In Situ Diversification of the Antibody Repertoire in Chronic Lyme Arthritis Synovium
http://www.100md.com
免疫学杂志 2005年第5期
Abstract
Lyme arthritis is initiated by the tick-borne spirochete, Borrelia burgdorferi. In a subset of patients, symptoms do not resolve in response to standard courses of antibiotics. Chronic joint inflammation may persist despite spirochetal killing, suggesting an autoimmune etiology. The pathogenic mechanisms that sustain chronic Lyme arthritis have not been fully elucidated, although T cells are believed to play a role. The synovial lesion contains elements of a peripheral lymph node, with lymphoid aggregates, plasma cells and follicular dendritic cells. An analysis of activated cells at the site of injury could yield clues regarding the nature of the response and the identity of potential autoantigens. Using laser-capture microdissection, we have isolated plasma cells from the joint tissue of chronic Lyme arthritis patients who underwent synovectomy. Expressed Ig V regions were amplified by RT-PCR. A majority of isolated cells expressed H chains, which is indicative of a class-switched response. There were a large number of nucleotide substitutions from germline, with a higher fraction of replacement mutations in the CDRs, suggesting a process of Ag-driven selection. We have recovered clonal clusters of cells containing identical junctions and V(D)J rearrangements. Sequence analysis reveals a hierarchy of shared somatic mutations between members of a given clone. Intraclonal diversity among plasma cells of close physical proximity points toward an ongoing process of diversification and affinity maturation, possibly driven by the chronic presence of an autoantigen.
Introduction
Lyme disease is a complex, multisystem illness caused by infection with the tick-borne spirochete Borrelia burgdorferi (Bb)3 (1, 2). Arthritis is a prominent, late-stage manifestation of the disease (3). In a subset of patients, arthritic symptoms do not resolve despite 2–3 mo courses of antibiotics, both oral and i.v. (4). Bb DNA usually cannot be detected in the synovial fluid and tissue of these "treatment-resistant" patients by PCR (5). Persistence of inflammation in the apparent absence of the initiating pathogen points toward an autoimmune phenomenon, downstream of the original bacterial infection.
Histologic examination of joint tissue from chronic Lyme arthritis patients who underwent synovectomy reveals a marked infiltration of mononuclear cells, with vascular proliferation and synovial hypertrophy. The arthritic lesion mimics that in rheumatoid arthritis and other chronic inflammatory arthritides. There are mixed aggregates of B cells, T cells, and follicular dendritic cells, surrounded by clusters containing large numbers of plasma cells (6). The cellular composition of synovial lymphocytic infiltrates is reminiscent of that observed in germinal centers in peripheral lymphoid organs. Data from rheumatoid arthritis have shown that V gene diversification, hypermutation, and terminal differentiation of B cells occur within synovial germinal center-like structures (7, 8, 9). This microenvironment seems to support longevity of plasma cells that accumulate in the chronically inflamed tissue (9, 10, 11, 12).
Bb uses various mechanisms of dissemination to home to specific sites in the body, including the joints (13). Spirochetal lipoproteins are potent inducers of proinflammatory cytokines like TNF- and IL-6 (14, 15, 16). In addition to promoting tissue damage, these cytokines may have a role in maintaining ectopic lymphoid tissue in the synovium (17, 18). As in rheumatoid or reactive arthritis, the joint becomes a site of intense inflammatory activity. The ectopic lymphoid follicles persist despite apparent spirochetal clearance. Hence, it is conceivable that cells within them are being triggered by a self-Ag, either joint-specific or exposed/up-regulated due to chronic tissue injury. We hypothesize that the inflammatory milieu in the diseased synovium supports a local immune response, characterized by plasma cell differentiation, somatic hypermutation, and the generation of affinity matured variants.
In this study, we have used laser-capture microdissection (LCM) to isolate single or groups of plasma cells from synovial tissue sections obtained from treatment-resistant Lyme arthritis patients (19). We have chosen to examine plasma cells because they represent terminal effectors derived from B cells that were recruited through appropriate activation signals and retained at the site of injury, as opposed to bystander or migrant B cells. Ig V regions were amplified by RT-PCR, enabling analysis of the expressed Ab repertoire in the context of the inflamed synovium. The data suggest a process of maturation and diversification of the Ab response, occurring in situ. Although such phenomena have been documented for both rheumatoid and reactive arthritis, this is the first description of a similar process in chronic Lyme arthritis, where T cells have been traditionally thought of as the dominant players, acting through a cascade of proinflammatory cytokines (20, 21, 22). Additionally, single cell RT-PCRs yield in vivo pairings of H and L chains that can be cloned together and expressed. This allows for the generation of Ab V region probes for the detection of potentially cross-reactive Ags that are relevant to disease pathogenesis.
Materials and Methods
Preparation of tissue samples
Patient tissue samples came from an archival synovial tissue collection, obtained from Lyme arthritis patients who had persistent arthritis after antibiotic treatment and had undergone synovectomy. Biopsy samples of inflamed tissue were extracted arthroscopically and snap-frozen in OCT (Miles Laboratories). Frozen tissue was cut into 6-μm sections in a cryostat and stored at –80°C. Synovial tissue from two patients, A and B, were selected for detailed analysis, based on a previous report documenting marked inflammatory and lymphocytic infiltration in the synovium (23). Duration of arthritis before antibiotic therapy was 1 mo for patient A and 6 mo for patient B. Both patients were treated with oral doxycholine for a period of 30 days, as well as oral amoxicillin (patient A) and i.v. ceftriaxone (patient B). After antibiotic therapy, synovial samples from A and B were negative in PCRs done to detect Bb DNA. Duration of arthritis posttreatment and before synovectomy was 4 mo in both of the cases.
Isolation of plasma cells
Plasma cells were identified by indirect immunofluorescence, using anti-CD138:FITC (Serotec) as primary Ab (1/50 dilution) and anti-fluorescein:Alexa 488 (Molecular Probes) as secondary Ab (1/100 dilution). Staining protocols were optimized for efficient LCM and maximal RNA recovery. Briefly, tissue sections were thawed and immediately fixed in cold acetone (4°C) for 4 min. Prolonged exposure to aqueous solutions can result in significant degradation of RNA. Therefore, relative to standard tissue-staining procedures, we used shorter Ab incubation times–10 min in primary Ab and 5 min in secondary Ab. Incubations were performed on a cold block (HistoGene Cold Block) to minimize background staining and in the presence of RNase inhibitor (RNase Out; Invitrogen Life Technologies) to preserve RNA integrity. The slides were washed in 1x PBS and dehydrated by sequential immersions in 75, 95, and 100% ethanol for 30 s each, followed by two 5-min washes in xylene. Staining was immediately followed by LCM (19), using a Pixcell IIe LCM instrument.
Noncoverslipped samples were visualized through the microscope ocular or on the live video monitor to locate areas of interest. Tightly clustered cells were not teased apart, to avoid mechanical tearing of the tissue. CD138+ cells were captured on High Sensitivity LCM caps (CapSure HS) either singly, in doublets/triplets, or as larger groups (ranging in size from 15 to 40 cells).
Extraction of captured material
A mixture containing 10 μl of 1x reverse transcriptase (RT) buffer, 1.4 μl of RNase inhibitor (40U/μl RNase Out; Invitrogen Life Technologies) and 3 μl of a 5% Nonidet P-40 solution was added to the LCM cap/microcentrifuge tube assembly, to lyse the captured cells on the transfer film. The assembly was inverted, spun at 4000 x g for 3 min and the extracted material was directly used in the RT-PCR. RT-PCR and subsequent rounds of PCR amplification were conducted, following the protocol and primers described by Wang et al. (24).
Synthesis of cDNA
cDNA was synthesized with a mixture of primers complementary to sequences in the C regions of μ-, -, -, and -chains (24). CμI, CI, CI, and CI, each at 5 pmol/μl, were added to the PCR tubes, which were incubated at 65°C for 3 min and then cooled to 25°C for 3 min. For the RT reaction, each tube received 2 μl of 5x RT buffer, 2 μl of 0.1 M DTT, 1 μl of 10 mM dNTP, and 0.5 μl of Superscript II RT (200 U/μl). Samples were incubated at 37°C for 1 h, heated to 70°C to inactivate the enzyme, and cooled to 4°C.
Amplification of Ig H and L chain V regions (VH and VL)
The cDNA was used as a template for two rounds of PCR amplification, using progressively nested gene or gene family specific primers (24). H and L chain PCRs were done separately. The first PCR was performed with a mixture of primers corresponding to conserved regions in the leader sequences of VH1-6, V1-4, or V1-3 families and corresponding CμII, CII, CII, or CII primers, complementary to sequences in the C domains.
Each amplification was performed in a 50-μl reaction volume containing 25 μl of 2x PFU Ultra Hotstart DNA Polymerase Master Mix, 1 μl of each H or L primer (at a working concentration of 20 pmol/μl), and 8 μl of the cDNA mixture. Cycling conditions included an initial 2-min denaturation step at 94°C, 3 cycles of preamplification (94°C for 45 s, 45°C for 45 s, 72°C for 1 min 45 s), and 30 cycles of amplification (94°C for 45 s, 50°C for 45 s, 72°C for 1 min 45 s). This was followed by a final 10-min incubation step at 72°C. In the second round, 3-μl aliquots from the first PCR were separately amplified for μ, , , or V regions, using the same cycling parameters (excluding the preamplification step). Primers used in the second PCR corresponded to the first six codons of the VH or VL family gene segments on the 5' end and more nested C region primers, CμIII, CIII, CIII, and CIII, on the 3' end. Amplified products run at around 400 bp on a 1% agarose gel. Gel bands were excised and purified, using the QIAquick gel extraction kit.
Yields of RNA from small numbers of cells, especially single cells, captured from a tissue section of 6-μm thickness, tend to be low. Therefore, sensitivity and efficiency of H and L PCRs were optimized at a single cell level, using hybridomas or lymphoblastoid cell lines as positive controls. In patient A, 9 of 14 (64%) laser-captured samples yielded both H and L chains. One sample (containing 10 cells) yielded only L chains and two single cell samples yielded H chains alone. Two LCM caps (also containing single cells) yielded neither an H nor an L product in the PCRs. A total of 24 LCM samples were obtained from patient B. Of these, 19 (79%) yielded both H and L chains while three samples (two were single cells and one contained a pair) gave only L chains. Two LCM caps did not yield any Ig product.
Sequence analysis
PCR products were cloned into the TOPO-TA sequencing vector (Invitrogen Life Technologies) and used to transform Escherichia coli. Plasmid minipreps were prepared (QIAprep; Qiagen) and sequenced, using the M13 reverse primer at the Tufts University Core Facility. The number of colonies picked for sequencing depended upon the number of plasma cells that yielded the corresponding PCR product. Five to 10 plasmids were sequenced for small groups (up to 4 cells) and 10–30 plasmids for larger groups (up to 40 cells). For the identification of the most homologous germline matches, VH and VL sequences were aligned against Ig genes in the IMGT database (25, 26) using the V-QUEST program (http://imgt.cines.fr/textes/vquest/). In addition, the Ig-basic local alignment search tool was used to perform searches within sequences listed in GenBank (www.ncbi.nlm.nih.gov/igblast/). D-segment assignations were based on a minimum of 10 nucleotide homology with the closest germline match. All sequences were deposited with GenBank under the accession numbers AY685251 through AY685387.
Analysis of mutations
Complementarity determining regions (CDRs) and framework regions (FWRs) were defined based on the Kabat numbering system (27). The first eight codons in FWR1, being primer-encoded, were excluded from analysis. Mutations were calculated for CDRs 1, 2, and 3 for VL sequences, and CDRs 1 and 2 for VH sequences. A multinomial distribution model (28) was used to assess the statistical significance of an excess of replacement mutations in CDRs and silent mutations in FWRs for each Ig V region. The corresponding germline gene sequence was imported into the JAVA applet at http://www-stat.stanford.edu/immunoglobulin provided by Lossos et al. (28) along with the associated set of r (number of replacement mutations) and s (number of silent mutations) values for FWRs and CDRs.
Results
Histology and immunofluorescence
Synovial tissue samples from two chronic Lyme arthritis patients (A and B; described in Materials and Methods) were used in this study. In H&E-stained sections, synovia from both patients exhibited marked lymphocytic infiltration (Fig. 1a). There were numerous lymphoid aggregates, some perivascular in localization. Plasma cells, identified by their extensive basophilic cytoplasm and low nucleus:cytoplasm ratio, were observed in clusters proximal to these aggregates or around the lumen of small blood vessels. By histologic staining, there appeared to be a larger number of plasma cells in patient A tissue than patient B tissue.
