Beryllium-Induced Hypersensitivity: Genetic Susceptibility and Neoantigen Generation

Beryllium (Be) is a rare alkaline earth metal that is used in a variety of high-technology industries, including aerospace, ceramics, electronics, and nuclear defense (1). Be exposure primarily occurs through inhalation of particulates by workers involved in the machining of Be-containing products. More than 1 million individuals have been exposed to Be in the workplace; in 2004, it was estimated that ∼140,000 United States workers were exposed to Be (2). The United States is the leading producer and consumer of Be products, using 250 tons in 2013 (3). The adverse health effects of Be exposure became apparent in the 1930s (4–7). With the introduction of Be-exposure standards in 1949, the occurrence of acute berylliosis was virtually eliminated, but cases of chronic Be disease (CBD) continue to occur. Depending on the nature of the exposure and the genetic susceptibility of the individual, CBD will develop in 1–16% of exposed subjects (reviewed in Ref. 1). Thus, CBD remains an important public health concern.

Workplace screening of Be-exposed workers identified individuals sensitized to Be but having no evidence of lung disease. These Be-sensitized (BeS) subjects have a Be-specific immune response in peripheral blood but no clinical or pathologic features of CBD (8). The rate of progression from Be sensitization to disease is difficult to assess; a subset of BeS subjects progressed to CBD at a rate of 6–8% per year (8, 9). Although Be sensitization is required for the development of CBD, not all BeS subjects progress to CBD, suggesting that differences in exposure and/or genetic factors may contribute to disease progression. CBD is characterized by noncaseating granulomatous inflammation and alveolitis composed of Be-specific CD4⁺ T cells. Granulomas primarily occur in the lung, although other organ systems may be involved (10). Diagnosis of CBD requires the detection of a Be-specific immune response in blood and/or lung (11) and the presence of noncaseating granulomatous inflammation on a biopsy specimen (12). The pathology of CBD is identical to that seen in sarcoidosis, a more common granulomatous lung disease of unknown etiology (13). Because of the persistence of Be in the lung years after exposure cessation (14), the natural history of disease is characterized by a gradual decline in lung function, with one third of untreated patients historically progressing to end-stage respiratory insufficiency (15).

Over the past decade, major advances in our understanding of the pathogenesis of Be-induced disease have occurred. This review focuses on recent advances in our understanding of T cell recognition of Be and the interaction between environmental exposure and genetic susceptibility in the genesis of granulomatous inflammation.

In addition to its ability to serve as an antigenic stimulus, Be functions as an adjuvant in immune responses (Fig. 1A). For example, rabbits vaccinated with trichostrongylus extracts combined with Be demonstrated increased protection against parasitic challenge compared with mice vaccinated with extract alone (16). Lee et al. (17) showed that the adjuvant properties of Be were due to its ability to increase IFN-γ secretion. Adjuvants typically operate via engagement of pattern recognition receptors that drive activation and maturation of APCs. Exposure of macrophages and dendritic cells (DCs) to Be induced the release of inflammatory chemokines, cytokines, and reactive oxidative species (18–22). Li et al. (23) showed that exposure of monocyte-derived DCs induced phosphorylation of MAPK p38, resulting in NF-κB activation and enhanced production of IFN-γ and TNF-α by Be-specific CD4⁺ T cells. In mice, pulmonary Be exposure rapidly induced cellular death and release of the alarmins DNA and IL-1α into the lung, followed by IL-1R–dependent expression of KC and neutrophil infiltration (24).

These innate pathways may drive acute berylliosis and contribute to Be sensitization in response to low-dose exposures in genetically susceptible individuals. Under steady-state conditions, DCs maintain tolerance to innocuous substances (25). In contrast, Be exposure enhanced migration of DCs to draining lymph nodes, upregulated the costimulatory molecules CD80 and CD86 on migratory DCs, and enhanced primary and memory CD4⁺ T cell responses to a model Ag (24). These adjuvant effects of Be required MyD88-dependent signaling pathways, unlike the vaccine adjuvant aluminum hydroxide, which enhances Ab production to bystander Ag via MyD88-independent pathways (24, 26). Thus, Be has a critical impact on DC function through innate receptor pathways that may have important contributions to disease pathogenesis.

The lungs of CBD patients are characterized by an influx of activated CD4⁺ T cells that play a critical role in the pathogenesis of CBD (11, 27–30). After T cell recognition of Be, Ag-specific CD4⁺ T cells undergo clonal proliferation, differentiate into memory T cell subsets, and home to the lung (Fig. 1B, 1C). Be-specific CD4⁺ T cells in bronchoalveolar lavage fluid (BALF) express markers of previous activation (28, 31), recognize Be in a CD28 costimulation–independent manner (32), and exhibit an effector memory T cell phenotype (31, 33). In addition, Be-specific CD4⁺ T cells in BALF upregulate PD-1 (34) and CTLA-4 (35), coinhibitory receptors that negatively regulate T cell function (36). Upregulation of PD-1 on Be-specific CD4⁺ T cells dampens proliferation of these cells, thus playing a key role in preserving lung function in the presence of persistent Ag exposure.

