lymphocyte choriomeningitis virus
Central and peripheral tolerance mechanisms were for a long time the only regulatory circuits known in autoimmunity. It is now becoming clear that the target tissue itself may have the capacity to control its own destiny. Here we review mechanisms by which the target tissue regulates local inflammation, and the way this could influence progression to overt autoimmunity. Moreover, we discuss recent data showing that physiological properties of the target tissue can determine the organ specificity of autoimmune disease. These recent discoveries and ideas concerning the regulatory potential of the target tissue may, in the future, add a new dimension to our concept of regulatory circuits in autoimmunity.
Autoimmune disease results from a pathological immune response to self, mediated by autoreactive B or T lymphocytes. The ferocity of the immune response evident once clinical manifestations become visible initially made it difficult to imagine that an apparently uncontrolled attack on self could be restrained once underway. However, some autoimmune diseases, such as multiple sclerosis (MS), undergo cycles of remission 1 and it is known for many autoimmune diseases that tissue inflammation can exist for long periods without overt disease (e.g., in 2, 3). These observations suggest that disease progression is regulated even after inflammation is present in the tissue and initial damage has occurred. This review discusses recent studies highlighting that the targeted tissue can itself modulate inflammation, and therefore has the potential to play an important role in regulating disease susceptibility and severity. This may lead to a further de-centralization and broader understanding of immune tolerance, from central, to peripheral, to local regulation by the target tissue.
The response of the target cell to inflammation determines tissue damage
Inflammation induces a program of changes within the tissue that promote immune surveillance and recognition. Cytokines play a major role in mediating these events, and evidence is now emerging which demonstrates the importance of the target tissue response in conferring susceptibility to autoimmune destruction. This was clearly illustrated in a recent study by Zinkernagel and co-workers 4. The authors used a transgenic mouse model in which pancreatic β cells express a dominant epitope (GP) of lymphocyte choriomeningitis virus (LCMV). In this model, diabetes is induced by LCMV infection. The virus does not extensively infect β cells, but it activates LCMV-GP-specific T cells that recognize the neo-self LCMV-GP antigen in β cells. However, a high frequency of activated and proliferating LCMV-GP-specific CD8 T cells, induced by immunization, is not sufficient to cause disease. Instead, progression to diabetes correlates with the up-regulation of MHC class I by β cells during viral infection. LCMV infection activates TLR signaling and IFN-α production by immune cells, and the subsequent action of IFN-α on β cells enables effective T cell recognition and killing. Similarly, autoimmune liver destruction in mice requires local innate immune signaling that induces pro-inflammatory changes in gene expression in the target liver cells 5. Target cell responses may, therefore, be necessary to allow the progression of autoimmunity, and viral infection is effective at converting target tissue status 4.
The magnitude of the target tissue response to pro-inflammatory cytokines may also determine disease. This has, for example, been demonstrated in studies using the suppressor of cytokine signaling 1 (SOCS-1), a powerful negative regulator of IFN signaling 6. Nonobese diabetic (NOD) mice expressing SOCS-1 specifically in pancreatic β cells are protected from developing spontaneous diabetes. Interestingly, SOCS-1 expression does not prevent insulitis, indicating that target cell cytokine responses are required specifically for the transition from insulitis to diabetes 7, 8. Similar to the study by Zinkernagel and co-workers, these observations suggest that inhibition of target cell cytokine responses can contribute to local regulation of the autoimmune process.
Inflammation-associated changes in target cells: The importance of a healthy dialogue with immune cells
Parenchymal cells, in turn, regulate inflammation by the release of soluble cytokines and chemokines, as well as via cell-cell interactions. While many changes in the tissue induced by inflammation may promote further inflammation and target cell killing, counter-regulatory pathways that affect the resolution of inflammation and protect against tissue damage are also activated. In this way the tissue provides both positive and negative signals that could regulate the progression of an autoimmune response. Some examples of these pathways are given below.
