MECHANISMS OF AUTOIMMUNE DISEASE INDUCTION

Transcription

1 458 ARTHRITIS & RHEUMATISM Volume 38 Number 4, April pp , American College of Rheumatology REVIEW MECHANISMS OF AUTOIMMUNE DISEASE INDUCTION The Role of the Immune Response to Microbial Pathogens SAMUEL M. BEHAR and STEVEN A. PORCELLI Autoimmune diseases are clinical syndromes in which tissue damage appears to result from aberrant responses of the immune system to normal self antigens. Many well-recognized conditions, including a majority of the systemic rheumatic disorders, are currently grouped in this etiologic category. The central mechanism of these diseases is thought to be a defect in immunologic tolerance, resulting in the activation and expansion of self antigen-specific T and B lymphocyte clones and the production of circulating autoantibodies and a myriad of cytokines and other inflammatory mediators. Although much information has accumulated concerning the detailed immunopathology of autoimmune diseases, it is still not known what causes them. Our failure to understand the events triggering these diseases is a serious obstacle to the development of more effective preventive or therapeutic measures. The idea that microbial species, including bacteria, viruses, and parasites, may represent the major environmental factors that initiate and sustain anti-self immune responses continues to exert a powerful influence on autoimmune disease research. This idea arises naturally from the obvious fact that the immune system has evolved mainly to recognize and respond to microbial pathogens. Unlike the conceptually straightforward role of microbial organisms in infectious diseases, the mechanisms by which these agents may circumvent immunologic tolerance to self antigens and Dr. Behar s work was supported in part by NIH training grant AR Dr. Porcelli s work was supported by NIAMS Clinical Investigator Award AR Samuel M. Behar, MD, PhD, Steven A. Porcelli, MD: Brigham and Women s Hospital and Harvard Medical School, Boston, Massachusetts. Address reprint requests to Steven A. Porcelli, MD, Department of Rheumatology and Immunology, Room 508 Seeley G. Mudd Building, 250 Longwood Avenue, Boston, MA Submitted for publication August 29, 1994; accepted in revised form November 15, trigger autoimmune disease remain difficult to fathom. In this article we summarize some of the areas of current research that may eventually clarify the mechanisms by which infectious agents could act as the inciting agents of autoimmune diseases. Molecular mimicry by microbial antigens: a triggering mechanism for autoimmune disease? Microorgansims produce a multitude of foreign antigens that collectively comprise the major set of determinants recognized by the immune system. These antigens potentially include a wide variety of carbohydrates, lipids, and proteins that can be recognized by specific receptors of inflammatory cells. One set of specific immune receptors, the B cell receptors or immunoglobulins, can and does recognize all of these substances. In contrast, the antigen receptors of T cells, the major regulatory cells for most specific immune responses, have so far been found to recognize almost exclusively peptides derived from protein antigens, although a few exceptions to this general rule have recently been brought to light (1-3). The processes leading to the production of these antigenic peptides and their presentation to T cells by the class I and I1 major histocompatibility complex (MHC) molecules are now quite well understood at least in a general way (for review, see ref. 4), and serve as the cornerstone of our present understanding of specific cellular immunity. In one frequently cited model of autoimmune pathogenesis, microbial antigens with structural similarity to self antigens are viewed as stimuli that induce immune responses that cross-react with self tissue antigens, leading in some way to a subversion of normal tolerance mechanisms and a self-perpetuating autoimmune response (for review, see refs. 5 and 6). The first part of this hypothesis (i.e., the sharing of

2 INDUCTION OF AUTOIMMUNITY BY MICROBIAL PATHOGENS 459 antigenic determinants by a microbe and its host species, commonly termed molecular mimicry [7]) is now well established as a frequent occurrence in nature. Immunologic cross-reactivity between microbial antigens and self tissues by serum antibodies has been demonstrated experimentally in many situations in humans and laboratory animals (for review, see ref. 6). In addition, surveys of the growing database of protein sequences have revealed many similar potential T cell epitopes (i.e., peptides with related or identical sequences) that are shared between microbial and mammalian species (5,6,8). Although it is still impossible to accurately predict which of these potential T cell epitopes are actually processed and presented to T cells in vivo, synthetic peptides corresponding to some of these epitopes have been shown experimentally to result in cross-reactive cellmediated immunity (i.e., the expansion of T cells that recognize both the immunizing peptide and the similar self peptide that it mimics) (6). Thus, it has been proposed that one mechanism by which microorganisms may trigger anti-self responses is by stimulating strong immune responses to microbial antigens that bear some similarity to self tissue components. In the presence of these strong immune responses, it is believed that cross-reactive T and B cell clones arise that actually react against homologous self antigens that would normally not be immunogenic. In this way a protective immune response against an invading microbe is envisioned to set off an autoimmune response that may be either transient and self-limited, or under the influence of various host and environmental factors, chronic or relapsing. The clinical features of the resulting autoimmune disease are thought to depend largely on qualitative features of the immune response (i.e., cellular versus humoral, which cytokines predominate, etc.) and on the tissue distribution of the particular target self antigens. Although conceptually straightforward, the importance of molecular mimicry in the actual pathogenesis of autoimmune disease is still debated. This hypothesis seems to provide a plausible explanation for inflammatory conditions associated with autoimmune responses that occur in the setting of recent prior infection by a distinct microorganism, such as the nonsuppurative sequelae of group A streptococcal infections (i.e., rheumatic fever or glomerulonephritis) or epidemic Reiter s syndrome and reactive arthritis. Nevertheless, it must be conceded that in spite of considerable research on a variety of diseases and disease models, no example of an infectious process triggering autoimmunity through a mechanism based on molecular mimicry has yet been conclusively demonstrated. Furthermore, the model as it currently stands leaves unanswered a number of important questions. Autoimmune disease models based on molecular mimicry presuppose that autoreactive T cells exist prior to autoimmune disease induction but are maintained in a nonresponsive state of anergy or ignorance. Although this does appear to be the case (for review, see ref. 9), it is necessary to explain why these T cells escape the major mechanism of tolerance induction by thymic deletion of autoreactive T cells, and also how they are normally maintained in a silent state. Furthermore, models of autoimmune disease initiated by infectious agents must explain how the autoimmune state is sustained after the apparent disappearance of the inciting agent, leading to the chronic or relapsing clinical course of many autoimmune syndromes. Refining the molecular mimicry hypothesis: new insights from animal models of autoimmunity. Many excellent animal models have been developed that support the idea that under appropriate experimental conditions, immunization by proteins with similarity to self antigens can stimulate an immune response that ultimately leads to autoimmune disease. Most of these models, several of which are summarized in Table 1, involve the hyperimmunization of genetically susceptible animals with foreign proteins that are homologous to tissue-specific self proteins. Although strictly speaking this is not molecular mimicry (which involves immunization by nonhomologous microbial antigens that structurally resemble unrelated antigens of the host), the mechanism leading to autoimmune disease induction in such models is generally believed to parallel that which comes into play during molecular mimicry of a self antigen by an infecting microorganism. A noteworthy feature of most of these models is the requirement for coadministration of microbial products (usually Freund s complete adjuvant, an emulsion of heat-killed Mycobacterium tuberculosis suspended in an oil base) to efficiently stimulate the anti-self response. The precise role these bacterial products play in induction of autoimmunity is not well understood. In fact, the widespread use of bacterial adjuvants in such animal models and in immunologic research in general often receives relatively little attention, prompting one noted expert to refer to this as the immunologist s dirty little secret (10). Although a few animal models in which autoimmunity appears to

3 BEHAR AND PORCELLI Table 1. Animal models of autoimmune disease induction by molecular mimicry Disease induction Human disease Animal model Species Antigen Cofac tors Target tissue (antigen) counterpart Experimental allergic Rodents, Foreign myelin basic Freund s complete Central nervous system Multiple sclerosis encephalomyelitis primates protein (MBP), adjuvant (FCA), (self MBP) Experimental acute Rodents self MBP peptide Foreign nicotinic pertussis toxin FCA (proteolipid protein) Neuromuscular Myasthenia gravis myasthenia gravis acetylcholine junction (self receptor (nachr) nachr) Collagen-induced Rodents Foreign type I1 FCA Joints (self type I1 Rheumatoid arthritis arthritis collagen collagen) Adjuvant-induced Rats Mycobacteria (heat FCA Joints (self heat-shock Rheumatoid arthritis, arthritis killed) protein 60?) reactive arthritis Coxsackievirus-induced Mice Coxsackievirus Myocardium (cardiac Viral or idiopathic myocarditis infection myosin?) myocarditis follow a more natural form of immunization (e.g., coxsackievirus myocarditis in mice) also exist, the role of molecular mimicry in these models is less well established (for review, see ref. 6). An important concept that has recently emerged primarily from studies of these animal models is that of the existence of an extensive repertoire of self antigens that is normally not accessible to the immune system, but becomes so upon induction of an inflammatory response directed against self antigens. Earlier versions of this idea focused on immunologically privileged or sequestered sites, i.e., anatomic locations that were not accessible to the immune system because of avascularity (e.g., the cornea) or because of anatomic barriers (e.g., Bowman s capsule in the kidney, and possibly some or all of the central nervous system). More recently it has been recognized that antigens exist in all anatomic sites that are normally not recognized by the immune system because antigen-presenting cells do not normally process and present them efficiently for recognition by T cells. These self antigens that are not normally perceived by the immune system are collectively referred to as cryptic self (1 1). The existence of cryptic epitopes has been clearly demonstrated by examining the T cell responses of normal mice to both foreign or self protein antigens (for review, see ref. 11). For example, immune responses of normal mice to autologous cytochrome c have been studied to explore the role of cryptic self in the generation of autoreactive T cells, and these studies are summarized here to highlight the major implications of this model for autoimmune disease pathogenesis. Cytochrome c is a heme-containing protein that is ubiquitously expressed in all tissues as a component of the mitochondria1 electron transport chain. As expected, normal mice are tolerant to this self protein and do not mount T cell or antibody responses following immunization with autologous cytochrome c, although they do respond when immunized with cytochrome c from other species (e.g., pigeon, horse, rabbit, or dog) (12). Since mice immunized with foreign cytochrome c proteins generate significant levels of antibodies that bind both foreign and self cytochrome c molecules, it appears that tolerance to self cytochrome c is maintained mainly at the T cell level (12,13). Additional studies in this system show that this tolerance is not due exclusively to deletion of autoreactive T cell precursors responsive to mouse cytochrome c. In fact, T cells specific for self cytochrome c are present in normal animals and can be elicited by at least two different manipulations. For example, concurrent immunization of mice with a mixture of foreign (e.g., human) plus autologous mouse cytochrome c proteins leads to the activation of autoreactive T cells specific for determinants in the amino terminal portion of mouse cytochrome c (e.g., amino acid residues 1-80) (13). Alternatively, although mice are tolerant to immunization with intact autologous cytochrome c, a synthetic peptide corresponding to the carboxy terminus of this self antigen (i.e., residues ) is an effective T cell immunogen. The ability to respond to this peptide is probably not due to complex immunoregulatory factors, such as the absence of suppressor T cell epitopes (14). Instead, it appears that the likely explanation for these results is that the epitope is normally not generated by antigenpresenting cells at levels sufficient to allow recognition by autoreactive T cells after processing of the intact