FIGURE 1. Histology and immunofluorescent detection of plasma cells in synovial tissue sections. a, H&E-stained synovial sections from patients A (i and ii) and B (iii and iv). P, plasma cell; L, lymphocyte; V, blood vessel. Pictures were taken at x40 magnification, except for i which was at x20. Lymphoid aggregates and proximal plasma cell groups are shown (i, ii, iii). iv, Large lymphocytic infiltrate in patient B synovium. b, Immunofluorescent detection (x50) of plasma cells by staining for CD138 (syndecan-1). A monoclonal anti-CD138:FITC was used as primary Ab and the FITC tag was detected by an anti-fluorescein:Alexa488 (green) Ab. The negative control was a nonspecific FITC-conjugated Ab (i and iv). Plasma cells were observed and laser-captured either singly/in small groups (indicated by arrows; ii and v) or as larger clusters (iii and vi).
For the purposes of LCM, we detected plasma cells by indirect immunofluorescence for syndecan-1 or CD138, a phenotypic marker (Fig. 1b). Distinct membrane staining for CD138 distinguishes plasma cells, which express this marker, from B cells, which do not (29, 30). CD138+ cells were distributed through the tissue singly or in small cellular groups, as well as in larger clusters, the latter being especially prominent in patient A. Synovial plasma cells were isolated by LCM and subjected to RT-PCR for the amplification of Ig H and L chain V region (VH and VL) sequences.
Repertoire of synovial plasma cells
Analysis of 59 unique H and 78 unique L sequences expressed within the arthritic lesions of both Lyme patients revealed a diverse, but nonrandom repertoire. In patient A, the VH repertoire was dominated by gene segments belonging to the VH3 family (10 of 18 sequences, 55%) (Table I). VH3-7 was the most prevalent, recurring four times. In patient B, the VH3 and VH4 families were about equally represented (Table I)–VH3-23 (five clones), VH3-30 (five clones), and VH4-61 (three clones) were the most recurrent individual gene segments. The majority of plasma cells isolated from patient synovia expressed H chains–100% in patient A and 66% in patient B (the remaining 34% expressed μ), which is indicative of an Ag-driven response. Interestingly, we identified six examples in which identical V regions were expressed with both μ and C regions, suggesting in situ class switching and the involvement of T cell-derived cytokines.
Table I. Distribution of VH and V families in the expressed H or repertoire of synovial plasma cells isolated from patients A and B
The VL repertoire from patient A mainly consisted of L chains, with a : ratio of 2.7 compared with the expected ratio of 1.5 for peripheral blood (31). V1 and V3 were the most commonly used gene families, and V1-5 (five clones) was the most recurrent gene segment (Table I). In synovial tissue from patient B, the VL repertoire had a much lower : ratio of 0.9, the V1 family being the most prominent among gene segments used (Table I). Neither the H nor the -chain distributions differed significantly from those that have already been reported for the normal, peripheral Ig repertoire (32, 33, 34, 35).
Restricted use of L chains
We observed a marked repertoire restriction in the V regions recovered from patient B. A total of 27 unique sequences were obtained from 3 single cells, 10 groups comprising 2–4 cells each and 2 clusters of 20 and 30 cells (summarized in Table II). Fifteen of the 27 (>55%) used a single V gene segment, V3-21. Two other segments, V1-47 and V1-44 each accounted for 19 and 15% of recombinants, respectively (5 and 4 of 27). These three gene segments together accounted for 89% of the captured repertoire. Furthermore, all 27 rearrangements from this patient used J3. For patient A, in whom the synovial Ig repertoire was dominated by -chains, a total of seven distinct sequences were recovered. Four of them contained V3-21, V1-47, or V1-44, the segments that were prominent in patient B, and all of them also contained J3.
Table II. Germline gene usage of V regions isolated from patient Ba
The 5'-V PCR primers used in our study were designed to identify gene segments belonging to the V1, 2, and 3 families, which account for >96% of expressed V gene segments in the peripheral repertoire (36). Under these PCR conditions, we detected just 2 and 6 rearrangements, respectively, in the 20- and 30-cell clusters of sample nos. 9 and 10 (Table II). Because this discrepancy between cell number and the number of unique sequences recovered did not occur with either the single cells or the two to four cell groups, samples 9 and 10 appeared to be oligoclonal cellular clusters. Of a total of seven J gene segments (organized as J-C pairs), four are considered to be functional (J 1, 2, 3, and 7). We used a 3' C primer which is complementary to a 16 nucleotide stretch common to all four J-C pairs, five codons downstream of the beginning of the C regions. Hence, there is no primer-induced bias at the 3' end. The marked overrepresentation of J3, V3-21, and V1-47 in these synovial samples was statistically significant, as assessed by a 2 test for "goodness-of-fit" between observed and expected frequencies (Table III).
Table III. Bias in the use of V and J gene segments in synovial plasma cells isolated from patient Ba
The V-J junctions isolated from patient B displayed limited diversity. Junctional analysis of the 15 V3-21:J3 sequences, isolated from multiple clusters, revealed two separate recombination events, each associated with a characteristic set of mutations and using different J3 alleles. This resulted in two distinct CDR3 sequences, with a similar number of nucleotide substitutions from germline but different amino acid replacements (Fig. 2, a and b). The other major V-J recombination group is comprised of V1-44:J3 (15%) and V1-47:J3 (19%) recombinants. Germline V1-44 and V1-47 gene segments share 97% identity, differing in CDR3 by a single amino acid–asparagine in V1-44 and serine in V1-47 (26). Hence, rearrangement to the same J3 gene segment yielded junctions with limited variability and a strong conservation of similar groups at the amino acid level (Fig. 2c). We also recovered two V1:J3 junctions from patient A, which displayed a high degree of homology to those observed in patient B synovium. Despite the convergent use of L chains, the corresponding H chain partners were diverse. Possibly, these junctions contribute key paratopic determinants to the ligand-binding site and an Ag-mediated bias allows for their selection, even when associated with different H chains.
FIGURE 2. Conserved junctions isolated from synovial plasma cells. a, Proposed mechanism of generation of V3-21:J3 junctions in patient B synovium. Arrows denote mutations from the corresponding germline gene segment. N-nucleotide insertion in junction (Jtn) 2 is indicated. Junctions were analyzed using the IMGT V-QUEST program (25 ). b, Amino acid homology between junctions derived from V3-21. Jtns 1 and 2 were present in 67 and 33% of V3-21:J3 recombinants, respectively. c, Homologous V1-47/44:J3 junctions. Jtn 3 was present in 33% and Jtns 4 and 5 were each present in 22% of V1:J3 sequences from patient B. [A]1 and 2 represent junctions isolated from patient A. An asterisk (*) indicates identity, double dots (:) denote a strong conservation of similar groups, and a single dot (.) denotes weak conservation of similar groups.
Mutational analysis of V regions
Somatic mutations are believed to occur randomly within the V regions. However, mutations that cause amino acid replacements could result in a selective advantage or disadvantage to the cells that harbor them (37, 38). Ag-mediated selection is evidenced by a higher frequency of replacement mutations in the CDRs, which bind Ag, relative to the FWRs, which maintain the -sheet scaffold of the V domain. Activated B cells thus undergo multiple rounds of mutation and selection, eventually resulting in Ig receptors with greatly enhanced affinity and selectivity for particular Ags, thereby fine tuning the response.
The majority of V regions isolated from patient synovia contained a large number of nucleotide substitutions from the closest germline matches. In general, the H chains appeared to be more mutated than the L chains (with an average of 26 and 16 nucleotide mutations each). The fraction of substitutions that resulted in amino acid replacements was calculated for all Ig sequences, recovered from both patients (Table IV). Overall, we observed a higher mean frequency of replacements in the CDRs, relative to the FWRs–a trend indicative of affinity maturation, as opposed to nonspecific, mitogenic expansion. This difference was statistically significant (p < 0.05; Student’s t test) for H and sequences from both patients and sequences from patient A. Further, we separately analyzed μ and H chain sequences from patient B synovium, where both isotypes were recovered. The sequences, but not μ, exhibited a significantly higher frequency of replacement mutations in the CDRs compared with the FWRs (p < 0.02; Student’s t test) (Table V). Thus, Ig V regions belonging to the traditionally more pathogenic IgG isotype appear to be more Ag selected than those belonging to the IgM isotype.
Table IV. Mean fractions of replacement mutations (Rfrn) within H or L CDRs and FWRsa
Table V. Mean fractions of replacement mutations (Rfrn) within μ or CDRs and FWRs in patient B
We used a multinomial distribution model to estimate whether Ag-selective pressure was operating on individual V regions (28). The model equation was used to determine whether the clustering of replacement mutations in CDRs and their relative scarcity in FWRs occurred at a frequency greater than that expected by random chance alone, taking into account the inherent susceptibility for amino acid substitutions in the corresponding germline gene segment, based on its codon composition. Because Ig mutations can have a total of four distribution possibilities in this model (replacement or silent in FWRs or CDRs), we obtained two separate sets of p values, characterizing each of the two regions (pfwr and pcdr) (Table IV). As expected, the majority of synovial H and L chain FWRs (>80%) had a paucity of replacement and an excess of silent mutations (pfwr < 0.05), consistent with requirements for the maintenance of structural integrity of the V domain. Of these sequences, an average of 26, 45, and 37% of H, -, and -chains respectively, preferentially accumulated replacement mutations in their CDRs (pcdr and pfwr < 0.05) (Table IV), indicative of ligand-dependent selection. Hence, the clustering of replacement mutations within the synovial Ig repertoire followed a nonrandom pattern, being significantly scarce in the FWRs of most V regions and abundant in the CDRs of a smaller population of sequences. The latter group comprises the subset of V regions that seem to have undergone affinity maturation in each patient. The multinomial analysis yielded similar results when applied to published sequences associated with well-defined, Ag-selected Ab sets, induced by immunization (Table VI).
Table VI. Application of the multinomial model to published Ig sequences deposited in the NCBI databasea
Intraclonal diversification of V domains
We recovered a number of clonally related V regions with identical V(D)J rearrangements and junctions, i.e., CDR3 sequences, but different somatic mutations. For example, a single V1-5:J1 recombination event recurred in three separate LCM samples within a localized tissue area–a single plasma cell (no. 1-1), a doublet (no. 2-1), and a cluster of 20 plasma cells (no. 1-2). All of these samples were captured from a single synovial section from patient A, within the same field of view in the microscope ocular (Fig. 3). For the cellular cluster, sample no. 1-2, a single V1-5:J1 sequence was present in 8 of 15 VL sequencing plasmids analyzed (>50%; the remaining seven plasmids contained diverse V:J rearrangements). Sample nos. 1-1 and 2-1 yielded a single and a pair of V1-5:J1 sequences, respectively. These V domains shared an identical CDR3 sequence, indicating that they were derived from a common B cell precursor, but exhibited a stepwise accumulation of nucleotide substitutions from germline (Fig. 3a). The majority of replacement mutations were clustered in the CDRs while a total of three FWR replacements (two conservative amino acid substitutions in FWR3 and a less conservative AlaThr mutation in FWR2) were observed in downstream members of the clone. A genealogical tree was constructed, based on the hierarchy of somatic mutations some of which are shared at each level, while others are unique to individual clone members (Fig. 3b). The high replacement:silent mutation (R/S) ratios in the CDRs, compared with FWRs, and highly significant multinomial p values associated with each chain (Fig. 3c) indicate that the cells synthesizing these V mRNAs are probably the members of an Ag-selected clone.
FIGURE 3. Intraclonal diversification of V regions in synovial plasma cells. a, Stepwise accumulation of somatic mutations in synovial V domains isolated from patient A. V1-5 (allele 03) was the closest germline match. Nucleotide and amino acid sequences of the germline gene segment are shown. CDRs 1, 2, and 3 and sections of the FWRs have been depicted. The sequences were identical through the rest of the V region. Intervening regions are indicated by dots. Dashes represent sequence identity. Amino acid replacements have been underlined. indicates conservation of hydrophilicity/hydrophobicity and * indicates conservation of size of the amino acid replacement (defined as being in the same hydropathy or volume class) (60 ). b, Genealogical tree and schematic representation of the areas from which each LCM sample (denoted by dotted rectangles) was isolated within one field of view in the microscope ocular (indicated by the double circle). 2-1K2 and 2-1K3 were both recovered from a single LCM sample (box 1 in the figure). 1-2K and 1-1K were identical sequences but were obtained from two separate LCM samples (boxes 2 and 3). c, R/S ratios in the FWRs and CDRs; p values, as determined by a multinomial distribution model (described in Materials and Methods). d, Clonally related sequences isolated from sample no. 2-1, identical throughout the VH domain, except for a single amino acid substitution, one residue upstream of CDR3.
