Identification of LFA-1 as a Candidate Autoantigen in Treatment-Resistant Lyme Arthritis

See allHide authors and affiliations

Science  31 Jul 1998:
Vol. 281, Issue 5377, pp. 703-706
DOI: 10.1126/science.281.5377.703


Treatment-resistant Lyme arthritis is associated with immune reactivity to outer surface protein A (OspA) of Borrelia burgdorferi, the agent of Lyme disease, and the major histocompatibility complex class II allele DRB1*0401. The immunodominant epitope of OspA for T helper cells was identified. A homology search revealed a peptide from human leukocyte function–associated antigen-1 (hLFA-1) as a candidate autoantigen. Individuals with treatment-resistant Lyme arthritis, but not other forms of arthritis, generated responses to OspA, hLFA-1, and their highly related peptide epitopes. Identification of the initiating bacterial antigen and a cross-reactive autoantigen may provide a model for development of autoimmune disease.

Lyme disease is a multisystem illness caused by infection with the spirochete Borrelia burgdorferi (1). A prominent late manifestation of the disease is Lyme arthritis (1, 2). About 10% of patients with Lyme arthritis develop what we have termed antibiotic treatment–resistant Lyme arthritis, which typically affects one knee for months to years after multiple courses of antibiotics (1). Such patients have no detectable spirochetal DNA in joint fluid after antibiotic therapy, which suggests that the spirochete has been eliminated by this treatment (3). Because there is increased frequency of the HLA-DRB1*0401allele in these patients (4), an autoimmune etiology should be considered. The hypervariable 3 region (HVR3) at residues 67 to 74 of DRB1*0401 is associated with susceptibility to rheumatoid arthritis (RA) and is contained in at least 15 differentDRB1 alleles (5). Most patients with prolonged treatment-resistant Lyme arthritis have one of these homologous alleles (4). What antigen are these class II molecules presenting?

Borrelia burgdorferi induces an immune response of expanding reactivity to an array of spirochetal proteins over months to years (6). Antibody reactivity to outer surface protein A (OspA) typically develops near the beginning of prolonged episodes of arthritis (7). T cell lines from patients with treatment-resistant Lyme arthritis preferentially recognize OspA, compared with patients with treatment-responsive disease. OspA-reactive type 1 T helper (TH1) cells are detectable in the synovial fluid of individuals with treatment-resistant arthritis years after antibiotic treatment (7). Thus, these patients may have progressed into an autoimmune state by developing a cross-reactive response between OspA and a self-antigen.

We used the DRB1*0401 peptide-binding algorithm (8) to determine the scores for all nine-residue peptides in the OspA protein sequence that contained an appropriate pocket 1 anchor residue—F, I, M, L, T, V, or Y—necessary for binding in the DRB1*0401 peptide-binding cleft. According to this algorithm, only peptides with scores greater than 2 are likely to bind and be able to be presented by the DRB1*0401 molecule (8). The highest scoring peptide that was identified, OspA165–173, had a predicted binding score of 6.5, and the next best scoring peptide, OspA237–245, achieved a score of 3.7. To verify that these peptides can bind to DRB1*0401 in vitro, the binding of125I-labeled m1–7 (YRAMATL; predicted DRB1*0401 binding score = 5.9), which has the consensus binding motif for DRB1*0401 (9), was measured when in competition with unlabeled 20-residue peptides from OspA. Only OspA154–173, which contains the DRB1*0401-predicted dominant epitope OspA165–173, inhibited binding of the radiolabeled peptide m1–7 to purified DRB1*0401 (Table 1), confirming the algorithm's prediction.

Table 1

Inhibition of m1–7 peptide binding to DRB1*0401 (15) by 20-residue peptides of OspA.