4 INDUCTION OF AUTOIMMUNITY BY MICROBIAL PATHOGENS 46 1 Table 2. Proposed mechanisms by which microbial infections could augment presentation of cryptic self epitopes Mechanism Cause Effect Activation of antigen-presenting cells (APCs) Generation of autoreactive B cells Cellular stress response Access to sequestered antigens Direct infection of APCs, cytokine effects, response to microbial products Molecular mimicry of self antigen by microbial antigen Direct infection of APCs, metabolic effects of infection: fever, hypoxia, etc. Destruction of anatomic barriers, increased vascular permeability * MHC = major histocompatibility complex; CAMS = cellular adhesion molecules. Increased phagocytosis and antigen uptake, altered antigenprocessing machinery, up-regulation of T cell-activating molecules (i.e., antigen-presenting molecules [class I and I1 MHC, CDl], adhesion molecules [integrins, Ig-CAMS, selectins, others], costimulatory molecules [B7-1, B7-2])* Uptake of self antigen by autospecific B cell receptors, increased delivery of self antigen to endocytic compartments Up-regulation of heat-shock proteins (hsp), presentation of self heat-shock protein peptides to T cells, binding of peptides to class I1 MHC facilitated by hsp70 Increased uptake and processing of self antigens by APCs, influx of APCs and self antigen-specific lymphocytes cytochrome c protein. The peptide is thus an example of a cryptic epitope, i.e., one that is not normally generated by antigen-presenting cells at levels sufficient to activate T cells. The concept of cryptic self is useful in explaining the existence of potentially autoreactive but silent T cells in the blood and peripheral lymphoid organs of normal animals. These T cells are neither deleted in the thymus nor tolerized (e.g., rendered nonresponsive or anergic ) extrathymically because the self antigens that they recognize are normally not presented at levels that can be recognized. Technically, these T cells are not tolerant since they are capable of responding when confronted with an appropriately presented antigen. Rather, they are considered to be ignorant since they are functionally unaware of the existence of the antigens they recognize, even though these may actually exist in abundance all around them. Recognition of this situation allows us to ask a central question relevant to the activation of autoreactive T cells: What is it that leads to the presentation of previously cryptic epitopes, thereby activating these previously ignorant autoreactive T cells? As a corollary, one can ask how exposure to various infectious organisms might modify the processing and presentation of antigens by antigen-presenting cells, thus leading to the presentation of cryptic epitopes and the activation of previously ignorant autoreactive T cells. Although it is probable that there are multiple mechanisms that lead to the presentation of cryptic self, studies carried out in the cytochrome c system indicate that tolerance is broken by the generation of cross-reactive B cells elicited by the foreign cytochrome c immunogen (12). These B cells are potent antigen-presenting cells that efficiently bind and inter- nalize self antigen because they express, on their surfaces, immunoglobulin receptor molecules that bind to the self antigen with high affinity. Once bound to the B cell receptor, these self antigen molecules are rapidly endocytosed and efficiently directed into antigen-processing compartments within the endosoma1 network of the B cell (15,16). This pathway is believed to up-regulate the generation of cryptic self epitopes, leading to the activation of autoreactive T cells. The generation of such cross-reactive B cells thus provides one general mechanism by which molecular mimicry by the antigens of a microbial pathogen could act as a trigger of autoimmunity. In this case, it is the ability of a similar microbial antigen to break tolerance at the B cell level that gives rise to the generation of the cross-reactive B cells, the first step in a chain of events that leads to the presentation of cryptic self epitopes and the activation of autoreactive T cell clones. A variety of other mechanisms by which microbial infection could alter antigen processing and presentation to promote cryptic self recognition have been suggested, some of which are summarized in Table 2 (for review, see refs. 11 and 17). Experimental evidence for most of these is only rudimentary at present, and it remains to be determined which, if any, are actually relevant to the generation of autoimmunity. It is noteworthy, however, that some studies suggest that substances derived from pathogenic bacteria or even normal body flora might augment the presentation of cryptic self, leading to the activation of previously ignorant autoaggressive T cells. An impressive example of this is seen in a transgenic mouse line that spontaneously develops experimental allergic encephalomyelitis (EAE), an inflammatory demyelinat-

5 462 BEHAR AND PORCELLI ing condition that has been widely studied as a model for human multiple sclerosis (MS). Transgenic mice in which the vast majority of T cells express an antigen receptor specific for the central nervous system (CNS) self antigen myelin basic protein (MBP) were constructed. EAE was observed to spontaneously develop in these mice if they were bred and housed in a nonsterile facility, but not if they were bred and maintained in a sterile, specific pathogen-free facility. This suggested that the presentation of MBP leading to activation of transgenic T cells was somehow augmented by colonization of the animals with normal microbial flora, or perhaps by contact with ubiquitous pathogens. In support of this idea, injection of these mice with several different bacterial products, including pertussis toxin and lipopolysaccharide, also led to an increased frequency of spontaneous develpment of EAE (18). The manner in which these and possibly other bacterial products lead to the overt disruption of tolerance in these animals is not yet known, but might involve an up-regulation of antigen presentation and the enhanced display of a previously cryptic MBP epitope. Interestingly, it has recently been reported that among HLA-B27 transgenic rats that develop a spontaneous arthropathy with other associated pathologic features similar to those of human Reiter s syndrome and ankylosing spondylitis (19), clinical disease and pathologic changes are greatly attenuated in animals bred and maintained in sterile, germfree conditions (20). This suggests that autoimmune disease induction in these animals may also be triggered by products of ubiquitous microbial flora, possibly through a mechanism that induces presentation of previously cryptic self epitopes. Epitope spreading following induction of autoimmunity. A major obstacle that must be overcome in order to establish molecular mimicry as a mechanism leading to spontaneous autoimmune diseases is the identification of the precise epitope or epitopes that initiate the putative cross-reactive immune responses in susceptible individuals. In other words, what is the precise foreign antigen and what is the structurally related self antigen that are relevant to autoimmune disease induction? The great difficulty in answering these questions arises partly from the common occurrence of molecular mimicry, which makes it likely that numerous similarities will be found between any particular collection of tissue antigens and whatever microorganisms one chooses to examine. This in turn makes it difficult to predict which potential cross- reactions are the important ones in terms of disease induction, as opposed to the many that probably do not give rise to autoimmune pathology. Recent studies in animal models suggest an additional level of complexity to this problem by showing that the array of autoantigens recognized during an autoimmune response expands over time, a process now referred to as epitope spreading (for review, see ref. 17). Studies of the mouse model of EAE have provided a precise description of the sequence of epitope spreading following immunization with a single autoantigenic T cell epitope (21). Immunization of disease-susceptible mice with a synthetic peptide corresponding to the major encephalitogenic site of mouse MBP (amino acids 1-11 of the native protein with the N-terminus modified by acetylation, referred to as Acl-11) has been shown to be sufficient to induce inflammatory CNS disease. Interestingly, examination of the reactivity of T cells arising as a result of this immunization reveals an initial response directed at Acl-1 1, followed subsequently by strong responses to peptide epitopes corresponding to other portions of the MBP molecule (e.g., MBP 3547, , and ). This spreading of the T cell response to encompass other portions of the MBP molecule has been termed intramolecular spreading of the autoimmune repertoire. In addition, following induction of EAE by immunization with Acl-11, T cell responses against proteolipid protein, another myelin-associated self protein, can also be detected. Thus, in this welldefined animal model of autoimmune disease, induction of an immune response against a singie normally cryptic self epitope (MBP Acl-11) triggers an autoimmune response that spreads in waves, first to new epitopes on the antigen bearing the immunizing epitope (intramolecular spreading) and then to epitopes on other antigens expressed at the site of the resulting tissue inflammation (intermolecular spreading). Intra- and intermolecular spreading of the autoimmune T cell repertoire has also been elegantly demonstrated in recent studies of the nonobese diabetic (NOD) mouse model of spontaneous autoimmune pancreatic islet disease that results in diabetes mellitus. Studies from two different laboratories show that the autoimmune T cells of NOD mice become activated and expand in vivo against a defined group of islet cell antigens in an orderly sequential manner (22,23). Measurement of T cell responses against a panel of known islet cell antigens in young (34weekold) NOD mice reveals strong responses to one isoform of the enzyme glutamic acid decarboxylase