Another such clonally related cluster (sample no. 8) recovered from this patient contained a total of three somatically mutated variants of a single V3-15:J4 recombination event (Fig. 4a). Identical VJ junctions and shared mutations in CDR1 (Asn Lys) and FWR2 (silent) suggest that 8K1 and 8K2 were derived from a common precursor. The sequences seem to have then diverged, each accumulating unique mutations. The 8K10 V region appears to have evolved from 8K2, diversifying further by acquiring additional amino acid replacements in CDR1. Notably, the positively charged lysine residue common to 8K1 and 8K2 was replaced by negatively charged aspartate in 8K10.
FIGURE 4. In situ diversification of V regions in synovial lesions. Clonally related sequences obtained from (a) a single LCM sample (no. 8) isolated from patient A, comprising V domains, and (b) a single LCM sample (no. 5-1) from patient B, comprising V domains. The junctional proline residue in CDR3 is boxed in a. Amino acid replacements are indicated on top of the corresponding nucleotide sequence and are underlined. indicates conservation of hydrophilicity/hydrophobicity and * indicates conservation of size of the amino acid replacement (60 ). c, Insets, Schematic representation of genealogical relationships between sequences. The unidentified, intermediary level is depicted in black.
Similar instances of V region diversification were also observed in patient B synovium. One such example included three separate -chains, all isolated from the same LCM sample of approximately four cells (sample no. 5-1; Fig. 4b). The 5-1L1 sequence recurred in four other LCM samples (nos. 3-2, 4-2, 4-3, and 10). These V domains appear to have arisen from a common precursor, containing one of the two V3-21:J3 junctions predominantly expressed in this patient (Jtn 1; Fig. 2).
Recurrent use of specific VH-VL combinations
Analysis of VH-VL combinations provided some examples of probable clonal expansion and some examples of convergence of nonclonal, but similar, V domain structures. Two groups of plasma cells (comprising 5 and 30 cells) isolated from patient A appeared to be clonal, each yielding a single VH-VL pair (Table VII).
Table VII. Clonal clusters of synovial plasma cells isolated from patient A
The related V3-15:J4 sequences (Fig. 4a) were coupled with a single rearrangement (VH1-3:D3-3:JH4), isolated from this probably clonal cluster (sample no. 8). The V1-5:J1 somatic variants (Fig. 3a) were found to be associated with related, but nonidentical H chains. The single plasma cell (no. 1-1) yielded a single H sequence, containing a VH3-7:D2-21:JH3 rearrangement. The plasma cell doublet (no. 2-1) contained two H sequences (VH3-7:D1-1:JH6), identical to each other except for a single replacement mutation, one residue upstream of the CDR3 region (Fig. 3d). This represents an instance of clonal expansion, wherein both H and L chains in one cell were inherited from the other, but with the accumulation of additional somatic mutations in the progeny. Finally, the larger cluster of cells (no. 1-2) yielded a total of five VH rearrangements, of which two used VH3-15. Because VH3-7 was used in V domains isolated from the other two samples, we repeated the H-PCR, using VH3 family specific primers alone. Of 10
additional plasmid sequences, 8 used VH3-15 (80%) and 2 used VH3-7 (recombined to D1-1:JH6). Assignments of H-L pairings are not possible within a cluster of cells. In the one and two cell samples (nos. 1-1 and 2-1), at least, VH3-7 appeared to be associated with the corresponding V1-5:J1 sequences. As observed with the junctions (mentioned in a previous section), it appeared that clonally derived VL domains were paired with similar, but not clonal VH domains. One explanation could be that the mRNAs encoding the relevant H chains were not detected by the PCR. This did not seem to be the case for both samples nos. 1-1 and 2-1, which yielded one and two H sequences, respectively. The second possibility is a remarkable convergence in the usage and somatic diversification of the -chains, potentially mediated by a strong, selective bias that seems to operate preferentially on VL domains.
Discussion
Classically, Lyme arthritis has been considered a hallmark of late-stage disease, developing months after disease onset. Arthritic episodes tend to be intermittent, self-limiting, and readily subside in response to antibiotics (4, 39). However, 10% of patients with arthritis develop chronic inflammation of the joints, which is resistant to standard courses of antibiotic treatment (40). At this stage, Bb DNA cannot usually be detected by PCR in either synovial tissue or fluid samples from affected patients (5). Lymphoid foci persist in the joints of chronic Lyme arthritis patients, long after the apparent elimination of Bb by antibiotic treatment. Our analysis of plasma cells isolated from synovial lesions reveals an ongoing process of maturation and restriction of the Ab response in the inflamed synovia of the patients examined, specifically, expansion and diversification of certain clones, somatic hypermutation, and Ag-mediated selection of Ig V domains.
The portal of entry of Bb is the dermis, at the site of infective tick attachment (3). The spirochete expresses a range of receptors that bind host components, allowing it to disseminate and seed other tissues, including the synovium (13). Bacterial lipoproteins act as potent inflammatory triggers for a variety of cells that are associated with joint pathology, such as neutrophils, monocytes, synovial fibroblasts, and endothelial cells (14, 15, 41, 42, 43). The strong innate response and resulting tissue injury elicited by the spirochete could have an adjuvant effect and set the stage for a later, self-perpetuating autoimmune reaction (44).
The occurrence of "treatment-resistant" Lyme arthritis correlates with an increased frequency of certain rheumatoid arthritis-associated MHC class II alleles, particularly HLA-DRB1*0401, as well as HLA-DRB1*0101, establishing a role for T cells and Ag presentation in chronic inflammation (45).
Molecular mimicry arising from a shared linear epitope between a Bb surface lipoprotein (outer surface protein A (OspA)) and human LFA-1 has been proposed in the pathogenesis of chronic Lyme arthritis (21), OspA-reactive T cells being significantly more abundant in the synovial fluid of treatment-resistant patients, compared with treatment-responsive ones (22).
Previous reports exploring Ab autoreactivity in chronic Lyme arthritis have focused mainly on serum Ig and isotype-specific rheumatoid factors (RFs). Elevated levels of IgA RFs, but not IgG or IgM RFs, have been found in 25% of Lyme arthritis patients, although they did not correlate with either severity or duration of arthritis (46). Early studies on the humoral response document a role for locally generated immune complexes in the pathophysiology of Lyme arthritis. Immune complexes in arthritis patients, as measured by C1q-binding assays, were found to disappear from circulation and gradually accumulate in the synovial fluid, as joint inflammation became progressively more chronic (47, 48). Histopathologic lesions in the affected joints of treatment-resistant patients contain nodular aggregates of T-, B-, and follicular dendritic cells, surrounded by plasma cell clusters (6). These pseudolymphoid follicles, also observed in rheumatoid and reactive arthritis (10), resemble germinal centers with respect to their cellular constituents, but lack the characteristic architecture of the latter (6). The persistence of such structures in the apparent absence of infection suggests that a local response, originally elicited by Bb, could be sustained by cross-reactive self-determinants that are exposed or up-regulated in the prevailing inflammatory milieu.
This study describes the repertoire and mutational status of Abs produced at the site of injury by synovial plasma cells, presumably the progeny of B cell precursors that were driven to hypermutate and differentiate via appropriate activation triggers. Although Bb lipoproteins, like Osps A and B, can be mitogenic for murine B cells, a similar, direct effect has not been documented in humans (49). Elevated total IgM levels in the blood and synovial fluid of early Lyme disease patients possibly arise due to the polyclonal activation of B cells by Bb during this initial phase (50). Our results reveal a somatically mutated, largely IgG response associated with the lesions that develop in chronic Lyme synovitis.
No single H or gene dominated the repertoire in either of the two patients examined. In patient B, 30% of the entire VL repertoire consisted of two prominent V3-21:J3 clones, representing a skewing of the response in the joint of this patient. Because the generation of V(D)J junctions is a stochastic process, identical rearrangements, along with identical junctions can be considered as markers of clonal origin. These are not typical clones–despite clonality in the usage of VL domains, the VH partners were disparate and did not share a common lineage. There could be several explanations for this. First, as mentioned previously, the H chain PCR might have failed to detect all possible VH rearrangements within a given sample. Although this might have occurred in the two large clusters (sample nos. 9 and 10), it did not seem the case for the single/two to four cell groups, where the number of recovered H chains matched the number of cells contained in the LCM sample. A second possibility could be the immunodominance of these V:J domains in mediating binding to a particular, as yet undetermined, Ag. A selective bias would then enable the expansion of cells expressing these L chains, based upon their contribution to the epitope binding site. Although there are several instances of VH restriction in different systems (such as human cold agglutinins) (51), examples of convergence in VL usage are relatively rare. Precedents for the latter have been observed in responses to Haemophilus influenzae type B polysaccharide and also in human fetal splenic B cells where such convergence was hypothesized to be mediated by self-Ag(s) (52, 53). Finally, a third possibility is suggested by recent reports of elevated levels of Ig-free L chains (IgLCs) in the synovial fluid of rheumatoid arthritis patients, as well as in other bodily fluids of patients with multiple sclerosis and Sjogren’s disease (54). IgLCs can bind Ag, albeit at lower affinities than tetrameric Ig, and mediate effector functions like activation of the alternative complement pathway, specific proteolysis, and mast cell-dependent hypersensitivity (54, 55). Potentially, the clonally related -chains observed in patient B could be secreted into the joint space and bind Ag independently of the VH domains. However, at this time, there are no reports of the detection or involvement of IgLCs in chronic Lyme arthritis.
The occurrence of affinity maturation was statistically inferred for each sequence, using a multinomial distribution model (28). A higher fraction of VL sequences than VH sequences appeared to have undergone Ag-driven selection (pcdr < 0.05; Table IV), reflecting a greater selective pressure on the former. We compared these results with those obtained in other documented affinity matured responses, applying the multinomial model to published Ig sequences in the NCBI database. Our values were similar to those calculated for several sets of hybridoma Abs generated in mice by primary immunization with the hapten 4-(hydroxyl-3-nitrophenyl)acetyl (NP) (1; Table VI) (56) or elicited by influenza virus hemagglutinin (HA) under a variety of primary and secondary immunization protocols (2 and 3; Table VI) (57, 58). The last example (4; Table VI) is from a study of the Ab response to the H. influenzae capsular polysaccharide–tetanus toxoid conjugate vaccine (HibCP-TT) in a human subject (59). The response had several unusual features, compared with conventional protein vaccines, such as a highly restricted V gene repertoire, virtually the entire response being dominated by the progeny of a few virgin B cells and extensive intraclonal affinity maturation, reflected by the high percentage of sequences with significant pcdr values (4; Table VI). In our study, the response was much more heterogeneous, possibly reflecting the involvement of more than one Ag, with unequal degrees of selection of different V regions within the repertoire. All of the five V1-5 rearrangements in patient A, 7 of 15 V3-21:J3 rearrangements from patient B, as well as both V3-21:J3 sequences from patient A had a significant excess of CDR replacements, indicating a selective expansion of cells expressing these V domains due to a possible Ag-mediated bias.
Several of the V regions discussed here contained replacement mutations in their FWRs, but the replacements appeared to be functionally conservative or occurred at sites that can accommodate substitutions without major effect on FWR structure (60). For example, in the V1-5:J1 somatic variants from patient A, the AlaThr transition at position 43 maintained the small residue size required at this site, the GlnLys replacement conserved the hydrophilic character of position 79, while position 95, which had a closely related ThrSer mutation, is not significantly conserved with regard to the nature/size of the amino acids that it can harbor (60). Conservation of either size, or hydrophobicity, or both have been indicated in Figs. 3 and 4.
Finally, the recovery of clonally derived sequences with a hierarchical pattern of mutations, all isolated from a localized tissue region, points toward a process of in situ differentiation. Such plasma cells must have arisen from precursors that divided and diversified within the tissue. Potentially, this local immune response, in the absence of infection, is sustained by putative self-Ag(s) that mimic a protein (or other moieties) in Bb. Borrelial flagellin was found to cross-react with heat shock protein 60, which in addition to being an axonal component, is also expressed in considerable amounts in the synovium, both under normal and pathologic conditions such as osteoarthritis and rheumatoid arthritis (61, 62). Another bacterial candidate for such mimicry is OspA, a highly immunogenic spirochetal lipoprotein which may be up-regulated in response to inflammatory cues (63). The development of both T cell and Ab reactivity to the protein can be correlated with the progression from episodic to chronic, treatment-resistant arthritis (22). In this study, we have demonstrated a process of diversification and, to some extent, restriction of the Ab repertoire within the arthritic lesions of two treatment-resistant Lyme arthritis patients. We have generated single chain V region fragments (scFvs) (64) from some of the H-L chain pairs that were recovered (our unpublished data). These scFvs will be useful as probes for the identification of potentially cross-reactive Ags in this model of chronic arthritis, where the etiologic agent that triggers disease is known. This would allow us to assess a pathogenic role for molecular mimicry between the initiating pathogen, Borrelia, and a putative autoantigen that sustains disease in a population of susceptible individuals.
