**3.4 The thymic germinal centre response in myasthenia gravis**

#### **3.4.1 Germinal centres in the thymus**

Thymi from 5 EOMG patients were examined by immunohistology. All 5 contained large numbers of germinal centres with typical mantle zones within the thymic medulla, histologically indistinguishable from germinal centres in human tonsil controls. The mantle zones contained densely packed CD20+ B-cells surrounding the germinal centre B-cells (Fig. 3A). These were interspersed with a network of follicular dendritic cells extending throughout the dark and light zones (Fig. 3C) and a crescent of T-cells can be seen at the apex of the light zone (Fig. 3B). Proliferating B-cells were distributed throughout the germinal centre but in larger numbers within the dark zone (Fig. 3D). Autoradiography with 125I-α-bungarotoxin alone, which binds to AChR, diffusely labelled c.50% of germinal centres and appeared to be associated with the follicular dendritic cell processes. No labelling was seen in human tonsils or thymi from two seronegative myasthenia patients and bungarotoxin binding was blocked by the cholinergic drug, carbamyl choline, which is structurally similar to AChR, indicating that the follicular dendritic network contained membrane-bound antigen or immune complexes.

In contrast, 125I-α-bungarotoxin-labelled AChR bound to individual cells in 20% of germinal centres, including large numbers of moderately labelled centrocytes in the light zone, smaller numbers in the dark zone, and heavily labelled plasmablasts/plasma cells in and around the germinal centres (Fig. 3E & F).

The Ectopic Germinal Centre Response in Autoimmune Disease and Cancer 407

The distribution of VH-gene families was similar in the follicular mantle and all four germinal centres, with predominant use of members of the VH3 gene family compared with the number of germline VH-genes in this family (Fig. 4A). However, since this gene family is also predominantly used by the peripheral blood B-cells of healthy individuals, it was not considered to be significant. No VH6, VH7 or JH2 genes were isolated and JH1 was underused, but the rarely used VH5-51 gene and the JH4 exon (Fig. 4B) were over-represented, both being used by rearranged V-genes from three different germinal centres in combination with different D exons (DH2-2, DH5-12 and DH-6-19). In many cases the D exon could not be identified due to junctional diversity and removal of bases during recombination. Of those that could be identified, DH3 and DH6 were the most commonly used. These data strongly imply selection for B-cells expressing antigen receptors using particular combinations of V, D and J segments, despite the heterogeneity of the germinal centre B-cells. This would be even more apparent if individual members of the same B-cell clone were counted separately.

Fig. 4. Frequency of usage of VH and JH germline genes by thymic germinal centre B-cells. A: The frequency of VH gene family usage in all germinal centres analysed differs significantly from predictions from the number of members of each family in the germline. Members of clonally related sets were only counted once. B: The frequency of individual JH exon usage in all germinal centres analysed differs significantly from the number of JH exons in the germline (\*p<0.01). Reproduced from

A B

**3.4.3 Somatic hypermutation and clonal proliferation of B-cells in thymic germinal** 

All the germinal centres and the follicular mantle contained B-cells expressing both mutated and unmutated Ig V-genes, the latter presumably coming from naïve B-cells. The majority of V-genes from the germinal centres and the follicular mantle were mutated, with considerable variation in the number of mutations, ranging from 0 – 52. Some of the rearranged V-genes in clonally related sets had high ratios of replacement to silent mutations from 4.7:1 to 7.0:1 in the CDRs, which form the antigen-binding site, suggesting affinity selection of mutated antigen receptors is taking place in the germinal centre. Some other sequences had low numbers of replacement mutations, suggesting selection against replacement mutations, as also found by (Zuckerman *et al.*, 2010). 18 different sets of functionally rearranged VH-genes included two or more related sequences sharing the same V, D and J segments and junctional sequences but with significantly more than one mutation per V-gene and, in three cases, they were cloned in separate amplifications from different sections through the same germinal centre and therefore could only be from different B-cells.

Sims *et al* (2001).

**centres** 

Fig. 3. Immunohistochemically stained serial sections through thymic germinal centres.

