**2. Methods**

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

It has now been shown by combined immunohistochemistry, identification of antigen specificity of B-cells and plasma cells in and around ectopic germinal centres, and sequence analysis of expressed, rearranged Ig V-genes and their somatic mutations, that germinal centre B-cells in the target tissues of several autoimmune diseases are undergoing clonal expansion, somatic hypermutation and affinity selection, in a similar manner to that seen in the germinal centres of secondary lymphoid organs (Table 1 and section 1.2). This has been demonstrated in Sjögren's syndrome, rheumatoid arthritis, psoriatic arthritis, myasthenia gravis, multiple sclerosis and also in breast cancer. In some of these cases, expression of RAG1 and 2 have been observed (Armengol *et al.*, 2001), indicating that receptor revision also takes place in ectopic germinal centres and therefore the generation and attempted elimination of self-reactive B-cells. The signals involved in tertiary lymphoid organ neogenesis appear to be similar to those in development of secondary lymphoid organs, although the temporal and causal relationship between appearance of these structures in the target tissue and autoimmune pathology-related tissue damage is unclear. One scenario is that an initial event in the tissue, which could, in some cases, include microbial infection, leads to the release of molecules seen by the immune system as "danger signals" (Matzinger, 2007) thereby inducing infiltration of inflammatory cells and subsequent lymphoid neogenesis, causing further tissue damage with concomitant release of selfantigens, more danger signals and a vicious cycle, perpetuating a chronic autoimmune reaction. Alternatively, initial tissue damage may be caused by an autoimmune response commencing in the secondary lymphoid organs, with subsequent events following a similar course to that described above. Lymphotoxins α, β, α1β2 and TNFα have been shown to be required for development of ectopic germinal centres. Growth-factor receptor-bound protein-2 (Grb2) has recently been shown to control orthotopic lymphoid follicle organisation and the germinal centre response by inducing production of lymphotoxin-α via CXCR5 signalling (Jang *et al.*, 2011). These molecules are secreted by infiltrating B and Th1 cells and activated NK cells; on binding to their receptor on stromal cells they induce expression of adhesion molecules and secretion of chemokines which induce further lymphocyte infiltration and segregation into B-cell follicles, formation of a follicular dendritic cell network and T-cell areas. It has also recently been proposed that overexpression of costimulatory molecules on Tfh-cells may contribute to overcoming B-cell tolerance (Patakas *et al.*, 2011). This may be a contributory factor in ectopic as well as orthotopic germinal centres. Primary B-cell follicles are rarely seen in autoimmune disease target tissues but this may be because chronic antigen stimulation has been in progress for a considerable time before biopsies are taken. For example, in type I diabetes mellitus there is evidence that the autoimmune response develops long before overt disease is diagnosed. Whether ectopic germinal centres are initiated by naïve or memory B-cells is unclear but recent evidence shows that at least some B-cell clones arise *de novo* from naïve B-cells (Sims

*et al.*, 2001; Nzula, Going, & Stott, 2003b; Nzula, Going, & Stott, 2003a).

The frequency of ectopic germinal centres varies markedly between autoimmune diseases; as one might expect, the highest incidence is in diseases where pathogenic autoantibodies are most strongly implicated. Thus, they have been identified in thyroid tissues of 100% of Hashimoto's thyroiditis patients and 54 – 63% of Graves' disease cases; in rheumatoid arthritis the figure is 25 – 50% but in Sjögren's syndrome it is only 17%, although variations may to some extent reflect differences in the difficulty of finding the germinal centres. In Sjögren's syndrome, the source is usually biopsies of the small labial salivary glands of which there is a large number; as g.c.s are only present in some of the many small labial