Disclosures
The authors have no financial conflict of interest.
Acknowledgments
We thank Charles Vanderburg at the Harvard Center for Neurodegeneration and Repair and the Tufts University Imaging Facility for help with LCM. We also thank Michael Berne and his staff at the Tufts University Core Facility for all the sequencing work, Leanne Hoffman for technical assistance, Tatyana Vorobyova, Abbie L. Meyer, and Bettina P. Iliopoulou for helpful discussions.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by National Institutes of Health Grant AR45386, the Eshe Fund and the Center for Gastroenterology Research on Absorptive and Secretory Processes (MO1-RR0054).
2 Address correspondence and reprint requests to Dr. Brigitte T. Huber, Department of Pathology, Tufts University School of Medicine, 150 Harrison Avenue, Boston, MA 02111. E-mail address: brigitte.huber{at}tufts.edu
3 Abbreviations used in this paper: Bb, Borrelia burgdorferi; LCM, laser-capture microdissection; RT, reverse transcriptase; CDR, complementarity determining region; FWR, framework region; R/S, replacement:silent mutation; Osp, outer surface protein; RF, rheumatoid factor; IgLC, Ig-free L chain; NP, 4-(hydroxyl-3-nitrophenyl)acetyl; HA, hemagglutinin; scFv, single chain V region fragment.
Received for publication August 17, 2004. Accepted for publication December 6, 2004.
References
Ackermann, R., J. Kabatzki, H. P. Boisten, A. C. Steere, R. L. Grodzicki, S. Hartung, U. Runne. 1984. Ixodes ricinus spirochete and European erythema chronicum migrans disease. Yale J. Biol. Med. 57:573.
Ackermann, R., J. Kabatzki, H. P. Boisten, A. C. Steere, R. L. Grodzicki, S. Hartung, U. Runne. 1984. [Spirochete etiology of erythema chronicum migrans disease]. Dtsch. Med. Wochenschr. 109:92.
Rahn, D. W.. 1998. Natural history of Lyme disease. D. W. Rahn, and J. Evans, eds. Lyme Disease 35. American College of Physicians, Philadelphia.
Steere, A. C.. 2001. Lyme disease. N. Engl. J. Med. 345:115
Carlson, D., J. Hernandez, B. J. Bloom, J. Coburn, J. M. Aversa, A. C. Steere. 1999. Lack of Borrelia burgdorferi DNA in synovial samples from patients with antibiotic treatment-resistant Lyme arthritis. Arthritis Rheum. 42:2705
Steere, A. C., P. H. Duray, E. C. Butcher. 1988. Spirochetal antigens and lymphoid cell surface markers in Lyme synovitis: comparison with rheumatoid synovium and tonsillar lymphoid tissue. Arthritis Rheum. 31:487.
Zhang, Z., X. Wu, B. H. Limbaugh, S. L. Bridges, Jr. 2001. Expression of recombination-activating genes and terminal deoxynucleotidyl transferase and secondary rearrangement of immunoglobulin light chains in rheumatoid arthritis synovial tissue. Arthritis Rheum 44:2275.
Itoh, K., V. Patki, R. A. Furie, E. K. Chartash, R. I. Jain, L. Lane, S. E. Asnis, N. Chiorazzi. 2000. Clonal expression is a characteristic feature of the B-cell repetoire of patients with rheumatoid arthritis. Arthritis Res. 2:50.
Kim, H. J., V. Krenn, G. Steinhauser, C. Berek. 1999. Plasma cell development in synovial germinal centers in patients with rheumatoid and reactive arthritis. J. Immunol. 162:3053.
Berek, C., H. J. Kim. 1997. B-cell activation and development within chronically inflamed synovium in rheumatoid and reactive arthritis. Semin. Immunol. 9:261.
Burdin, N., L. Galibert, P. Garrone, I. Durand, J. Banchereau, F. Rousset. 1996. Inability to produce IL-6 is a functional feature of human germinal center B lymphocytes. J. Immunol. 156:4107
Liu, Y. J., J. Banchereau. 1996. Mutant mice without B lymphocyte follicles. J. Exp. Med. 184:1207.[Medline]
Emmert-Buck, M. R., R. F. Bonner, P. D. Smith, R. F. Chuaqui, Z. Zhuang, S. R. Goldstein, R. A. Weiss, L. A. Liotta. 1996. Laser capture microdissection. Science 274:998.
Gross, D. M., A. C. Steere, B. T. Huber. 1998. T helper 1 response is dominant and localized to the synovial fluid in patients with Lyme arthritis. J. Immunol. 160:1022.
Gross, D. M., T. Forsthuber, M. Tary-Lehmann, C. Etling, K. Ito, Z. A. Nagy, J. A. Field, A. C. Steere, B. T. Huber. 1998. Identification of LFA-1 as a candidate autoantigen in treatment- resistant Lyme arthritis. Science 281:703.
Trollmo, C., A. L. Meyer, A. C. Steere, D. A. Hafler, B. T. Huber. 2001. Molecular mimicry in Lyme arthritis demonstrated at the single cell level: LFA-1 L is a partial agonist for outer surface protein A-reactive T cells. J. Immunol. 166:5286.
Costes, V., V. Magen, E. Legouffe, L. Durand, P. Baldet, J. F. Rossi, B. Klein, J. Brochier. 1999. The Mi15 monoclonal antibody (anti-syndecan-1) is a reliable marker for quantifying plasma cells in paraffin-embedded bone marrow biopsy specimens. Hum. Pathol. 30:1405.
Stahlberg, A., P. Aman, B. Ridell, P. Mostad, M. Kubista. 2003. Quantitative real-time PCR method for detection of B-lymphocyte monoclonality by comparison of and immunoglobulin light chain expression. Clin. Chem. 49:51.
Foster, S. J., H. P. Brezinschek, R. I. Brezinschek, P. E. Lipsky. 1997. Molecular mechanisms and selective influences that shape the gene repertoire of IgM+ B cells. J. Clin. Invest. 99:1614.
Brezinschek, H. P., R. I. Brezinschek, P. E. Lipsky. 1995. Analysis of the heavy chain repertoire of human peripheral B cells using single-cell polymerase chain reaction. J. Immunol. 155:190.
Huang, C., A. K. Stewart, R. S. Schwartz, B. D. Stollar. 1992. Immunoglobulin heavy chain gene expression in peripheral blood B lymphocytes. J. Clin. Invest. 89:1331.
Guigou, V., A. M. Cuisinier, C. Tonnelle, D. Moinier, M. Fougereau, F. Fumoux. 1990. Human immunoglobulin VH and VK repertoire revealed by in situ hybridization. Mol. Immunol. 27:935.
Ignatovich, O., I. M. Tomlinson, P. T. Jones, G. Winter. 1997. The creation of diversity in the human immunoglobulin V() repertoire. J. Mol. Biol. 268:69.
Shlomchik, M. J., A. H. Aucoin, D. S. Pisetsky, M. G. Weigert. 1987. Structure and function of anti-DNA autoantibodies derived from a single autoimmune mouse. Proc. Natl. Acad. Sci. USA 84:9150.
Shlomchik, M. J., A. Marshak-Rothstein, C. B. Wolfowicz, T. L. Rothstein, M. G. Weigert. 1987. The role of clonal selection and somatic mutation in autoimmunity. Nature 328:805
Steere, A. C., R. T. Schoen, E. Taylor. 1987. The clinical evolution of Lyme arthritis. Ann. Intern. Med. 107:725.
Steere, A. C., R. E. Levin, P. J. Molloy, R. A. Kalish, J. H. r. Abraham, N. Y. Liu, C. H. Schmid. 1994. Treatment of Lyme arthritis. Arthritis Rheum. 37:878
Norgard, M. V., L. L. Arndt, D. R. Akins, L. L. Curetty, D. A. Harrich, J. D. Radolf. 1996. Activation of human monocytic cells by Treponema pallidum and Borrelia burgdorferi lipoproteins and synthetic lipopeptides proceeds via a pathway distinct from that of lipopolysaccharide but involves the transcriptional activator NF-B. Infect. Immun. 64:3845.
Wooten, R. M., T. B. Morrison, J. H. Weis, S. D. Wright, R. Thieringer, J. J. Weis. 1998. The role of CD14 in signaling mediated by outer membrane lipoproteins of Borrelia burgdorferi. J. Immunol. 160:5485.
Hardin, J. A., A. C. Steere, S. E. Malawista. 1984. The pathogenesis of arthritis in Lyme disease: humoral immune responses and the role of intra-articular immune complexes. Yale J. Biol. Med. 57:589.
Hardin, J. A., L. C. Walker, A. C. Steere, T. C. Trumble, K. S. Tung, R. C. Williams, Jr, S. Ruddy, S. E. Malawista. 1979. Circulating immune complexes in Lyme arthritis: detection by the 125I-C1q binding, C1q solid phase, and Raji cell assays. J. Clin. Invest. 63:468.
Ma, Y., J. J. Weis. 1993. Borrelia burgdorferi outer surface lipoproteins OspA and OspB possess B-cell mitogenic and cytokine-stimulatory properties. Infect. Immun. 61:3843.
Steere, A. C., J. A. Hardin, S. Ruddy, J. G. Mummaw, S. E. Malawista. 1979. Lyme arthritis: correlation of serum and cryoglobulin IgM with activity, and serum IgG with remission. Arthritis Rheum. 22:471.
Potter, K. N.. 2000. Molecular characterization of cold agglutinins. Transfus. Sci. 22:113.
Lucas, A. H., K. D. Moulton, D. C. Reason. 1998. Role of II-A2 light chain CDR-3 junctional residues in human antibody binding to the Haemophilus influenzae type b polysaccharide. J. Immunol. 161:3776.
Lee, J., N. L. Monson, P. E. Lipsky. 2000. The VJ repertoire in human fetal spleen: evidence for positive selection and extensive receptor editing. J. Immunol. 165:6322
Redegeld, F. A., F. P. Nijkamp. 2003. Immunoglobulin free light chains and mast cells: pivotal role in T-cell-mediated immune reactions?. Trends Immunol. 24:181.
Redegeld, F. A., M. W. van der Heijden, M. Kool, B. M. Heijdra, J. Garssen, A. D. Kraneveld, H. Van Loveren, P. Roholl, T. Saito, J. S. Verbeek, et al 2002. Immunoglobulin-free light chains elicit immediate hypersensitivity-like responses. Nat. Med. 8:694.
Boersch-Supan, M. E., S. Agarwal, M. E. White-Scharf, T. Imanishi-Kari. 1985. Heavy chain variable region: multiple gene segments encode anti-4-(hydroxy-3-nitro-phenyl)acetyl idiotypic antibodies. J. Exp. Med. 161:1272.
Caton, A. J., S. E. Stark, J. Kavaler, L. M. Staudt, D. Schwartz, W. Gerhard. 1991. Many variable region genes are utilized in the antibody response of BALB/c mice to the influenza virus A/PR/8/34 hemagglutinin. J. Immunol. 147:1675.
Clarke, S., R. Rickert, M. K. Wloch, L. Staudt, W. Gerhard, M. Weigert. 1990. The BALB/c secondary response to the Sb site of influenza virus hemagglutinin: nonrandom silent mutation and unequal numbers of VH and V mutations. J. Immunol. 145:2286.
Hougs, L., L. Juul, H. J. Ditzel, C. Heilmann, A. Svejgaard, T. Barington. 1999. The first dose of a Haemophilus influenzae type b conjugate vaccine reactivates memory B cells: evidence for extensive clonal selection, intraclonal affinity maturation, and multiple isotype switches to IgA2. J. Immunol. 162:224
Pommie, C., S. Levadoux, R. Sabatier, G. Lefranc, M. P. Lefranc. 2004. IMGT standardized criteria for statistical analysis of immunoglobulin V-REGION amino acid properties. J. Mol. Recognit. 17:17.
Schett, G., M. Tohidast-Akrad, G. Steiner, J. Smolen. 2001. The stressed synovium. Arthritis Res. 3:80.
Sigal, L. H., S. Williams, B. Soltys, R. Gupta. 2001. H9724, a monoclonal antibody to Borrelia burgdorferi’s flagellin, binds to heat shock protein 60 (HSP60) within live neuroblastoma cells: a potential role for HSP60 in peptide hormone signaling and in an autoimmune pathogenesis of the neuropathy of Lyme disease. Cell. Mol. Neurobiol. 21:477.