A & B: Germinal centres stained (red) with anti-CD20 and anti-CD3 for B and T-cells respectively. The arrow (B) shows a crescent of T-cells in the light zone. C: A network of follicular dendritic cell processes is spread throughout the germinal centre, including both the light and dark zones. D: Germinal centre cells stained with an antibody against proliferating cell nuclear antigen, revealing dividing B-cells in both areas but more concentrated in the dark zone. E & F: 125I-α-bungarotoxin-labelled AChR reveals individual AChR-specific plasmablasts/plasma cells in and around germinal centres (detected by autoradiography). Diffuse labelling in the light zone probably indicates binding of free 125I-αbungarotoxin to AChR trapped on follicular dendritic cells (see text). Reproduced from Sims *et al* (2001).

#### **3.4.2 Ig V-gene expression by thymic germinal centre B-cells**

In order to determine whether thymic germinal centre B-cells are subjected to antigendriven clonal proliferation, somatic hypermutation and affinity selection, as seen in the orthotopic germinal centres of secondary lymphoid organs, we cloned and sequenced rearranged Ig heavy chain V-genes from multiple sections through four thymic germinal centres (A – D) and the follicular mantle surrounding one of them (A). 216 rearranged VHgenes, derived from 61 independently rearranged, functional sequences, were obtained from the four germinal centres and 46 VH-genes from the follicular mantle were derived from 24 functional VH-genes. Since the PCR error rate in control experiments was estimated to be less than one base per four VH-genes, only sequences using the same V, D & J exons, the same junctional sequences and a minimum of one base difference per gene, were accepted as mutated members of a clonally related set of B-cells. This conservative assessment almost certainly discards some B-cell clones with low frequencies of somatic mutation and therefore underestimates the true B-cell diversity. When calculating the number of VH-genes used, members of the same B-cell clone were counted only once. However, this reflects the number of individual B-cell clones and non-dividing B-cells rather than the total number of B-cells using a particular gene.

Fig. 3. Immunohistochemically stained serial sections through thymic germinal centres.

**3.4.2 Ig V-gene expression by thymic germinal centre B-cells** 

B-cells using a particular gene.

A & B: Germinal centres stained (red) with anti-CD20 and anti-CD3 for B and T-cells respectively. The arrow (B) shows a crescent of T-cells in the light zone. C: A network of follicular dendritic cell processes is spread throughout the germinal centre, including both the light and dark zones. D: Germinal centre cells stained with an antibody against proliferating cell nuclear antigen, revealing dividing B-cells in both areas but more concentrated in the dark zone. E & F: 125I-α-bungarotoxin-labelled AChR reveals individual AChR-specific plasmablasts/plasma cells in and around germinal centres (detected by autoradiography). Diffuse labelling in the light zone probably indicates binding of free 125I-αbungarotoxin to AChR trapped on follicular dendritic cells (see text). Reproduced from Sims *et al* (2001).

In order to determine whether thymic germinal centre B-cells are subjected to antigendriven clonal proliferation, somatic hypermutation and affinity selection, as seen in the orthotopic germinal centres of secondary lymphoid organs, we cloned and sequenced rearranged Ig heavy chain V-genes from multiple sections through four thymic germinal centres (A – D) and the follicular mantle surrounding one of them (A). 216 rearranged VHgenes, derived from 61 independently rearranged, functional sequences, were obtained from the four germinal centres and 46 VH-genes from the follicular mantle were derived from 24 functional VH-genes. Since the PCR error rate in control experiments was estimated to be less than one base per four VH-genes, only sequences using the same V, D & J exons, the same junctional sequences and a minimum of one base difference per gene, were accepted as mutated members of a clonally related set of B-cells. This conservative assessment almost certainly discards some B-cell clones with low frequencies of somatic mutation and therefore underestimates the true B-cell diversity. When calculating the number of VH-genes used, members of the same B-cell clone were counted only once. However, this reflects the number of individual B-cell clones and non-dividing B-cells rather than the total number of The distribution of VH-gene families was similar in the follicular mantle and all four germinal centres, with predominant use of members of the VH3 gene family compared with the number of germline VH-genes in this family (Fig. 4A). However, since this gene family is also predominantly used by the peripheral blood B-cells of healthy individuals, it was not considered to be significant. No VH6, VH7 or JH2 genes were isolated and JH1 was underused, but the rarely used VH5-51 gene and the JH4 exon (Fig. 4B) were over-represented, both being used by rearranged V-genes from three different germinal centres in combination with different D exons (DH2-2, DH5-12 and DH-6-19). In many cases the D exon could not be identified due to junctional diversity and removal of bases during recombination. Of those that could be identified, DH3 and DH6 were the most commonly used. These data strongly imply selection for B-cells expressing antigen receptors using particular combinations of V, D and J segments, despite the heterogeneity of the germinal centre B-cells. This would be even more apparent if individual members of the same B-cell clone were counted separately.