6 INDUCTION OF AUTOIMMUNITY BY MICROBIAL PATHOGENS 463 (GAD65), but not to the other antigens tested. These responses are initially limited to one region of the GAD65 protein spanned by two adjacent synthetic peptides, but spread intramolecularly as the mice age, to involve other regions of the protein. With increasing age of the animals and progression of islet cell inflammation, the T cell responses also spread intermolecularly to involve other islet cell proteins, including heat-shock protein 60, carboxypeptidase H, and insulin. The important basic conclusion of these studies is that the expressed autoimmune repertoire is not fixed, but evolves during the course of an autoimmune disease. Thus, exposure to foreign antigens that mimic self epitopes may generate a response that spreads both intra- and intermolecularly to ultimately involve a much wider range of autoantigens than could have been predicted by examining the inciting antigen. It is important to note that in the case of EAE, immunization of susceptible animals with representative second-wave peptides can elicit clinical disease, indicating that T cells responding to secondary epitopes may contribute importantly to the autoimmune pathogenesis. It is also noteworthy that the late responses to second-wave determinants may involve antigen presentation by different MHC molecules than did the response to the initial disease-inciting epitope, thus complicating the interpretation of MHC-linked disease susceptibility markers (e.g., HLA-DR4 and related alleles in rheumatoid arthritis [RA], or HLA- B27 in ankylosing spondylitis, reactive arthritis, or Reiter s syndrome). Furthermore, analysis of the expressed repertoire of autoimmune T and B cell receptors at the site of disease will also be complicated by the recruitment of clones reactive to secondary wave determinants, which may be part of the explanation for the inconsistent results and lack of consensus in interpretation that have so far emerged from studies of lymphocyte antigen receptors expressed in human autoimmune disease tissue specimens (24). Persistent antigens and persistent pathogens: the merging of infectious and autoimmune disease The mechanisms discussed above consider the introduction of an infectious agent and the immune response to its antigens as a trigger for autoimmune diseases. However, also implicit in this model is the elimination of the infectious agent in the setting of continuing inflammatory disease. Thus, in true autoimmune disease, it is believed that the immune re- sponse is actually directed against self components, and that the continuing recognition of these self antigens accounts for the chronic phase of these diseases. However, a substantial number of situations have now been brought to light in which inflammatory disease believed to be autoimmune in nature may actually be caused by an immune response directed against nonviable but persistent microbial antigens present in the target tissues, or even against occult viable infectious organisms. Such situations blur the distinction between autoimmune and infectious diseases, and raise the possibility that some syndromes now classified as autoimmune diseases may be more correctly viewed as chronic infections. Persistent microbial antigens. Persistence or repeated deposition of microbial antigens is a potential mechanism by which pathogenic organisms could cause chronic inflammatory disease. A possible example of this mechanism is the joint inflammation that is sometimes observed following nonarticular infections, and which has traditionally been classified as either postinfectious or reactive arthritis. Postinfectious arthritis, which occurs during the acute phase of an infectious illness, is a noninfectious synovitis believed to be mediated primarily by deposition of foreign antigens in the form of immune complexes. Examples include the arthritis associated with acute hepatitis B (25), and the sterile joint effusions sometimes seen during acute disseminated infections with meningococci (26) or gonococci (27). In contrast, reactive arthritis occurs in a subset of patients several weeks after acute bacterial diarrhea (postenteric arthritis) (28) or urogenital infection with Chlamydia trachomatis (sexually acquired arthritis) (29,30). This is generally thought to occur after the clearance of infectious agents and their products, and is presumed to arise from an autoimmune mechanism such as molecular mimicry. However, recent improvements in our ability to detect the presence of microbes and their products have begun to erode the traditional distinction between postinfectious and reactive arthritis by suggesting that the latter may also result from the persistence of microbial products in the inflamed tissues. The role of persistent microbial antigens in reactive arthritis has been extensively studied in a well-defined population of patients who developed articular symptoms following enteric infection by Yersinia enterocolitica (31,32). In two-thirds of these patients, Yersinia antigens are detected in 1-10% of synovial fluid (SF) cells (mostly neutrophils and some

7 464 BEHAR AND PORCELLI mononuclear cells) by immunofluorescence staining using a Yersinia-specific rabbit antiserum or a monoclonal antibody specific for a polysaccharide antigen of Yersinia, and these results are further supported by Western blot analysis (31). These studies have been confirmed by a second group of investigators who, using immunofluorescence, detected particles that were thought to represent Yersinia-derived antigens in the synovial membrane of 4 HLA-B27+ patients with Yersinia-triggered reactive arthritis (32). The antigens appear to be localized in the cytoplasm of large mononuclear cells, although the precise cell type has not been identified. Some of the biopsy specimens used in these studies were obtained long after the onset of arthritis (5 months to 17 years), indicating that these bacterial antigens can persist for extended periods of time (31,32). Although less well studied, similar findings have been reported in patients with reactive arthritis associated with other infectious agents. For example, Salmonella antigens have been detected in 10-50% of the SF cells from 8 HLA-B27+ patients with Salmonella-triggered reactive arthritis, in studies using rabbit antisera specific for Salmonella species and monoclonal antibodies specific for Salmonella lipopolysaccharide (33). Chlamydia antigens have also been detected in the SF and synovial tissue of patients with sexually acquired reactive arthritis and negative culture findings (34,35). The detection of persistent Yersinia antigens in the synovium of patients with Yersinia-triggered reactive arthritis weeks, months, or even years after the sentinel enteric infection raises the possibility that a chronic infection by this organism has been established in the inflamed joints. This issue has been addressed in many of the above studies by performing cultures of SF or tissues, the results of which have proven to be uniformly negative. Furthermore, in studies using the polymerase chain reaction (PCR), bacterial DNA has been sought in SF and tissue as a surrogate marker for the presence of viable organisms. Although this method is exceedingly sensitive (DNA from as few as 10 bacteria per lo5 cells can be detected), no Yersinia DNA has been found in samples of SF or tissue, even when simultaneously performed immunofluorescence studies demonstrated Yersinia antigens in 1% of the SF cells (36,37). Similarly, despite the identification of immunoreactive antigens and particles thought to represent Chlamydia, PCR failed to detect chlamydia1 DNA in the SF of patients with sexually acquired reactive arthritis (38). Negative findings can never be viewed as conclusive, but the results of these experiments, together with the inability to culture the organisms, indicate that there may be persistent bacterial antigens within the joint in the absence of viable organisms. The relevance of the detection of bacterial antigens in the joint to the development of chronic arthritis is unproven at this time. For example, it is not known if patients who recover from bacterial enteritis without developing arthritis also have bacterial antigens in their joints, or if the antigens that have been detected are present in a form that is recognized by the immune system. However, the possible significance of these findings has been highlighted recently by the isolation of T cell clones specific for Yersinia from the SF and tissue of patients with reactive arthritis. T cell clones from 1 patient recognized antigens from Y enterocolitica and Yersinia pseudotuberculosis, but not other bacterial species (39). Such T cells presumably arose in vivo as a consequence of persistent stimulation by antigen in the joint. Interestingly, although the patient was HLA-B27+, these Yersiniureactive T cells were CD4+ and restricted by HLA- DR4. More recently, CD8+ T cell clones have been derived from SF of several patients with Yersiniatriggered reactive arthritis. These T cells are restricted by HLA-B27 and lyse Yersinia-infected cells, but not cells that have been exposed to killed Yersinia organisms (40). This implies that active intracellular infection, and not simply phagocytosis of bacterial antigens, may be required for antigen presentation to HLA-B27-restricted T cells, as appears to be the case in general for T cells that recognize bacterial antigens in the context of class I MHC molecules (41). Based on these preliminary findings, it can be speculated that HLA-B27 and other class I MHC molecules may be involved in the presentation of bacterial antigens to class I-restricted (i.e., predominantly CD8+) T cells during the initiation of reactive arthritis, when viable organisms are present within macrophages and possibly other types of cells. In contrast, during the chronic and relapsing phases of arthritis, which appear to be characterized by the presence of persistent bacterial antigens but no viable organisms, antigen presentation may be predominantly by class I1 MHC molecules with activation of CD4+ T cells. A true test of the validity of such a model awaits the detailed analysis of T cell clones isolated from the synovium of reactive arthritis patients at various times after the onset of their illness. This area of research could lead to new therapeutic strategies based on the elimination of either intracel-

8 INDUCTION OF AUTOIMMUNITY BY MICROBIAL PATHOGENS 465 lular infections or persistent antigens, depending upon the stage of the disease (42). In this regard, differentiation between persistent antigens and chronic infections may be critical. Persistent pathogens: lessons from Lyme disease. Lyme arthritis provides an excellent example of how inflammatory arthritis with features suggesting an autoimmune pathogenesis can in fact be caused by an occult microbial pathogen. Many of the initial cases of Lyme disease were mistaken for juvenile rheumatoid arthritis before a series of landmark epidemiologic and laboratory investigations determined that the disease results from infection with the spirochete Borrelia burgdorferi, which is transmitted by the bite of the deer tick Zxodes dammini (for review, see ref. 43). The many studies that have followed from this discovery have made Lyme disease perhaps the best example of a chronic arthritis with a known etiology. Arthritis occurs in approximately 60% of patients with untreated Lyme disease, and becomes chronic in 10% of cases, with erosions of cartilage and bone and synovial histopathologic features similar in many ways to those of RA (44,45). In contrast to reactive arthritis, Lyme arthritis may be the result of an immune response in tissues that are persistently infected by viable, replicating microorganisms. Strong indirect evidence for this view is provided by the observation that treatment with antibiotics cures many cases of established Lyme arthritis, and prevents the development of this condition if given early after infection. More direct evidence for this hypothesis comes from silver staining of synovium from affected joints, which has revealed the presence of intact spirochetes (46). Antigens specific for B burgdorferi also have been detected in the synovium. In one report, synovial biopsy samples from 12 patients with chronic Lyme arthritis were investigated by immunohistochemistry (45). Monoclonal antibodies specific for the B burgdorferi 31-kd outer membrane polypeptide and the 41-kd flagellar antigen detected spirochetes and globular antigen deposits in and around blood vessels in areas of lymphocytic infiltration in 50% of biopsy specimens. Nevertheless, it has proven difficult to culture the organisms from synovial specimens, and relatively few instances of successful isolation of B burgdorferi have been reported (4749). Recent application of detection methods based on PCR now provides strong additional support for the presence of intact viable B burgdorferi in the joints during chronic Lyme arthritis. Using a nested PCR technique with which DNA can be detected from as few as 10 spirochetes in 1 ml of fluid (50-53), B burgdorferi DNA has been detected in most synovial specimens. In one of the largest reported series, PCR was performed on 92 SF samples from 88 patients with Lyme arthritis and 64 controls with other articular diseases, using primers specific for the plasmidencoded Osp A gene and the genomic 16s ribosomal RNA gene of B burgdorferi (52). Seventy-five of the patients with Lyme arthritis had a positive PCR result, whereas all of the controls had negative results. Of the 73 patients who had not been treated or had received only short courses of oral antibiotics, 96% had a positive PCR reaction. In contrast, only 37% of the 19 patients who had received appropriate antispirochetal antibiotic treatment had a positive PCR reaction. Remarkably, DNA from B burgdorferi was detected by PCR in 12 untreated patients up to 7 years after the onset of arthritis. The detection of pathogen-specific DNA sequences at the site of inflammation in chronic Lyme arthritis stands in marked contrast to the results reported for patients with reactive arthritis associated with Yersinia and Chlamydia species, as described above. Although it may be an immune response to bacterial antigens that gives rise to chronic inflammation in all of these diseases, it appears that in the case of Lyme arthritis this process is usually sustained by the persistence of living microorganisms, and not by the failure to clear nonviable antigenic material from the synovium. However, it may also be that both of these mechanisms are operating in some cases. For example, of 10 patients who had chronic Lyme arthritis despite multiple courses of antibiotics, 7 had no B burgdorferi DNA detected by PCR in posttreatment synovial samples (52), indicating that chronic arthritis may continue even after the eradication of viable spirochetes. An immunogenetic basis for this is likely, since certain class I1 MHC alleles (HLA-DR2 or HLA-DR4) are found with increased frequency in patients with chronic Lyme arthritis compared with patients with Lyme arthritis of short duration (54). However, it is not yet known if these patients harbor persistent B burgdorferi antigens in their joints in the absence of viable spirochetes. These studies of Lyme disease have increased the need to consider persistent infection by slowgrowing or fastidious bacterial pathogens as an etiology for idiopathic diseases with autoimmune features. A variety of other conditions have already been suggested to belong in this category, although in few of