Crowley, H., B. T. Huber. 2003. Host-adapted Borrelia burgdorferi in mice expresses OspA during inflammation. Infect. Immun. 71:4003
Polymenis, M., B. D. Stollar. 1995. Single-chain anti-DNA FV. Methods Mol. Biol. 51:265
Farner, N. L., T. Dorner, P. E. Lipsky. 1999. Molecular mechanisms and selection influence the generation of the human VJ repertoire. J. Immunol. 162:2137.(Srimoyee Ghosh, Allen C. )
Lyme arthritis is initiated by the tick-borne spirochete, Borrelia burgdorferi. In a subset of patients, symptoms do not resolve in response to standard courses of antibiotics. Chronic joint inflammation may persist despite spirochetal killing, suggesting an autoimmune etiology. The pathogenic mechanisms that sustain chronic Lyme arthritis have not been fully elucidated, although T cells are believed to play a role. The synovial lesion contains elements of a peripheral lymph node, with lymphoid aggregates, plasma cells and follicular dendritic cells. An analysis of activated cells at the site of injury could yield clues regarding the nature of the response and the identity of potential autoantigens. Using laser-capture microdissection, we have isolated plasma cells from the joint tissue of chronic Lyme arthritis patients who underwent synovectomy. Expressed Ig V regions were amplified by RT-PCR. A majority of isolated cells expressed H chains, which is indicative of a class-switched response. There were a large number of nucleotide substitutions from germline, with a higher fraction of replacement mutations in the CDRs, suggesting a process of Ag-driven selection. We have recovered clonal clusters of cells containing identical junctions and V(D)J rearrangements. Sequence analysis reveals a hierarchy of shared somatic mutations between members of a given clone. Intraclonal diversity among plasma cells of close physical proximity points toward an ongoing process of diversification and affinity maturation, possibly driven by the chronic presence of an autoantigen.
Introduction
Lyme disease is a complex, multisystem illness caused by infection with the tick-borne spirochete Borrelia burgdorferi (Bb)3 (1, 2). Arthritis is a prominent, late-stage manifestation of the disease (3). In a subset of patients, arthritic symptoms do not resolve despite 2–3 mo courses of antibiotics, both oral and i.v. (4). Bb DNA usually cannot be detected in the synovial fluid and tissue of these "treatment-resistant" patients by PCR (5). Persistence of inflammation in the apparent absence of the initiating pathogen points toward an autoimmune phenomenon, downstream of the original bacterial infection.
Histologic examination of joint tissue from chronic Lyme arthritis patients who underwent synovectomy reveals a marked infiltration of mononuclear cells, with vascular proliferation and synovial hypertrophy. The arthritic lesion mimics that in rheumatoid arthritis and other chronic inflammatory arthritides. There are mixed aggregates of B cells, T cells, and follicular dendritic cells, surrounded by clusters containing large numbers of plasma cells (6). The cellular composition of synovial lymphocytic infiltrates is reminiscent of that observed in germinal centers in peripheral lymphoid organs. Data from rheumatoid arthritis have shown that V gene diversification, hypermutation, and terminal differentiation of B cells occur within synovial germinal center-like structures (7, 8, 9). This microenvironment seems to support longevity of plasma cells that accumulate in the chronically inflamed tissue (9, 10, 11, 12).
Bb uses various mechanisms of dissemination to home to specific sites in the body, including the joints (13). Spirochetal lipoproteins are potent inducers of proinflammatory cytokines like TNF- and IL-6 (14, 15, 16). In addition to promoting tissue damage, these cytokines may have a role in maintaining ectopic lymphoid tissue in the synovium (17, 18). As in rheumatoid or reactive arthritis, the joint becomes a site of intense inflammatory activity. The ectopic lymphoid follicles persist despite apparent spirochetal clearance. Hence, it is conceivable that cells within them are being triggered by a self-Ag, either joint-specific or exposed/up-regulated due to chronic tissue injury. We hypothesize that the inflammatory milieu in the diseased synovium supports a local immune response, characterized by plasma cell differentiation, somatic hypermutation, and the generation of affinity matured variants.
In this study, we have used laser-capture microdissection (LCM) to isolate single or groups of plasma cells from synovial tissue sections obtained from treatment-resistant Lyme arthritis patients (19). We have chosen to examine plasma cells because they represent terminal effectors derived from B cells that were recruited through appropriate activation signals and retained at the site of injury, as opposed to bystander or migrant B cells. Ig V regions were amplified by RT-PCR, enabling analysis of the expressed Ab repertoire in the context of the inflamed synovium. The data suggest a process of maturation and diversification of the Ab response, occurring in situ. Although such phenomena have been documented for both rheumatoid and reactive arthritis, this is the first description of a similar process in chronic Lyme arthritis, where T cells have been traditionally thought of as the dominant players, acting through a cascade of proinflammatory cytokines (20, 21, 22). Additionally, single cell RT-PCRs yield in vivo pairings of H and L chains that can be cloned together and expressed. This allows for the generation of Ab V region probes for the detection of potentially cross-reactive Ags that are relevant to disease pathogenesis.
Materials and Methods
Preparation of tissue samples
Patient tissue samples came from an archival synovial tissue collection, obtained from Lyme arthritis patients who had persistent arthritis after antibiotic treatment and had undergone synovectomy. Biopsy samples of inflamed tissue were extracted arthroscopically and snap-frozen in OCT (Miles Laboratories). Frozen tissue was cut into 6-μm sections in a cryostat and stored at –80°C. Synovial tissue from two patients, A and B, were selected for detailed analysis, based on a previous report documenting marked inflammatory and lymphocytic infiltration in the synovium (23). Duration of arthritis before antibiotic therapy was 1 mo for patient A and 6 mo for patient B. Both patients were treated with oral doxycholine for a period of 30 days, as well as oral amoxicillin (patient A) and i.v. ceftriaxone (patient B). After antibiotic therapy, synovial samples from A and B were negative in PCRs done to detect Bb DNA. Duration of arthritis posttreatment and before synovectomy was 4 mo in both of the cases.
Isolation of plasma cells
Plasma cells were identified by indirect immunofluorescence, using anti-CD138:FITC (Serotec) as primary Ab (1/50 dilution) and anti-fluorescein:Alexa 488 (Molecular Probes) as secondary Ab (1/100 dilution). Staining protocols were optimized for efficient LCM and maximal RNA recovery. Briefly, tissue sections were thawed and immediately fixed in cold acetone (4°C) for 4 min. Prolonged exposure to aqueous solutions can result in significant degradation of RNA. Therefore, relative to standard tissue-staining procedures, we used shorter Ab incubation times–10 min in primary Ab and 5 min in secondary Ab. Incubations were performed on a cold block (HistoGene Cold Block) to minimize background staining and in the presence of RNase inhibitor (RNase Out; Invitrogen Life Technologies) to preserve RNA integrity. The slides were washed in 1x PBS and dehydrated by sequential immersions in 75, 95, and 100% ethanol for 30 s each, followed by two 5-min washes in xylene. Staining was immediately followed by LCM (19), using a Pixcell IIe LCM instrument.
Noncoverslipped samples were visualized through the microscope ocular or on the live video monitor to locate areas of interest. Tightly clustered cells were not teased apart, to avoid mechanical tearing of the tissue. CD138+ cells were captured on High Sensitivity LCM caps (CapSure HS) either singly, in doublets/triplets, or as larger groups (ranging in size from 15 to 40 cells).
Extraction of captured material
A mixture containing 10 μl of 1x reverse transcriptase (RT) buffer, 1.4 μl of RNase inhibitor (40U/μl RNase Out; Invitrogen Life Technologies) and 3 μl of a 5% Nonidet P-40 solution was added to the LCM cap/microcentrifuge tube assembly, to lyse the captured cells on the transfer film. The assembly was inverted, spun at 4000 x g for 3 min and the extracted material was directly used in the RT-PCR. RT-PCR and subsequent rounds of PCR amplification were conducted, following the protocol and primers described by Wang et al. (24).
Synthesis of cDNA
cDNA was synthesized with a mixture of primers complementary to sequences in the C regions of μ-, -, -, and -chains (24). CμI, CI, CI, and CI, each at 5 pmol/μl, were added to the PCR tubes, which were incubated at 65°C for 3 min and then cooled to 25°C for 3 min. For the RT reaction, each tube received 2 μl of 5x RT buffer, 2 μl of 0.1 M DTT, 1 μl of 10 mM dNTP, and 0.5 μl of Superscript II RT (200 U/μl). Samples were incubated at 37°C for 1 h, heated to 70°C to inactivate the enzyme, and cooled to 4°C.
Amplification of Ig H and L chain V regions (VH and VL)
The cDNA was used as a template for two rounds of PCR amplification, using progressively nested gene or gene family specific primers (24). H and L chain PCRs were done separately. The first PCR was performed with a mixture of primers corresponding to conserved regions in the leader sequences of VH1-6, V1-4, or V1-3 families and corresponding CμII, CII, CII, or CII primers, complementary to sequences in the C domains.
Each amplification was performed in a 50-μl reaction volume containing 25 μl of 2x PFU Ultra Hotstart DNA Polymerase Master Mix, 1 μl of each H or L primer (at a working concentration of 20 pmol/μl), and 8 μl of the cDNA mixture. Cycling conditions included an initial 2-min denaturation step at 94°C, 3 cycles of preamplification (94°C for 45 s, 45°C for 45 s, 72°C for 1 min 45 s), and 30 cycles of amplification (94°C for 45 s, 50°C for 45 s, 72°C for 1 min 45 s). This was followed by a final 10-min incubation step at 72°C. In the second round, 3-μl aliquots from the first PCR were separately amplified for μ, , , or V regions, using the same cycling parameters (excluding the preamplification step). Primers used in the second PCR corresponded to the first six codons of the VH or VL family gene segments on the 5' end and more nested C region primers, CμIII, CIII, CIII, and CIII, on the 3' end. Amplified products run at around 400 bp on a 1% agarose gel. Gel bands were excised and purified, using the QIAquick gel extraction kit.
Yields of RNA from small numbers of cells, especially single cells, captured from a tissue section of 6-μm thickness, tend to be low. Therefore, sensitivity and efficiency of H and L PCRs were optimized at a single cell level, using hybridomas or lymphoblastoid cell lines as positive controls. In patient A, 9 of 14 (64%) laser-captured samples yielded both H and L chains. One sample (containing 10 cells) yielded only L chains and two single cell samples yielded H chains alone. Two LCM caps (also containing single cells) yielded neither an H nor an L product in the PCRs. A total of 24 LCM samples were obtained from patient B. Of these, 19 (79%) yielded both H and L chains while three samples (two were single cells and one contained a pair) gave only L chains. Two LCM caps did not yield any Ig product.
Sequence analysis
PCR products were cloned into the TOPO-TA sequencing vector (Invitrogen Life Technologies) and used to transform Escherichia coli. Plasmid minipreps were prepared (QIAprep; Qiagen) and sequenced, using the M13 reverse primer at the Tufts University Core Facility. The number of colonies picked for sequencing depended upon the number of plasma cells that yielded the corresponding PCR product. Five to 10 plasmids were sequenced for small groups (up to 4 cells) and 10–30 plasmids for larger groups (up to 40 cells). For the identification of the most homologous germline matches, VH and VL sequences were aligned against Ig genes in the IMGT database (25, 26) using the V-QUEST program (http://imgt.cines.fr/textes/vquest/). In addition, the Ig-basic local alignment search tool was used to perform searches within sequences listed in GenBank (www.ncbi.nlm.nih.gov/igblast/). D-segment assignations were based on a minimum of 10 nucleotide homology with the closest germline match. All sequences were deposited with GenBank under the accession numbers AY685251 through AY685387.
Analysis of mutations
Complementarity determining regions (CDRs) and framework regions (FWRs) were defined based on the Kabat numbering system (27). The first eight codons in FWR1, being primer-encoded, were excluded from analysis. Mutations were calculated for CDRs 1, 2, and 3 for VL sequences, and CDRs 1 and 2 for VH sequences. A multinomial distribution model (28) was used to assess the statistical significance of an excess of replacement mutations in CDRs and silent mutations in FWRs for each Ig V region. The corresponding germline gene sequence was imported into the JAVA applet at http://www-stat.stanford.edu/immunoglobulin provided by Lossos et al. (28) along with the associated set of r (number of replacement mutations) and s (number of silent mutations) values for FWRs and CDRs.
Results
Histology and immunofluorescence
Synovial tissue samples from two chronic Lyme arthritis patients (A and B; described in Materials and Methods) were used in this study. In H&E-stained sections, synovia from both patients exhibited marked lymphocytic infiltration (Fig. 1a). There were numerous lymphoid aggregates, some perivascular in localization. Plasma cells, identified by their extensive basophilic cytoplasm and low nucleus:cytoplasm ratio, were observed in clusters proximal to these aggregates or around the lumen of small blood vessels. By histologic staining, there appeared to be a larger number of plasma cells in patient A tissue than patient B tissue.
