alleles (type B Glu69-carrying alleles), alleles *1701, *0901, and *1001 appear to be more likely to confer greater susceptibility to CBD than the other non-*0201 Glu69 alleles, *1301, *0601, and *1901. Ten of 20 CBD patients have allele *1701, *0901, or *1001 while only 5 of 75 beryllium-exposed nondiseased individuals have the same alleles (χ2 = 22.3, corrected (6) p < 0.0006; odds ratio = 14; confidence interval, 4.0–49.4). The p value was corrected for the total number of six different non-*0201 Glu69 DPB1 alleles compared, indicated in the parentheses. This extremely high frequency (50% of alleles *1701, *0901, or *1001 in CBD patients) is well above any recorded previously (16, 17, 18, 19, 20, 21, 22). There are no significant differences in the allele distribution of DPB1 genes among our control individuals compared with the existing data bases with relevant ethnic groups (16, 17, 18, 19, 20, 21, 22). In both our control group and in the published literature, *0401 is the most common DPB1 allele in the total population, and *0201 is the most common Glu69-carrying DPB1 allele. Alleles *1701, *0901, and *1001 differ from other non-*0201 Glu69 alleles (*1301 and *0601) at the aa position 9, or both positions 9 and 55–56, suggesting that these positions specifically are related to the process of disease development among these non-*0201 Glu69 carriers.

The β-chain of the HLA-DPB1 gene must form a heterodimer complex with the α-chain of the DPA1 gene to have Ag-binding and -presenting properties. Because these two genes are physically separated by only 1–2 kb (23, 24) and show strong linkage disequilibrium (16), we also sequenced the DPA1 genes from all of the DNA samples that contained Glu69 alleles in their DPB1 loci. It was found that almost all of the DPB1 allele *0201 carriers (22/22 in the control group and 7/8 in the CBD group) have at least one allele *0103 of their DPA1 gene. By comparison, most non-*0201 Glu69 allele carriers (12/13 in the control group and 14/16 in the CBD group) have at least one DPA1 *02 allele (predominantly *02011). The DPB1 *0201 allele was reported to be exclusively associated with the DPA1 allele *01 (16), but among the 30 *0201 Glu69-carriers from our control and CBD groups, there was one exception observed, which does not have an *01 DPA1 allele but has a homozygous *02011 DPA1 allele. This very rare exception was a CBD case individual, which further suggests that the DPA1 *02 allele might facilitate disease development. The main differences between DPA1 *0103 and *02011 alleles involves three amino acids at positions 31, 50, and 83. All of these three amino acid changes are changes between different amino acid groups (from hydrophobic to hydrophilic or from uncharged to charged amino acids). These differences might affect the folding property of the α-chain of the HLA-DP genes and the dimer formation between the α-chain and the β-chain of the HLA-DP genes. We speculate that the dimers formed by both structurally different α-chain and β-chain may completely alter the properties of the binding pocket for the Ag, thereby causing different susceptibilities to CBD.

Another observation from our experiments is that homozygous Glu69 carriers were found almost exclusively in the CBD group (Table I). Six Glu69/Glu69 individuals out of 20 individuals were found in the CBD group, compared with only 1 of 75 in the control group. On the contrary, the heterozygous individuals, who have one Glu69-containing allele and one non-Glu69-containing allele, were found only slightly more frequently in the CBD group (13 of 20) than in the control group (33 of 74). This finding suggests to us that Glu69 alone is not sufficient for the disease susceptibility and that some other factors must also be involved in the disease development. Interestingly, among these six Glu69/Glu69 from the CBD group, only one of them was homozygous for *0201. The other five have one *0201 allele and one non-*0201 Glu69-carrying allele. If we remove the Glu69/Glu69 individuals from both groups, the frequency of heterozygous *0201 individuals in the CBD group (2 of 14) does not show an increase compared with the control group (22 of 74) but rather shows a substantial decrease. By comparison, the heterozygous non-*0201 Glu69 carriers do show a substantial increase in the CBD group (11 of 14) over the control group (11 out 74). This suggests that the DPB1 allele *0201 is not a major disease allele but the rare non-*0201 Glu69-carrying alleles are. The very high susceptibility of Glu69/Glu69 individuals to CBD is therefore mainly due to the presence of non-*0201 Glu69 alleles. It may also indicate the absence of a protective effect conferred by non-Glu69 alleles, predominantly DPB1 allele *0401, an allele that was reported to have some protective function against allergic asthma (25).

