IgVH genes from different anatomical regions, with different histopathological patterns, of a rheumatoid arthritis patient suggest cyclic re-entry of mature synovial B-cells in the hypermutation process

Although IgV genes in rheumatoid B cells have been intensivelyanalyzed, many questions concerning antigen driven B-cell maturation andrecirculation remain unanswered. It would be interesting to know whether B-cellmaturation in rheumatoid tissue is different from that in secondary lymphaticorgans. Moreover, it would be interesting to know whether there exists arestricted number of antigens that act on the lesions of different anatomicalsites of the RA patient, and whether B cells recirculate between the differentjoints.

RNA and genomic DNA were prepared from tissue sections from threedifferent anatomical sites, with different histopathologies and differentonsets (left and right peroneal tendons and cubita synovial membrane), of a RApatient. Genomic DNA was amplified by seminested polymerase chain reaction(PCR), and the cDNA corresponding to the RNA was amplified by PCR using primersspecific for each IgVH family. The obtained sequences were compared with theirgermline counterparts on the V-Base data Bank [1]. Animmunohisto-chemical analysis of the infiltrate and the clinical data of localdisease activity were also included.

In the locations with longer disease duration (right peronealtendon 5 months, left peroneal tendon 2 months) a very intense inflammatoryinfiltrate with germinal centers containing Ki-M4-positive follicular dendriticcells (FDC) was observed. In the location with shorter disease duration (rightcubita 2 weeks) a low, diffuse and nonfollicular infiltration with markedoedema was detected. From the 55 analyzed clones seven expressed nonfunctionalrearrangements (pseudogenes) with stop codons, and 48 were found to expressfunctional genes. Among the 48 clones that expressed functional genes, therewere two that had amino acid deletions on their complementarity determiningregion (CDR)2 - clones K194/1 and K194/111 - similar to theones described by Wilson et al [2] and Goossenset al [3]. Two types of mixed molecules werefound. Mixed molecules of the first type (k194/57, k194/67 and k194/109) are composed of rearrangements of two different IgVgenes. Mixed molecules of the second type (k194/126,k194/119, k194/30 and k194/99) are composed of a IgVgene rearrangement that is fragmented by insertions of small random sequences.These insertions are different from the ones described by Wilson et al[2] since they are not duplicates or parts of IgV genes.The ratio of replacement mutations to silent mutations (R/S ratio) increasedwith disease duration. There was strong heterogeneity among the CDR3segments.

The amino acid sequences that belonged to the VH1-family obtainedfrom the three anatomical regions were primarily compared with the amino acidsequences of their closest germline counterparts (Fig. 1a). One result from this comparison was the heterogeneity inthe CDR3 rearrangements. Moreover, sequences k194/58 andk194/82 are clonally related (confirmed at nucleotide level). Then,the 21 amino acid sequences were compared with the most widely used germlinecounterpart IgHV1-18 ^* 01 (Fig. 1b).All of these VH1 sequences had mainly conservative mutations in the frameworkregion (FR) and nonconservative mutations in the CDR. Also, there was an almostoverall conservation of the mutational cold spots and 'structural coldspots' [4] among the 19 VH1 segments. Thereplacement (11 from 19 replacements resulted in a proline residue) in position34 of CDR2 could be interpreted as an antigen-selected mutational hotspot.

The comparison of the five sequences belonging toIgHV4-30-1/4-31 ^* 02 resulted in two types of clonal relation(Fig. 2a). The first type of clonal relation, betweensequences k194/100 and k194/101 (Fig. 2b), suggests that both sequences are derived from a singleprogenitor cell. The second type of clonal relation is between sequencesk194/23, k194/102 and k194/103 (Fig. 2c). It suggests that an unmutated progenitor cell gave rise tok194/23 (left peroneal tendon), from which k194/103 (rightcubita) derived and later generated k194/102 (right cubita).

The analysis of the 55IgVH sequences corroborates the findings ofother groups that studied a singlelocation and RA B-cell hybridomas [5,6,7,8,9,10] and addsfurther information on B-cell distribution and activation in RA. First, aminoacid deletions and mixed molecules could be interpreted as novel pathways togenerate antibody specificities, leading, for instance, to autoreactiveantibodies that could contribute to the local and systemic tissue destruction.Second, an apparently conserved mutational pattern among the 19 amino acid VH1segments suggests that in all three RA lesions of this patient the synovial Bcells are dealing with a restricted number of antigens. Third, the existence ofclonally related B cells in the cubita and left peroneal tendon leaves no doubtthat in this patient there is a cyclic re-entry of mutated B cells in thehypermutation process [11]. The already mutated B cellsfrom the early RA lesions sequentially colonize new germinal centers insecondary lymphatic organs as proposed by Kepler et al [12]. These reactivated B-cells then invade new anatomicalregions, leading to the perpetuation of the chronic inflammation in RA.

