1.6 Endocrine autoimmunity

Dysfunction within the endocrine system can lead to a variety of diseases with autoimmune attack against individual components being some of the most common. Endocrine autoimmunity encompasses a spectrum of disorders including, e.g., common disorders such as type 1 diabetes, Graves’ disease, Hashimoto’s thyroiditis, and rarer disorders including Addison’s disease and the autoimmune polyendocrine syndromes type 1 (APS 1) and type 2 (APS 2) (see Table 1.6.1). Autoimmune attack within each of these diseases although aimed at different endocrine organs is caused by a breakdown in the immune system’s ability to distinguish between self and nonself antigens, leading to an immune response targeted at self tissues. Investigating the mechanisms behind this breakdown is vital to understand what has gone wrong and to determine the pathways against which therapeutics can be targeted. Before discussing how self-tolerance fails, we first have to understand how the immune system achieves self-tolerance.

The ability to be able to detect and destroy foreign molecules that have entered our bodies is essential for self-protection and survival. Antigens within the body have to be constantly monitored by the immune system to determine whether they are self, requiring no further action, or nonself, requiring activation by the immune system to ensure removal. This monitoring of endogenous and exogenous antigens occurs by two distinct routes.

Internally or endogenously derived proteins, including those such as tumour or viral antigens, are presented to the immune system by human leucocyte antigen (HLA) class I molecules. Before presentation by the HLA class I molecules, ubiquitin is added to endogenous antigens to enable them to enter and be degraded by the cytosolic pathway. This involves the antigens entering the proteasome, which is composed of several proteases and generates specific HLA class I peptides. Peptides are then translocated from the cytosol into the rough endoplasmic reticulum by Tip-associated protein (TAP), which has the highest affinity for 8–10 amino acid peptides that are optimally bound by HLA class I molecules. The HLA class I α chain and associated β₂ microglobulin (β₂M) chain are synthesized along the rough endoplasmic reticulum. Calnexin associates with free HLA class I α chain to promote folding and β₂M binding. Calnexin is then released and the class I molecule associates with chaperone proteins calrecticulin, PDIA3, and tapasin. Tapasin binds to the TAP transporter bringing the newly synthesized HLA class I molecules into proximity with peptide, aiding peptide capture. Peptide binding further increases HLA class I stability, causing dissociation of calreticulin, tapasin and PDIA3 and exit from the rough endoplasmic reticulum before proceeding to the cell surface of the antigen-presenting cell (APC) for recognition by CD8⁺ T lymphocytes and natural killer (NK) cells. If the peptide is recognized as nonself CD8⁺ T cells become activated and functional effector cytotoxic T lymphocytes (CTLs) are produced, which possess lytic capabilities and play a role in CD8⁺ T memory cell generation. Activated NK cells complement the CTL response by acting before T-cell expansion and differentiation of CD8⁺ T cells and produce lymphokines, including interferons, which aid in the recruitment of additional cells to the site of inflammation and also produce cytokines and chemokines that aid cell destruction (see Fig. 1.6.1).

Extracellular or exogenous antigens, such as bacterial antigens, are handled by a separate pathway involving HLA class II presenting molecules. Exogenous antigens are internalized into the APC via endocytosis or phagocytosis and enter the endocytic pathway. The endocytic pathway consists of a series of compartments termed early endosome, late endosome, and lysosome, and as the antigen progresses through the compartments they become more acidic, leading to a series of proteolytic processes, resulting in the breakdown of the protein into 13–18 amino acid peptides, which HLA class II molecules preferentially bind. The HLA class II α and β chains association within the endoplasmic reticulum, where the peptide-binding domain is occupied by the invariant chain (Ii) to prevent endogenous peptide binding. The Ii also aids the HLA class II molecules to exit the endoplasmic reticulum, traverse the Golgi, and enter the endocytic pathway, where they encounter antigenic peptides. As the HLA class II molecules progress through the increasingly acidic compartments of the endocytic pathway, the Ii is degraded by proteolysis, leaving class II associated invariant chain peptide (CLIP) to occupy the binding domain. Removal of CLIP and exchange for antigenic peptide occurs by the HLA-DM accessory molecule, whose role is inhibited by HLA-DO. Once the HLA class II molecule has acquired peptide it is pulled out of the endocytic pathway and shuttled to the plasma membrane in transport vesicles before being displayed on the cell surface for recognition by CD4⁺ T helper (Th) cells. If the CD4⁺ Th cells determine that the antigen is non-self, two responses are generated, a T helper response 1 (Th1) which leads to macrophage activation, to kill the invading pathogen and a T helper response 2 (Th2) which leads to activation of B cells, which can produce antibodies. Antibodies are soluble copies of the antigen receptor that can bind to and eliminate the invading antigen and also bind to macrophages, neutrophils, and NK cells, stimulating these cells to attack the tissue directly (1) (See Fig. 1.6.1).