Fig. 4. Frequency of usage of VH and JH germline genes by thymic germinal centre B-cells.

A: The frequency of VH gene family usage in all germinal centres analysed differs significantly from predictions from the number of members of each family in the germline. Members of clonally related sets were only counted once. B: The frequency of individual JH exon usage in all germinal centres analysed differs significantly from the number of JH exons in the germline (\*p<0.01). Reproduced from Sims *et al* (2001).

#### **3.4.3 Somatic hypermutation and clonal proliferation of B-cells in thymic germinal centres**

All the germinal centres and the follicular mantle contained B-cells expressing both mutated and unmutated Ig V-genes, the latter presumably coming from naïve B-cells. The majority of V-genes from the germinal centres and the follicular mantle were mutated, with considerable variation in the number of mutations, ranging from 0 – 52. Some of the rearranged V-genes in clonally related sets had high ratios of replacement to silent mutations from 4.7:1 to 7.0:1 in the CDRs, which form the antigen-binding site, suggesting affinity selection of mutated antigen receptors is taking place in the germinal centre. Some other sequences had low numbers of replacement mutations, suggesting selection against replacement mutations, as also found by (Zuckerman *et al.*, 2010). 18 different sets of functionally rearranged VH-genes included two or more related sequences sharing the same V, D and J segments and junctional sequences but with significantly more than one mutation per V-gene and, in three cases, they were cloned in separate amplifications from different sections through the same germinal centre and therefore could only be from different B-cells.

The Ectopic Germinal Centre Response in Autoimmune Disease and Cancer 409

These related sets of V-genes were therefore derived from members of the same proliferating B-cell clone, showing that antigen-driven B-cell proliferation and somatic hypermutation are taking place within the thymic germinal centres. All of the isolated B-cell

We do not know the total number of clones proliferating within a single germinal centre. We examined every tenth section but may have missed small B-cell clones, so there are likely to be more than we detected. Genealogical trees of related sets of V-genes were constructed by the most parsimonious relationship on the basis of shared and unshared mutations (section 2.5). Six of the 18 clones isolated from two germinal centres containing AChR-specific B-cells were derived from unmutated precursors, i.e. naïve B-cells, whereas the VH-genes of the earliest founder cells isolated from the other 12 clones contained from 5 to >30 mutations. Examples of six of these clones are shown in Fig. 5. Although we cannot rule out the possibility that some of these clones were also derived from unidentified naïve B-cells, it is most probable that the majority were generated from founder memory B-cells. Thus, both memory and naïve B-cells have been stimulated by antigen and are proliferating and mutating their antigen receptors within thymic germinal

**3.4.4 Evidence for selection of AChR-specific B-cell clones in thymic germinal centres**  Three B-cell clones from 3 different AChR positive germinal centres expressed the rarely used VH5-51 genes. Furthermore, these three independent sets of V-genes exhibited the same amino acid replacements at three positions. Comparison of our V-gene sequences with those of heavy chains from known AChR-specific hybridomas and Fabs revealed some common features. Germline genes used by four of our B-cell clones also encode anti-AChR antibodies cloned from other EOMG patients, and several of the amino acid substitutions in CDR1 and CDR2 from two B-cell clones were also present in an anti-AChR Fab (Sims *et al.*, 2001; Matthews *et al.*, 2002). It is therefore unlikely that these mutations occurred by chance, suggesting that a common selection process for mutants with high affinity for the AChR is