Inflammatory molecules released by the target tissue
The migration of activated lymphocytes into the target organ is dependent on the expression of chemokines and adhesion molecules within the tissue. Activated cells of the innate immune system are effective producers of chemokines 9, but studies on tissue from patients with autoimmune disease have demonstrated that tissue cells also produce chemokines (Table 1). Parenchymal cells express a range of chemokines after stimulation with pro-inflammatory cytokines in vitro (e.g., in 10, 11), suggesting that this is a typical tissue response to inflammation. That chemokines expressed by the target tissue are important for autoimmune disease progression has been suggested by findings in several animal models. For example, Frigerio et al. 12 showed that islet β cells strongly express the CXCR3 ligands CXCL9 (Mig) and CXCL10 (IP-10), and that CXCR3 deficiency delays the development of insulitis and diabetes in the rat insulin promoter-GP LCMV-induced model. More recently, van Loo et al. 13 demonstrated that inhibition of NF-κB activation selectively in CNS-resident cells prevents the expression of inflammatory molecules, including chemokines (e.g., RANTES and CXCL10), and protects from experimentally induced encephalomyelitis.
|Immunomodulatory molecule||Diseasea)||Tissue cell type||Proposed mechanism||Reference|
|Chemokines||CXCL10MCP-1||MS||Astrocytes in MS lesions||Recruitment of T cells Recruitment and activation of monocytes and T cells||717273|
|CXCL13||SS||Epithelial cells in acini and ducts of inflamed salivary glands||Recruitment of B and T cells||74|
|CXCL10Mig||HT||Thyroid follicular cells||Recruitment of T cells||75|
|Cytokines||IFN-α||T1D||Pancreatic beta cell||Up-regulation of MHC I expression, increased recognition by cytotoxic T cells||76|
|IL-15||CeD||Intestinal epithelium||Activation (incl. up-regulation of NKG2D expression) and survival of intraepithelial lymphocytes, induction of MICA expression on enterocytes||19,20|
|Costimulatory factors||Fractalkine||RA||Fibroblast-like synoviocytes||Providing a costimulatory signal to T cells||77|
|NKG2D ligands||RA||Proliferating synovial cells (fibroblast-like synoviocytes?)||Proliferation and cytokine production by CD4+CD28– T cells||30|
|NKG2D ligands||CeD||Intestinal epithelium||Activation of innate immune response, costimulatory signal to gliadin-specific CD8 T cells||29|
|Other molecules||B cell activating factor (BAFF)||RA||Fibroblast-like synoviocytes||Survival of B cells in the inflamed joint||78|
The target tissue also produces cytokines that promote growth and survival of recruited lymphocytes. Production of B cell-activating factor (BAFF) by fibroblast-like synoviocytes has been found in the inflamed joint of patients with rheumatoid arthritis (RA) 14, in salivary glands of patients with Sjögren's syndrome 15, and in lesions in patients with MS 16. BAFF regulates B cell survival and maturation, and may also affect T cell responses (reviewed in 17). IL-15 is a T cell growth factor and its expression in the tissue in many inflammatory diseases may promote chronic inflammation and local tissue damage 18. This cytokine is produced by intestinal epithelial cells from patients with Celiac disease (CeD) 19, 20 and pancreatic islets following in vitro cytokine exposure 10. It is also produced by myocytes during experimental myasthenia gravis 21. Collectively, these observations suggest that by responding to pro-inflammatory cytokines parenchymal cells may directly influence the recruitment, activation, and survival of potentially self-reactive B and T cells. Antagonists for a number of chemokine receptors (e.g., CCR1, CCR2 and CXCR3) are currently in Phase I–II clinical trials aimed at treating RA, MS and psoriasis (reviewed in 9), and their primary site of action is likely to be in the target tissue itself.
Regulation by direct interaction between target and immune cells
Once autoreactive lymphocytes have been recruited to the target tissue, their activity may be further influenced by direct interaction with target cells. One of the most evident changes associated with cytokine exposure in parenchymal tissues is the up-regulation of MHC class I molecules. As discussed above, up-regulation of MHC class I is necessary for conversion to a destructive autoimmunity (e.g., in 4, 8, 22). Elements of the immunoproteosome antigen-processing pathway are also induced by cytokines in autoimmune target cells (e.g., in 23, 24). Changes that the target tissue undergoes during inflammation may, therefore, modify the avidity of the cytotoxic immunological synapse, and facilitate recognition by low-affinity CD8 T cells 25. In this way, target cells may be able to control recognition by autoreactive T cells to limit cytotoxic killing 26, and increased expression of MHC class I and other molecules that increase recognition may play a significant role in promoting autoimmune tissue destruction.