Striking numbers of Be-responsive, Th1-polarized CD4⁺ T cells are present in the BALF of CBD patients (31, 37), whereas Th2- or Th17-polarized T cells have not been detected (31, 37). The frequency of IFN-γ–producing CD4⁺ T cells in the BALF of CBD patients after BeSO₄ stimulation ranged from 1.7 to 29% (31), whereas the frequency in blood ranged from undetectable to ∼1.0% (38, 39). Conversely, Be-responsive CD8⁺ T cells have not been detected in blood or BALF, suggesting that this T cell lineage plays a minor role, if any, in CBD. Although the vast majority of Be-specific cells are compartmentalized to the lung, greater numbers of circulating Be-specific cells strongly correlated with alveolar inflammation, as measured by BALF WBCs and lymphocyte counts (38, 39). Thus, the number of circulating Be-specific CD4⁺ T cells may provide a glimpse into the lung without the need for invasive procedures.

CD4⁺ T cells recognize Ag presented by MHC class II (MHCII) molecules via a surface receptor composed of α- and β-chains (40). TCR α-chain (TCRA) and TCR β-chain (TCRB) genes are formed through somatic rearrangement of germline gene segments, and the expressed TCRB genes are generated from rearrangement of variable (V) to diversity (D) to junctional (J) region gene segments. The highly variable junctional region forms the CDR3, which is critically involved in the TCR’s interaction with the MHC-peptide complex. With the αβTCR repertoire being estimated at >10⁷ possibilities (41), there is little chance that any two expanded T cell clones will express nearly identical TCRs unless selected by the same MHC-peptide complex.

Studies of TCR expression on CD4⁺ T cells from ex vivo BALF of CBD patients demonstrated the presence of oligoclonal T cell populations that were specific for CBD and not seen in other diseases, such as sarcoidosis (29, 30). Certain TCR β-chain variable (Vβ) region motifs were enriched in lung CD4⁺ T cells from CBD patients and persisted at high frequency in subjects with persistent disease (29, 30). In addition, we identified a public Vβ5.1⁺ TCR repertoire in BALF CD4⁺ T cells in all HLA-DP2–expressing CBD patients who were evaluated (42). These Ag-specific public Vβ5.1 chains were paired with different α-chains, and their frequency was inversely correlated with a loss of lung function and exercise capacity, suggesting a pathogenic role for this T cell subset in CBD (42). Public T cells are defined by the expression of identical TCR Vα and/or Vβ genes that are present in the majority of subjects in response to a specific epitope. Despite public repertoires being restricted in nature, they are typically dominant and dictate disease severity (43–47). Most public repertoires were identified in MHC class I–restricted CD8⁺ T cells (48–50). Conversely, public repertoires have rarely been detected in the CD4⁺ T cell subset as a result, in most cases, of unknown stimulatory Ags. For CBD, the use of Be-loaded HLA-DP2–peptide tetramers (described below) facilitated the identification of epitope-specific public CD4⁺ T cells and suggests that public CD4⁺ T cells are more common than previously thought.

In addition to workplace exposure to Be, genetic susceptibility plays an essential role in the development of Be-induced disease. Saltini et al. (27) demonstrated that BALF CD4⁺ T cells from CBD patients recognized Be in an MHCII–restricted manner and subsequently showed that genetic susceptibility was most strongly associated with a particular MHCII molecule, HLA-DP (51). This study demonstrated that HLA-DPB1 alleles with a glutamic acid (E) at position 69 of the β-chain (βGlu69) were strongly linked to disease susceptibility (51), with the most prevalent βGlu69-containing allele being HLA-DPB1*02:01. Since the initial report, multiple studies corroborated these findings, documenting the presence of βGlu69-containing DPB1 alleles in 73–95% of BeS subjects and CBD patients compared with 30–48% of exposed controls (reviewed in Ref. 1). In CBD patients who do not express a βGlu69-containing HLA-DP allele, an increased frequency of HLA-DRB1*13:01 alleles was identified (52, 53). Importantly, these alleles possess an analogous glutamic acid residue at position 71 of the β-chain (βGlu71). In addition, several HLA-DR alleles that share a phenylalanine at position 47 of the β-chain were associated with disease in individuals lacking a βGlu69-containing HLA-DP allele (54). A differential risk for disease development was associated with certain rare βGlu69-containing DPB1 alleles, such as HLA-DPB1*17:01 (52, 55–57). Thus, Be-induced disease is a classic example of a disorder resulting from gene-by-environment interactions, where both components are required for disease development. In this regard, the probability of CBD increases with HLA-DP βGlu69 copy number and increasing workplace exposure to Be (58), suggesting that genetic and exposure factors may have an additive effect on the risk for disease development (59).