Correct functioning of the antigen presentation pathways is vital to enable foreign antigens to be quickly detected and removed, whilst protecting self. Although the need to produce a large repertoire of T and B cells to respond to a variety of invading pathogens is obvious, education of this T and B cell population is also vital to protect against autoimmunity, with systems in place in both the thymus for T cells and the bone marrow/lymph system for B cells to achieve this.

Random T-cell receptor (TCR) rearrangement is employed to enable the generation of a vast T cell population with varying antigen specificities. The downside of the random nature of the rearrangements inevitably leads to some of these T cells being self or autoreactive. Consequently, central tolerance mechanisms are employed during T cell development to ensure these cells do not enter the periphery (Fig. 1.6.2).

Progenitor double-negative CD4^–/CD8^– T cells produced in the bone marrow progress to the thymus where random rearrangement of the TCR β and then TCR α chains (which compose the TCR) occurs, via recombination-activating gene 1 and gene 2 (RAG1 and RAG2, respectively), to become double-positive CD4⁺/CD8⁺ T cells (1). As RAG1 and RAG2 function is random, once the TCR is expressed, the TCR needs to be checked to make sure that it can recognize self HLA molecules (HLA restricted) in a process referred to as positive selection. Positive selection occurs in the cortical region of the thymus, where cortical thymic epithelial cells (cTECs) expressing HLA class I and class II molecules present antigens for recognition by these CD4⁺/CD8⁺ T cells. T cells that bind to the peptide presenting HLA molecules receive a survival signal and progress to the next selection stage, whereas those that do not interact receive no survival signal and die via neglect (1). Once T cells have completed positive selection, they are then subjected to negative selection in the medullary thymic epithelial cells (mTECs) to remove any autoreactive T cells. Any cells that recognize HLA and self-antigens too strongly are either deleted by apoptosis or undergo TCR editing, where additional TCR rearrangements occur to try to prevent them expressing an autoreactive TCR, before being retested for autoreactivity (2). Only T cells that are HLA restricted and are not autoreactive are allowed to mature into CD4⁺ Th or CD8⁺ T cells.

Although central tolerance mechanisms attempt to remove many autoreactive T cells, inevitably some do progress into the periphery, highlighting the requirement for peripheral autoreactivity prevention mechanisms. Along with CD4⁺ Th and CD8⁺ T cell, an additional form of CD4⁺ Th cells (which also express high levels of CD25 and foxp3) are generated known as T regulatory cells (Treg). These cells are mainly formed through interaction with cTECs (involved in positive selection) and are not believed to encounter mTECs (involved in negative selection) during their generation. T regs represent approximately 6–7% of the mature CD4⁺ Th cell population and function by monitoring the periphery for autoreactive T cell activity and suppress the activation and expansion of autoreactive T cells, although the exact mechanisms by which they achieve this are still being elucidated (3).

In a similar manner to T cells, the majority of B cells formed during development are polyreactive and can recognize self and nonself, so also encounter a series of checkpoints to check their activity and/or autoreactive potential (see Fig. 1.6.3). Bone marrow hematopoietic stem cells give rise to progenitor B cells that do not possess a functional B-cell receptor (BCR). The BCR is composed of a immunoglobulin M (IgM) heavy and light chain together with a Igα and Igβ heterodimer (4). Random rearrangement of the BCR heavy and light chain is essential for BCR diversity and development. The heavy chain first undergoes rearrangement mediated by RAG1 and RAG2 and along with a surrogate light chain is presented on the progenitor B cell surface. It then interacts with bone marrow stromal cells to receive survival signals, enabling the cell to start dividing and progress into a precursor B cell. Rearrangement of the precursor B-cell light chain then occurs and together with the rearranged BCR heavy is expressed on the B cell surface as IgM. Further interaction between the precursor B cells and the stromal cells enables them to receive additional survival signals, triggering further proliferation and progression to become IgM expressing immature B cells that exit the bone marrow (1, 4).

Once released from the bone marrow immature B cells need antigen-induced activation for survival and to generate IgG that can be secreted from mature B cells. Immature B cells together with naïve T cells enter the lymph nodes and temporarily sequester into primary follicles and T-cell zones which are found between the follicles, respectively (1). Immature B cells internalize antigen present within the lymph nodes and present the antigen for recognition by antigen-specific naïve T cells at the primary follicle/T zone interface. On interaction, naïve T cells produce lymphokines, causing rapid proliferation and clonal expansion of that specific BCR, which then leads to the generation of a germinal centre consisting of a dark and light zone where two distinct phases of B-cell proliferation and differentiation occur. Within the dark zone activated immature B cells spontaneously undergo random rearrangement of their antibody genes and intense proliferation, giving rise to an immature B-cell-expressing membrane IgG. As B cell numbers increase, they move from the dark zone into the light zone. Due to the random nature of the heavy and light chain rearrangements many of the membrane IgGs produced are autoreactive or nonfunctioning, so are exposed to antigen to check binding ability and affinity. B cells with high affinity receptors for antigen make close contact with antigen displayed on the long extensions of the follicular dendritic cells, which act as an antigen reservoir for both foreign and self-antigens. Centrocytes bearing low-affinity IgG genes do not interact with the presented antigen and are destroyed via apoptosis. Centrocytes bearing IgG bind to antigen presented by the follicular dendritic cells, ingest and process the antigen, and display the antigen on their surface via HLA class II molecules. The B cell then presents the antigen to a T cell, which recognizes and binds to the presented antigen. This enables T-cell-dependent B cell activation to occur which induces the necessary stop signal to prevent apoptosis occurring within the B cell. This enables it to undergo differentiation into a large plasmablast, which migrates to the medulla of the lymph node where it will develop into a plasma cell and begin to secrete antibody molecules, or a small memory B cell which can remain in the lymph node or recirculate to other parts of the body ready to reactivate on reencountering that antigen. This process leads to the generation of a large and diverse repertoire of mature B cells that are not self-reactive and can be activated to produce high-affinity antibodies against foreign antigens.