View this table:

To test for T cell reactivity in vivo, we made use of class II–deficient mice transgenic for a chimeric DRB1*0401 molecule (DRB1*0401-tg) (9). Any CD4+ T cell response generated in these mice can be directly attributed to the presence of the DRB1*0401 molecule. The ElisaSpot assay was used for measuring antigen-specific T cell reactivity, a sensitive and efficient technique that allows detection of cytokine production at the single cell level, which may occur in the absence of proliferation (10). We initially assayed for production of TH1 and TH2 cytokines, interferon-γ (IFN-γ), and interleukin-5 (IL-5), respectively. Both IFN-γ–producing and IL-5–producing cells were detected when cells were activated with a polyclonal stimulus [anti-CD3; monoclonal antibody (mAb) 145.2C11]. In contrast, when cells were stimulated with OspA antigen, IFN-γ production was dominant, with essentially no detectable IL-5 secretion (11). Therefore, detection of IFN-γ was used as the readout for antigen-specific T cell reactivity in all subsequent assays. DRB1*0401-tg mice were immunized with OspA and lymph node cells were stimulated with overlapping 20-residue peptides of OspA: the immunostimulatory epitopes correlated precisely with the epitopes predicted by the DRB1*0401 algorithm (Fig. 1A). Immunization of the DRB1*0401-tg mice with OspA165–173 resulted in a recall response to whole OspA in vitro (11). Hence, we have identified the immunodominant epitope of OspA in the context of DRB1*0401. To test the ability of OspA165–173 to be presented by DRB1 alleles related to DRB1*0401 (5), we performed the same experiment in mice transgenic for DRB1*0101 (12). These transgenic mice possess a full complement of murine class II genes, thereby providing distinct major histocompatibility complex (MHC) alleles for OspA peptide presentation. ElisaSpot analyses of OspA-immunized DRB1*0101-tg or (DRB1*0101-tg × SJL)F1 mice showed reactivity to OspA165–173 as well as to an array of other epitopes (Fig. 1B). In contrast to DRB1*0401-tg mice, reactivity toward OspA165–173 developed as a subdominant epitope, suggesting that alternative determinants are available for binding that could influence disease development. Interestingly, the F1 mice had a response to OspA165–173 that was three times the response of DRB1*0101-tg mice. This is likely because of expression of the murine I-Eβs chain, which is homologous in the HVR3 to DRB1*0401 (5), thereby providing twice the number of class II molecules for presentation of this particular peptide. Thus, we have identified the immunodominant OspA peptide recognized in the context of DRB1*0401 and found that DRB1 and murine class II alleles homologous to DRB1*0401 in their HVR3 can also present this epitope.

Figure 1

IFN-γ ElisaSpot analysis demonstrates OspA165–173 as the functional, immunodominant epitope of OspA in DR4-tg mice. (A) Class II–deficient, DRB1*0401-tg mice, immunized with whole OspA, but not a control protein, recall whole OspA and OspA164–183 specifically. DRB1*0401-tg mice were immunized in both hind footpads with either 50 μl of OspA (44 μg/ml) or human spinal chord extract (hSCE; 100 μg/ml) in complete Freund's adjuvant. Eight days later, draining popliteal lymph nodes were isolated and 5 × 105 cells were cultured with either a positive control stimulant, CD3 antibody, mAb 145.2C11, or one of the following test antigens: hSCE (50 μg/ml), OspA (10 μg/ml), overlapping OspA 20-mer peptides (10 μg/ml each), or medium alone. IFN-γ production was analyzed 24 hours later by ElisaSpot (10). Values from wells with medium alone were subtracted from values from wells that contained antigen. Antigens are listed as overlapping 20-mer peptides spanning OspA, beginning with amino acid 17. Residues 1 to 17 contain the leader sequence and are therefore cleaved during export through the bacterial membrane. Representative experiments of six OspA-immunized and two hSCE-immunized mice are shown. (B) DRB1*0101-tg and (DRB1*0101-tg × SJL)F1 mice immunized with whole OspA recall OspA165–173 as well as other epitopes. In contrast to the DRB1*0401-tg mice, the DRB1*0101-tg mice express murine class II; therefore, a broader array of OspA epitopes is recognized. Experiments were performed as described above. One of three and one of two representative experiments are shown for DRB1*0101-tg and (DRB1*0101-tg × SJL)F1 mice, respectively.