9 466 BEHAR AND PORCELLI A C D ~ TCR C D ~ TCR B peripheral T cells 1. proliferation 2. anergy 3. death w MHC II MHC II 1. deletion thy mocytes C,--. Naive/memory (resting) T cells 0 V88Val 0 Vp8 Va2 - Vp8Va3 0 VP8Va4 Stimulus: SEB nn OV0" 0 VP8Va5 0 Vp8Va6 Stimulus: MBP " Expanded (activated) T cells these is the evidence for a bacterial etiology anywhere near as convincing as for Lyme disease (55-61). Nonetheless, this concept has obvious clinical importance since, if correct, it would mandate that treatment strategies shift away from the use of immunosuppressive agents and toward the development of long-term antibiotic treatments, vaccines, and immunotherapy. Bacterial superantigens The discovery of bacterial superantigens and the clarification of their mechanism of action have had a profound influence on the study of immunity to infection, and more recently have also begun to have a Figure 1. Mechanism and consequences of T cell recognition of superantigens (SAg). A, T cell recognition of a superantigen (left) is contrasted with conventional recognition of a peptide antigen presented by class I1 major histocompatibility complex (MHC) (right). TCR = T cell receptor. B, Mature peripheral T cells frequently undergo proliferation and activation of their effector functions upon recognition of superantigens, although induction of anergy and cell death are other possible outcomes. Activation of immature T cells by superantigens in the thymus causes cell death (thymic deletion). C, High frequency of T cells reactive with a single superantigen. The example shown is for staphylococcal enterotoxin B (SEB), which in mice activates most T cells that express Vp8. This is contrasted to a typical protein antigen, myelin basic protein (MBP), which activates only certain rare T cells that express Vp8 in combination with V02. significant impact on the study of autoimmune diseases. Superantigens exert their effects on T cells by ligating the antigen receptors of the T cell to class I1 MHC molecules expressed on other cells without any requirement for antigen processing (for review, see ref. 62) (Figure IA). The recognition of a superantigen by a T cell can have several different consequences, including proliferation and expansion, or, conversely, the induction of prolonged nonresponsiveness (anergy) or even cell death (63,64) (Figure 1B). The factors that determine which of these alternative outcomes predominate remain incompletely understood. The ability of a superantigen to activate a T cell is in most cases dependent solely on expression by the T cell of

10 INDUCTION OF AUTOIMMUNITY BY MICROBIAL PATHOGENS 467 one of a subgroup of specific Vp elements (i.e., particular variable segments of the T cell receptor [TCR] p chain). Since activation is dependent only on Vp and not on other variable elements of the TCR, a single type of superantigen is able to activate a far larger population of T cells than do the peptides derived from even the most immunogenic conventional protein antigens (Figure 1C). This property makes superantigens among the most potent natural immunostimulatory substances known. Superantigens from a variety of bacterial species have been characterized, many of which now are associated with diseases in humans (Table 3). The pathogenic mechanisms by which superantigens trigger the clinical syndromes with which they are associated are not entirely clear, but are likely to be dependent on their ability to activate large numbers of T cells. The various disease manifestations that result may be primarily a consequence of the dysregulated production of T cell-derived cytokines and other soluble mediators of inflammation (65). At least two mechanisms by which superantigens could play a role in autoimmune disease have been proposed. Friedman and coworkers have suggested that superantigens could stimulate autoantibody production by activating normal T cells and providing a molecular bridge that facilitates their interaction with autoantibody-producing B cells (66,67). The superantigen is thus envisioned as giving rise to a situation similar to graft-verus-host disease, in which a large proportion of grafted T cells recognize alloantigens expressed on host B cells, a situation that has been observed in animal models and in humans to result in the production of autoantibodies with specificities similar to those seen in systemic lupus erythematosus (SLE) (68). Those investigators suggest that B cells producing self-reactive antibodies might be preferentially activated under these circumstances be- cause they receive additional synergizing growth signals from the interaction of their surface immunoglobulin molecules with abundantly expressed self antigens (66). The expansion of autoreactive B cells may then lead to a secondary stimulation of autoreactive T cells, as detailed in experiments on the mouse response to autologous cytochrome c discussed above. A second mechanism by which superantigens may trigger autoimmunity is by directly facilitating the activation of previously silent (i.e., anergic or ignorant) autoreactive T cell clones. Superantigens may do this by promoting cell division, which may drive an anergic T cell out of its nonresponsive state, or by lowering the threshhold level of a normally cryptic self antigen that is required for autoreactive T cell stimulation. Consistent with the hypothesis that superantigens recruit the helper activity of normal T cells for autoantibody production, it has been reported that the production of rheumatoid factor (RF) is induced by the bacterial superantigen staphylococcal enterotoxin D (SED) during coculture of human CD4+ T cell clones with purified normal B cells (69). This is in contrast to the effects of T cell activation by a mitogenic anti-cd3 monoclonal antibody bound to the surface of the culture vessel, which stimulates the production of polyclonal IgG and IgM but little RF. Induction of RF in these experiments requires physical contact between T and B cells, since separation of the T and B cells by a permeable membrane prevented the production of RF. Thus, in this in vitro system the activation of T cells by SED and their subsequent interaction with B cells results in a shift of the secreted immunoglobulins (and presumably of the B cell repertoire) toward a preponderance of antibodies specific for self immunoglobulins, suggesting that this mechanism favors the production of autoantibodies. Perhaps the most provocative findings relating a superantigen to autoimmune phenomena in animals Table 3. Microbial superantigens and reported disease associations Organism Staphylococcus Streptococcus Yersinia Pseudomonas Mycoplasma Mycobacteria Toxoplasma Superantigens Enterotoxins A, B, C, D, and E Toxic shock syndrome toxin Exfoliating toxins A and B Pyrogenic exotoxins M proteins Y pseudotuberculosis mitogen Y enterocolitica superantigen activity Exotoxin A M arthritidis mitogen M tuberculosis superantigen T aondii superantigen Possible disease association Food poisoning, shock, Kawasaki disease Toxic shock syndrome Staphylococcal scalded skin syndrome Scarlet fever, shock Rheumatic fever Izumi fever, Kawasaki disease?? Arthritis, shock??

11 468 BEHAR AND PORCELLI have come from studies of disease associated with infection by Mycoplasma arthritidis in rodents. This natural pathogen of rodents can become established as a chronic infection in the joints, with the subsequent development of an inflammatory arthritis that bears a striking resemblance histopathologically to RA (70,71). It is now known that this organism produces a soluble protein that is strongly stimulatory to T cells, originally described as Mycoplasma arthritidis mitogen, or MAM. More recent analysis, however, has shown that MAM is actually not a mitogen, but a superantigen that activates T cells expressing many different Vp genes in rats, mice, and humans (72,73). The injection of purified MAM into the joints of mice has been shown to trigger inflammatory arthritis with many of the features associated with actual infection by M arthritidis (74), suggesting that the arthritis associated with infection by this organism may be mediated by local production of a superantigen. Findings of other studies in animal systems have been supportive of the idea that infection with superantigen-producing bacteria could induce an exacerbation of previously established autoimmune disease, probably by reactivating autoreactive T cells. For example, the T cell response to MBP in mice with EAE is dominated by T cells expressing just a few Va and Vp gene segments, with VpS consistently appearing as one of the dominant V-region genes expressed by encephalitogenic CD4+ T cells (75). Injection of mice with staphylococcal enterotoxin B (SEB), which can activate T cells expressing Vp8, does not cause the development of EAE but can cause a clinical relapse in mice that have recovered from a previous episode of EAE (76). Furthermore, if SEB is administered to mice that have been infused with encephalitogenic T cell clones expressing VpS, severe disease and death result. This is not observed if SEB is given after infusion of non-encephalitogenic T cell clones. Conversely, SEB actually lessens disease severity if administered before disease induction (77), possibly because SEB induces anergy of Vp8+ T cells under these circumstances, and prevents their subsequent activation by antigen. Related findings have been reported from studies using the staphylococcal superantigen toxic shock syndrome toxin 1 (TSST- 1) in experimental bacterial cell wall-induced arthritis. In this model, rats immunized intraarticularly in the ankle with the cell wall of group A Streptococcus develop a transient destructive arthritis (78). If, after the animals recover, TSST-1 is given intravenously, the arthritis is reactivated. The reactivation shows a dose-responsiveness, and at high doses of TSST-1 a prolonged inflammatory response is seen, with development of pannus and erosions. Only modest inflammation occurs in control joints (i.e., those not injected with bacterial cell walls), and cyclosporin A suppresses all phases of the arthritis, indicating that it is T cell dependent. In a similar study investigating disease in mice with collagen-induced arthritis, intraarticular administration of the superantigen MAM has also been shown to trigger exacerbations of previously established inflammatory arthritis (79). There are as yet no compelling data showing that superantigens are responsible for causing human autoimmune diseases, or for inducing flares of disease activity as observed in the animal studies summarized above. However, several intriguing observations that have already emerged from this relatively new area of investigation are worthy of comment. One observation that has been put forth as evidence of a role for superantigens in autoimmune disease is that synovial infiltrates from some patients with RA appear to have a selective expansion of T cells expressing particular Vp genes (e.g., Vp14 [SO]). However, studies from multiple laboratories show considerable variability in the particular Vp genes that are expanded in RA synovium, indicating either that a variety of different superantigens are involved or that other factors must influence these T cell expansions (for concise review, see ref. 24). Other studies have begun to suggest possible similarities between the superantigen effects observed in the mouse model of EAE and in human MS. T cell clones that are specific for MBP and also responsive to bacterial superantigens have been isolated from patients with MS as well as from normal subjects. These T cell clones can be activated in vitro by bacterial superantigens at extremely low doses to proliferate or mediate cytotoxicity (81), suggesting that even minor exposures to superantigens in vivo could augment the inflammatory effects of these cells. A link between bacterial superantigens and Kawasaki disease has also been suggested by a recent clinical study in which 13 of 16 patients with acute untreated Kawasaki disease grew superantigenproducing bacteria from surveillance cultures, as opposed to only 1 of 15 controls (82). Eleven of the isolates were Staphylococcus aureus that produced TSST-1, and 2 were streptococci that produced pyrogenic exotoxins B and C. These toxins are all known to be superantigens that stimulate Vp2+ T cells, which are expanded in the blood of children with Kawasaki