### **2.1 Identification and cellular composition of ectopic germinal centres**

The methods we used to identify ectopic germinal centres, characterise their cellular composition, analyse the rearranged Ig V-gene sequences expressed by germinal centre Bcells and identify their antibody specificity have been described in detail in previously published papers (Nzula, Going, & Stott, 2003a; Sims *et al.*, 2001). Briefly, sections were cut from snap frozen tissue biopsies and every tenth section stained for B-cells with anti-CD20. Sections containing germinal centre-like structures or B-cell aggregates were further characterised by staining for T-cells (anti-CD3, CD4, CD8), regulatory T-cells (anti-FoxP3), follicular dendritic cells (anti-FDC (DAKO) or anti-CD35), plasma cells (DAKO), macrophages (anti-CD68) and proliferating cells (anti-Ki67). Double immunofluorescent staining with the above cell subset-specific antibodies and Ki67 was used to identify dividing cells. Acetylcholine receptor-specific B-cells in germinal centres from the thymus of myasthenia gravis patients were identified by 125I-α-bungarotoxin-labelled acetylcholine receptor and autoradiography (Shiono *et al.*, 2003; Hill *et al.*, 2008); other autoantibodyproducing cells were identified by immunofluorescence staining with the relevant antigen.

#### **2.2 Cloning and sequence analysis of rearranged Ig V-genes**

Ectopic germinal centres and B-cell aggregates were excised by microdissection, digested with proteinase K and the released DNA used as a template for amplification of the rearranged Ig V-genes by nested PCR. Details of the method and the primers are described in Sims *et al.* (2001) and Nzula *et al.* (2003). Amplified DNA was purified by agarose gel electrophoresis, ligated into plasmid DNA and cloned in *E. coli*. Cloned plasmid DNA was purified and the Ig V-genes sequenced in both directions using primers complementary to sequences flanking the cloning site. The best matching germline V, D & J sequences were identified initially by comparison with the VBASE directory of human Ig V-genes and later, after VBASE ceased to be updated, using the Immunogenetics (IMGT) Database of Human

The Ectopic Germinal Centre Response in Autoimmune Disease and Cancer 403

repeated until the eluate had become enriched with a small number of 'phage clones. These were recloned and their H & L chains sequenced and used to investigate the specificity and

The method of Hershberg *et al*. (2008), described in section 2.2 above, was used to determine the significance of replacement mutations in rearranged Ig V-genes cloned from germinal centre B-cells, as evidence for affinity maturation of the antibodies expressed by them. The distribution of VH gene families and individual V, D and J exons was assessed using two-

Myasthenia gravis is an organ-specific autoimmune disease characterised by weakness of striated muscles and thymic hyperplasia (Vincent, 2002). Patients are generally divided into subgroups with early-onset (EOMG, pre-40 years) or late onset (LOMG, post-40 years) forms of the disease, or with thymoma in about 10% of patients. It is a classic autoantibodymediated autoimmune disease, caused by autoantibodies directed against the postsynaptic nicotinic acetylcholine receptor (AChR) at the neuromuscular junction. Many thymoma patients and some late onset patients also have serum antibodies against striated muscle antigens, interferon-α and IL-12. Loss of functional AChRs leads to muscle weakness, usually first evident in weakness of eye movement. This can progress to other striated muscles of the body, causing problems with breathing due to effects on the diaphragm, swallowing difficulties and paralysis. These effects can be life-threatening if untreated. Evidence that the effects are mediated by autoantibodies against the AChR include induction of similar symptoms by: their transfer from mother to baby *in utero*; passive transfer from patients to mice; immunisation of animals with AChR; and marked improvement of symptoms in patients after removal of circulating IgG antibodies by plasmapheresis. Several pathogenic mechanisms are involved (Vincent, 2002; Drachman, 1994): (i) Cross-linking of the receptor by autoantibodies causes loss of AChR by antigenic modulation, leading to internalisation and degradation of the receptors; (ii) The majority of anti-AChR antibodies are of the IgG1 and IgG3 subclasses, which are particularly efficient at complement activation, resulting in lysis and damage to the muscle membrane; (iii) Less commonly, some antibodies cause direct inhibition of the ion channel function of the AChR; (iv) Antibody-dependent cell-mediated cytotoxicity has also been implicated, although there is little direct evidence for this mechanism. The IgG autoantibodies can cross the placenta of pregnant mothers with myasthenia gravis by an active transport mechanism involving the neonatal Fc receptor, FcRn, resulting in transient symptoms of myasthenia gravis in the newborn infant. The symptoms gradually ameliorate as the maternal antibodies are catabolised and replaced by the infant's own antibodies. More rarely, the autoantibodies produced by multiparous mothers can induce severe, often fatal, developmental abnormalities, termed arthrogryposis multiplex congenita, due to paralysis of fetal muscles

properties of their antigen-binding sites.