We searched the Genetics Computer Group gene bank for human proteins containing sequences homologous to OspA165–173. Of the 20 peptides retrieved with the highest identity and homology scores, two were of human origin: hLFA-1 (CD11a/CD18, integrin αLβ2) and 40S ribosomal protein. Only the peptide contained in hLFA-1, hLFA-1αL332–340, attained a significant DR4-binding score (7.3), with six–amino acid identity (YVIEGTSKQ; nonconserved residues in italics), suggesting hLFA-1 as a potential autoantigen. The peptide contained within the 40S ribosomal protein sequence (YVLEGKELE) attained a DR4-binding score of 0, mostly because of Lys at position p6, which is not tolerated in the DR4-HVR3 (13). The hLFA-1αL332–340 peptide is located extracellularly in the interactive or I-domain that mediates the binding interaction between LFA-1 and its ligand, intercellular adhesion molecule–1 (ICAM-1) (14). When the DR4-binding algorithm was applied to the entire I-domain (amino acids 170 to 349), hLFA-1αL332–340 achieved the highest predicted binding score (7.3), nearly twice that of the next highest scoring peptide, hLFA-1αL196–204 (binding score = 4.3), and higher than that of OspA165–173. We determined, by performing the peptide binding competition assay [median inhibitory concentration (IC50) = 0.7825 mM], that hLFA-1αL331–345, a 15-mer containing the core residues 332 to 340, was capable of binding DRB1*0401 in vitro.

To test the hypothesis that hLFA-1 is an autoantigen in patients with treatment-resistant Lyme arthritis, but not in other forms of chronic inflammatory arthritis, we mapped the immunodominant epitope of OspA in synovial fluid (SF) cells from a patient (4) with treatment-resistant Lyme arthritis (Fig. 2A) (10). As in the DRB1*0401-tg mouse, OspA164–183 was immunodominant. We then analyzed the antigen reactivity profile of SF T cells from patients with treatment-resistant Lyme arthritis as well as patients with other forms of chronic arthritis (15). ElisaSpot for IFN-γ production (10) and proliferation assays (16) showed that people in a panel consisting of only those with treatment-resistant Lyme arthritis have varying degrees of SF T cell reactivity to whole OspA, OspA164–183 as well as hLFA-1 (Fig. 2B). Reactivity to hLFA-1 is due to recognition of hLFA-1αL326–343, the region homologous with OspA164–183 (Fig. 2, C and D). This reactivity appears to develop over time, as patients who initially showed no response to hLFA-1 had marked reactivity when tested 1 to 3 months later (11).

Figure 2

SF T cells from patients with treatment-resistant Lyme arthritis generate a response to hLFA-1. (A) IFN-γ ElisaSpot analysis of 3 × 105 SF T cells per well, from patient 4, cultured with each of the overlapping OspA peptides at 10 μg/ml, revealed OspA164–183 as the immunodominant epitope (10,15). Reactivity to whole OspA was positive as determined by proliferation assay (medium, 2552 177 248 cpm; OspA, 24,497 177 2079 cpm) (16). (B) SF T cells from patients with treatment-resistant Lyme arthritis, but not other forms of chronic arthritis, produce IFN-γ in response to in vitro restimulation with OspA and hLFA-1. We cultured 3 × 105 SF cells with either a positive control, CD3 antibody hybridoma OKT3 supernatant, or one of the following test antigens: OspA (10 μg/ml), OspA164–183 (10 μg/ml), hLFA-1 (70 ng/ml), or medium alone for 24 hours. Reactivity was determined by performing an IFN-γ ElisaSpot assay. Values from medium-alone wells were subtracted from wells containing antigen. Because of technical limitations relating to the purification process of hLFA-1, the molar concentration of hLFA-1 used in these experiments is three orders of magnitude lower than the optimal concentration used for OspA. When equimolar amounts of OspA and hLFA-1 were tested, reactivity to OspA was depressed to levels comparable to those for hLFA-1 (11). Because of limited numbers of cells, controls not tested for reactivity to OspA164–183 were patients 17, 18, and 20; and, for hLFA-1, patients 17 and 18. (C) Treatment-resistant Lyme arthritis patient 11, who is homozygous for DRB1*0401, demonstrates SF T cell reactivity to the 20-mer containing the OspA homologous,DRB1*0401-defined dominant epitope within the I-domain, hLFA-1αL326–345. We cultured 3 × 105SF cells with hLFA-1αL326–345 (25 μg/ml). IFN-γ ElisaSpot assay was performed as described above. (D) Treatment-resistant Lyme arthritis patient 10, who is heterozygous for an RA-associated allele (DRB1*0102), demonstrates SF T cell reactivity to the 20-mer containing the OspA homologous,DRB1*0401-defined dominant epitope within the I-domain, hLFA-1αL326–345. We cultured 3 × 105SF cells with equimolar amounts of OspA164–183, hLFA-1, and hLFA-1αL326–345. IFN-γ ElisaSpot assay was performed as described above.