FIGURE 1. Histology and immunofluorescent detection of plasma cells in synovial tissue sections. a, H&E-stained synovial sections from patients A (i and ii) and B (iii and iv). P, plasma cell; L, lymphocyte; V, blood vessel. Pictures were taken at x40 magnification, except for i which was at x20. Lymphoid aggregates and proximal plasma cell groups are shown (i, ii, iii). iv, Large lymphocytic infiltrate in patient B synovium. b, Immunofluorescent detection (x50) of plasma cells by staining for CD138 (syndecan-1). A monoclonal anti-CD138:FITC was used as primary Ab and the FITC tag was detected by an anti-fluorescein:Alexa488 (green) Ab. The negative control was a nonspecific FITC-conjugated Ab (i and iv). Plasma cells were observed and laser-captured either singly/in small groups (indicated by arrows; ii and v) or as larger clusters (iii and vi).
For the purposes of LCM, we detected plasma cells by indirect immunofluorescence for syndecan-1 or CD138, a phenotypic marker (Fig. 1b). Distinct membrane staining for CD138 distinguishes plasma cells, which express this marker, from B cells, which do not (29, 30). CD138+ cells were distributed through the tissue singly or in small cellular groups, as well as in larger clusters, the latter being especially prominent in patient A. Synovial plasma cells were isolated by LCM and subjected to RT-PCR for the amplification of Ig H and L chain V region (VH and VL) sequences.
Repertoire of synovial plasma cells
Analysis of 59 unique H and 78 unique L sequences expressed within the arthritic lesions of both Lyme patients revealed a diverse, but nonrandom repertoire. In patient A, the VH repertoire was dominated by gene segments belonging to the VH3 family (10 of 18 sequences, 55%) (Table I). VH3-7 was the most prevalent, recurring four times. In patient B, the VH3 and VH4 families were about equally represented (Table I)–VH3-23 (five clones), VH3-30 (five clones), and VH4-61 (three clones) were the most recurrent individual gene segments. The majority of plasma cells isolated from patient synovia expressed H chains–100% in patient A and 66% in patient B (the remaining 34% expressed μ), which is indicative of an Ag-driven response. Interestingly, we identified six examples in which identical V regions were expressed with both μ and C regions, suggesting in situ class switching and the involvement of T cell-derived cytokines.
Table I. Distribution of VH and V families in the expressed H or repertoire of synovial plasma cells isolated from patients A and B
The VL repertoire from patient A mainly consisted of L chains, with a : ratio of 2.7 compared with the expected ratio of 1.5 for peripheral blood (31). V1 and V3 were the most commonly used gene families, and V1-5 (five clones) was the most recurrent gene segment (Table I). In synovial tissue from patient B, the VL repertoire had a much lower : ratio of 0.9, the V1 family being the most prominent among gene segments used (Table I). Neither the H nor the -chain distributions differed significantly from those that have already been reported for the normal, peripheral Ig repertoire (32, 33, 34, 35).
Restricted use of L chains
We observed a marked repertoire restriction in the V regions recovered from patient B. A total of 27 unique sequences were obtained from 3 single cells, 10 groups comprising 2–4 cells each and 2 clusters of 20 and 30 cells (summarized in Table II). Fifteen of the 27 (>55%) used a single V gene segment, V3-21. Two other segments, V1-47 and V1-44 each accounted for 19 and 15% of recombinants, respectively (5 and 4 of 27). These three gene segments together accounted for 89% of the captured repertoire. Furthermore, all 27 rearrangements from this patient used J3. For patient A, in whom the synovial Ig repertoire was dominated by -chains, a total of seven distinct sequences were recovered. Four of them contained V3-21, V1-47, or V1-44, the segments that were prominent in patient B, and all of them also contained J3.
Table II. Germline gene usage of V regions isolated from patient Ba
The 5'-V PCR primers used in our study were designed to identify gene segments belonging to the V1, 2, and 3 families, which account for >96% of expressed V gene segments in the peripheral repertoire (36). Under these PCR conditions, we detected just 2 and 6 rearrangements, respectively, in the 20- and 30-cell clusters of sample nos. 9 and 10 (Table II). Because this discrepancy between cell number and the number of unique sequences recovered did not occur with either the single cells or the two to four cell groups, samples 9 and 10 appeared to be oligoclonal cellular clusters. Of a total of seven J gene segments (organized as J-C pairs), four are considered to be functional (J 1, 2, 3, and 7). We used a 3' C primer which is complementary to a 16 nucleotide stretch common to all four J-C pairs, five codons downstream of the beginning of the C regions. Hence, there is no primer-induced bias at the 3' end. The marked overrepresentation of J3, V3-21, and V1-47 in these synovial samples was statistically significant, as assessed by a 2 test for "goodness-of-fit" between observed and expected frequencies (Table III).
Table III. Bias in the use of V and J gene segments in synovial plasma cells isolated from patient Ba
The V-J junctions isolated from patient B displayed limited diversity. Junctional analysis of the 15 V3-21:J3 sequences, isolated from multiple clusters, revealed two separate recombination events, each associated with a characteristic set of mutations and using different J3 alleles. This resulted in two distinct CDR3 sequences, with a similar number of nucleotide substitutions from germline but different amino acid replacements (Fig. 2, a and b). The other major V-J recombination group is comprised of V1-44:J3 (15%) and V1-47:J3 (19%) recombinants. Germline V1-44 and V1-47 gene segments share 97% identity, differing in CDR3 by a single amino acid–asparagine in V1-44 and serine in V1-47 (26). Hence, rearrangement to the same J3 gene segment yielded junctions with limited variability and a strong conservation of similar groups at the amino acid level (Fig. 2c). We also recovered two V1:J3 junctions from patient A, which displayed a high degree of homology to those observed in patient B synovium. Despite the convergent use of L chains, the corresponding H chain partners were diverse. Possibly, these junctions contribute key paratopic determinants to the ligand-binding site and an Ag-mediated bias allows for their selection, even when associated with different H chains.
FIGURE 2. Conserved junctions isolated from synovial plasma cells. a, Proposed mechanism of generation of V3-21:J3 junctions in patient B synovium. Arrows denote mutations from the corresponding germline gene segment. N-nucleotide insertion in junction (Jtn) 2 is indicated. Junctions were analyzed using the IMGT V-QUEST program (25 ). b, Amino acid homology between junctions derived from V3-21. Jtns 1 and 2 were present in 67 and 33% of V3-21:J3 recombinants, respectively. c, Homologous V1-47/44:J3 junctions. Jtn 3 was present in 33% and Jtns 4 and 5 were each present in 22% of V1:J3 sequences from patient B. [A]1 and 2 represent junctions isolated from patient A. An asterisk (*) indicates identity, double dots (:) denote a strong conservation of similar groups, and a single dot (.) denotes weak conservation of similar groups.
Mutational analysis of V regions
Somatic mutations are believed to occur randomly within the V regions. However, mutations that cause amino acid replacements could result in a selective advantage or disadvantage to the cells that harbor them (37, 38). Ag-mediated selection is evidenced by a higher frequency of replacement mutations in the CDRs, which bind Ag, relative to the FWRs, which maintain the -sheet scaffold of the V domain. Activated B cells thus undergo multiple rounds of mutation and selection, eventually resulting in Ig receptors with greatly enhanced affinity and selectivity for particular Ags, thereby fine tuning the response.
The majority of V regions isolated from patient synovia contained a large number of nucleotide substitutions from the closest germline matches. In general, the H chains appeared to be more mutated than the L chains (with an average of 26 and 16 nucleotide mutations each). The fraction of substitutions that resulted in amino acid replacements was calculated for all Ig sequences, recovered from both patients (Table IV). Overall, we observed a higher mean frequency of replacements in the CDRs, relative to the FWRs–a trend indicative of affinity maturation, as opposed to nonspecific, mitogenic expansion. This difference was statistically significant (p < 0.05; Student’s t test) for H and sequences from both patients and sequences from patient A. Further, we separately analyzed μ and H chain sequences from patient B synovium, where both isotypes were recovered. The sequences, but not μ, exhibited a significantly higher frequency of replacement mutations in the CDRs compared with the FWRs (p < 0.02; Student’s t test) (Table V). Thus, Ig V regions belonging to the traditionally more pathogenic IgG isotype appear to be more Ag selected than those belonging to the IgM isotype.
Table IV. Mean fractions of replacement mutations (Rfrn) within H or L CDRs and FWRsa
Table V. Mean fractions of replacement mutations (Rfrn) within μ or CDRs and FWRs in patient B
We used a multinomial distribution model to estimate whether Ag-selective pressure was operating on individual V regions (28). The model equation was used to determine whether the clustering of replacement mutations in CDRs and their relative scarcity in FWRs occurred at a frequency greater than that expected by random chance alone, taking into account the inherent susceptibility for amino acid substitutions in the corresponding germline gene segment, based on its codon composition. Because Ig mutations can have a total of four distribution possibilities in this model (replacement or silent in FWRs or CDRs), we obtained two separate sets of p values, characterizing each of the two regions (pfwr and pcdr) (Table IV). As expected, the majority of synovial H and L chain FWRs (>80%) had a paucity of replacement and an excess of silent mutations (pfwr < 0.05), consistent with requirements for the maintenance of structural integrity of the V domain. Of these sequences, an average of 26, 45, and 37% of H, -, and -chains respectively, preferentially accumulated replacement mutations in their CDRs (pcdr and pfwr < 0.05) (Table IV), indicative of ligand-dependent selection. Hence, the clustering of replacement mutations within the synovial Ig repertoire followed a nonrandom pattern, being significantly scarce in the FWRs of most V regions and abundant in the CDRs of a smaller population of sequences. The latter group comprises the subset of V regions that seem to have undergone affinity maturation in each patient. The multinomial analysis yielded similar results when applied to published sequences associated with well-defined, Ag-selected Ab sets, induced by immunization (Table VI).
Table VI. Application of the multinomial model to published Ig sequences deposited in the NCBI databasea
Intraclonal diversification of V domains
We recovered a number of clonally related V regions with identical V(D)J rearrangements and junctions, i.e., CDR3 sequences, but different somatic mutations. For example, a single V1-5:J1 recombination event recurred in three separate LCM samples within a localized tissue area–a single plasma cell (no. 1-1), a doublet (no. 2-1), and a cluster of 20 plasma cells (no. 1-2). All of these samples were captured from a single synovial section from patient A, within the same field of view in the microscope ocular (Fig. 3). For the cellular cluster, sample no. 1-2, a single V1-5:J1 sequence was present in 8 of 15 VL sequencing plasmids analyzed (>50%; the remaining seven plasmids contained diverse V:J rearrangements). Sample nos. 1-1 and 2-1 yielded a single and a pair of V1-5:J1 sequences, respectively. These V domains shared an identical CDR3 sequence, indicating that they were derived from a common B cell precursor, but exhibited a stepwise accumulation of nucleotide substitutions from germline (Fig. 3a). The majority of replacement mutations were clustered in the CDRs while a total of three FWR replacements (two conservative amino acid substitutions in FWR3 and a less conservative AlaThr mutation in FWR2) were observed in downstream members of the clone. A genealogical tree was constructed, based on the hierarchy of somatic mutations some of which are shared at each level, while others are unique to individual clone members (Fig. 3b). The high replacement:silent mutation (R/S) ratios in the CDRs, compared with FWRs, and highly significant multinomial p values associated with each chain (Fig. 3c) indicate that the cells synthesizing these V mRNAs are probably the members of an Ag-selected clone.
FIGURE 3. Intraclonal diversification of V regions in synovial plasma cells. a, Stepwise accumulation of somatic mutations in synovial V domains isolated from patient A. V1-5 (allele 03) was the closest germline match. Nucleotide and amino acid sequences of the germline gene segment are shown. CDRs 1, 2, and 3 and sections of the FWRs have been depicted. The sequences were identical through the rest of the V region. Intervening regions are indicated by dots. Dashes represent sequence identity. Amino acid replacements have been underlined. indicates conservation of hydrophilicity/hydrophobicity and * indicates conservation of size of the amino acid replacement (defined as being in the same hydropathy or volume class) (60 ). b, Genealogical tree and schematic representation of the areas from which each LCM sample (denoted by dotted rectangles) was isolated within one field of view in the microscope ocular (indicated by the double circle). 2-1K2 and 2-1K3 were both recovered from a single LCM sample (box 1 in the figure). 1-2K and 1-1K were identical sequences but were obtained from two separate LCM samples (boxes 2 and 3). c, R/S ratios in the FWRs and CDRs; p values, as determined by a multinomial distribution model (described in Materials and Methods). d, Clonally related sequences isolated from sample no. 2-1, identical throughout the VH domain, except for a single amino acid substitution, one residue upstream of CDR3.