Negative selection appears to occur at several stages during B cell maturation. Any developing B cell that binds self-antigen too strongly and is autoreactive is destroyed by either receptor editing (where BCR heavy chain undergoes further rearrangement to alter antigen specificity, making the cell less self-reactive), clonal deletion (where BCRs with high affinity for self-antigen are deleted by apoptosis) or clonal anergy (where autoreactive B cells are converted to a state that is no longer responsive to BCR engagement, due to constant receptor occupancy, making the B cell less able to compete for survival signal). Negative selection processes are, however, not complete and some autoreactive B cells do enter the periphery. A subset of regulatory B (Breg) cells that produce interleukin 10 (IL-10) have been identified in mice and have been proposed to play a role in down regulating peripheral B-cell autoreactivity by inhibiting their antigen-presentation abilities and proinflammatory cytokine production but to date identification of these Breg cells in humans is still awaited (5).

Although mechanisms are in place to prevent autoimmunity, over 5% of the general population have autoimmune disease (AID), with the endocrine AIDs representing some of the most common AIDs, suggesting that a breakdown in these mechanisms can occur. Different endocrine diseases have differing population frequencies and vary with respect to the organ/s targeted (see Table 1.6.1). The generation of autoantibodies is common to all endocrine AIDs and tissue destruction is also found in several diseases, with GD being the major exception.

The first evidence that disruption of immune pathways is important in autoimmunity and could be caused by inherited genetic factors came from studies in families which showed clustering of AIDs in family members at a greater rate than would be expected in the general population. Twin studies showed further evidence of a genetic link with autoimmunity. In Graves’ disease monozygotic twins, which share the same DNA, demonstrated concordance levels of about 30–40% compared with rates of 0–7% in dizygotic twins. The incomplete disease penetrance in monozygotic twins suggests that environmental factors also play a role. The relative contribution of genetic and environmental factors is difficult to estimate, with some studies suggesting that genetic factors may contribute up to 80% of the susceptibility. It has also been clearly established that with the exception of APS endocrine diseases are not simple monogenic disorders but are instead polygenic disorders caused by several different genetic effects. The co-existence of different endocrine AIDs and other nonendocrine AIDs (e.g. rheumatoid arthritis) has also been shown to run in individuals and families, suggesting that these diseases share a series of common AID pathways. However, as each of these diseases presents with autoimmunity against a different set of organs, it also suggests mechanisms unique to each disease. Taken together three distinct areas have been identified into which autoimmune endocrine mechanisms fall:

Autoantibody profiles vary disease by disease and even within a given disease autoantibodies can be detected against a variety of different molecules (Table 1.6.2). Variation in thymic expression of antigens and apoptotic clearance of autoantigens has been proposed to explain why certain antigens are targeted for autoimmune attack.

Many of the molecules against which autoantibodies are raised are only expressed in a limited number of tissues. As a result of this limited expression of tissue-restricted antigens (TRAs) it was believed that they were not expressed by mTECs and, therefore, were not tolerated during negative selection of T cells. Investigation into the genetic causes of monogenic APS 1 determined that disease onset was caused by deletions within the autoimmune regulatory gene (AIRE1). APS 1 is characterized by autoimmune polyendocrinopathies, chronic mucocutaneous candidiasis and ectodermal dystrophies, suggesting a potential role for AIRE in controlling autoreactivity. This has been confirmed by findings in AIRE-deficient mice as they have been found to lack a subset of TRAs present in normal mice (6). This indicates that AIRE plays a key role in transcribing TRAs promiscuously in mTECs so that they can also form part of the repertoire of antigens that developing T cells are negatively selected against (see Fig. 1.6.2). AIRE expression is mainly restricted to mTECs and thymic dendritic cells and by mediating transcriptional regulation from multiple sites can transcribe up to 3000 genes including several thyroid, liver, and pancreas TRAs (6). Low levels of AIRE expression has also been detected in the periphery in the gonads, liver, central nervous system, and bone marrow and it has been proposed that the peripheral expression of self-antigens could complement AIRE’s role in the thymus and aid peripheral tolerance (7). Interestingly cTECs, which are involved in positive selection of CD4⁺ Th cells and T regs do not express AIRE (6). A study on interindividual variation in thymic expression of several type 1 diabetes autoantigens, including insulin, glutamate decarboxylase 67 (GAD₆₇), and IA-2, demonstrated that there was up to a 50-fold variation in expression levels (8). An individual’s own variation in TRA expression in the thymus could determine whether certain self-antigens are tolerated or not. AIRE, however, only acts on a limited set of antigens. Certain endocrine autoantigens including thyroid peroxidase and GAD₆₅ are not under AIRE control, suggesting that other genes may also control thymic expression (6).