Anomalous expression of NKG2D ligands by the target tissue is also emerging as a common feature of many immune-mediated diseases. NKG2D is an activating receptor on NK cells and a costimulatory receptor on T cells 27. Up-regulated expression of NKG2D ligands occurs in the intestine of patients with active CeD. Locally produced IL-15 (see above) increases NKG2D expression in intraepithelial lymphocytes (IEL). This induces a phenotypic change that enables IEL to mediate epithelial cell damage in an antigen-independent way, by recognition of the NKG2D ligand MIC 28, 29. Synoviocytes in RA patients also express MIC molecules. CD4+CD28– effector cells within inflamed synovia up-regulate NKG2D in response to IL-15 and TNF, and MIC ligands induce proliferation and cytokine production in these cells 30. MHC class II expression has rarely been seen in parenchymal cells, but it is possible that CD4 T cells may also acquire direct cytotoxic capacity through TCR-independent, NKG2D-mediated mechanisms. Ogasawara et al. 31 have shown that NKG2D is highly expressed on pancreatic islet-infiltrating CD8 T cells in the NOD mouse. The murine NKG2D ligand Rae-1 is not generally expressed in normal tissues but is induced during stress, infection and transformation, and is expressed in the islets of NOD mice. NKG2D ligation was shown to be important for diabetes development since blocking NKG2D in NOD mice prevented diabetes. Moreover, the accumulation of proliferating CD8 T cells was reduced specifically in the pancreas, suggesting that activated CD8 T cells in the target tissue are uniquely dependent on NKG2D costimulation. Taken together, these observations suggest that NKG2D-NKG2D ligand interactions contribute to T cell activation and increase cytotoxicity of T cells within the target tissue. The inappropriate expression of NKG2D ligands by the target tissue, in conjunction with the induction of NKG2D by cytokines such as IL-15 and TNF, may therefore be important steps in the development of immune-mediated tissue damage and some autoimmune diseases.
The negative costimulatory molecule PD-L1 (also known as B7-H1) is also up-regulated in murine islets during diabetes progression 32, 33, and its expression in β cells is highly inducible by cytokine treatment in vitro (N.H. and N.S., unpublished data). PD-L1 deficiency greatly accelerates the onset of diabetes in NOD mice, but chimeric mice expressing PD-L1 only in hematopoietic cells remain susceptible to rapid onset diabetes. This indicates that PD-L1 expression in a non-immune cell type is key for diabetes regulation. Islets lacking PD-L1 transplanted into diabetic NOD mice are more rapidly destroyed than wild-type islets, demonstrating that islet PD-L1 expression confers protection against autoimmune destruction 34. It is not yet known how islet PD-L1 inhibits tissue destruction. However, systemic deficiency in PD-L1 and PD-L2 was shown to potentiate the acquisition of effector function during T cell priming, raising the intriguing possibility that this may be linked to the lack of PD-L1 expression in islets 34.
Many other receptor-ligand systems that may play an immunoregulatory role within the tissue during autoimmunity are also coming to light. In a recent publication, Liu et al. 35 suggest that neuron-T cell interactions can convert CD4 effector T cells into Foxp3-expressing regulatory T cells (Treg), promoting neuron survival in vitro and suppressing EAE in vivo. The mechanism of Treg induction is thought to be MHC independent and occur via the interactions of TGF-β and B7 molecules expressed by neurons with their co-receptors expressed on effector CD4 T cells. Moreover, CD40L binding to target cell expressed CD40 molecules may lead to chemokine expression by the target cell 36. A regulatory role for the ligand CD200 (previously denoted OX2) has also been suggested, since CD200 deficiency increases the severity or onset of multiple autoimmune diseases in mouse models, including EAE, collagen-induced arthritis (CIA) 37, autoimmune alopecia 38 and uveitis 39, 40. CD200 interacts with CD200R on myeloid cells, and is thought to modulate the myeloid cell activation threshold 41, 42. Besides being expressed by B and T cells, CD200 is expressed by endothelial cells and neurons 43, and future studies will hopefully demonstrate whether lymphoid or tissue deficiency of the ligand is accountable for the augmented disease severities discussed above.
These examples provide support for the idea that interactions between immune cells and the tissue modulate both innate and adaptive responses. While this field is still developing, both cytokine secretion and cell-contact dependent mechanisms have been shown to occur (Fig. 1), and this communication may affect both the susceptibility of the target to killing, and the potential of self-reactive lymphocytes to destroy.