Tissue samples (one from the right and another from the left tendon of musculus peroneus longus and one from the cubita synovial tissue) from a 48-year-old female patient with confirmed seropositive RA [18] were obtained at synovectomy and were snap frozen. The patient was receiving antirheumatic medication (gold, methotrexate and sulphasalazine). In the present investigation, the degree of local disease activity was scored according to the method of Fuchs et al [19] and Krenn et al [20], on the basis of warmth, effusion and swelling. The patient was seropositive for rheumatoid factors. All tissues (approximately 50%) were fixed in formalin and embedded in paraffin (Giemsa, haematoxylin and eosin staining) to be used for diagnosis and scoring of the degree of inflammation. The only material used for immunohistochemical analyses was tissue that exhibited macroscopic signs of inflammation, taken from at least three different regions of the resected synovial membrane.

For immunohistochemical staining, 7-μ m cryosections (mounted on poly-L-lysine-coated slides) were used. Immediately before staining, the cryosections were treated with acetone for 10min, air dried at room temperature (10-20min) and double immunohistochemical staining was performed, as described by Krenn et al [7]. Briefly, the indirect immunoperoxidase technique (Ki-M4; DAKO, Hamburg, Germany) was combined with the alkaline phosphatase/antialkaline phosphatase technique (CD20; DAKO). No counterstaining was performed. In all cases, control staining was performed and single stainings were compared with double stainings in order to ensure that the pattern of immunohistochemical reaction remained unaltered.

A portion of tissue (approximately 50%) was fixed in formalin and paraffin-embedded (Giemsa, haematoxylin and eosin staining) for use in diagnosis and scoring of the degree of the inflammatory infiltration, which in the present study was performed according to the method of Krenn et al [7] on a semiquantitative 1-5 scale. Very low inflammatory infiltration was indicated by 1 on the scale: the synovial intima is slightly enlarged (two to three cell layers thick); the degree of lymphocytic infiltration is very low, showing a diffuse pattern; and the subsynovial region exhibits chronic tissue granulation with slight fibrosis. Low inflammatory infiltration was indicated by 2 on the scale: the synovial intima is slightly enlarged (two to three cell layers thick), and the degree of inflammatory infiltration is low, with a diffuse perivascular lymphocytic and plasma cell infiltration; and the subsynovial region shows chronic tissue granulation with moderate fibrosis. Moderate inflammatory infiltration was indicated by 3 on the scale: the synovial intima is moderately enlarged (three to five cell layers thick), and the degree of lymphocytic infiltration is moderate, with small follicle-like aggregates near small blood vessels; and there is moderate cellularity of the subsynovial region, which exhibits slight fibrosis. Strong inflammatory infiltration was indicated by 4 on the scale: the synovial intima is extensively enlarged (five to 10 cell layers), and lymphocytes exhibit a dense follicle-like pattern; and the 'interfollicular' area exhibits very high cellularity without fibrosis. Very strong inflammatory infiltration was indicated by 5 on the scale: the synovial intima is extensively enlarged, and the distribution of lymphocytes exhibits a dense follicle-like pattern with formation of germinal centers; and granulomas and hemigranulomas can be seen in the subsynovial region. In each histopathological analysis, 10 fields were examined, and the most prominent finding in a given field determined the score.

Total RNA from about 50 tissue sections of 20 μ m was prepared using the method of Chomczynski and Sacchi [21]. cDNA synthesis was performed with 5 μ g RNA using Gibco BRL (Karlsruhe, Germany) M-MLV reverse transcriptase according to the supplier's manual. The amplification of the VH genes was carried out in a 25 μ l volume containing 1.75mmol/l MgCl₂, 0.4pmol/l primer,1U Taq polymerase (MBI Fermentas, St Leon-Rot, Germany) and 200 μ mol/l of each dNTP. The cycle profile for amplification consisted of DNA denaturation at 95°C for 2min, followed by 45 cycles of 94°C for 30s, primer annealing at 65°C for 30s for VH3 and VH4 primers (60°C for VH1, VH2 and VH5), and extension at 72°C for 80s. Primer sequences were described previously [22], and are located from codons 17 to 24 (according to V-Base [1] sequence comparison). In brief, the following primers, given in the 5' -3' direction, were used:

V_H1 5' CCT CAG TGA AGT YTC CTG CAA GGC 3'

V_H2 5' GTC CTG CGC TGG TGA AAS CCA CAC A 3'