12 INDUCTION OF AUTOIMMUNITY BY MICROBIAL PATHOGENS 469 disease (83). A T cell activator from Ypseudotuberculosis has recently been characterized and may represent a novel superantigen (84). This molecule, currently named Y pseudotuberculosis mitogen (YPM), stimulates proliferation and interleukin-2 (IL-2) production by human T cells expressing Vp3, VpS, Vp13.1, and Vp13.2. Clinical observations suggest possible links between YPM and cases of Kawasaki disease and other inflammatory diseases (85). Retroviruses and autoimmunity Infectious retroviruses have been studied in rodent species for many years, but only relatively recently have these agents been identified as pathogens in humans. Because of their ability to integrate permanently into the genome of host cells and persist for the life span of the host animal as an intracellular parasite (86), these agents can create a situation in which the distinction between self and nonself becomes extremely unclear. Although direct evidence connecting retrovirus infection to autoimmune diseases in a causative manner is still lacking, several remarkable animal models demonstrate the paradigm of occult retrovirus infection leading to a chronic inflammatory disease. Most notable in this regard are the diseases produced by the caprine arthritis encephalitis virus in goats and by the closely related maedi/ visna virus in sheep (87,88). These viruses cause persistent infections that result in arthritis, encephalitis, and interstitial pneumonitis. The pathologic lesions that develop in infected animals are considered to be similar in many ways to certain spontaneous human autoimmune diseases, particularly RA and MS. Both of these viruses are ubiquitous and infect a majority of their host species, but only a small proportion of infected animals develop disease. This seems to indicate the interaction of the viruses with genetic or environmental susceptibility factors that determine the eventual development of autoaggressive tissue pathology, as has been postulated to occur for human autoimmune disease syndromes. It is also noteworthy that, without prior knowledge of the structure of the causative virus and specific probes that can be used to identify its antigens or nucleic acid sequences, these agents are virtually impossible to detect. The discovery of human infectious retroviruses (i.e., the human immunodeficiency viruses [HIV] and human T cell leukemia viruses [HTLV I and 111) in the 1980s and their subsequent association with certain autoimmune disease syndromes (for review, see refs. 89 and 90) has led to renewed interest in the possibility of a retroviral etiology for many human autoimmune diseases. Although it seems unlikely on epidemiologic grounds that any of the currently known human retroviruses is a major causative factor for commonly recognized autoimmune diseases of humans, the possibility that other, more ubiquitous, infectious retroviruses not yet isolated may give rise to some or many of these diseases is not at all ruled out at this time. In addition, it is well known that the genomes of all vertebrate animals, including humans, contain many integrated retrovirus-related sequences (so-called endogenous retroviruses, or ERVs) (91). These ERVs, thousands of which are scattered throughout the human genome, are believed to have arisen by the accumulation throughout evolution of infectious retroviruses that were incorporated into the germline by rare infectious events involving early embryos or germ cells. It is clear that most ERVs represent defective proviruses (i.e., double-stranded DNA sequences encoding the RNA genome of the infectious virus) that have been inactivated by partial deletions or insertion of stop codons in their structural genes, rendering them incapable of giving rise to new infectious retrovirus particles. Nevertheless, even these noninfectious proviruses are now known to profoundly influence the immune systems of some animals in ways.that could conceivably be linked to the generation of autoimmune diseases (92), as briefly summarized below. Retroviruses and persistent antigenemia. Both infectious retroviruses and endogenous retroviruses represent potential sources of antigens that can be recognized as foreign by the immune system. Retroviral infections are, as a rule, persistent for the life of the host, and may lead to continual antigenemia. Infected individuals often respond to these antigens by production of specific antibodies, which in general also requires the participation of specific T cells. In addition, although it is clear that all ERVs so far examined in humans and most ERVs in rodents are defective and unable to replicate, some of these do contain intact retroviral genes that can be transcribed and translated, giving rise to immunogenic retroviral products. One exhaustively studied example of this is the response of lupus-prone mouse strains to the major mouse ERV envelope protein SU, or gp70. All mice express the ERV-encoded gp70 envelope protein, but the level of expression is higher in lupus-prone strains, probably because expression of the gp70 gene is induced in activated B cells, which are numerous in such mice (93). In (New Zealand black x New Zealand white)f,

13 BEHAR AND PORCELLI mice, which develop a severe accelerated form of murine lupus, high levels of anti-gp70 antibodies are present, and the antigen circulates in the form of immune complexes (94). Similar gp70/anti-gp70 immune complexes can also be recovered from the kidneys of animals with nephritis. Although the relationship between the immune response to gp70 and the induction of murine lupus remains unclear, these studies demonstrate how ERVs may give rise to persistent target antigens. The search for retroviral antigens in human tissues affected by diseases of presumed autoimmune etiology has yielded some provocative although still inconclusive results. Expression of antigens reactive with monoclonal antibodies specific for p19 and p24 gag (group-specific antigen) proteins of HTLV-I (95) and p24 and p17 gag proteins of HIV-1 (96) has been observed, by immunohistochemical analysis, in a substantial fraction of rheumatoid synovial specimens, but not in control samples. A similar study has documented the expression of an antigen reactive with one monoclonal antibody against HTLV-1 p19 gag protein in salivary gland epithelial cells of one-third of patients with Sjogren s syndrome (SS), although other antibodies against HTLV-1 gag proteins (including another against p19) or HIV proteins did not react (97). Definitive studies have not yet appeared that would establish whether these serologic reactivities represent the true expression of retroviral antigens (presumably encoded by ERVs, although conceivably they could also be from as-yet-unidentified infectious retroviruses) or simply a chance cross-reaction of the antibodies used with unidentified cross-reactive host proteins. Another exciting but still incomplete line of investigation involves a direct approach to the isolation of retroviruses from patients with SS (98). To accomplish this, a human T lymphoblastoid cell line that is known to be susceptible to infection by other human retroviruses was cocultured with homogenates of inflamed salivary gland tissue from SS patients. Upon screening with antibodies against the p24 gag protein of HIV, 2 of 6 lymphoblastoid lines that had been treated in this way were found to have converted to positive, suggesting the presence of a virus related to HIV. Electron microscopy of the 2 positive cell lines revealed retroviral A-type particles (i.e., a morphology usually associated with noninfectious defective retroviruses) in vacuoles or cisternae within the cells, and manganese-dependent reverse transcriptase activity (a characteristic retroviral enzymatic activity) was detected. The authors of this study suggested that these so-called human intracisternal A-type particles (HIAP) may provide the antigenic stimulus for the production of gag-reactive antibodies that have been detected in the sera of a subset of patients with SS and other autoimmune diseases. However, until these findings have been replicated and extended to a larger sample of SS patients, and the putative agent giving rise to the HIAPs identified by cloning and sequencing of its genome, the results must be viewed with a cautious blend of enthusiasm and skepticism. Retroviruses and molecular mimicry. As described for the antigens of other microbial pathogens, retroviral antigens can potentially mimic self epitopes. Although the frequency with which this occurs and its relevance to autoimmune pathogenesis are still not known, some interesting examples have been documented and partially explored. In one detailed study, an antigenic cross-reactivity between a murine leukemia virus p30 gag protein and a human 70-kd protein was identified (99). Cloning of the complementary DNA encoding the human protein revealed it to be a subunit of the U1 small nuclear RNP (snrnp) complex, a nuclear protein-rna complex that bears the determinants recognized by anti-sm and anti-rnp antibodies associated with human SLE and related diseases. In addition, the 70-kd protein is homologous to a core consensus sequence that occurs as a tandem repeat in p3ogag of many mammalian retroviruses, suggesting that autoantibodies to U1 snrnp may arise by exposure to one or more occult human retroviruses through a mechanism of molecular mimicry. Another study identified the presence of antibodies reactive to p24gag of HIV-1 in a significant fraction of SS and SLE patients, but not in a variety of control subjects (100,101). In a subset of SLE patients, these antibodies were shown to cross-react with Sm antigen (101,102), again suggesting that some anti-sm responses may arise through molecular mimicry by retroviral gag proteins. Retroviral superantigens. Studies focusing on the biology of the Mls, or minor lymphocytestimulating, locus in the mouse have recently led to the fascinating insight that many murine retroviruses encode proteins with superantigen activity. This important line of research began with the identification of a genetic locus (Mfs) in mice that encodes an antigen that is strongly stimulatory to T cells from Mlsdisparate strains of mice (103). Subsequently, it was shown that Mfs acts like the superantigens produced by bacteria in that its ability to activate T cells depends