**3.4.5 Evidence for immunisation by the fetal form of acetylcholine receptor** 

Babies born to mothers with myasthenia gravis can develop transient symptoms due to transplacental transfer of the maternal autoantibodies and, in rare cases, they have severe developmental abnormalities, arthrogryposis multiplex congenita (AMC), caused by maternal anti-AChR antibodies that inhibit the ion channel function of the fetal AChR, causing paralysis during fetal development *in utero*, whereas the adult form of the mother's AChR is relatively unaffected. Fetal AChR-specific antibodies are particularly prevalent in women who have had babies, suggesting that they may be induced by the

In order to determine the specificity and clonal origins of fetal AChR-specific autoantibodies, combinatorial Fab libraries were constructed from cDNA prepared from thymic cells of two mothers (M2 and M6) of AMC babies. 25 Fab clones were isolated and two clonally related sets were examined in greater detail. All Fabs bound specifically to the γ subunit of fetal AChR, except one that recognised the β subunit also present in the adult receptor. Sequencing of the fetal-specific Fabs revealed clearly restricted usage of VH, JH, Vκ & Jκ gene segments and convergent coding mutations. All the Fabs from AMC mother M2

clones were small, containing a maximum of five members.

centres of myasthenia patients.

in operation.

fetus.

Fig. 5. Examples of clonally related sets of rearranged VH genes isolated from thymic germinal centres C and D.

The genealogical trees were constructed using the most parsimonious parent-daughter cell relationships of clonally related sets of sequences (section 2.2). The best matching germline gene is depicted as an ellipse. Letters in the circles refer to individual sequences from the same clonally related set. Deduced intermediates are shown as dotted circles. Numbers at the side of the arrows indicate the minimum number of mutations required to generate a daughter cell from the immediate parental cell. The results show that the B-cells from which they were derived are undergoing antigen-driven clonal proliferation and somatic hypermutation from both naïve and memory B-cell precursors. A: B-cell clones from germinal centre C; B: B-cell clones from germinal centre D; C: Frequency of mutations in the VH genes expressed by the most probable progenitor cells of all 18 B-cell clones analysed. Reproduced from Sims *et al.* (2001).

Fig. 5. Examples of clonally related sets of rearranged VH genes isolated from thymic

The genealogical trees were constructed using the most parsimonious parent-daughter cell relationships of clonally related sets of sequences (section 2.2). The best matching germline gene is depicted as an ellipse. Letters in the circles refer to individual sequences from the same clonally related set. Deduced intermediates are shown as dotted circles. Numbers at the side of the arrows indicate the minimum number of mutations required to generate a daughter cell from the immediate parental cell. The results show that the B-cells from which they were derived are undergoing antigen-driven clonal proliferation and somatic hypermutation from both naïve and memory B-cell precursors. A: B-cell clones from germinal centre C; B: B-cell clones from germinal centre D; C: Frequency of mutations in the VH genes expressed by the most probable progenitor cells of all 18 B-cell clones analysed. Reproduced from Sims

germinal centres C and D.

*et al.* (2001).

These related sets of V-genes were therefore derived from members of the same proliferating B-cell clone, showing that antigen-driven B-cell proliferation and somatic hypermutation are taking place within the thymic germinal centres. All of the isolated B-cell clones were small, containing a maximum of five members.

We do not know the total number of clones proliferating within a single germinal centre. We examined every tenth section but may have missed small B-cell clones, so there are likely to be more than we detected. Genealogical trees of related sets of V-genes were constructed by the most parsimonious relationship on the basis of shared and unshared mutations (section 2.5). Six of the 18 clones isolated from two germinal centres containing AChR-specific B-cells were derived from unmutated precursors, i.e. naïve B-cells, whereas the VH-genes of the earliest founder cells isolated from the other 12 clones contained from 5 to >30 mutations. Examples of six of these clones are shown in Fig. 5. Although we cannot rule out the possibility that some of these clones were also derived from unidentified naïve B-cells, it is most probable that the majority were generated from founder memory B-cells. Thus, both memory and naïve B-cells have been stimulated by antigen and are proliferating and mutating their antigen receptors within thymic germinal centres of myasthenia patients.