V_H3 5' GGG GTC CCT GAG ACT CTC CTG TGC AG 3'

V_H4 5' GAC CCT GTC CCT CAC CTG CRC TGT C 3'

V_H5 5' AAA AAG CCC GGG GAG TCT CTG ARG A 3'

V_H6 5' ACC TGT GCC ATC TCC GGG GAC AGT G 3'

J_H1-5 5' GGT GAC CAG GGT BCC YTG GCC CCA G 3'

J_H6 5' GGT GAC CGT GGT CCC TTG CCC CCA G 3'

DNA extraction and amplification of IgVH genes was performed according to the method of Kim et al [16] with minor modifications. In short, DNA was prepared by incubating 10×5μ m tissue sections at 50°C for 1h with proteinase K (Boehringer Mannheim, Mannheim, Germany), which was inactivated by heating at 95°C. To improve the specificity of the PCR amplification, seminested PCR reactions were carried out as follows. In the first step, amplification with Taq polymerase was performed with VH 5' primers and external J_H region-specific 3' primers [16]. In the second round,aliquots were specifically amplified for the heavy-chain genes using the same 5' V_H region primers, but internal J_H region primers (seminested PCR). In brief, the following primers, given in the 5' -3' direction, were used:

V_H1 5' CCA TGG ACT GGA CCT GGA 3'

V_H2 5' ATG GAC ATA CTT TGT TCC AC 3'

V_H3 5' CCA TGG AGT TTG GGC TGA GC 3'

V_H4 5' ATG AAA CAC CTG TGG TTC TT 3'

V_H5 5' ATG GGG TCA ACC GCC ATC CT 3'

V_H6 5' ATG TCT GTC TCC TTC CTG AT 3'

J_Hexternal 5' CTC ACC TGA GGA GAC GGT GAC C 3'

J_Hinternal

5' TGA (AG)GA GAC GGT GAC C(AG)(GT) GT(GCT) CC 3'

The final concentrations of the reagents were 0.1mmol/l MgCl₂, 200 μ mol/l of each dNTP, 10pmol/l of each primer and 2U Taq DNA polymerase. The cycle program consisted of a denaturation step at 95°C for 5min followed by five cycles at 95°C for 40s, 65°C for 40s and 72°C for 1min and 50s; five cycles at 95°C for 40s, 60°C for 40s and 72°C for 1min and 50s; and 25 cycles at 95°C for 40s, 55°C for 40s and 72°C for 1min and 50s. The cycles were followed by a final 10-min incubation at 72°C.

Aliquots of the final PCR products were separated by electrophoresis using a 2% low melting agarose gel (Roth, Karlsruhe, Germany), and DNA bands in the range of 350bp were purified from the agarose gel using High-Pure DNA gel extraction kit (Boehringer Mannheim). Cloning of PCR fragments was performed using the pCR-Script Amp SK(+) cloning kit (Stratagene, Heidelberg, Germany). Positive clones were sequenced using the DyeDeoxy Termination Cycle Sequencing Kit (Applied BioSystems Inc, Weiterstadt, Germany), and analyzed using an automated DNA sequencer ABIPrism373 (Applied BioSystems Inc). Both strands were sequenced using T3 and T7 primers. The sequences were analyzed using DNAman for Windows software (Lynon BioSoft, Vaudreuil, Quebec, Canada), Genebank and v-Base databases [1].

The female patient, who suffered from a confirmed seropositive RA with involvement of tendon sheaths, exhibited severe signs of local disease activity, with tenosynovitis of the right distal peroneus longus tendon and left distal peroneus longus tendon, and synovialis of the right cubita (Table 1). The durations of local disease were 5 months, 2 months and 2 weeks, respectively.

A heterogeneous inflammatory infiltrate could be observed in the different locations. In both locations with longer disease duration (right peroneal tendon [5 months] and left peroneal tendon [2 months]) a very intense inflammatory infiltrate, with Ki-M4-positive FDC-containing germinal centers (Fig. 3a,3b and insert), could be observed (inflammatory score 5). However, in the right cubita (disease duration 2 weeks) a low, diffuse and nonfollicular infiltration with marked oedema (inflammatory score 2) was detected (Fig. 3c). The latter synovialitis histopathologically showed a more acute inflammatory reaction, whereas in the right and left peroneal tendons the morphological pattern of a typical chronic tendosynovitis was present. Immunohistochemically, the right and left peroneal tendons exhibited a dense follicular-like infiltration, with Ki-M4-positive FDC and peripherally located CD20-positive B lympocytes representing germinal centers (insert in Fig. 3a and 3b). In the right cubita only a very low and diffuse nonfollicular distribution of lymphocytes without Ki-M4-positive FDC could be recognized.