Borrelia burgdorferi sensu stricto is the only spirochetal strain associated with treatment-resistant Lyme arthritis (17) and the sole strain that contains the OspA165–173 sequence that is highly related to hLFA-1αL332–340. Murine LFA-1a differs significantly from hLFA-1 at this particular epitope, providing an explanation for why chronic Lyme arthritis does not develop in DRB1*0401-tg mice exposed to B. burgdorferi (12).

Our demonstration of autoreactivity against hLFA-1 (in particular, the predicted cross-reactive epitope) in patients with treatment-resistant Lyme arthritis suggests that this disease involves an autoimmune process. However, although the genetic predisposition for development of treatment-resistant Lyme arthritis has been correlated with DR4, we cannot rule out other genetic, environmental, and infectious factors that might be involved. As mentioned above, the HVR3 of the DRB1 chains associated with RA possesses a shared epitope at residues 67 to 74 (5). Most patients with severe RA carry at least one allele that contains the shared epitope sequence of DRB1*0401, henceforth referred to as an RA-associated allele (5). Individuals who develop the most severe form of RA typically have two RA-associated alleles (18). HLA typing of our panel of 11 treatment-resistant Lyme arthritis patients revealed that 7 possessed at least one RA-associated allele (15), and 9 made a response to hLFA-1. Patient 11, who was homozygous for DRB1*0401, responded four times more vigorously to both OspA and hLFA-1 than the next highest responder. In patients with other forms of arthritis, the presence of an RA-associated allele by itself was not sufficient for induction of an OspA or hLFA-1 response, as at least five of the nine control patients possessed an RA-associated allele (15) yet made no response to OspA or hLFA-1. Thus, priming by B. burgdorferi infection or at least with OspA may be required for development of an autoimmune response to hLFA-1. Other factors may also be involved in development of treatment-resistant Lyme arthritis, as some treatment-resistant patients who do not possess an RA-associated allele make a response to hLFA-1 and some patients with treatment-resistant Lyme arthritis do not respond to either OspA or hLFA-1 (Fig. 2B).

On the basis of our DRB1*0401-restricted OspA T cell epitope mapping data, as well as previous work on immune reactivity and cytokine production in response to infection with B. burgdorferi (7), we propose a model on how an immune reaction to B. burgdorferi might result in development of an autoimmune response against hLFA-1: B. burgdorferi enters the host via a tick bite and disseminates to multiple tissues. Months later, a highly inflammatory immune response develops in the joint, and this response is dominated by TH1 IFN-γ–producing cells that contain OspA reactive cells. We propose that the high local concentration of IFN-γ up-regulates expression of ICAM-1 (19) on synoviocytes and synovial fibroblasts as well as of MHC class II molecules on the local professional and nonprofessional antigen-presenting cells (APCs) (19). This enhanced ICAM-1 expression leads to recruitment of LFA-1 expressing cells, in particular activated TH1 cells. The combination of elevated LFA-1 expression on T cells and macrophages plus MHC class II up-regulation on APCs may result in increased LFA-1 peptide presentation by macrophages and synoviocytes that have processed either endogenous or phagocytosed LFA-1 (20). Hence, a vicious cycle is initiated so that, even after elimination of the spirochetes by antibiotic therapy, the OspA-primed T cells remain activated by stimulation with LFA-1. The release of inflammatory cytokines by these activated T cells and macrophages may then result in tissue damage and joint destruction (21).


View Abstract

Navigate This Article