By using the techniques of 1) two rounds of direct automated DNA sequencing to obtain precise allele assignment and 2) accurate haplotype determination of all Glu69 carriers to examine the HLA-DPB1 region, we have found that the specific Glu69-carrying HLA-DPB1 alleles and their copy number (heterozygous or homozygous) appear to be the primary factors underlying development of CBD. Moreover, the most common Glu69-carrying allele, *0201, was not highly associated with CBD, but the relatively rare non-*0201 Glu69-carrying alleles were. In addition, we found that specific HLA-DPA1 alleles in the HLA-DPB1 Glu69 carriers also appeared to be associated with the disease development

In the previous landmark study (12), the Glu69 marker was shown to be highly associated with CBD (97%), but also showed a high association with the control group. The specific major disease allele was claimed to be the most common Glu69-carrying HLA-DPB1 allele, *0201. There are several possible reasons for the different conclusions reached here in this study and in the Richeldi et al. study. (12). First, a less sensitive technique was used in the original studies by Richeldi et al. (partial regional group-specific hybridization vs two rounds whole-exon two DNA sequencing). Second, a limited set of probes was used and a limited number of regions were determined in the Richeldi et al. study. To correctly assign the HLA-DP alleles by hybridization with group-specific oligonucleotides, more probes and more variable regions in addition to regions C (coding for aa 55–57) and D (coding for aa 69), and especially regions A (coding for aa 8–11) and F (coding for aa 84–87), should have been used, as reported in the literature before 1993 by other researchers (14, 26, 27). Third, the unnecessary single nucleotide insertion compared with the previous literature (14, 26, 27) in three (DB14, DB18 and DB19) of four group-specific oligonucleotides used by Richeldi et al. probably further reduced the accuracy of the allele assignment.

Fig. 2 is a schematic illustration of several typical haplotypes commonly seen in CBD patients (Glu69 and non-Glu69 individuals) and the control group (Glu69 and non-Glu69 individuals). The rarest allele combination seen in the control group or total population, the Glu69/Glu69 individual, has a very high frequency in CBD and is followed by a relatively rare allele combination, heterozygous non-*0201 Glu69 alleles, which also has a higher frequency in the CBD sample. The sum of these two rare groups (Glu69/Glu69 and non-*0201 Glu69 heterozygotes) accounts for 85% of the CBD sample, but only for 16% of the control sample (χ2 = 35.446, corrected (4) p < 0.0004; odds ratio = 29.8; confidence interval = 7.6–117.5). On the contrary, the most common allele combination (non-Glu69 individuals, predominantly Lys69/Lys69) seen in the control group (54.7%) or total population is not found frequently in CBD (5%). It is followed by the relatively common allele combination, *0201 Glu69/non-Glu69 individuals, which is found at substantially lower frequencies in CBD (10%) than in controls (29.3%). The dramatic differences in Glu69 allele frequencies between CBD and control groups strongly implicate a genetic risk factor in the development of CBD and further may explain why CBD incidence does not correlate very well with the exposure levels. Based on our results, the disease predictive value of having the Glu69 marker is only 0.36. By comparison, the CBD predictive value for having non-*0201 Glu69 alleles is 0.57, and the predictive value of having two Glu69-containing alleles is 0.85. These results support the hypothesis that it is not the mere presence of Glu69, per se, but the allele types and their copy number (homozygous or heterozygous) that confer greatest susceptibility to development of CBD in beryllium-exposed individuals.

FIGURE 2.

Schematic risk estimation of individuals with typical allele combinations in their HLA-DPB1 and DPA1 genes commonly seen in CBD (left two) and in controls or in the total population (right two). All filled symbols represent HLA-DPA1 genes and their protein products, α-chains, while all open symbols represent HLA-DPB1 genes and their protein products, β-chains. The p values in columns 1, 2, and 4 were corrected for four different HLA-DPB1 haplotype categories.

FIGURE 2.

Schematic risk estimation of individuals with typical allele combinations in their HLA-DPB1 and DPA1 genes commonly seen in CBD (left two) and in controls or in the total population (right two). All filled symbols represent HLA-DPA1 genes and their protein products, α-chains, while all open symbols represent HLA-DPB1 genes and their protein products, β-chains. The p values in columns 1, 2, and 4 were corrected for four different HLA-DPB1 haplotype categories.