The comparison was restricted to the VH1 and VH4 families, because they yielded the more relevant results.

Recently, the introduction of deletions and duplications, in addition to nucleotide exchanges, has been described as a feature of the somatic hypermutation process [2,3]. Amino acid deletions in the IgV genes have only been described in lymphomas and healthy secondary lymphatic tissue, however [2,3]. In the present study we report the existence of such amino acid deletions in the IgV genes from synovial B cells of an autoimmune disease. For the first time, amino acid deletions in the VH genes were found in B cells of an autoimmune disease. The detection of these deletion events in RA synovialitis stresses the functional homology of the synovial membrane to secondary lymphatic tissue.

Also for the first time, IgV gene mixed molecules were found that are formed either by two segments of different IgV genes or by a IgV gene rearrangement that is fragmented by random insertions. The mechanisms that underlie the formation of these mixed molecules could be modified pathways to the unified model for somatic hypermutation, as proposed by Maizels [37] (Fig. 6); an initiating lesion could lead to a hypermutation either templated (using another germline gene as template) or untemplated (by inserting small random sequences). In some cases the process of untemplated hypermutation could lead to the insertion of STOP codons, rendering the gene nonfunctional.

Hence, the production of mixed molecules and the introduction of deletions could represent novel pathways for RA synovial B cells to generate new specificities that lead, for instance, to autoreactive antibodies that could contribute to the local and systemic tissue destruction.

The comparison of the amino acid sequences of the 19 VH1 segments from the different locations provided some valuable data on the interaction of the RA synovial B cells and their target antigen(s).

All of the VH1 sequences had mainly conservative mutations in the FR and nonconservative in the CDR, thus, agreeing with the results from Wedemayer et al [38]. When resolving at 2.1Å, the crystal structure of a germline antibody Fab fragment and its complex with hapten, they observed an expansion of the binding potential of the primary antibody repertoire. This expansion derived from configurational stability due to antigen binding and somatic mutations, nonconservative mutations in the CDR that raised the affinity for the hapten, and conservative mutations in the FR.

Another important finding was the almost overall conservation of the mutational cold spots and 'structural cold spots' [4] among the 19VH1 segments. The interesting absence of mutations in positions 17, 23, 28, 30, 48, 56 and 71 suggests the existence of more 'structural cold spots' in the VH1 family than those described.

During the germinal center reaction (for instance in the follicles of the two earlier lesions of this patient), rearranged B cells with low-affinity receptors improve their affinity by somatic hypermutation [39,40]. Nevertheless, these mutations can also decrease the affinity instead of expanding it. Therefore, as reported by Meffre et al [41], under appropriate regulation VDJ rearrangements take place in mature B cells of human tonsil. Hence, the heterogenity observed on the CDR3 of the 19 VH1 sequences could be due to a reactivation of the rearrangement process in order to rescue these RA synovial B cells from deleterious somatic mutations, or to further increase their binding affinity.

Based on the above findings, there appears to be a conserved mutational pattern among all 19 VH1 segments, hence suggesting that in all three RA lesions of this patient the synovial B cells were activated by a restricted number of antigens. This is strengthened by the replacement in position 34 of CDR2, which could be interpreted as an antigen-selected mutational hotspot.

The increment of mutations with antigen dose [42] possibly indicates that the maturation of the immune response is a continuous process with the production of an increasing number of hypermutated memory B cells with time. In the special case of RA, the local joint destruction may release antigens that lead to the hypermutation process. Characteristic for B-cell hypermutation are the elevated R/S ratios in the CDR. In the present study there was a direct association of the overall R/S ratios with the duration of local disease. Synovial B cells were shown to undergo a germinal center-like reaction in RA [5,16]. Therefore, we suggest that the activated B cells from this patient have undergone a local maturation in the germinal center-like structures detected in the peroneal tendons (left and right). On the other hand, the fact that one lesion was free from FDCs and exhibited only an acute inflammatory infiltrate could support the model proposed by Oprea and Perelson [43]; the already mutated germinal center B cells from the peroneal tendons might have migrated into the cubita synovial tissue, as has been shown for closely located finger joints [44], and re-entered in a cyclic hypermutation process. The apparent existence of a mutational pattern on amino acid level of clones obtained from the different regions could support this hypothesis. However, the existence of clonally related B-cells in the cubita and left peroneal tendon leaves no doubts that, in the patient studied, there is cyclic re-entry of the mutated B cells from the early RA lesions in the hypermutation process [11] that sequentially colonize new germinal centers, as proposed by Kepler and Perelson [12]. These reactivated B cells then invade new anatomical regions, leading to the perpetuation of the chronic inflammation in RA.