14 INDUCTION OF AUTOIMMUNITY BY MICROBIAL PATHOGENS 47 1 on their expression of particular Vp gene products (104,105). This was followed by the extraordinary finding that Mls and other related genetic loci in the mouse are tightly linked to ERVs of the mouse rnammary tumor virus family (MMTV) ( ). Detailed analysis of the genetic structure of these MMTV ERVs revealed that all contain open translational reading frames in their 3 long terminal repeat (LTR) sequences that encode a series of related proteins with superantigen activity (110). At least one example of an infectious MMTV that encodes a similar superantigen has also been described (1 11). These retrovirus-encoded superantigens clearly have a riiajor effect on shaping the T cell antigen receptor repertoire in mice, since their expression in the thymus leads to the deletion of large numbers of developing T cells depending on the Vp genes that they express (104,105,109). Although no direct involvement of retrovirus-encoded superantigens has been shown for autoimmune diseases in mice, it remains a plausible hypothesis that these molecules could trigger or augment autoaggressive responses in the same way as has been suggested for bacterial superantigens. However, it is important to point out that the phenomenon of retroviral superantigens appears so far to be entirely restricted to the mouse, and similar results have not been reported in any other species. One study of human T cells in HIV-infected subjects has been claimed to provide evidence for a superantigen associated with this human retrovirus (112), but this result has not been confirmed by subsequent reports. Unique genetic effects of retroviruses: transactivation, insertional inactivation, and activation of autogenes. In addition to their ability to stimulate the immune system through antigenic and superantigenic challenges, retroviruses have been shown to activate immunity through several other mechanisms that may be unique to this class of microbial pathogens. These additional effects arise from the ability of retroviruses to insert their genomes into that of the host cell they infect, and from the tendency of these agents to exert effects on the regulation of genes in infected cells (illustrated in Figure 2, and described below). One mechanism that may be almost unique to the retroviruses is that of trans-activation of immunoregulatory genes. The genomes of all human retroviruses characterizer to date encode proteins that act on the viral LTR promoter and stimulate transcription (86). These regulators are called trans-acting transcrip- - c\ Lnl Infectious Retrovirus c Host Cdl DNA 1) Disruptionflnactivation 2) Cis-activation a) 5 LTR promotion b) 3 LTR promotion enhancer effect 3) Trans-activation Figure 2. Potential mechanisms by which infectious retroviruses could alter the expression of autogenes. The early events following retroviral infection are depicted at the top, followed by the effects of retroviral integration on autogene expression. Horizontal arrows indicate transcription of the autogene. LTR indicates the viral long terminal repeats, which contain the promoters for viral gene transcription. Insertion of the viral DNA into the host cell chromosome within an autogene causes transcription to terminate prematurely (asterisk) or produces an aberrant read-through transcript (dotted arrow), both resulting in inactivation of the gene. Insertion of the virus near the 5 end of the autogene can up-regulate transcription from several different promoters, leading to increased autogene expression. Trans-activator proteins produced from viral genes can act on autogenes at sites distant from the insertion site of the virus, also leading to increased autogene expression. tional activators (tat in HIV, or tux in HTLV-I) (113,114). Although the main function of these transactivators appears to be the enhancement of viral gene transcription, it has been shown that these proteins can also up-regulate the transcription of certain nonviral genes within the infected host cell, while simultaneously repressing the expression of other cellular genes. Trans-activation by HIV tat or HTLV-1 tax proteins has been shown to induce a number of genes that are known to contribute to inflammatory reactions, including those for IL-2 ( ), IL-2 receptor (Y subunit (115,117,118), granulocyte-macrophage colony-stimulating factor (1 19), IL-3 (120), transforming growth factor pl (121), and IL-6 (122).

15 472 BEHAR AND PORCELLI Recent studies in transgenic mice have generated support for a model of retroviral induction of inflammatory diseases through a mechanism of genetic trans-activation. A disease resembling SS has been reported in mice transgenic for the HTLV-1 tax gene, in which exocrinopathy of the salivary and lachrymal glands occurred in conjunction with lymphocytic infiltration of these tissues (123). Subsequently, a second group has generated two lines of mice transgenic for a larger portion of the HTLV-1 genome, including the env (envelope protein) and tax genes (124). These mice showed a high incidence of clinical arthritis affecting the ankles, feet, and knees, and histologic analysis revealed a pattern similar to RA, with pannus formation and erosion of bone. Unlike the tax transgenic mice initially reported, these mice did not show abnormalities of the salivary or lachrymal glands, despite significant levels of tax transcription in these tissues. The reason for the difference in the clinical and pathologic syndromes seen in these mice is not yet known. However, since these mice appeared to be immunologically tolerant to HTLV-1 env and tax proteins (as indicated by the absence of serum antibody responses), it has been proposed that both the inflammatory exocrinopathy and the joint disease result from direct effects of tax on cellular gene expression, rather than from an immune response to HTLV-1 proteins. Consistent with this hypothesis, the level of messenger RNA encoding tax appears to correlate with the severity of tissue inflammation in both tax transgenic mouse models. Other mechanisms related to the effects of retrovirus insertion into the host genome have been suggested to result in altered immune regulation, perhaps leading to autoimmunity. It is generally believed that there must be a large number of genes that exert critical regulatory effects on the immune system, and that the inactivation or up-regulation of one or more of these genes may be part of the pathway to autoimmune disease. These critical immunoregulatory genes have been termed autogenes by one group of investigators (125), who suggest an analogy with oncogenes, a class of normal cellular genes that may give rise to cancers when their function is altered. When a retrovirus inserts itself into the DNA of a host cell, it may either inactivate a gene by disrupting its coding sequence or enhance the transcription of a cellular gene through the activity of viral promoters and transcriptional enhancers. If such an effect is exerted on an important autogene, the net result could conceivably be autoimmune disease. Although it is likely that the identification and characterization of autogenes has only just begun, one interesting example relating an alteration in autogene function to retroviral insertional activity has already been found in mice. This is the Fas protein deficiency that results in systemic autoimmune and lymphoproliferative disease in mice bearing the lpr gene, now known to be a mutant allele of the gene encoding the Fas protein (126). The Fas protein (also known as Apo-1) is normally expressed on the surface of lymphocytes and is important for the induction of apoptosis, or programmed cell death. In mice homozygous for the lpr gene, the absence of Fas protein expression results in a defect in apoptosis, which leads to a massive accumulation of lymphocytes and associated autoimmune disease. Molecular analysis of the defective Fas gene in mice of the MRLllpr strain has revealed that the gene is disrupted by the insertion of a retrovirus-related transposable element known as ETn into the coding sequence of the gene (127). Thus, this may represent the first example of a more general mechanism by which retroviruses and their related endogenous sequences may alter immune function by changing the expression of autogenes, with one potential result being autoimmune disease. Conclusions Understanding autoimmunity and autoimmune diseases remains a daunting challenge in the 1990s. It is unclear at this point whether there exists a single unifying mechanism or a small number of mechanisms that account for the development of these diseases, or if there exist many different routes that culminate in the same final condition. Since the immune system is widely believed to have evolved mainly to recognize patterns of microbial structure, it is not surprising that a connection between microorganisms has been frequently suggested. The concept that molecular mimicry of self antigens by infectious agents or even normal microbial flora may represent the initiating step in the development of autoimmunity continues to dominate much of the thinking in this area. Nevertheless, while it is clear that many antigenic similarities do exist between the tissues of higher organisms and the many microbial species to which they are exposed, it has not yet been possible to demonstrate a clear link between cross-reactive immune responses resulting from molecular mimicry and the development of a spontaneous autoimmune disease. In this review we have highlighted some of the

16 INDUCTION OF AUTOIMMUNITY BY MICROBIAL PATHOGENS 473 recent insights that have emerged from the study of disease models based on molecular mimicry, and we have also summarized other mechanisms that are now suspected to link infectious agents to the development of autoimmunity. New and unexpected findings of startling significance, such as the discovery of bacterial and viral superantigens and the variety of effects that retroviruses can exert on the immune system, continue to enhance our appreciation of the many subtle ways in which the immune system and the microbial world interact with each other. Undoubtedly, there remain many important but as-yetundiscovered aspects to this story. As our knowledge of this area and of the precise mechanisms governing self tolerance continues to grow, it will be continually necessary to reassess our concepts of autoimmune disease pathogenesis. In particular, the possibility that some of these disorders may actually result from persistent infections or repeated exposure to foreign antigens may require a reclassification of certain autoimmune diseases, or a broader definition of autoimmunity. REFERENCES 1. Constant P, Davodeau F, Peyrat M, Poquet Y, Puzo G, Bonneville M, Fournie J: Stimulation of human ys T cells by nonpeptidic mycobacterial ligands. Science 264: , Tanaka Y, Sano S, Nieves E, DeLibero G, Rosa D, Modlin R, Brenner M, Bloom B, Morita C: Nonpeptide ligands for human ys T cells. Proc Natl Acad Sci U S A 91: , Beckman EM, Porcelli S, Morita C, Behar S, Furlong S, Brenner MB: Recognition of a lipid antigen by CDl restricted crp T cells. Nature 372: , Germain RN, Margulies DH: The biochemistry and cell biology of antigen processing and presentation. Annu Rev Immuno1 11:403450, Oldstone MBA: Molecular mimicry as a mechanism for the cause and as a probe uncovering etiologic agent($ of autoimmune disease. Curr Top Microbiol Immunol 145: , Porcelli S: Molecular mimicry and the generation of autoimmune diseases. Rheumatol Rev 2:41-50, Damian RT: Molecular mimicry: antigen sharing by parasite and host and its consequences. Am Naturalist 98: , Baum H, Butler P, Davies H, Sternberg MJE, Burroughs AK: Autoimmune disease and molecular mimicry: an hypothesis. Trends Biochem Sci 18: , Miller JFAP, Morahan G: Peripheral T cell tolerance. Annu Rev Immunol 10:51-70, Janeway CA Jr: Approaching the asymptote? Evolution and revolution in immunology. Cold Spring Harb Symp Quant Biol 54:1-13, Sercarz E, Lehmann PV, Arnetani A, Benichou G, Miller A, Moudgil K: Dominance and crypticity of T cell antigenic determinants. Annu Rev Immunol , Lin RH, Mamula MJ, Hardin JA, Janeway CA: Induction of autoreactive B cells allows priming of autoreactive T cells. J Exp Med 173: , Mamula MJ, Lin RH, Janeway CA, Hardin JA: Breaking T cell tolerance with foreign and self co-immunogens: a study of autoimmune B and T cell epitopes of cytochrome c. J lmmunol 149: , I Mamula MJ. The inability to process a self-peptide allows autoreactive T cells to escape tolerance. J Exp Med , Lanzavecchia A: Antigen-specific interactions between T and B cells. Nature , Rock KL, Benacerraf B, Abbas A: Antigen presentation by hapten-specific B lymphocytes. I. Role of surface immunoglobulin receptors. J Exp Med 160: , Lehmann PV, Sercarz EE, Forsthuber T, Dayan C, Gammon G: Determinant spreading and the dynamics of the autoimmune T-cell repertoire. Immunol Today 14: , Goverman J, Woods A, Larson L, Weiner L, Hood L, Zaller DM: Transgenic mice that express a myelin basic proteinspecific T cell receptor develop spontaneous autoimmunity. Cell , Hammer RE, Maika SD, Richardson JA, Tang 1-P, Taurog JD: Spontaneous inflammatory disease in rats expressing HLA- B27 and human p2-m: an animal model of HLA-B27-associated human disorders. Cell 63: , Taurog JD, Richardson JA, Croft JT, Simmons WA, Zhou M, Fernandez-Sueiro JL, Balish E, Hammer RE: The germ free state prevents development of gut and joint inflammatory disease in HLA-B27 transgenic rats. J Exp Med 180: , Lehmann PV, Forsthuber T, Miller A, Sercarz EE: Spreading of T-cell autoimmunity to cryptic determinants of an autoantigen. Nature 368: , Kaufman DL, Clare-Saizler M, Tian J, Forsthuber T, Ting GSP, Robinson P, Atkinson MA, Sercarz EE, Tobin AJ, Lehmann PV: Spontaneous loss of T-cell tolerance to glutamic acid decarboxylase in murine insulin-dependent diabetes. Nature 366:69-72, Tisch R, Yang X-D, Singer SM, Liblau R, Fugger L, McDevitt HO: Immune response to glutamic acid decarboxylase correlates with insulitis in non-obese diabetic mice. Nature 366:72-75, Capra JD, Natvig JB: Is there V region restriction in autoimmune disease? Immunologist 1:16-19, Wands JR, Mann E, Alpert E, Isselbacher KJ: The pathogenesis of arthritis associated with acute hepatitis-b surface antigen-positive hepatitis: complement activation and characterization of circulating immune complexes. J Clin Invest 55: , Jarrett MP, Moses S, Barland P, Miller MH: Articular complications of meningococcal meningitis: an immune complex disorder. Arch Intern Med 140: , Goldenberg DL: Postinfectious arthritis: new look at an old concept with particular attention to disseminated gonococcal infection. Am J Med 74: , Granfors K: Do bacterial antigens cause reactive arthritis. Rheum Dis Clin North Am 18:3748, Rahman MU, Hudson AP, Schumacher HR Jr: Chlamydia and Reiter s syndrome (reactive arthritis). Rheum Dis Clin North Am 18:67-80, Silveira LH, Gutierrez F, Scopelitis E, Cuellar ML, Citera G, Espinoza LR: Chlamydia-induced reactive arthritis. Rheum Dis Clin North Am 19: , Granfors K, Jalkanen S, Von Essen R, Lahesmaa-Rantala R, Isomaki R, Pekkola-Heino K, Merilahti-Palo R, Saario R, Isomaki H, Toivanen A: Yersinia antigens in synovial fluid from patients with reactive arthritis. N Engl J Med 320: , Hammer M, Zeidler H, Klimsa S, Heesemann J: Yersinia