Another such clonally related cluster (sample no. 8) recovered from this patient contained a total of three somatically mutated variants of a single V3-15:J4 recombination event (Fig. 4a). Identical VJ junctions and shared mutations in CDR1 (Asn Lys) and FWR2 (silent) suggest that 8K1 and 8K2 were derived from a common precursor. The sequences seem to have then diverged, each accumulating unique mutations. The 8K10 V region appears to have evolved from 8K2, diversifying further by acquiring additional amino acid replacements in CDR1. Notably, the positively charged lysine residue common to 8K1 and 8K2 was replaced by negatively charged aspartate in 8K10.
FIGURE 4. In situ diversification of V regions in synovial lesions. Clonally related sequences obtained from (a) a single LCM sample (no. 8) isolated from patient A, comprising V domains, and (b) a single LCM sample (no. 5-1) from patient B, comprising V domains. The junctional proline residue in CDR3 is boxed in a. Amino acid replacements are indicated on top of the corresponding nucleotide sequence and are underlined. indicates conservation of hydrophilicity/hydrophobicity and * indicates conservation of size of the amino acid replacement (60 ). c, Insets, Schematic representation of genealogical relationships between sequences. The unidentified, intermediary level is depicted in black.
Similar instances of V region diversification were also observed in patient B synovium. One such example included three separate -chains, all isolated from the same LCM sample of approximately four cells (sample no. 5-1; Fig. 4b). The 5-1L1 sequence recurred in four other LCM samples (nos. 3-2, 4-2, 4-3, and 10). These V domains appear to have arisen from a common precursor, containing one of the two V3-21:J3 junctions predominantly expressed in this patient (Jtn 1; Fig. 2).
Recurrent use of specific VH-VL combinations
Analysis of VH-VL combinations provided some examples of probable clonal expansion and some examples of convergence of nonclonal, but similar, V domain structures. Two groups of plasma cells (comprising 5 and 30 cells) isolated from patient A appeared to be clonal, each yielding a single VH-VL pair (Table VII).
Table VII. Clonal clusters of synovial plasma cells isolated from patient A
The related V3-15:J4 sequences (Fig. 4a) were coupled with a single rearrangement (VH1-3:D3-3:JH4), isolated from this probably clonal cluster (sample no. 8). The V1-5:J1 somatic variants (Fig. 3a) were found to be associated with related, but nonidentical H chains. The single plasma cell (no. 1-1) yielded a single H sequence, containing a VH3-7:D2-21:JH3 rearrangement. The plasma cell doublet (no. 2-1) contained two H sequences (VH3-7:D1-1:JH6), identical to each other except for a single replacement mutation, one residue upstream of the CDR3 region (Fig. 3d). This represents an instance of clonal expansion, wherein both H and L chains in one cell were inherited from the other, but with the accumulation of additional somatic mutations in the progeny. Finally, the larger cluster of cells (no. 1-2) yielded a total of five VH rearrangements, of which two used VH3-15. Because VH3-7 was used in V domains isolated from the other two samples, we repeated the H-PCR, using VH3 family specific primers alone. Of 10
additional plasmid sequences, 8 used VH3-15 (80%) and 2 used VH3-7 (recombined to D1-1:JH6). Assignments of H-L pairings are not possible within a cluster of cells. In the one and two cell samples (nos. 1-1 and 2-1), at least, VH3-7 appeared to be associated with the corresponding V1-5:J1 sequences. As observed with the junctions (mentioned in a previous section), it appeared that clonally derived VL domains were paired with similar, but not clonal VH domains. One explanation could be that the mRNAs encoding the relevant H chains were not detected by the PCR. This did not seem to be the case for both samples nos. 1-1 and 2-1, which yielded one and two H sequences, respectively. The second possibility is a remarkable convergence in the usage and somatic diversification of the -chains, potentially mediated by a strong, selective bias that seems to operate preferentially on VL domains.
Discussion
Classically, Lyme arthritis has been considered a hallmark of late-stage disease, developing months after disease onset. Arthritic episodes tend to be intermittent, self-limiting, and readily subside in response to antibiotics (4, 39). However, 10% of patients with arthritis develop chronic inflammation of the joints, which is resistant to standard courses of antibiotic treatment (40). At this stage, Bb DNA cannot usually be detected by PCR in either synovial tissue or fluid samples from affected patients (5). Lymphoid foci persist in the joints of chronic Lyme arthritis patients, long after the apparent elimination of Bb by antibiotic treatment. Our analysis of plasma cells isolated from synovial lesions reveals an ongoing process of maturation and restriction of the Ab response in the inflamed synovia of the patients examined, specifically, expansion and diversification of certain clones, somatic hypermutation, and Ag-mediated selection of Ig V domains.
The portal of entry of Bb is the dermis, at the site of infective tick attachment (3). The spirochete expresses a range of receptors that bind host components, allowing it to disseminate and seed other tissues, including the synovium (13). Bacterial lipoproteins act as potent inflammatory triggers for a variety of cells that are associated with joint pathology, such as neutrophils, monocytes, synovial fibroblasts, and endothelial cells (14, 15, 41, 42, 43). The strong innate response and resulting tissue injury elicited by the spirochete could have an adjuvant effect and set the stage for a later, self-perpetuating autoimmune reaction (44).
The occurrence of "treatment-resistant" Lyme arthritis correlates with an increased frequency of certain rheumatoid arthritis-associated MHC class II alleles, particularly HLA-DRB1*0401, as well as HLA-DRB1*0101, establishing a role for T cells and Ag presentation in chronic inflammation (45).
Molecular mimicry arising from a shared linear epitope between a Bb surface lipoprotein (outer surface protein A (OspA)) and human LFA-1 has been proposed in the pathogenesis of chronic Lyme arthritis (21), OspA-reactive T cells being significantly more abundant in the synovial fluid of treatment-resistant patients, compared with treatment-responsive ones (22).
Previous reports exploring Ab autoreactivity in chronic Lyme arthritis have focused mainly on serum Ig and isotype-specific rheumatoid factors (RFs). Elevated levels of IgA RFs, but not IgG or IgM RFs, have been found in 25% of Lyme arthritis patients, although they did not correlate with either severity or duration of arthritis (46). Early studies on the humoral response document a role for locally generated immune complexes in the pathophysiology of Lyme arthritis. Immune complexes in arthritis patients, as measured by C1q-binding assays, were found to disappear from circulation and gradually accumulate in the synovial fluid, as joint inflammation became progressively more chronic (47, 48). Histopathologic lesions in the affected joints of treatment-resistant patients contain nodular aggregates of T-, B-, and follicular dendritic cells, surrounded by plasma cell clusters (6). These pseudolymphoid follicles, also observed in rheumatoid and reactive arthritis (10), resemble germinal centers with respect to their cellular constituents, but lack the characteristic architecture of the latter (6). The persistence of such structures in the apparent absence of infection suggests that a local response, originally elicited by Bb, could be sustained by cross-reactive self-determinants that are exposed or up-regulated in the prevailing inflammatory milieu.
This study describes the repertoire and mutational status of Abs produced at the site of injury by synovial plasma cells, presumably the progeny of B cell precursors that were driven to hypermutate and differentiate via appropriate activation triggers. Although Bb lipoproteins, like Osps A and B, can be mitogenic for murine B cells, a similar, direct effect has not been documented in humans (49). Elevated total IgM levels in the blood and synovial fluid of early Lyme disease patients possibly arise due to the polyclonal activation of B cells by Bb during this initial phase (50). Our results reveal a somatically mutated, largely IgG response associated with the lesions that develop in chronic Lyme synovitis.
No single H or gene dominated the repertoire in either of the two patients examined. In patient B, 30% of the entire VL repertoire consisted of two prominent V3-21:J3 clones, representing a skewing of the response in the joint of this patient. Because the generation of V(D)J junctions is a stochastic process, identical rearrangements, along with identical junctions can be considered as markers of clonal origin. These are not typical clones–despite clonality in the usage of VL domains, the VH partners were disparate and did not share a common lineage. There could be several explanations for this. First, as mentioned previously, the H chain PCR might have failed to detect all possible VH rearrangements within a given sample. Although this might have occurred in the two large clusters (sample nos. 9 and 10), it did not seem the case for the single/two to four cell groups, where the number of recovered H chains matched the number of cells contained in the LCM sample. A second possibility could be the immunodominance of these V:J domains in mediating binding to a particular, as yet undetermined, Ag. A selective bias would then enable the expansion of cells expressing these L chains, based upon their contribution to the epitope binding site. Although there are several instances of VH restriction in different systems (such as human cold agglutinins) (51), examples of convergence in VL usage are relatively rare. Precedents for the latter have been observed in responses to Haemophilus influenzae type B polysaccharide and also in human fetal splenic B cells where such convergence was hypothesized to be mediated by self-Ag(s) (52, 53). Finally, a third possibility is suggested by recent reports of elevated levels of Ig-free L chains (IgLCs) in the synovial fluid of rheumatoid arthritis patients, as well as in other bodily fluids of patients with multiple sclerosis and Sjogren’s disease (54). IgLCs can bind Ag, albeit at lower affinities than tetrameric Ig, and mediate effector functions like activation of the alternative complement pathway, specific proteolysis, and mast cell-dependent hypersensitivity (54, 55). Potentially, the clonally related -chains observed in patient B could be secreted into the joint space and bind Ag independently of the VH domains. However, at this time, there are no reports of the detection or involvement of IgLCs in chronic Lyme arthritis.
The occurrence of affinity maturation was statistically inferred for each sequence, using a multinomial distribution model (28). A higher fraction of VL sequences than VH sequences appeared to have undergone Ag-driven selection (pcdr < 0.05; Table IV), reflecting a greater selective pressure on the former. We compared these results with those obtained in other documented affinity matured responses, applying the multinomial model to published Ig sequences in the NCBI database. Our values were similar to those calculated for several sets of hybridoma Abs generated in mice by primary immunization with the hapten 4-(hydroxyl-3-nitrophenyl)acetyl (NP) (1; Table VI) (56) or elicited by influenza virus hemagglutinin (HA) under a variety of primary and secondary immunization protocols (2 and 3; Table VI) (57, 58). The last example (4; Table VI) is from a study of the Ab response to the H. influenzae capsular polysaccharide–tetanus toxoid conjugate vaccine (HibCP-TT) in a human subject (59). The response had several unusual features, compared with conventional protein vaccines, such as a highly restricted V gene repertoire, virtually the entire response being dominated by the progeny of a few virgin B cells and extensive intraclonal affinity maturation, reflected by the high percentage of sequences with significant pcdr values (4; Table VI). In our study, the response was much more heterogeneous, possibly reflecting the involvement of more than one Ag, with unequal degrees of selection of different V regions within the repertoire. All of the five V1-5 rearrangements in patient A, 7 of 15 V3-21:J3 rearrangements from patient B, as well as both V3-21:J3 sequences from patient A had a significant excess of CDR replacements, indicating a selective expansion of cells expressing these V domains due to a possible Ag-mediated bias.
Several of the V regions discussed here contained replacement mutations in their FWRs, but the replacements appeared to be functionally conservative or occurred at sites that can accommodate substitutions without major effect on FWR structure (60). For example, in the V1-5:J1 somatic variants from patient A, the AlaThr transition at position 43 maintained the small residue size required at this site, the GlnLys replacement conserved the hydrophilic character of position 79, while position 95, which had a closely related ThrSer mutation, is not significantly conserved with regard to the nature/size of the amino acids that it can harbor (60). Conservation of either size, or hydrophobicity, or both have been indicated in Figs. 3 and 4.
Finally, the recovery of clonally derived sequences with a hierarchical pattern of mutations, all isolated from a localized tissue region, points toward a process of in situ differentiation. Such plasma cells must have arisen from precursors that divided and diversified within the tissue. Potentially, this local immune response, in the absence of infection, is sustained by putative self-Ag(s) that mimic a protein (or other moieties) in Bb. Borrelial flagellin was found to cross-react with heat shock protein 60, which in addition to being an axonal component, is also expressed in considerable amounts in the synovium, both under normal and pathologic conditions such as osteoarthritis and rheumatoid arthritis (61, 62). Another bacterial candidate for such mimicry is OspA, a highly immunogenic spirochetal lipoprotein which may be up-regulated in response to inflammatory cues (63). The development of both T cell and Ab reactivity to the protein can be correlated with the progression from episodic to chronic, treatment-resistant arthritis (22). In this study, we have demonstrated a process of diversification and, to some extent, restriction of the Ab repertoire within the arthritic lesions of two treatment-resistant Lyme arthritis patients. We have generated single chain V region fragments (scFvs) (64) from some of the H-L chain pairs that were recovered (our unpublished data). These scFvs will be useful as probes for the identification of potentially cross-reactive Ags in this model of chronic arthritis, where the etiologic agent that triggers disease is known. This would allow us to assess a pathogenic role for molecular mimicry between the initiating pathogen, Borrelia, and a putative autoantigen that sustains disease in a population of susceptible individuals.
Disclosures
The authors have no financial conflict of interest.