Several groups showed that Be presentation occurs primarily through HLA-DP, with HLA-DR playing a minor role, particularly in subjects lacking a βGlu69-containing HLA-DP molecule (54, 60, 61). In individuals expressing HLA-DP βGlu69, Be-specific T cells were restricted only by HLA-DP alleles that contain βGlu69, and amino acid substitution at this position abolished T cell responses. Thus, the molecular mechanism for the genetic association of particular MHCII genes with disease was based on the ability of those proteins to bind and present Ag to pathogenic CD4⁺ T cells (60, 61).

Longstanding questions in CBD and other metal-induced hypersensitivities include the nature of metal interactions with MHCII molecules and the role of peptide in creating a ligand recognized by metal-specific TCRs. Saltini and colleagues (62) proposed that the properties of the p4 pocket of the HLA-DP–peptide–binding region, together with electron-donating amino acids derived from HLA-DP–binding peptides, could coordinate the positively charged Be ion. In addition to Be, specific peptides are required to complete the Be-specific αβTCR ligand (63, 64). However, a set of known HLA-DP2–binding peptides (65) did not induce IL-2 secretion by T cell hybridomas expressing Be-specific TCRs (64). We (64) identified Be-dependent mimotopes that bind to HLA-DP2 and form a complex with Be recognized by pathogenic CD4⁺ T cells in CBD. These Be-dependent mimotopes expressed negatively charged aspartic and glutamic acid residues at p4 (pD4) and p7 (pE7) of the peptide (64), and the location of these amino acids, in addition to βGlu69 contributed by the HLA-DP2 β-chain, suggested their role in Be coordination for T cell recognition.

Using human protein databases to identify endogenously derived peptides with homology to the mimotope sequences, plexin A peptides that bound HLA-DP2/Be and stimulated pathogenic CD4⁺ T cells from CBD patients were identified (64). Plexins are transmembrane proteins encoded by nine genes (PLXNA1–4, B1–3, C1, and D1) that are involved in cell movement and response (66). Only the plexin A family contains the stimulatory epitope that includes acidic amino acids at both the p4 and p7 positions (66). Using Be-loaded HLA-DP2–plexin A4 tetramers, we (64) identified tetramer-binding CD4⁺ T cells in the BALF of all HLA-DP2–expressing CBD patients who exhibited a Be-specific immune response in lung. Interestingly, the CD4⁺ T cells expressing the public Vβ5.1⁺ TCR were also specific for the HLA-DP2–plexin A/Be complex (42), strongly implicating plexin A as a relevant endogenous Ag in CBD.

To characterize the structural features of βGlu69-containing HLA-DP molecules that explain disease association, multiple HLA-DP2 (DPA1*01:03, DPB1*02:01) molecules were crystallized with self-peptides derived from HLA-DR α-chain, Ras, or HLA-A28 (63, 67). The overall structure of these HLA-DP2–peptide complexes was similar to that of other MHCII-peptide complexes; however, several unique features of this molecule likely contribute to the development of CBD. First, there was a widening of the peptide-binding groove between the peptide and the β-chain α-helix (63, 67), suggesting that the α-helix is flexible in this region and can roll away from the peptide and the floor of the binding groove. The net effect of this widening was a solvent-exposed acidic pocket composed of three glutamic acid residues on the HLA-DP2 β-chain: βGlu68 and βGlu69 from the β-chain α-helix and βGlu26 from the floor of the peptide-binding groove (63, 67). In addition to βGlu69, the HLA-DP2 crystal structure suggested that βGlu26 and βGlu68 may be involved in Be coordination and presentation. Site-directed mutagenesis of each of these glutamic acids to alanines abrogated the ability of Be-pulsed HLA-DP2–expressing fibroblasts to stimulate Be-specific TCRs (63). Because βGlu26 and βGlu68 are invariant among HLA-DP alleles (68), they were not identified in genetic analyses of linkage between HLA-DPB1 alleles and CBD, and their presence is not sufficient for Be presentation in the absence of βGlu69. Because βGlu69 is the most important polymorphism associated with the genetic susceptibility to Be-induced disease and solved structures of other Be-associated proteins show Be coordination by acidic amino acids (69), these findings suggested that this acidic pocket is the Be binding site within the TCR footprint of HLA-DP2.