Close modal

It is not clear at this stage how different genotypes confer differential susceptibilities to CBD. Although CBD occurs on average in 1–6% of the exposed population, for some occupations, the incidence is as high as 16% (4), which is coincidentally the same percentage of the individuals with the greatest susceptible allele combination (Glu69/Glu69 and non-0201 Glu69/non Glu69) from our control group shown in Fig. 2. It would be of interest retrospectively to determine the allele type combination of those CBD patients as well as their histories of beryllium exposure. The approach of allele designation used here is powerful for addressing future mechanistic and structural studies of beryllium interactions with tissue. Screening of detailed allele designation at HLA-DP loci among beryllium workers would provide a very useful tool for measuring genetic risk. Such a screening tool could be incorporated into existing CBD prevention programs, which now focus mainly on exposure risk and early disease detection with the lymphocyte proliferation test. It may also prove to be useful for the development of diagnostic tools for susceptibility to other DPB1 Glu69-related diseases (28, 29, 30) where a higher predictive value than dichotomous testing for the presence or absence of a single marker would be desirable. By far the most important contribution from our findings is the potential impact that they may have on the use of genetic information to prevent further cases of CBD from occurring. Although legal and ethical issues must be addressed, the CBD-HLA case may, in the future, lead to the first example for use of a genetic screening test in the workplace to virtually eliminate or drastically reduce the occurrence of an occupational disease.

We thank Ms. Yulin Shou, Usha Sadivisan-Nair, Nancy Lehnert, and Elizabeth Barker for their skilled technical assistance, and Dr. Hugh Smith of the Occupational Medicine group for coordination of the samples.

1

This work was supported by the U.S. Department of Energy.

3

Abbreviations used in this paper: CBD, chronic beryllium disease; LPT, lymphocyte proliferation test.