Whether influenced by AIRE or not, thymic expression does seem to be an important trigger for AID. This concept is eloquently highlighted by the role of the variable number of tandem repeats (VNTR) located in the 5′ region upstream from the insulin (INS) gene in type 1 diabetes. The INS-VNTR consists of tandem repeats of a 14–15 base-pair consensus sequence which clusters into sets of 30–60 repeats (class 1), 60–120 repeats (class II), and 120–170 repeats (class III). Homozygosity of class I alleles was found to be associated with type 1 diabetes, whereas presence of the class III allele offered protection from disease onset (9). When investigating thymic INS expression transcriptional activity was found to be approximately 200–300% higher in INS transcripts encoded by the resistant class III alleles compared to levels of INS transcripts produced by the class I predisposing alleles (a). Mouse models which express low levels of insulin in the thymus presented with spontaneous peripheral reactivity to insulin, whereas mice with normal insulin levels did not, providing further support for this mechanism (10). Similarly the generation of autoantibodies to the thyroid-stimulating hormone receptor (TSHR) in Graves’ disease has also been linked to differential expression of TSHR isoforms. After screening the TSHR for association with Graves’ disease, a number of single nucleotide polymorphisms (SNPs) have been shown to be strongly associated with disease onset. Preliminary mRNA studies have shown differences in relative levels of full length TSHR (flTSHR) and two known TSHR isoforms, ST4 and ST5, in the thyroid between those with and without the associated SNPs. If differences between thyroid and thymic expression of these transcripts is demonstrated, as with INS in type 1 diabetes, this could have an effect on how these isoforms are tolerated during negative selection and/or their availability to be presented to the immune system in the periphery (11).

Apoptosis is a tightly controlled process that maintains homoeostasis in the immune system by deleting potentially autoreactive or nonfunctioning T and B cells, tumour cells, and virally-infected cells and by performing controlled destruction of dead or dying cells. Cells designated for apoptosis go through three phases, triggering, signalling, and execution. The triggering phase involves either engagement of death receptor machinery (including Fas (CD95) and Fas ligand (FasL) or tumour necrosis factor α (TNFα) and TNF-related apoptosis-inducing ligand (TRAIL)) or lack of survival signals triggered by growth factor deprivation, cellular stress, or cytotoxic drugs. Signalling is a multistep process achieved through a series of different accessory molecules including cytochrome c, mitogen-activated protein kinase (MAPK), c-Jun N-terminal kinase (JNK), and protein kinase A (PKA) and PKB, which enable activation of caspase 3, 7, and 8, key initiators of cellular destruction (12). These signalling pathways result in a programme of plasma membrane blebbing, cytoplasmic and organelle contraction and shrinkage, nuclear chromatin condensation, DNA and RNA degradation, and cytoskeletal rearrangements. To prevent proinflammatory cytokine release, apoptosed cells are removed by phagocytes, including macrophages and immature dendritic cells, which also present apoptosed antigens on their cell surface for recognition by the immune system. Apoptotic regulation can occur at several different levels including regulation of death receptor expression levels, expression of proapoptotic and antiapoptotic proteins, including the B-cell lymphoma (BCL) family, and changes in intracellular signalling (12).

Increased apoptosis or defects in phagocytosis of apoptotic cell debris, could lead to increased apoptotic debris that could accumulate and overwhelm the system providing an increased source of autoantigens which if presented to the immune system could trigger an immune response. Lack of or disrupted apoptosis has been proposed as a mechanism for triggering AID, in particular the thyroid autoimmune diseases’ Graves’ disease and Hashimoto’s thyroiditis. Normal thyroid cells express low levels of death receptors and priming by cytokines is needed to trigger these receptors. Immunohistochemical staining of thyroid glands in Graves’ disease suggests that Fas is upregulated and FasL downregulated by infiltrating immune cells (13). This suggests that thyroid cells are less resistant to Fas-mediated apoptosis and lose cytotoxic abilities against invading T cells. Interestingly, in Hashimoto’s thyroiditis, which unlike Graves’ disease displays autoimmune destruction of the thyroid gland, thyrocytes are committed to apoptosis by inappropriate Fas and TNFα mediated signalling. This suggests that on thyroid infiltration by immune cells, which precedes both diseases, there could be a battle in the thyroid gland between apoptosis and thyroid cell production/survival and which one wins could determine whether a person develops Hashimoto’s thyroiditis or Graves’ disease, respectively (14).

Variation or disruption of antigen presentation and T- and B-cell recognition/activation is also key to autoimmune onset. Presentation of antigens by HLA class I and class II encoded molecules during central tolerance and in the periphery dictate how antigens are presented to CD4⁺ Th, CD8⁺ T and Treg cells. Variation within these molecules enables variation between different individual’s immune systems to enable the human race to encounter new threats and survive. This variation occurs throughout the HLA molecules but tends to cluster within the antigen binding domains, suggesting that natural variations have not only protected us from diseases in the past but now may be aiding AID development.

The HLA region on chromosome 6 contains several important immune response genes and is split into three parts, the HLA class I, class II, and class III region. The HLA class II encoded DRB1, DQB1, and DQA1 molecules were the first to be investigated for association with AID, with association between these molecules being detected for most endocrine AIDs. Association of this region was first detected with type 1 diabetes, with the presence of aspartic acid at position β57 of the DQB1 chain shown to confer type 1 diabetes resistance, whereas presence of a neutral residue such as alanine or serine conferred susceptibility (15). Association was also reported at the DRB1 locus, with DRB1*04 and DRB1*03 shown to strongly predispose to type 1 diabetes. DRB1*04 has also been associated with Hashimoto’s thyroiditis and DRB1*03 has been associated with Graves’ disease, Hashimoto’s thyroiditis and Addison’s disease (16). These genes also form haplotypes encompassing DRB1-DQB1-DQA1, with the DR3 haplotype (containing DRB1*03) termed the ‘autoimmunity haplotype’ due to its association with so many AIDs, and the DR4 haplotype (containing DRB1*04) strongly associated with several endocrine AIDs. Strong linkage disequilibrium (in which variation in one gene is also linked with other variants in the same or neighbouring genes) between DRB1-DQA1-DQB1 has made it difficult to determine which individual gene and, in turn, which molecule is the most important. A regression analysis performed in Graves’ disease on the DRB1-DQB1-DQA1 haplotype revealed that DQB1 was unable to explain association of this haplotype with Graves’ disease and that the association was due to DRB1 or DQA1 (17). Further work in Graves’ disease comparing the predisposing DRB1*03 allele against the protective DRB1*07 revealed that DRB1*03 contained a positively charged arginine at position β74, compared with DRB1*07, whch contained a noncharged glutamine (17). DRB1 position β74 has been shown to vary between the lower risk DRB1*0403 and DRB1*0406 T1D alleles, which contains a negatively charged glutamic acid compared with the high risk noncharged polar alanine (16). Position β74 is also part of the shared epitope that is highly associated with RA and is composed of DRB1 positions β70–β74. DRB1-encoded position β74 spans several amino peptide binding domains which are important for antigen/autoantigen binding and TCR receptor docking and interaction, so any variations within this binding domain could be affecting how peptides are presented to the immune system and whether an immune response is mounted against them.

Several hypotheses have been put forward to explain how the DR and DQ molecules could be associated with autoimmunity (16).

Although originally investigation of the HLA region was limited to the HLA class II DR/DQ molecules, other parts of the HLA region also encode key parts of the immune system, none more than the HLA class I encoded A, B, and C molecules, which present endogenous antigen for recognition by CD8⁺ T cells. Association of the HLA class I encoded HLA-B*27 with ankylosing spondylosis has been long established, but it was not until recent advances in statistical modelling for it to be possible to model the HLA class II effects and determine if HLA class I associations are still exerting a primary effect. Analysing over 1729 markers across the whole HLA region in several white Caucasian type 1 diabetes datasets revealed a secondary peak of association after accounting for HLA class II effects and demonstrated independent type 1 diabetes associations for the HLA-B locus and some evidence of association with HLA-A (18). In Graves’ disease, when HLA-B and -C were screened for association and subjected to logistic regression to see if the effects were independent of HLA-DR/DQ, HLA-C and to a lesser extent HLA-B produced stronger association signals than that seen at the HLA class II region (19).

Several hypotheses have been suggested to try and explain these newly detected associations. Unlike HLA class II molecules, HLA class I molecules play a key role in presenting viruses to the immune system. There has been evidence to suggest that viruses could be one of the key environmental triggers for autoimmunity, with several different viruses proposed to play a role in endocrine AIDs (Table 1.6.3). Several different mechanisms have been proposed by which viruses could trigger disease including those listed below (16).

There are also some potential nonviral mechanisms proposed including conversion of the HLA class I molecules themselves into peptides which when presented by HLA class II could cause an autoimmune response to be triggered and cross presentation of exogenous antigen by HLA class I molecules with further studies needed to decipher the exact mechanism at play in endocrine autoimmunity.

T-cell activation is a two-stage process whereby first, the T cell has to recognize and bind to peptides being displayed by a given HLA molecule and second, costimulatory signals are required from accessory molecules on the T cell surface to enable the signal to be transduced and the T cell to become activated. These signals are mediated and controlled by a balance between the T cell surface molecules CD28 and CTLA-4. CTLA-4 appears to downregulate T cell signalling whereas CD28 promotes T-cell signalling. CD28 is always expressed on T cells whereas CTLA-4 is normally up regulated during T-cell signalling but Tregs, unlike other T cells, consistently express CTLA-4 on their surface (20). CTLA-4 could function either by blocking positive signalling pathways or through initiating negative signalling pathways.

CTLA-4 has been proposed to block positive signalling through various mechanisms. CTLA-4 and CD28 both bind to CD80 (B7-1) and CD86 (B7-2) on the surface of APCs. CTLA-4 possess a 50–100-fold greater affinity for these molecules suggesting that CTLA-4 could either out compete CD28 for its ligands or could sequester available ligands, preventing CD28 binding and blocking CD28 positive signalling, causing T-cell anergy (20). It has also been suggested that CTLA-4 reduces lipid raft and microcluster formation that occurs after TCR ligation to increase adaptor molecules and local enzyme numbers essential for T-cell signalling, thereby preventing strong costimulation. More recently a reverse-stop signal model has also been suggested. T cells normally transit rapidly through the lymph node, scanning for APCs displaying antigens, ‘sniffs’ the antigen carefully and quickly moves on unless the antigen shows strong affinity for the TCR, preventing T cells slowing down for weakly bound antigen and weak TCR signalling (21). If a ‘strong’ antigen is detected, increased clustering of adhesion molecule lymphocyte function-associated antigen 1 (LFA1) on T cell surfaces occurs to reduce T cell speed to enable TCR/APC complex (also known as the immunological synapse) stabilization (22). This is known as inside out signalling as TCR binding to the HLA presented antigen, signals to within the cells to produce more LFA1 adhesion molecules which bind to intercell adhesion molecule 1 (ICAM-1) on the surface of the APC to further strength the interaction at the immunological synapse (21). Stable immunological synapse formation is key for TCR engagement and scanning of HLA presented peptide as there is minimal half-life between the HLA-TCR interaction necessary to produce a productive TCR signal. CTLA-4 controls LFA1 production thereby controlling T cell motility, which is proposed to reverse or override the TCR-induced upregulation of adhesion factors prematurely disrupting immunological synapse formation (22). Limiting TCR/APC contact time could result in more avid interactions still occurring but less reactive, low affinity antigens may be ignored, suggesting that not every antigen could be screened during central tolerance and that autoreactive T cells against low affinity peptides could escape central tolerance.

CTLA-4 is also believed to directly activate negative signals to prevent or dampen down TCR signalling by binding several protein tyrosine phosphatases including SHP2 and PP2A, which inhibit cell signalling proteins recruited to the TCR by dephosphorylation (23). CTLA-4 also inhibits JNK and elk-related tyrosine kinase (ERK) leading to reduced production of several transcription factors including nuclear factor κ-B (NF-κB), nuclear factor of activated T cells (NF-AT) and activator protein 1 (AP1). CTLA-4 can also up regulate the tryptophan degrading enzyme IDO (EC number 1.13.11.52), which can breakdown tryptophan in a manner that can inhibit T-cell activation (20).

Association of CTLA-4 has been consistently reported with most AIDs, with fine mapping studies revealing that association was due to a small number of SNPs located within a 6.1 kb block (24). CTLA-4 exists in humans as both a full length version anchored to the T-cell membrane (flCTLA-4) and soluble form containing no transmembrane domain and, therefore, not anchored to the cell (sCTLA-4). Studies comparing flCTLA-4 and sCTLA-4 mRNA levels in serum and plasma samples demonstrated that possession of the susceptibility haplotype of these SNPs affected efficiency and splicing of sCTLA-4 producing less sCTLA-4 than the protective haplotype. Increased sCTLA-4 could be a marker of increased T cell activity, suggesting that possession of the susceptibility haplotype increased T cell function (24). These results could also indicate downregulation of Treg function. Other studies failed to detect sCTLA-4 in serum or replicate this effect suggesting the mechanism for action requires further confirmation.

PTPN22 is another inhibitor of T-cell signalling, but acts further downstream than CTLA-4 on several molecules including lymphocyte-specific protein-tyrosine kinase (Lck), ζ-chain associated kinase (Zap-70), CD3ε/TCRζ-chains and valosin containing protein that all control T-cell signalling. The C1858T SNP within PTPN22 has been consistently associated with several endocrine AIDs including type 1 diabetes, Graves’ disease, and Hashimoto’s thyroiditis and a series of other nonendocrine AIDs including rheumatoid arthritis (25). The C1858T variation encodes an amino acid change from arginine to tryptophan in the PTPN22-encoded LYP molecule at amino acid position 620 (R620W). The R620W variation is located within the first of four proline rich regions (P1) within LYP and interacts with the SH3 domain of Csk, an important intracellular tyrosine kinase. Csk suppresses the negative regulatory tyrosine in the c terminus of Lck (and Fyn) by dephosphorylation, leading to inhibition of Lck kinase activity which plays a role in T-cell signal transduction. Presence of LYP*620W severely impairs Lyp-Csk complexes and acts as a gain of function mutation by causing increased T-cell inhibition by dephosphorylating Lck and other signalling proteins more efficiently than LYP*620R (26).

Several mechanisms have been proposed to explain the LYP*620W gain of function. It has been suggested that stronger downstream inhibition of TCR signalling seen in those with LYP*W620 could effect autoreactive T-cell negative selection signals, particularly those with moderate autoreactive affinity, leading to a failure to delete these molecules prior to entry into the periphery (26). Presence of LYP*620W in Tregs may also inhibit their signalling pathways potentially preventing peripheral autoreactive T cell deletion (25). PTPN22 also interacts with other adaptor molecules including c-Cbl, a proto-oncogene which becomes phosphorylated after T cell stimulation and whose expression is reduced when LYP is overexpressed, and growth factor receptor bound protein 2 (Grb2), which like Csk has a SH3 domain binding site for LYP and is involved in negative regulation of the CD28 signalling pathway (27). Interestingly, interaction between CTLA-4 and LYP has been postulated. CTLA-4 and LYP both interact with Fyn, Lck and Zap70, with CTLA-4 believed to use LYP complexed with Grb2 to aid in downregulating T cell activation (27). LYP has also been proposed to play a role in lipid raft formation which CTLA-4 is believed to downregulate (25). PTPN22 is also expressed in other cell types including B cells, NK cells, macrophages, and dendritic cells and could have an, as yet, unidentified role in controlling their signalling (26). Unsurprisingly, potential additional effects independent of R620W have also been detected in PTPN22, suggesting that there could be other variations in PTPN22 leading to disease onset but due to a lack of replication between different studies further evidence is required to confirm these additional affects (26).

Between 60 and 70 of the over 100 PTPNs encoded within the human genome act as positive or negative regulators of T-cell activation (25), suggesting that further family members may too be playing a role in AID onset. PTPN2 is one such family member. The PTPN2 knockout mouse (lacking homologous TCPTP) exhibits defective T- and B-cell development and activation and when investigated within humans, variations within PTPN2 were associated with type 1 diabetes, Graves’ disease, and coeliac disease, with further work being performed to decipher the underlying disease mechanism (28).

Disrupted Treg function is believed to be an important factor in preventing/controlling autoimmunity onset and specific mechanisms that control Treg function on top of CTLA-4 and PTPN22 have been identified. IL-2 mainly produced by activated T cells, promotes proliferation and enhances cytokine production. In Tregs IL-2 influences development and enhances Tregs ability to induce apoptosis of autoreactive T cells. IL-2 signals through the IL-2 receptor, which is composed of three subunits, an α chain (CD25 or IL-2 receptor α (IL-2Ra)) whose expression is restricted to T cells, in particular T regs, and a β (CD122) and γ chain (CD132) which are expressed on a variety of tissues and are involved in several cytokine signalling pathways. Screening of IL-2Ra and the surrounding region in type 1 diabetes showed strong evidence of association of two IL-2Ra SNPs with disease. Investigating individuals carrying two copies of the predisposing allele of either SNP had lower log concentrations of soluble IL-2Ra (sIl-2Ra), a marker of cell proliferation, than those carrying one or no copies, suggesting that reduced IL-2 signalling correlates with reduced T cell and, in particular, Treg function, which can in turn effect how the periphery is policed for autoreactive T cells (29).

Autoantibody production by B cells is key to autoimmune onset and can be either directed through binding to a receptor (such as TSHR autoantibodies in Graves’ disease binding the TSHR) or through formation of immune complexes in tissues that locally activate the complement cascade (1). Hypermutation in the BCR during affinity maturation, failure to remove autoreactive B cells in the bone marrow/lymph nodes and periphery, and perturbations in signalling thresholds could all play a role in disease onset. Disruptions in several molecules that control B-cell signalling has been suggested, including B-cell activating factor (BAFF), whose expression in secondary lymphoid tissue is vital for providing prosurvival signals that enable transition from immature to mature B cell and sustaining long-term memory B cell survival (30). Inappropriate overexpression of BAFF can promote autoreactive B cells survival rather than deletion (1) with animals that express high levels of BAFF experiencing a number of autoimmune manifestations including high circulating antibody levels and immune complex formation in serum and kidneys. Variations within Fc receptor like 3 (FCRL3), which encodes a member of the FC receptor-like family of proteins involved in regulating B cell signalling, have been detected in Graves’ disease and other nonendocrine AIDs, including rheumatoid arthritis, which has been shown to disrupt gene expression and has been proposed to lead to unregulated B-cell activation.