Recently, we (67) crystallized an HLA-DP2 mimotope/Be-specific AV22 TCR complex to a resolution of 2.8 Å (PDB ID code 4P4R). In this structure, the acidic pocket included two additional acidic amino acids contributed by the peptide at positions p4 and p7 (67). Surface plasmon resonance TCR binding experiments and mutational studies confirmed that the acidic properties of pD4, pE7, and βGlu69 were essential for Be presentation (67). Upon Be binding to HLA-DP2, the acidic pocket underwent a conformational rearrangement that captured the Be²⁺ and an accompanying cation through interactions via the carboxylates of βGlu69 and other HLA-DP2 and mimotope oxygens (67). Surprisingly, neither Be²⁺ nor Na⁺ was accessible on the surface of the complex for direct TCR interaction (Fig. 2A). However, the presence of these cations reduced the electrostatic surface potential and subtly altered the surface topology of the HLA-DP2 mimotope complex over the cation binding site where the AV22 TCR Vβ CDR3 interacts (Fig. 2), indicating that both of these changes likely contributed to creation of the αβTCR ligand (67). It is widely believed that nonpeptide moieties, such as metals, trinitrophenol, or fluorescein, are recognized by T cells as haptens (i.e., bind to the surface of the MHC-peptide complex and participate in TCR engagement) (70, 71). However, our structural data show that Be is not functioning as a hapten but rather indirectly induces changes in surface charge and topology that convert a tolerized self-peptide into a neoantigen. In essence, Be becomes part of the internal structure of the complex and represents a novel posttranslational modification.

Exposure of multiple inbred strains of mice to Be failed to lead to the development of a viable murine model of CBD (72). With the generation of HLA-DP2–transgenic mice (73), we exposed these mice to BeO and noted the development of peribronchovascular mononuclear infiltrates and an HLA-DP2–restricted, Th1-polarized Be-specific immune response (74). Using Be-loaded HLA-DP2–plexin A4 tetramers, CD4⁺ T cells derived from the lungs of BeO-exposed HLA-DP2–transgenic mice recognized identical αβTCR ligands as T cells from HLA-DP2–expressing CBD patients (74). This study confirmed the importance of HLA-DP2 and likely other βGlu69-containing HLA-DP molecules in the generation of CBD.

Metal ions, such as Ni, Co, and Cu, can induce allergic hypersensitivity. Ni is the most common contact allergen, with 10% of the white population having positive skin reactions (75). Unlike Be-specific T cells, some Ni-reactive T cell clones cross-react with other transitional metals, such as Cu and Pb (76), with no particular MHCII allelic association noted in some nickel-allergic subjects (77). However, MHCII-restricted CD4⁺ T cells have been identified (78), and studies suggest that Ni can bind to histidine residues derived from either the peptide (79) or the MHCII molecule (80). In this regard, we (80) showed that T cell recognition of Ni required HLA-DR52c with a specific unknown peptide(s) and was dependent on a histidine residue at position 81 of the MHCII β-chain. Thus, unlike Be, Ni acts as a hapten and directly participates in TCR engagement.

Severe allergic reactions to abacavir, a reverse-transcriptase inhibitor used in the treatment of HIV infection, were described recently and are strongly linked to HLA-B alleles (81). Structural and biochemical studies showed that abacavir binds within the HLA-B peptide-binding groove and restricts the repertoire of bound self-peptides, resulting in the generation of an exuberant polyclonal CD8⁺ T cell response that resembles an allogeneic response (82, 83). This is reminiscent of T cell recognition of Be, in which the Be²⁺ cation also binds within the peptide-binding groove of HLA-DP2 without being part of the TCR interface. However, unlike abacavir, Be modifies the αβTCR ligand without changing the HLA-DP2 bound self-peptide or restricting the peptide repertoire, as evidenced by the ability of fixed Be-pulsed APCs to stimulate Be-specific CD4⁺ T cells (84). Collectively, the recent findings in Be-, Ni-, and abacavir-induced hypersensitivity show how the addition of diverse small molecules can alter the topology of the MHC-peptide complex, generating neoantigens and Ag-specific immune responses directed against these previously tolerized self-peptides.

Recent progress defining the adjuvant properties of Be, unique structural features of HLA-DP2, stimulatory peptides that capture and coordinate Be, and structural changes induced by Be to the MHCII-peptide complex provides an explanation for the gene-by-environment interactions that lead to CBD. Similarity exists in the manner in which small molecules, such as drugs, can associate with certain HLA molecules and induce idiosyncratic reactions. The ability of Be and other small molecules to generate neoantigens also suggests similarity to autoimmunity, in which posttranslational modifications can alter peptide binding to the MHCII molecule and potentially T cell recognition. Thus, recent findings suggest that allergy hypersensitivities and autoimmunity may not be distinct disease processes but exist on a continuum.

The authors have no financial conflicts of interest.