17 BEHAR AND PORCELLI enterocolitica in the synovial membrane of patients with Yersinia-induced arthritis. Arthritis Rheum 33: , Granfors K, Jalkanen S, Lindberg AA, Maki-Ikola 0, Von Essen R, Lahesmaa R, Isomaki H, Saario R, Arnold WJ, Toivanen A: Salmonella lipopolysaccharide in synovial cells from patients with reactive arthritis. Lancet , Keat A, Thomas B, Dixey J, Osborn M, Sonnex C, Taylor- Robinson D: Chlamydia trachomatis and reactive arthritis: the missing link. Lancet 1:72-74, Schumacher HR Jr, Magge S, Cherian PV, Sleckman J, Rothfuss S, Clayburne G, Sieck M: Light and electron microscopic studies on the synovial membrane in Reiter s syndrome: immunocytochemical identification of chlamydia1 antigen in patients with early disease. Arthritis Rheum 31: , Viitanen A-M, Arstila TP, Lahesmaa R, Granfors K, Skurnik M, Toivanen P: Application of the polymerase chain reaction and immunofluorescence techniques to the detection of bacteria in Yersinia-triggered reactive arthritis. Arthritis Rheum 34:89-96, Nikkari S, Merilahti-Palo R, Saario R, Soderstrbm K-0, Granfors K, Skurnik M, Toivanen P: Yersinia-triggered reactive arthritis. Use of polymerase chain reaction and immunocytochemical staining in the detection of bacterial components from synovial specimens. Arthritis Rheum 35:682487, Wordsworth BP, Hughes RA, Allan I, Keat AC, Bell JI: Chlamydia1 DNA is absent from the joints of patients with sexually acquired reactive arthritis. Br J Rheumatol 29:20% 210, Viner NJ, Bailey LC, Life PF, Bacon PA, Gaston JSH: Isolation of Yersinia-specific T cell clones from the synovial membrane and synovial fluid of a patient with reactive arthritis. Arthritis Rheum 34: , Hermann E, Yu DTY, zum Buschenfelde K-HM, Fleischer B: HLA-B27-restricted CD8 T cells from synovial fluids of patients with reactive arthritis and ankylosing spondylitis. Lancet 343:646450, Kaufmann SHE: Immunity to intracellular bacteria. Annu Rev Immunol 11: , Lauhio A, Leirisalo-Rep0 M, Laahdevirta J, Saikku P, Rep0 H: Double-blind, placebo-controlled study of three-month treatment with lymecycline in reactive arthritis, with special reference to Chlamydia arthritis. Arthritis Rheum , Steere AC: Lyme disease. N Engl J Med , Steere AC, Schoen RT, Taylor E: The clinical evolution of Lyme arthritis. Ann Intern Med 107: , Steere AC, Duray PH, Butcher EC: Spirochetal antigens and lymphoid cell surface markers in Lyme synovitis: comparison with rheumatoid synovium and tonsillar lymphoid tissue. Arthritis Rheum 31:487495, Johnston YE, Duray PH, Steere AC, Kashgarian M, Buza J, Malawista SE, Askenase PW: Lyme arthritis: spirochetes found in synovial microangiopathic lesions. Am J Pathol 118: 2634, Schmidli J, Hunziker T, Moesli P, Schaad UB: Cultivation of Borrelia burgdorferi from joint fluid 3 months after treatment of facial palsy due to Lyme borreliosis. J Infect Dis 158: , Snydman DR, Schenkein DP, Beradi VP, Lastavica CC, Pariser KM: Borrelia burgdorferi in joint fluid in chronic Lyme arthritis. Ann Intern Med 104: , Haupl T, Hahn G, Rittig M, Krause A, Schoerner C, Schonherr U, Kalden JR, Burmester GR: Persistence of Borrelia burgdorferi in ligamentous tissue from a patient with chronic Lyme borreliosis. Arthritis Rheum 36: , Williams WV, Callegari P, Freundlich B, Keenan G, Eldridge D, Shin H, Kreitman M, McCallus D, Weiner DB: Molecular diagnosis of Borrelia burgdorferi infection (Lyme disease). DNA Cell Biol 11: , Liebling MR, Nishio MJ, Rodriguez A, Sigal LH, Jin T, Louie JS: The polymerase chain reaction for the detection of Borrelia burgdorferi in human body fluids. Arthritis Rheum 36:665475, Nocton JJ, Dressler F, Rutledge BJ, Rys PN, Persing DH, Steere AC: Detection of Borrelia burgdorferi DNA by polymerase chain reaction in synovial fluid from patients with Lyme arthritis. N Engl J Med 330: , Bradley JF, Johnson RC, Goodman JL: The persistence of spirochetal nucleic acids in active lyme arthritis. Ann Intern Med 120: , Steere AC, Dwyer E, Winchester R: Association of chronic Lyme arthritis with HLA-DR4 and HLA-DR2 alleles. N Engl J Med 323: , Rook CAW, Stanford JL: Slow bacterial infections or autoimmunity? Immunol Today 13: , Wilson KH, Blitchington R, Frothingham R, Wilson JAP: Phylogeny of the Whipple s-disease-associated bacterium. Lancet 338: , Relman DA, Schmidt TM, MacDermott RP, Falkow S: Identification of the uncultured bacillus of Whipple s disease. N Engl J Med 327: , Burnham WR, Lennard-Jones JE, Stanford JL, Bird RG: Mycobacteria as a possible cause of inflammatory bowel disease. Lancet 2: , Chiodini RJ, van Kruiningen HJ, Thayer WR, Coutu J: Spheroplastic phase of mycobacteria isolated from patients with Crohn s disease. J Clin Microbiol 24: , Wall S, Kunze ZM, Saboor S, Soufleri I, Seechum P, Chiodini R, McFadden JJ: Identification of spheroplast-like agents isolated from tissues of patients with Crohn s disease and control tissues by polymerase chain reaction. J Clin Microbiol 31: , Saboor S, Johnson N, McFadden J: Detection of mycobacterial DNA in sarcoidosis and tuberculosis with polymerase chain reaction. Lancet 339: , Marrack P, Kappler J: The staphylococcal enterotoxins and their relatives. Science 248: I, MacDonald HR, Lees RK, Baschieri S, Herrmann T, Lussow AR: Peripheral T-cell reactivity to bacterial superantigens in vivo: the response/anergy paradox. Immunol Rev 133: , Rammensee H-G, Kroschewsky R, Frangoulis B: Clonal anergy induced in mature Vp6 T lymphocytes on immunizing Mls-lb mice with MIS-I a expressing cells. Nature , Miethke T, Wahl C, Heeg K, Echtenacher B, Krammer PH, Wagner H: T cell-mediated lethal lethal shock triggered in mice by the superantigen stapylococcal enterotoxin B: critical role of tumor necrosis factor. J Exp Med 175:91-98, Friedman SM, Posnett DN, Tumang JR, Cole BC, Crow MK: A potential role for microbial superantigens in the pathogenesis of systemic autoimmune disease. Arthritis Rheum 34:468480, Tumang JR, Posnett DN, Cole BC, Crow MK, Friedman SM: Helper T cell-dependent human B cell differentiation mediated by a mycoplasmal superantigen bridge. J Exp Med 171: , Gleichman E, Van Elven EH, Vanden Veen JPW: A systemic lupus erythematosus (SLE)-like disease in mice induced by abnormal T-B cooperation: preferential formation of autoantibodies characteristic of SLE. Eur J Immunol 12: , He XW, Goronzy J, Weyand C: Selective induction of rheumatoid factor by superantigens and human helper T cells. J Clin Invest 89: , 1992