Genetic variation and target resistance to autoimmunity
In both humans and in mouse models of autoimmune disease, a single genetic locus has sometimes been linked with susceptibility to multiple diseases. The genes underlying such loci are likely to affect a general predisposition to the failure of immune tolerance and development of an autoaggressive immune response, for example, AIRE44, FoxP345, Ctla446 and PTPN2247. However, other loci are clearly disease specific, and presumably modify a generalized predisposition to lack of self-tolerance to confer organ/disease specificity. Some of these genes may be expected to act within the target tissue. In mice, susceptibility to antibody-mediated autoimmune myocarditis was shown some years ago to be due to genetically determined differences in the presence of myosin, the primary autoantigen, in the extracellular matrix of cardiac tissue 48. There is also evidence to suggest that variation in the responsiveness of the target tissue to cytokines and stress can determine genetic susceptibility to autoimmunity. Diabetes resistance in the ALR (alloxan-resistant) mouse strain is associated with a dominantly inherited ability to dissipate free-radical stress, and islets from ALR mice are resistant to CD8 T cell-mediated cytotoxicity and to cytokine- or glucose-induced stress 49. Interestingly, it was recently shown using an NOD congenic strain that alleles at the Idd9 genetic locus that protect against diabetes exacerbate the symptoms of EAE 50. Consistent with this, our data suggest that Idd9 genes act within the target islet tissue to confer protection against diabetes (N.H. and N.S., unpublished data). While these studies suggest that a component of autoimmune susceptibility may, in some cases, map to the target tissue, the gene variants that cause such phenotypes have yet to be determined. By discovering the identity of these genes and how they affect autoimmune susceptibility we may learn much more about physiological mechanisms of target regulation.
Target tissue responses–a fine line between death and survival
The studies discussed until now suggest that the tissue response to pro-inflammatory cytokines (e.g., IL-15, TNF and the IFN) can promote pathological tissue destruction in autoimmunity. However, it is important to keep in mind that these responses also have a biological function by providing protection from infections and tumors, and by contributing to wound healing and tissue regeneration. Indeed, mice harboring SOCS-1-expressing β cells develop diabetes following infection with Coxsackievirus B (CVB), a virus linked to the onset of diabetes in humans 51. Without an intact IFN response, β cells become permissive to CVB infection and infected mice develop diabetes as a result of virus and NK cell-dependent β cell destruction 52, 53. Similarly, SOCS-1 expression in cardiac myocytes worsens CVB3-induced cardiac injury, while expression of dominant-negative SOCS-1 is protective 54. Therefore, intact cytokine responses may be required to prevent loss of infected tissue but, if unregulated, they may promote the progression of tissue infiltration to a pathological autoimmune response, and enable effective killing by cells of the adaptive immune system. This raises the possibility that mechanisms of tissue regulation that normally inhibit excessive damage are defective in individuals susceptible to autoimmune disease. Indeed, protection against autoimmunity may entail both the ability to regulate pro-inflammatory changes and/or to adequately activate inhibitory pathways. Furthermore, the ability of the tissue to fine tune internal survival pathways, or to activate normally dormant regenerative pathways, may provide an additional level of regulation by determining target cell survival in the presence of cytotoxins and inflammatory stress. All these aspects have to be taken into account when new therapies with an aim to prevent tissue inflammation and thereby autoimmunity are designed.
Tissue specific characteristics govern susceptibility
The studies discussed above suggest that the target tissue inflammatory response regulates autoimmunity. With this in mind, it is important to discuss recent studies highlighting that tissue properties provide an additional level of regulation, and may in some cases explain organ specificity. Binstadt et al. 55 attempted to find an explanation for the joint-specific pathology of RA caused by autoreactivity to a constitutively expressed antigen. In the K/BxN TCR transgenic arthritis model, mice develop a T and B cell-dependent autoreactivity against the ubiquitous glucose-6-phosphate isomerase (G6PI) protein. Although the autoantigen in this model is unlikely to be involved in the majority of cases of human RA 56, 57, studies using this model have generated some potentially useful paradigms as to why certain autoantibodies have the potential to cause joint-specific pathology.