Traditionally for many endocrine AIDs it has been viewed that B cells initiate autoimmunity and T cells progress disease. In the NOD mouse, a model for type 1 diabetes, for example, B cells are the first molecules to infiltrate the pancreas (31). The view of B cells as just producing autoantibodies and acting as bystanders in autoimmune disease such as type 1 diabetes, Graves’ disease, and Hashimoto’s thyroiditis, progression has been revised recently due to their ability to act as APCs to CD4⁺ Th cells in low-antigen environments and their abilities to regulate inflammation through cytokine production. These features point to a larger and more active role for B cells in autoimmunity. It has also been suggested that T-cell-independent B-cell activation can also occur whereby antigens function as direct mitogenic stimuli causing antigen-specific B-cell activation through toll-like receptors (TLRs) or polysaccharides that directly engage the BCR (1). As TLR ligands, such as bacterial DNA and stimulatory CpG-oligodeoxyribonucleotides, are potent activators of B cells this could suggest another a way in which bacteria could be triggering autoimmunity.

Although many of the mechanisms behind disease onset have so far focused on disruptions to specific molecules within the immune system, there are several ‘external’ factors that can also impact on the immune system, potentially triggering autoimmunity.

Many AIDs have a strong female preponderance (see Table 1.6.1). Increased immune responsiveness in females, sex hormones, fetal microchimerism, and the presence of susceptibility loci on the sex chromosomes have all been put forward in an attempt to explain the female preponderance, although no single hypothesis has been confirmed. More recently, skewed X inactivation (XCI) has also been proposed as contributing to the female preponderance. During early development, females inherit one X chromosome from their father (XF) and one from their mother (XM). Males only inherit one X from their mother and a Y chromosome from their father. To enable dosage compensation to occur in females, one of the two X chromosomes present is randomly inactivated via methylation. Although XF:XM should be inactivated in a ratio of 50:50, skewed XCI can occur whereby more than 80% of one parent’s X chromosome is inactivated. Evidence for higher rates of skewing have been detected in several Graves’ disease datasets with 34–49% skewing seen in Graves’ disease cases versus only 1–12% in control subjects (32). It has been proposed that in skewed XCI individuals, antigens on one X chromosome may fail to be expressed at a sufficiently high level in the thymus, preventing the immune system tolerating these antigens. When these antigens are presented to the immune system later in life they may, therefore, be recognized as foreign and an autoimmune response mounted, although further study is required to confirm these effects.

Detecting the environmental contribution to disease is not easy because of the problems inherent in studying environmental impact during human development including the need for long-term follow-up and the reliance on patient recall. Even with these caveats in place, numerous potential environment factors have been suggested, including viruses (Table 1.6.3) and bacteria, chemicals, and stress (Table 1.6.4). These environmental factors are believed to impact upon the immune system in several different ways. First, fetal/maternal features such as birth weight, weight gain during pregnancy, and caesarean birth have all been proposed to contribute to onset of type 1 diabetes. Second, simply by the introduction of foreign particles into the body so that when the immune system tries to remove them autoimmunity is triggered as a side effect, as proposed for viruses or bacteria (Table 1.6.3 and HLA class I associations section). This is further supported by seasonal variation in the presentation of type 1 diabetes and Graves’ disease. In the general population the majority of births occur within the spring or summer. This pattern is altered in type 1 diabetes and Graves’ disease, with higher numbers born in the autumn or winter period, when an increased incidence of viral and bacterial infection occurs. Finally it can be affected by altering how the immune system functions. Stress and smoking are known to have immunosuppressive effects by stimulating the hypothalamo–pituitary–axis, which downregulates immune responsiveness. For example, an increase in the number of Graves’ disease cases has been noted during wartime and in type 1 diabetes both parental separation and bullying have been investigated as risk factors for disease onset. In type 1 diabetes there has been much debate concerning the benefits of breastfeeding over bottle feeding. It has been proposed that babies who are fed on cow’s milk get more exposure to cow insulin leading to antibody formation. These antibodies could cross-react and attack an individual’s own insulin-producing cells, whereas those who are breastfed would not get such early exposure to cow insulin. Although several studies have now been performed the data are inconclusive and further studies are required (33).

A lack of challenges to the immune system by foreign environmental factors has also been suggested as a potential cause of autoimmunity. The hygiene hypothesis suggests that changes in social behaviour combined with greater access to cleaning products could be contributing to the increased rates of AIDs. More sterile environments with a reduction in invading organisms could lead to autoimmune attack as our highly primed immune systems with less ‘foreign’ material to focus on could start to attack self-components.

Several large, long-term follow-up studies are currently being performed to evaluate the contribution of environmental factors, including investigating why different populations have variable disease rates, to see if changes in these differing populations’ environments could provide further insights.

In summary, this chapter highlights some of the key mechanisms that are at play to prevent autoimmunity and describes how disruptions within the immune system, both internally and externally can lead to the development of endocrine autoimmunity. As a result of advances in new genetic screening methodologies and long-term studies into environmental factors, our understanding of these mechanisms are constantly being updated and expanded on, with each new discovery helping to further identify the complex underlying pathologies involved in these diseases.