#### **3.4.4 Evidence for selection of AChR-specific B-cell clones in thymic germinal centres**

Three B-cell clones from 3 different AChR positive germinal centres expressed the rarely used VH5-51 genes. Furthermore, these three independent sets of V-genes exhibited the same amino acid replacements at three positions. Comparison of our V-gene sequences with those of heavy chains from known AChR-specific hybridomas and Fabs revealed some common features. Germline genes used by four of our B-cell clones also encode anti-AChR antibodies cloned from other EOMG patients, and several of the amino acid substitutions in CDR1 and CDR2 from two B-cell clones were also present in an anti-AChR Fab (Sims *et al.*, 2001; Matthews *et al.*, 2002). It is therefore unlikely that these mutations occurred by chance, suggesting that a common selection process for mutants with high affinity for the AChR is in operation.

#### **3.4.5 Evidence for immunisation by the fetal form of acetylcholine receptor**

Babies born to mothers with myasthenia gravis can develop transient symptoms due to transplacental transfer of the maternal autoantibodies and, in rare cases, they have severe developmental abnormalities, arthrogryposis multiplex congenita (AMC), caused by maternal anti-AChR antibodies that inhibit the ion channel function of the fetal AChR, causing paralysis during fetal development *in utero*, whereas the adult form of the mother's AChR is relatively unaffected. Fetal AChR-specific antibodies are particularly prevalent in women who have had babies, suggesting that they may be induced by the fetus.

In order to determine the specificity and clonal origins of fetal AChR-specific autoantibodies, combinatorial Fab libraries were constructed from cDNA prepared from thymic cells of two mothers (M2 and M6) of AMC babies. 25 Fab clones were isolated and two clonally related sets were examined in greater detail. All Fabs bound specifically to the γ subunit of fetal AChR, except one that recognised the β subunit also present in the adult receptor. Sequencing of the fetal-specific Fabs revealed clearly restricted usage of VH, JH, Vκ & Jκ gene segments and convergent coding mutations. All the Fabs from AMC mother M2

The Ectopic Germinal Centre Response in Autoimmune Disease and Cancer 411

Since the EOMG thymus contains many typical germinal centres surrounded by mantle zones, including clones of proliferating, AChR-specific B-cells, it is clearly a site of autoantibody diversification. The B-cells are undergoing somatic hypermutation and affinity selection by cognate binding to AChR bound to the membrane processes of follicular dendritic cells, which form a network surrounding B and T-cells in the germinal centres. Bcells expressing high affinity, AChR-specific antigen receptors receive rescue signals from follicular dendritic and helper T-cells inducing them to differentiate into antibody-secreting plasmablasts and plasma cells that migrate out of the follicles and leave the thymus to enter the circulation, in a classical germinal centre type response. The numerous mutations seen in some V-gene sequences suggests that Ig class-switched memory B-cells are also generated and either leave the thymus, or some may recirculate within the germinal centre, undergoing further rounds of somatic hypermutation and affinity selection, consistent with the observations of others that recirculation of centrocytes between the light and dark zones occurs in orthotopic germinal centres. Several of the B-cell clones we identified were derived from highly mutated precursors, which supports this hypothesis. The autoantibodies produced by mothers of AMC babies included antibodies specific for the fetal form of the

The reason why large numbers of germinal centres producing AChR-specific B-cells and plasma cells develop in the thymi of myasthenia gravis patients is unclear, but it is possible that the rare thymic myoid cells may be involved. The induction of fetal AChRspecific antibodies in parous women suggests, at least in some cases, that expression of the fetal AChR on thymic epithelial and myoid cells may initiate an immune response, for which the patient's immune system has not been tolerised. Furthermore, generation of thymic germinal centres may be induced by a pre-existing proinflammatory cytokine environment, including IFNγ and TNFα. These molecules have been shown to upregulate expression of AChR subunits in thymic epithelial cells and on the membranes of myoid cells (Poea-Guyon *et al.*, 2005). The chemokines CXCL13 AND CCL21 are produced by endothelial cells of the afferent lymphatic vessels in the thymus, attracting activated T and B-cells, including naïve B-cells, suggesting that this is the mechanism of induction of thymic germinal centres (Berrih-Aknin *et al.*, 2009; Le Panse *et al.*, 2006; Meraouna *et al.*,