1
Hardy, H. L., J. R. Tabershaw.
1946
. Delayed chemical pneumonitis occurring in workers to beryllium compounds.
J. Ind. Hyg. Toxicol.
28
:
197
2
Newman, L. S., J. Lloyd, E. Daniloff.
1996
. The natural history of beryllium sensitization and chronic beryllium disease.
Environ. Health Perspect.
104
:
937
3
Kreiss, K., F. Miller, L. S. Newman, E. A. Ojo-Amaize, M. D. Rossman, C. Saltini.
1994
. Chronic beryllium disease: from the workplace to cellular immunology, molecular immunogenetics, and back.
Clin. Immun. Immunopath.
71
:
123
4
Kreiss, K., S. Wasserman, M. M. Mroz, L. S. Newman.
1993
. Beryllium disease screening in the ceramics industry: blood lymphocyte test performance and exposure-disease relations..
J Occup. Med.
35
:
267
5
Newman, L. S..
1995
. Beryllium disease and sarcoidosis: clinical and laboratory links.
Sarcoidosis
12
:
7
6
Sterner, J., M. Eisenbud.
1951
. Epidemiology of beryllium intoxication.
Arch. Ind. Hyg. Occup. Med.
4
:
123
7
Eisenbud, M..
1998
. The standard for control of chronic beryllium disease.
Appl. Occup. Environ. Hygiene
13
:
25
8
Rossman, M. D., J. A. Kern, J. A. Elias, M. R. Cullen, P. E. Epstein, O. P. Preuss, T. N. Markham, R. P. Daniele.
1988
. Proliferative response of bronchoalveolar lymphocytes to beryllium: a test for chronic beryllium disease. [Published erratum appears in 1989 Ann. Intern. Med. 110:672.].
Ann. Intern. Med.
108
:
687
9
Kreiss, K., L. S. Newman, M. M. Mroz, P. A. Campbell.
1989
. Screening blood test identifies subclinical beryllium disease.
J. Occup. Med.
31
:
603
10
Saltini, C., K. Winestock, M. Kirby, P. Pinkston, R. G. Crystal.
1989
. Maintenance of alveolitis in patients with chronic beryllium disease by beryllium-specific helper T cells.
N. Engl. J. Med.
320
:
1103
11
Fontenot, A. P., B. L. Kotzin, C. E. Comment, L. S. Newman.
1998
. Expansions of T-cell subsets expressing particular T-cell receptor variable regions in chronic beryllium disease.
Am. J. Respir. Cell Mol. Biol.
18
:
581
12
Richeldi, L., R. Sorrentino, C. Saltini.
1993
. HLA-DPB1 glutamate 69: a genetic marker of beryllium disease.
Science
262
:
242
13
Newman, L. S..
1993
. To Be2+ or not to Be2+: immunogenetics and occupational exposure.
Science
262
:
197
14
Bugawan, T. L., A. B. Begovich, H. A. Erlich.
1990
. Rapid HLA-DPB typing using enzymatically amplified DNA and nonradioactive sequence-specific oligonucleotide probes. [Published erratum appears in 1991Immunogenetics 34:413.].
Immunogenetics
32
:
231
15
Marsh, S. G., J. G. Bodmer.
1995
. HLA class II region nucleotide sequences, 1995.
Tissue Antigens
45
:
258
16
Sage, D. A., P. R. Evans, W. M. Howell.
1994
. HLA DPA1-DPB1 linkage disequilibrium in the British Caucasoid population.
Tissue Antigens
44
:
335
17
al Daccak, R., F. Q. Wang, D. Theophille, P. Lethielleux, J. Colombani, P. Loiseau.
1991
. Gene polymorphism of HLA-DPB1 and DPA1 loci in caucasoid population: frequencies and DPB1-DPA1 associations.
Hum. Immunol.
31
:
277
18
Grubic, Z., R. Zunec, A. Naipal, A. Kastelan, M. J. Giphart.
1995
. Molecular analysis of HLA class II polymorphism in Croatians.
Tissue Antigens
46
:
293
19
Yelamos, J., J. R. Garcia-Lozano, I. Moreno, M. Romero, A. Garcia, B. Sanchez.
1994
. Frequency of HLA-DPB1 alleles in a Spanish population: their contribution to rheumatoid arthritis susceptibility.
Eur. J. Immunogenet.
21
:
91
20
Magzoub, M. M., H. A. Stephens, J. A. Sachs, P. A. Biro, S. Cutbush, Z. Wu, G. F. Bottazzo.
1992
. HLA-DP polymorphism in Sudanese controls and patients with insulin-dependent diabetes mellitus.
Tissue Antigens
40
:
64
21
Savage, D. A., D. Middleton, F. Trainor, A. Taylor, P. G. McKenna, C. Darke.
1992
. Frequency of HLA-DPB1 alleles, including a novel DPB1 sequence, in the Northern Ireland population.
Hum. Immunol.
33
:
235
22
Grundschober, C., A. Mazas, A. Sanchez, L. Excoffier, A. Langaney, M. Jeannet, J. M. Tiercy.
1994
. HLA-DPB1 DNA polymorphism in the Swiss population: linkage disequilibrium with other HLA loci and population genetic affinities.
Eur. J. Immunogenet.
21
:
143
23
Lawrence, S. K., H. K. Das, J. Pan, S. M. Weissman.
1985
. The genomic organisation and nucleotide sequence of the HLA-SB(DP) α gene.
Nucleic Acids Res.
13
:
7515
24
Kelly, A., J. Trowsdale.
1985
. Complete nucleotide sequence of a functional HLA-DP β gene and the region between the DP β 1 and DP α 1 genes: comparison of the 5′ ends of HLA class II genes.
Nucleic Acids Res.
13
:
1607
25
Caraballo, L., J. Marrugo, S. Jimenez, G. Angelini, G. B. Ferrara.
1991
. Frequency of DPB1*0401 is significantly decreased in patients with allergic asthma in a mulatto population.
Hum. Immunol.
32
:
157
26
Bugawan, T. L., G. T. Horn, C. M. Long, E. Mickelson, J. A. Hansen, G. B. Ferrara, G. Angelini, H. A. Erlich.
1988
. Analysis of HLA-DP allelic sequence polymorphism using the in vitro enzymatic DNA amplification of DP-α and DP-β loci.
J. Immunol.
141
:
4024
27
Angelini, G., T. L. Bugawan, L. Delfino, H. A. Erlich, G. B. Ferrara.
1989
. HLA-DP typing by DNA amplification and hybridization with specific oligonucleotides.
Hum. Immunol.
26
:
169
28
Potolicchio, I., P. A. Brookes, A. Madrigal, R. I. Lechler, R. Sorrentino.
1996
. HLA-DPB1 mismatch at position 69 is associated with high helper T lymphocyte precursor frequencies in unrelated bone marrow transplant pairs.
Transplantation
62
:
1347
29
Lympany, P. A., M. Petrek, A. M. Southcott, A. J. Newman Taylor, K. I. Welsh, R. M. du Bois.
1996
. HLA-DPB polymorphisms: Glu 69 association with sarcoidosis.
Eur. J. Immunogenet.
23
:
353
30
Potolicchio, I., G. Mosconi, A. Forni, B. Nemery, P. Seghizzi, R. Sorrentino.
1997
. Susceptibility to hard metal lung disease is strongly associated with the presence of glutamate 69 in HLA-DP β chain.
Eur. J. Immunol.
27
:
2741