18 INDUCTION OF AUTOIMMUNITY BY MICROBIAL PATHOGENS Cole BC, Washbum LR, Taylor-Robinson D: Mycoplasmainduced arthritis. In, The Mycoplasmas. Volume IV. Edited by S Razin, MF Barile. New York, Academic Press, Cole BC, Ward JR, Jones KS, Cahill JF: Chronic proliferative arthritis of mice induced by Mycoplasma arthritidis. I. Induction of disease and histopathological characteristics. Infect Immun 4: , Baccala R, Smith LR, Vestberg M, Peterson PA, Cole BC, Theofilopoulos AN: Mycoplasma arthritidis mitogen: V, engaged in mice, rats, and humans, and requirement of HLA- DRa for presentation. Arthritis Rheum , Bhardwaj N, Hodtsev AS, Nisanian A, Kabak S, Friedman SM, Cole BC, Posnett D: Human T-cell responses to Mycoplasma arthritidis-derived superantigen. Infect Immun 62: , Cannon GW, Cole BC, Ward JR, Smith JL, Eichwald EJ: Arthritogenic effects of Mycoplasma arthritidis T-cell mitogen in rats. J Rheumatol 14: , Zamvil SS, Steinman L: The T lymphocyte in experimental allergic encephalitis. Annu Rev Immunol , Brocke S, Gaur A, Piercy C, Gautam A, Gijbels K, Fathman CG, Steinman L: Induction of relapsing paralysis in experimental autoimmune encephalomyelitis by bacterial superantigen. Nature 365S2-644, Racke MK, Quigley L, Cannella B, Raine CS, McFarlin DE, Scott DE: Superantigen modulation of experimental allergic encephalomyelitis: activation or anergy determines outcome. J Immunol 152: , Schwab JH, Brown RR, Anderle SK, Schlievert PM: Superantigen can reactivate bacterial cell wall-induced arthritis. J Immunol 150: , Cole BC, Griffiths MM: Triggering and exacerbation of autoimmune arthritis by the Mycoplasma arthritidis superantigen MAM. Arthritis Rheum , Paliard X, West SG, Lafferty JA, Clements JR, Kappler JW, Marrack P, Kotzin BL: Evidence for the effect of a superantigen in rheumatoid arthritis. Science 253: , Bums J, Littlefield K, Gill J, Trotter JL: Bacterial toxin superantigens activate human T lymphocytes reactive with myelin autoantigens. Ann Neurol 32: , Leung DY, Meissner HC, Fulton DR, Murray DL, Kotzin BL, Schlievert PM: Toxic shock syndrome toxin-secreting Staphylococcus aureus in Kawasaki syndrome. Lancet 342: , Abe J, Kotzin BJ, Jujo K, Melish ME, Glode MP, Kohsaka T, Leung DY: Selective expansion of T cells expressing T-cell receptor variable regions Vp2 and VpS in Kawasaki disease. Proc Natl Acad Sci U S A 89: , Uchiyama T, Miyoshi-Akiyama T, Kato H, Fujimaki W, Imanishi K, Yan X-J: Superantigenic properties of a novel mitogenic substance produced by Yersinia pseudotuberculosis isolated from patients manifesting acute and systemic symptoms. J Immunol 151: , Sato K, Ouchi K, Taki M: Yersinia pseudotuberculosis infection in children resembling Izumi fever and Kawasaki syndrome. Pediatr Infect Dis J 2: , Schupbach J: Classification, structure, and biology of retroviruses. Curr Top Microbiol Immunol , McGuire TC, O'Rourke KI, Knowles DP, Cheevers WP: Caprine arthritis encephalitis lentivirus transmission and disease. Curr Top Microbiol Immunol 160:61-75, Kennedy-Stoskopf S: Pathogenesis of lentivirus induced arthritis: a review. Rheumatol Int 9: , Schattner A, Bentwich Z: Autoimmunity in human immunodeficiency virus infection. Clin Aspects Autoimmun 5: 19-27, Venables P, Brookes S: Retroviruses: Potential aetiological agents in autoimmune rheumatic disease. Br J Rheumatol 31:841446, Larsson E, Kato N, Cohen M: Human endogenous proviruses. Curr Top Microbiol Immunol 148: , Krieg AM, Gourley MF, Per1 A: Endogenous retroviruses: potential etiologic agents in autoimmunity. FASEB J 6: , Krieg AM, Steinberg AD: Retroviruses and autoimmunity. J Autoimmun 3: , Izui S, Lange G: Enzyme-linked immunosorbent assay for detection of retroviral gp70 and gp70-anti-gp70 immune complexes in sera from SLE mice. Clin Exp Immunol 71:45-49, Ziegler B, Gay RE, Huang G, Fassbender H, Gay S: Immunohistochemical localization of HTLV- 1 p19- and p24-related antigens in synovial joints of patients with rheumatoid arthritis. Am J Pathol 135:l-5, Ziegler BL, Thomas CA: Detection of HIV-I related antigens in rheumatoid synovial tissues. Acta Histochem Cytochem 24~1-9, Shattles WG, Venables PJW, Brookes SM, Clarke DA, Maini RN: Expression of an antigen cross-reactive with HTLV-1 in salivary glands in Sjogren's syndrome. Clin Exp Immunol 89:4651, Garry RF, Fermin CD, Hart DJ, Alexander SS, Donehower LA, Luo-Zhang H: Detection of a human intracisternal A-type retroviral particle antigenically related to HIV. Science 250: , Query CC, Keene JD: A human autoimmune protein associated with U1 RNA contains a region of homology that is cross-reactive with retroviral p3wg antigen. Cell 51: , Talal N, Dauphin6e MJ, Dang H, Alexander SS, Hart DJ, Garry RF Detection of serum antibodies to retroviral proteins in primary Sjogren's syndrome (autoimmune exocrinopathy). Arthritis Rheum 33: , Talal N, Gany RF, Schur PH, Alsexander S, Dauphinee MJ, Livas IH, Ballester A, Takei M, Dang H: A conserved idiotype and antibodies to retroviral proteins in systemic lupus erythematosus. J Clin Invest 85: , Talal N, Flescher E, Dang H: Are endogenous retroviruses involved in human autoimmune disease? J Autoimmun 5 (Suppl A):6146, Festenstein H: Immunogenetic and biological aspects of in vitro lymphocyte allotransformation (MLR) in the mouse. Transpl Rev 1k62-88, Kappler JW, Staerz U, White J, Marrack P: Self-tolerance eliminates T cells specific for Mls-modified products of the major histocompatibility complex. Nature 332:3540, MacDonald HR, Scheider R, Lees RK, Howe RC, Acha-Orbea H, Festenstein H, Zinkernagel R, Hengartner H: T-cell receptor Vp use predicts reactivity and tolerance to MIS"-encoded antigens. Nature 332:40-45, Frankel WN, Rudy C, Coffin JM, Huber BT: Linkage of MIS genes to endogenous mammary tumor viruses of inbred mice. Nature , Woodland DL, Happ MP, Gollob KJ, Palmer E: An endogenous retrovirus mediating deletion of ap T cells? Nature , Dyson PJ, Knight AM, Fairchild S, Simpson E, Tomonari K: Genes encoding ligands for deletion of Vpl 1 T cells cosegregate with mammary tumor virus genomes. Nature , Janeway C: Mls: makes a little sense. Nature 349:459460, 1991

19 476 BEHAR AND PORCELLI 110. Choi Y, Kappler JW, Marrack P: A superantigen encoded in the open reading frame of the 3' long terminal repeat of mouse mammary tumour virus. Nature 350: , Marrack P, Kushnir E, Kappler J: A maternally inherited superantigen encoded by a mammary tumour virus. Nature , Imberti L, Sottini A, Bettinardi A, Puoti M, Primi D: Selective depletion in HIV infection of T cells that bear specific T cell receptor Vp sequences. Science 254:860462, Green PL, Chen ISY: Regulation of human T cell leukemia virus expression. FASEB J 4: , Rice AP, Matthews MB: Transcriptional but not translational regulation of HIV-1 by the tat gene product. Nature , lnoue J, Seiki M, Taniguchi T, Tsuru S, Yoshida M: Induction of interleukin 2 receptor gene expression by p4ox encoded by human T-cell leukemia virus type 1. EMBO J 5: , Maruyama M, Shibuya H, Harada H, Hatakeyama M, Seiki M, Fujita T, Inoue J, Yoshida M, Taniguchi T: Evidence for aberrant activation of the interleukin 2 autocrine loop by HTLV-1-encoded p4ox and T3/Ti complex triggering. Cell 48: , I Siekevitz M, Feinberg MB, Holbrook N, Wong-Staal F, Green WC: Activation of interleukin 2 and interleukin 2 receptor (Tac) promoter expression by the transactivator (tat) gene product of human T-cell leukemia virus, type I. Proc Natl Acad Sci U S A 84:538!%5393, Cross SL, Feinberg JB, Wolf NJ, Holbrook F, Wong-Staal F, Leonard W: Regulation of the human interleukin-2 receptor alpha chain promoter: activation of a nonfunctional promoter by the transactivator gene of HTLV-I. Cell 49:47-56, Miyatake S, Seiki M, Yoshida M, Arai K: T-cell activation signals and human T-cell leukemia virus type I-encoded p4ox protein activate the mouse granulocyte-macrophage colony- stimulating factor gene through a common DNA element. Mol Cell Biol , Miyatake S, Seike M, Malefijt RD, Hike T, Fujisawa J, Takebe Y, Nishida J, Shlomai J, Yokota T, Yoshida M, Arai K, Arai N: Activation of T cell drived lymphokine genes in T cells and fibroblast: effects of human T cell leukemia virus type I p40" protein and bovine papilloma virus encoded E2 protein. Nucleic Acids Res 16: , Kim S-J, Kehrl JH, Burton J, Tendler CL, Jeang K-T, Danielpour D, Thevenin C, Kim KY, Sporn M, Roberts AB: Transactivation of the transforming growth factor p 1 (TGFPl) gene by human T lymphotropic virus type 1 tax: a potential mechanism for the increased production of TGFPl in adult T cell leukemia. J Exp Med 172: , Scala G, Ruocco MR, Ambrosino C, Mallard0 M, Giordano V, Baldassarre F, Dragonetti E, Quinto I, Venuta S: The expression of the interleukin 6 gene is induced by the human immunodeficiency virus 1 TAT protein. J Exp Med 179: , Green JE, Hinrichs SH, Vogel J, Jay G: Exocrinopathy resembling Sjogren's syndrome in HTV-I tax transgenic mice. Nature 341:72-74, Iwakura Y, Tosu M, Yoshida E, Takiguchi M, Sat0 K, Kitajima I, Nishioka K, Yamamoto K, Takeda T, Hatanaka M, Yamamoto H, Sekiguchi T: Induction of inflammatory arthropathy resembling rheumatoid arthritis in mice transgenic for HTLV-1. Science 253: , Mountz JD, Tala1 N: Retroviruses, apoptosis and autogenes. Immunol Today , Watanabe-Fukunaga R, Brannan CI, Copeland NG, Jenkins NA, Nagata S: Lymphoproliferation disorder in mice explained by defects in Fas antigen that mediates apoptosis. Nature 356:31&317, Wu J, Zhou T, He J, Mountz JD: Autoimmune disease in mice due to integration of an endogenous retrovirus in an apoptosis gene. J Exp Med 178: , 1993