We therefore propose a two step process for initiation of the autoimmune response in myasthenia gravis (SHIONO *et al.*, 2003). In step 1, there is hyperplasia of thymic epithelial cells expressing linear AChR epitopes, including the α and ε subunits, in the context of the MHC Class II antigen HLA-DR52a, a susceptibility factor for EOMG patients. Whereas these peptides would normally induce tolerance, an imbalance in regulatory factors and expression of costimulatory molecules results in activation of thymic Th-cells against AChR epitopes. In step 2, the Th-cells induce an early B-cell response against the linear peptides and some of the resulting antibodies cross-react with conformational epitopes of the native AChR expressed on the thymic myoid cells, leading to myoid cell damage, release of AChR/antibody immune complexes, danger signals, and recruitment of professional antigen presenting cells. These stimulate an enhanced B-cell response accompanied by formation of germinal centres with production of high affinity, class switched, pathogenic autoantibodies. Although some aspects of this hypothesis are conjectural, they are also

**3.5 Conclusions** 

AChR, directed against the γ subunit.

2006; Le Panse *et al.*, 2010).

testable.

used the VH3-07 gene recombined with JH6b and an unidentified D exon in combination with various Vκ genes, suggesting that the heavy chain is the major contributor to AChR binding, at least in this case. The VH3-07 segments were mutated and clonally related. Most of the Fabs from AMC mother M6 used the same combination of mutated VH3-21 and JH5b, with an unidentified D exon, plus a Vκ02-12/Jκ4 light chain, which was also used in two of the Fabs from M2. In this case, both the VH3-21 and Vκ02-12 sets of sequences were clonally related, suggesting that they may both be derived from the same B-cell clone. The clonally related sequences from both sets of Fabs from M2 and M6 contained many shared and unshared coding mutations. The apparent founder member of each set of sequences had a large number of base differences from the best matching germline V-gene, suggesting that the clones were derived from mutated memory B-cells (Fig. 6).

Fig. 6. Clonally related sets of rearranged VH & Vκ genes encoding AChR-binding Fabs cloned from thymic cells from AMC mother M6.

The heavy and light chains were from the same set of Fabs. Genealogical trees were constructed and mutations numbered as described in the legend to figure 5. Fab names are shown in the circles, H referring to the heavy chain and K referring to the κ light chain. Dotted circles represent hypothetical intermediates. Numbers at the side of arrows show the minimum number of mutations required to generate a daughter cell from its immediate parent. Reproduced from Matthews *et al.* (2002).

Several sequences from both clonally related sets of Fabs had many more replacement mutations than expected by chance, indicating affinity selection. There was also clear evidence of convergent mutations. Many independently rearranged sequences from both patients shared consensus mutations. All the Fabs using VH3 genes contained the same 31S→T mutation and most of the Vκ02-12 genes contained a 22SRASET28 motif in CDR1. A search of the GenBank database of all human Ig sequences found only two other κ chains containing this motif, suggesting that it is important in determining the specificity of the anti-AChR autoantibodies.

#### **3.5 Conclusions**

410 Autoimmune Disorders – Current Concepts and Advances from Bedside to Mechanistic Insights

used the VH3-07 gene recombined with JH6b and an unidentified D exon in combination with various Vκ genes, suggesting that the heavy chain is the major contributor to AChR binding, at least in this case. The VH3-07 segments were mutated and clonally related. Most of the Fabs from AMC mother M6 used the same combination of mutated VH3-21 and JH5b, with an unidentified D exon, plus a Vκ02-12/Jκ4 light chain, which was also used in two of the Fabs from M2. In this case, both the VH3-21 and Vκ02-12 sets of sequences were clonally related, suggesting that they may both be derived from the same B-cell clone. The clonally related sequences from both sets of Fabs from M2 and M6 contained many shared and unshared coding mutations. The apparent founder member of each set of sequences had a large number of base differences from the best matching germline V-gene, suggesting that

Fig. 6. Clonally related sets of rearranged VH & Vκ genes encoding AChR-binding Fabs

generate a daughter cell from its immediate parent. Reproduced from Matthews *et al.* (2002).

The heavy and light chains were from the same set of Fabs. Genealogical trees were constructed and mutations numbered as described in the legend to figure 5. Fab names are shown in the circles, H referring to the heavy chain and K referring to the κ light chain. Dotted circles represent hypothetical intermediates. Numbers at the side of arrows show the minimum number of mutations required to

Several sequences from both clonally related sets of Fabs had many more replacement mutations than expected by chance, indicating affinity selection. There was also clear evidence of convergent mutations. Many independently rearranged sequences from both patients shared consensus mutations. All the Fabs using VH3 genes contained the same 31S→T mutation and most of the Vκ02-12 genes contained a 22SRASET28 motif in CDR1. A search of the GenBank database of all human Ig sequences found only two other κ chains containing this motif, suggesting that it is important in determining the specificity of the

the clones were derived from mutated memory B-cells (Fig. 6).

cloned from thymic cells from AMC mother M6.

anti-AChR autoantibodies.

Since the EOMG thymus contains many typical germinal centres surrounded by mantle zones, including clones of proliferating, AChR-specific B-cells, it is clearly a site of autoantibody diversification. The B-cells are undergoing somatic hypermutation and affinity selection by cognate binding to AChR bound to the membrane processes of follicular dendritic cells, which form a network surrounding B and T-cells in the germinal centres. Bcells expressing high affinity, AChR-specific antigen receptors receive rescue signals from follicular dendritic and helper T-cells inducing them to differentiate into antibody-secreting plasmablasts and plasma cells that migrate out of the follicles and leave the thymus to enter the circulation, in a classical germinal centre type response. The numerous mutations seen in some V-gene sequences suggests that Ig class-switched memory B-cells are also generated and either leave the thymus, or some may recirculate within the germinal centre, undergoing further rounds of somatic hypermutation and affinity selection, consistent with the observations of others that recirculation of centrocytes between the light and dark zones occurs in orthotopic germinal centres. Several of the B-cell clones we identified were derived from highly mutated precursors, which supports this hypothesis. The autoantibodies produced by mothers of AMC babies included antibodies specific for the fetal form of the AChR, directed against the γ subunit.

The reason why large numbers of germinal centres producing AChR-specific B-cells and plasma cells develop in the thymi of myasthenia gravis patients is unclear, but it is possible that the rare thymic myoid cells may be involved. The induction of fetal AChRspecific antibodies in parous women suggests, at least in some cases, that expression of the fetal AChR on thymic epithelial and myoid cells may initiate an immune response, for which the patient's immune system has not been tolerised. Furthermore, generation of thymic germinal centres may be induced by a pre-existing proinflammatory cytokine environment, including IFNγ and TNFα. These molecules have been shown to upregulate expression of AChR subunits in thymic epithelial cells and on the membranes of myoid cells (Poea-Guyon *et al.*, 2005). The chemokines CXCL13 AND CCL21 are produced by endothelial cells of the afferent lymphatic vessels in the thymus, attracting activated T and B-cells, including naïve B-cells, suggesting that this is the mechanism of induction of thymic germinal centres (Berrih-Aknin *et al.*, 2009; Le Panse *et al.*, 2006; Meraouna *et al.*, 2006; Le Panse *et al.*, 2010).