**3.1 Pathology of myasthenia gravis** 

tailed 2 analysis, corrected for multiple comparisons.

**3. The ectopic germinal centre response in myasthenia gravis** 

**2.4 Statistical analysis** 

*in utero* (see section 3.4.5).

**3.2 Structure and epitopes of the acetylcholine receptor** 

The AChR is a pentameric transmembrane glycoprotein found almost exclusively at the muscle endplate, comprising two α polypeptide subunits, one β, one δ and, in the adult, one

Immunoglobulin Sequences (http://www.imgt.org/). Sequences were analysed using JOINSOLVER (http://joinsolver.niaid.nih.gov/) and IMGTV-QUEST. Silent and replacement somatic mutations were identified by comparison with the germline gene sequence; in early experiments the ratio of replacement to silent mutations was considered to be evidence of affinity selection if significantly higher than 3:1. To correct for the inherent bias towards replacement mutations in CDRs, we have more recently applied the method of Hershberg to determine whether affinity selection has occurred in B-cell clones from ectopic germinal centres (Hershberg *et al.*, 2008). This method employs an algorithm that allows for the effects of microsequences in the complementarity determining regions (CDRs) and the bias towards transition mutations. Clonally related sets of rearranged V-genes were identified by their use of the same germline V, (D) and J exons and shared junctional sequences. Genealogical trees were constructed by analysis of shared and unshared mutations using the parsimony method of phylogenetic analysis (PAUP, (Swofford, 1993)), enabling the assignment of sequences from parent and daughter cells that have been produced during clonal proliferation, thus providing clear evidence of the presence of clonally proliferating, somatically mutating B-cells within the germinal centre.

#### **2.3 Cloning antigen-specific autoantibodies from germinal centre B-cells by 'phage display**

In order to confirm the antigen specificity of B-cells generated in ectopic germinal centres, and to analyse in detail the relationship between their mutations and antigen specificity, we reconstituted the rearranged Ig V-genes as single chain Fv (scFv) or Fab antibodies by 'phage display. Single chain Fv antibodies comprise the heavy and light chain variable region domains linked by a short peptide. Although linked together by a short additional peptide sequence, the VH and VL domains are able to fold into their natural 3-dimensional conformation and pair correctly, as the antibody produced by a B-cell or plasma cell. They contain the antigen binding site, and therefore mimic the antigen specificity of the original antibodies from which they were derived. A caveat that must be born in mind is that the original H and L chain pairings are unknown, except when both genes are cloned from a single cell. The detailed methodology has been described elsewhere (Stott & Sims, 2000; Matthews *et al.*, 2002). Rearranged VH and VL-genes amplified either from microdissected germinal centres or pooled V-genes from the same B-cell clone, were used to construct scFvs using a (Gly4Ser)3 linker DNA. The resulting scFv library, comprising a pool of randomly linked VH-VL genes, was then inserted into the phagemids pCANTAB6 or pHEN2 continuously with the gene encoding the bacteriophage coat protein P3, and grown in *E. coli* in the presence of helper phage. The resulting scFv-P3 fusion protein was expressed on the surface of bacteriophage or as soluble scFv by transfection into a non-permissive, or permissive, strain of *E. coli* respectively. Alternatively, Fab libraries were constructed using whole light chain cDNA and DNA encoding the VH region and the first constant region domain of the heavy chain by similar techniques (Matthews et al., 2002). An advantage of amplifying directly from genomic DNA is that the distribution of cloned V-genes reflects the usage of those genes by B-cells and plasma cells more accurately than amplification from cDNA, which is biased towards plasma cells. The library of scFvs or Fabs attached to bacteriophage by the P3 'phage coat protein was then panned on plastic plates coated with either a whole extract of the target tissue, or purified recombinant antigen, to identify selfreactive antibodies. Bound 'phage were washed, eluted, re-grown in *E. coli* and panning repeated until the eluate had become enriched with a small number of 'phage clones. These were recloned and their H & L chains sequenced and used to investigate the specificity and properties of their antigen-binding sites.