Acknowledgments
We thank Charles Vanderburg at the Harvard Center for Neurodegeneration and Repair and the Tufts University Imaging Facility for help with LCM. We also thank Michael Berne and his staff at the Tufts University Core Facility for all the sequencing work, Leanne Hoffman for technical assistance, Tatyana Vorobyova, Abbie L. Meyer, and Bettina P. Iliopoulou for helpful discussions.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by National Institutes of Health Grant AR45386, the Eshe Fund and the Center for Gastroenterology Research on Absorptive and Secretory Processes (MO1-RR0054).
2 Address correspondence and reprint requests to Dr. Brigitte T. Huber, Department of Pathology, Tufts University School of Medicine, 150 Harrison Avenue, Boston, MA 02111. E-mail address: brigitte.huber{at}tufts.edu
3 Abbreviations used in this paper: Bb, Borrelia burgdorferi; LCM, laser-capture microdissection; RT, reverse transcriptase; CDR, complementarity determining region; FWR, framework region; R/S, replacement:silent mutation; Osp, outer surface protein; RF, rheumatoid factor; IgLC, Ig-free L chain; NP, 4-(hydroxyl-3-nitrophenyl)acetyl; HA, hemagglutinin; scFv, single chain V region fragment.
Received for publication August 17, 2004. Accepted for publication December 6, 2004.
References
Ackermann, R., J. Kabatzki, H. P. Boisten, A. C. Steere, R. L. Grodzicki, S. Hartung, U. Runne. 1984. Ixodes ricinus spirochete and European erythema chronicum migrans disease. Yale J. Biol. Med. 57:573.
Ackermann, R., J. Kabatzki, H. P. Boisten, A. C. Steere, R. L. Grodzicki, S. Hartung, U. Runne. 1984. [Spirochete etiology of erythema chronicum migrans disease]. Dtsch. Med. Wochenschr. 109:92.
Rahn, D. W.. 1998. Natural history of Lyme disease. D. W. Rahn, and J. Evans, eds. Lyme Disease 35. American College of Physicians, Philadelphia.
Steere, A. C.. 2001. Lyme disease. N. Engl. J. Med. 345:115
Carlson, D., J. Hernandez, B. J. Bloom, J. Coburn, J. M. Aversa, A. C. Steere. 1999. Lack of Borrelia burgdorferi DNA in synovial samples from patients with antibiotic treatment-resistant Lyme arthritis. Arthritis Rheum. 42:2705
Steere, A. C., P. H. Duray, E. C. Butcher. 1988. Spirochetal antigens and lymphoid cell surface markers in Lyme synovitis: comparison with rheumatoid synovium and tonsillar lymphoid tissue. Arthritis Rheum. 31:487.
Zhang, Z., X. Wu, B. H. Limbaugh, S. L. Bridges, Jr. 2001. Expression of recombination-activating genes and terminal deoxynucleotidyl transferase and secondary rearrangement of immunoglobulin light chains in rheumatoid arthritis synovial tissue. Arthritis Rheum 44:2275.
Itoh, K., V. Patki, R. A. Furie, E. K. Chartash, R. I. Jain, L. Lane, S. E. Asnis, N. Chiorazzi. 2000. Clonal expression is a characteristic feature of the B-cell repetoire of patients with rheumatoid arthritis. Arthritis Res. 2:50.
Kim, H. J., V. Krenn, G. Steinhauser, C. Berek. 1999. Plasma cell development in synovial germinal centers in patients with rheumatoid and reactive arthritis. J. Immunol. 162:3053.
Berek, C., H. J. Kim. 1997. B-cell activation and development within chronically inflamed synovium in rheumatoid and reactive arthritis. Semin. Immunol. 9:261.
Burdin, N., L. Galibert, P. Garrone, I. Durand, J. Banchereau, F. Rousset. 1996. Inability to produce IL-6 is a functional feature of human germinal center B lymphocytes. J. Immunol. 156:4107
Liu, Y. J., J. Banchereau. 1996. Mutant mice without B lymphocyte follicles. J. Exp. Med. 184:1207.[Medline]
Emmert-Buck, M. R., R. F. Bonner, P. D. Smith, R. F. Chuaqui, Z. Zhuang, S. R. Goldstein, R. A. Weiss, L. A. Liotta. 1996. Laser capture microdissection. Science 274:998.
Gross, D. M., A. C. Steere, B. T. Huber. 1998. T helper 1 response is dominant and localized to the synovial fluid in patients with Lyme arthritis. J. Immunol. 160:1022.
Gross, D. M., T. Forsthuber, M. Tary-Lehmann, C. Etling, K. Ito, Z. A. Nagy, J. A. Field, A. C. Steere, B. T. Huber. 1998. Identification of LFA-1 as a candidate autoantigen in treatment- resistant Lyme arthritis. Science 281:703.
Trollmo, C., A. L. Meyer, A. C. Steere, D. A. Hafler, B. T. Huber. 2001. Molecular mimicry in Lyme arthritis demonstrated at the single cell level: LFA-1 L is a partial agonist for outer surface protein A-reactive T cells. J. Immunol. 166:5286.
Costes, V., V. Magen, E. Legouffe, L. Durand, P. Baldet, J. F. Rossi, B. Klein, J. Brochier. 1999. The Mi15 monoclonal antibody (anti-syndecan-1) is a reliable marker for quantifying plasma cells in paraffin-embedded bone marrow biopsy specimens. Hum. Pathol. 30:1405.
Stahlberg, A., P. Aman, B. Ridell, P. Mostad, M. Kubista. 2003. Quantitative real-time PCR method for detection of B-lymphocyte monoclonality by comparison of and immunoglobulin light chain expression. Clin. Chem. 49:51.
Foster, S. J., H. P. Brezinschek, R. I. Brezinschek, P. E. Lipsky. 1997. Molecular mechanisms and selective influences that shape the gene repertoire of IgM+ B cells. J. Clin. Invest. 99:1614.
Brezinschek, H. P., R. I. Brezinschek, P. E. Lipsky. 1995. Analysis of the heavy chain repertoire of human peripheral B cells using single-cell polymerase chain reaction. J. Immunol. 155:190.
Huang, C., A. K. Stewart, R. S. Schwartz, B. D. Stollar. 1992. Immunoglobulin heavy chain gene expression in peripheral blood B lymphocytes. J. Clin. Invest. 89:1331.
Guigou, V., A. M. Cuisinier, C. Tonnelle, D. Moinier, M. Fougereau, F. Fumoux. 1990. Human immunoglobulin VH and VK repertoire revealed by in situ hybridization. Mol. Immunol. 27:935.
Ignatovich, O., I. M. Tomlinson, P. T. Jones, G. Winter. 1997. The creation of diversity in the human immunoglobulin V() repertoire. J. Mol. Biol. 268:69.
Shlomchik, M. J., A. H. Aucoin, D. S. Pisetsky, M. G. Weigert. 1987. Structure and function of anti-DNA autoantibodies derived from a single autoimmune mouse. Proc. Natl. Acad. Sci. USA 84:9150.
Shlomchik, M. J., A. Marshak-Rothstein, C. B. Wolfowicz, T. L. Rothstein, M. G. Weigert. 1987. The role of clonal selection and somatic mutation in autoimmunity. Nature 328:805
Steere, A. C., R. T. Schoen, E. Taylor. 1987. The clinical evolution of Lyme arthritis. Ann. Intern. Med. 107:725.
Steere, A. C., R. E. Levin, P. J. Molloy, R. A. Kalish, J. H. r. Abraham, N. Y. Liu, C. H. Schmid. 1994. Treatment of Lyme arthritis. Arthritis Rheum. 37:878
Norgard, M. V., L. L. Arndt, D. R. Akins, L. L. Curetty, D. A. Harrich, J. D. Radolf. 1996. Activation of human monocytic cells by Treponema pallidum and Borrelia burgdorferi lipoproteins and synthetic lipopeptides proceeds via a pathway distinct from that of lipopolysaccharide but involves the transcriptional activator NF-B. Infect. Immun. 64:3845.
Wooten, R. M., T. B. Morrison, J. H. Weis, S. D. Wright, R. Thieringer, J. J. Weis. 1998. The role of CD14 in signaling mediated by outer membrane lipoproteins of Borrelia burgdorferi. J. Immunol. 160:5485.
Hardin, J. A., A. C. Steere, S. E. Malawista. 1984. The pathogenesis of arthritis in Lyme disease: humoral immune responses and the role of intra-articular immune complexes. Yale J. Biol. Med. 57:589.
Hardin, J. A., L. C. Walker, A. C. Steere, T. C. Trumble, K. S. Tung, R. C. Williams, Jr, S. Ruddy, S. E. Malawista. 1979. Circulating immune complexes in Lyme arthritis: detection by the 125I-C1q binding, C1q solid phase, and Raji cell assays. J. Clin. Invest. 63:468.
Ma, Y., J. J. Weis. 1993. Borrelia burgdorferi outer surface lipoproteins OspA and OspB possess B-cell mitogenic and cytokine-stimulatory properties. Infect. Immun. 61:3843.
Steere, A. C., J. A. Hardin, S. Ruddy, J. G. Mummaw, S. E. Malawista. 1979. Lyme arthritis: correlation of serum and cryoglobulin IgM with activity, and serum IgG with remission. Arthritis Rheum. 22:471.
Potter, K. N.. 2000. Molecular characterization of cold agglutinins. Transfus. Sci. 22:113.
Lucas, A. H., K. D. Moulton, D. C. Reason. 1998. Role of II-A2 light chain CDR-3 junctional residues in human antibody binding to the Haemophilus influenzae type b polysaccharide. J. Immunol. 161:3776.
Lee, J., N. L. Monson, P. E. Lipsky. 2000. The VJ repertoire in human fetal spleen: evidence for positive selection and extensive receptor editing. J. Immunol. 165:6322
Redegeld, F. A., F. P. Nijkamp. 2003. Immunoglobulin free light chains and mast cells: pivotal role in T-cell-mediated immune reactions?. Trends Immunol. 24:181.
Redegeld, F. A., M. W. van der Heijden, M. Kool, B. M. Heijdra, J. Garssen, A. D. Kraneveld, H. Van Loveren, P. Roholl, T. Saito, J. S. Verbeek, et al 2002. Immunoglobulin-free light chains elicit immediate hypersensitivity-like responses. Nat. Med. 8:694.
Boersch-Supan, M. E., S. Agarwal, M. E. White-Scharf, T. Imanishi-Kari. 1985. Heavy chain variable region: multiple gene segments encode anti-4-(hydroxy-3-nitro-phenyl)acetyl idiotypic antibodies. J. Exp. Med. 161:1272.
Caton, A. J., S. E. Stark, J. Kavaler, L. M. Staudt, D. Schwartz, W. Gerhard. 1991. Many variable region genes are utilized in the antibody response of BALB/c mice to the influenza virus A/PR/8/34 hemagglutinin. J. Immunol. 147:1675.
Clarke, S., R. Rickert, M. K. Wloch, L. Staudt, W. Gerhard, M. Weigert. 1990. The BALB/c secondary response to the Sb site of influenza virus hemagglutinin: nonrandom silent mutation and unequal numbers of VH and V mutations. J. Immunol. 145:2286.
Hougs, L., L. Juul, H. J. Ditzel, C. Heilmann, A. Svejgaard, T. Barington. 1999. The first dose of a Haemophilus influenzae type b conjugate vaccine reactivates memory B cells: evidence for extensive clonal selection, intraclonal affinity maturation, and multiple isotype switches to IgA2. J. Immunol. 162:224
Pommie, C., S. Levadoux, R. Sabatier, G. Lefranc, M. P. Lefranc. 2004. IMGT standardized criteria for statistical analysis of immunoglobulin V-REGION amino acid properties. J. Mol. Recognit. 17:17.
Schett, G., M. Tohidast-Akrad, G. Steiner, J. Smolen. 2001. The stressed synovium. Arthritis Res. 3:80.
Sigal, L. H., S. Williams, B. Soltys, R. Gupta. 2001. H9724, a monoclonal antibody to Borrelia burgdorferi’s flagellin, binds to heat shock protein 60 (HSP60) within live neuroblastoma cells: a potential role for HSP60 in peptide hormone signaling and in an autoimmune pathogenesis of the neuropathy of Lyme disease. Cell. Mol. Neurobiol. 21:477.
Crowley, H., B. T. Huber. 2003. Host-adapted Borrelia burgdorferi in mice expresses OspA during inflammation. Infect. Immun. 71:4003
Polymenis, M., B. D. Stollar. 1995. Single-chain anti-DNA FV. Methods Mol. Biol. 51:265
Farner, N. L., T. Dorner, P. E. Lipsky. 1999. Molecular mechanisms and selection influence the generation of the human VJ repertoire. J. Immunol. 162:2137.(Srimoyee Ghosh, Allen C. )