We therefore propose a two step process for initiation of the autoimmune response in myasthenia gravis (SHIONO *et al.*, 2003). In step 1, there is hyperplasia of thymic epithelial cells expressing linear AChR epitopes, including the α and ε subunits, in the context of the MHC Class II antigen HLA-DR52a, a susceptibility factor for EOMG patients. Whereas these peptides would normally induce tolerance, an imbalance in regulatory factors and expression of costimulatory molecules results in activation of thymic Th-cells against AChR epitopes. In step 2, the Th-cells induce an early B-cell response against the linear peptides and some of the resulting antibodies cross-react with conformational epitopes of the native AChR expressed on the thymic myoid cells, leading to myoid cell damage, release of AChR/antibody immune complexes, danger signals, and recruitment of professional antigen presenting cells. These stimulate an enhanced B-cell response accompanied by formation of germinal centres with production of high affinity, class switched, pathogenic autoantibodies. Although some aspects of this hypothesis are conjectural, they are also testable.

The Ectopic Germinal Centre Response in Autoimmune Disease and Cancer 413

with other autoimmune diseases. Most bind to protein or ribonucleoprotein complexes involved in protein synthesis, translocation or elongation; MAA target antigens are primarily located in the nucleoplasm or nucleolus. The most prevalent MSAs are directed against amino-acyl-tRNA-synthetases (ARS), and are associated with a distinctive clinical phenotype, anti-synthetase syndrome, characterised by myositis, Raynaud's phenomenon and interstitial lung disease, with a higher mortality. Anti-Jo-1 (anti-histidyl-tRNA synthetase) antibodies are the most prevalent in myositis patients (20-30% of patients), while the other anti-ARS antibodies are only present in 1-3% of IM patients, and are a diagnostic

and prognostic marker for disease severity (Mielnik *et al.*, 2006; Zampieri *et al.*, 2005).

studies which showed an abundance of immunoglobulin transcripts.

**4.2 The muscle infiltrating B-cell response in myositis** 

clonal expansion of infiltrating, autoantibody producing B-cells *in situ* in IM.

**4.2.1 The cellular composition of infiltrating lymphoid cells in myositis** 

As described above B-cells have been found to be prominent within the muscle infiltrating cell populations of DM patients and are rarely found, or absent, in the inflamed muscle of PM and IBM patients. CD138+ plasma cells have been identified within the infiltrating populations, predominantly in the endomysial areas of muscle tissue of PM and IBM patients (Greenberg *et al.*, 2002; Greenberg *et al.*, 2005). This was confirmed by sequence analysis of immunoglobulin V-genes expressed by laser dissected cells as well as microarray

The role for B-cells and plasma cells in IM is still currently unresolved, with continuing studies providing further insight into the immune mechanisms. The identification of muscle infiltrating B-cells, plasma cells and autoantibodies suggests that these diseases may be at least partly driven by a loss of B-cell tolerance and, in the case of PM and IBM patients, not solely by the oligoclonal expansion of T-cells. We therefore investigated whether there is

To determine whether specific, antigen-driven, B-cell adaptive immune responses were occurring *in situ*, we used the methods described in section 2 to study the cellular composition of muscle infiltrating cells in twelve different muscle samples (2 DM, 9 PM, 1 IBM); we also examined their Ig V-gene repertoire and the processes of somatic hypermutation and clonal diversification of the rearranged V-genes. In contrast with other autoimmune diseases (see above), no classical ectopic germinal centre structures were observed within the inflamed muscle; instead, muscle–infiltrating cells were present in cellular aggregations which varied from loose to dense in the appropriate perivascular/perimysial or endomysial locations, as in previous studies. B-cells were a significant component of the inflammatory infiltrate in all samples examined for all three myositis subsets, either as CD20+ B-cells or differentiated plasma cells (Figure 7A-D), although the most significant infiltration of CD20+ B-cells was observed within the muscles of the two DM patients. FDCs were rare, and were seen only in one IBM and three PM samples, and not at all in DM. In addition to these cell phenotypes, CD3+, CD4+, CD8+, CD68+ and FoxP3+ cells were also present. Double immunofluorescence staining of cell phenotypes with the proliferating cell marker Ki67 identified proliferating cells within the infiltrating population. In addition to CD20+ B-cells (Figure 7E & F), proliferating CD3+, CD4+, CD8+ and CD68+ cells were observed, as well as FoxP3+ cells in one DM patient.

**4.1.2 Muscle-infiltrating B-cells in myositis** 