Joint-specific disease can be transferred by serum alone, indicating that antibodies are sufficient to initiate disease. Using intravital imaging of circulating fluorogenic probes, it was observed that vascular leak induced by serum transfer occurs specifically in the paws. G6PI antibodies aggregate in the circulation and this characteristic is essential for inducing vascular leak 55. However, the accumulation of anti-G6PI antibodies also occurs primarily in the distal limbs 58, and this may be due to the presence of the normally intracellular G6PI protein at the extracellular surface in the articular cavity 59. These studies in the K/BxN TCR transgenic arthritis model show that peculiarities in the physiology of the joint may contribute to the organ specificity of autoimmune attack in RA 60, both in terms of the preferential accumulation of pre-aggregated antibodies and the vascular leak response 55.
In human RA, circulating autoantibodies recognizing systemically expressed self antigens are also thought to cause joint pathology. Recent studies have convincingly demonstrated that citrullinated amino acids are a specific antigenic determinant frequently recognized by autoantibodies in RA patients 61, 62. Citrullination occurs as a post-translational modification, and can be increased during apoptosis 63. Apoptosis is a feature of many physiological situations, such as tissue damage, remodeling and infection. It has been suggested that the antibody response to citrullinated residues may arise during inflammation in genetically susceptible individuals. However, additional event(s) must occur before chronic arthritis becomes established 2. It is possible that the mechanical function of the joint makes it particularly vulnerable to damage and the presence of increased citrullinated residues; tissue injury has been shown to increase autoantibody binding and promote immune-mediated damage in other disease models 64. While the presence of citrullinated residues is increased in the joint, they are also present systemically 65. However, as discussed above, there appear to be other aspects unique to the physiology of the joint that perhaps also contribute to confer joint specificity in RA.
There are additional examples of abnormal or excessive post-translational modification of antigens within the target tissue that suggest that this may be a feature involved in several autoimmune and immune-mediated diseases. For example, in CeD, the immune response is triggered by gluten, and many of the gluten epitopes that are recognized by T cells contain glutamine residues that have been deamidated by tissue transglutaminase (tTG) 66. Other examples include the presence of antibody responses to epitopes modified by glycosylation or oxidation in patients with systemic lupus erythematosus, RA and type 1 diabetes 67, 68. Moreover, tTG activity is increased during active CeD 69. It is possible that autoimmunity, in some cases, reflects a dysregulation of modifying enzymes in target tissue, for example PADI4 in RA 70. Therefore, organ specificity in autoimmunity is perhaps not only regulated by central and peripheral tolerance mechanisms but also by tissue-specific properties.
The studies cited in this review suggest an increasing awareness of the potential role that the target tissue may play during autoimmunity. Parallels with immune regulation in lymphoid tissues are becoming clear. Target cells express chemokines, pro-inflammatory cytokines and positive and negative costimulatory molecules that may influence B and T cell recruitment, survival and cytotoxicity within the target tissue. Furthermore, some studies discussed here suggest that tissue regulation may be effective in controlling bystander damage during inflammation, but also in the inhibition of immune responses directed specifically against self. Other studies indicate that physiological genetic variants expressed in the target tissue can potentially determine susceptibility to disease, and that specific tissue properties confer organ specificity. Novel insights into the link between infection and autoimmunity have also been gained.
The observations discussed above raise the hypothesis, that the target cells/tissues are active participants in autoimmunity, possessing the capacity to regulate inflammation, and that an imbalance in tissue intrinsic pro- and anti-inflammatory responses may occur in individuals susceptible to autoimmunity. It is still unclear if and to what degree the target response is capable of impacting the outcome of a self-reactive immune response. However, we believe that the observations discussed here provide sufficient basis to suggest that this is a relevant question to consider. It is clear that different regulatory mechanisms may exist in different target organs, and the influence of tissue-specific characteristics is likely to be specific for each particular autoimmune disease. Still, some pathways may be more generalizable. Taking CeD as an example, here the intestinal exposure to gluten and its deamidation by tTG is specific for this particular disease, while the tissue expression of IL-15 and MIC may be a more general tissue response to stress.
As listed in Table 1, there are several examples of regulatory molecules that have been detected, directly ex vivo, in the target tissue of humans affected by autoimmune disease. Many of these molecules are already now considered therapeutic targets and, as this new field of target tissue regulation grows, it is possible that entirely novel approaches to therapeutic treatment of autoimmunity will be discovered.
Note added in proof
The manuscript referred to in the text as ‘unpublished data’, which presents evidence that genetic variation at the Idd9 diabetes susceptibility locus determines the resilience of the targets of autoimmunity, the islets, to destruction has now been published. The full reference is: