Absence of β₂ Integrins Impairs Regulatory T Cells and Exacerbates CD4⁺ T Cell-Dependent Autoimmune Carditis

Leukocytic infiltration of multiple tissue types is the hallmark of systemic autoimmune diseases. The cardiovascular system is frequently targeted in patients with systemic autoantibody-associated disorders, leading to increased morbidity and mortality. Well-known examples include the association of coronary artery inflammation with rheumatoid arthritis and systemic lupus erythematosus and the occurrence of cardiac valve inflammation (endocarditis) in patients with rheumatic fever, systemic lupus erythematosus, antiphospholipid Ab syndrome, and occasionally rheumatoid arthritis (1–5). How is the cardiovascular system targeted for attack in these disorders? More broadly, in systemic autoimmune diseases, are different immunopathogenic mechanisms at work in the various target organs?

We have taken advantage of the coexistence of autoimmune endocarditis in the K/BxN TCR transgenic mouse model of arthritis to begin to address these questions (6). In this model, T and B cell autoreactivity against the ubiquitously expressed Ag glucose-6-phosphate isomerase (GPI) results in the sustained production of high-titer arthritogenic anti-GPI Abs (7, 8). Interruption of the immunologic events leading up to anti-GPI autoantibody production prevents inflammation in both the joints and the heart. For instance, mice lacking B cells develop disease in neither tissue (6). However, the pathogenic effector mechanisms in the two target organs diverge downstream of autoantibody production. Specifically, arthritis in K/BxN TCR transgenic mice relies on complement component C5 but not activating Fcγ receptors, whereas endocarditis depends primarily on activating Fcγ receptors rather than C5. Additionally, although arthritis can easily be transferred via injection of serum (containing anti-GPI Abs) from a K/BxN TCR transgenic mouse into a naive recipient, endocarditis cannot (6). These findings support the notion that the immunopathogenic mechanisms responsible for end-organ inflammation in systemic autoimmune diseases can indeed vary between target organs in a single organism.

Particular attention has been paid to the role of CD4⁺ T lymphocytes in the pathogenesis of rheumatic carditis. CD4⁺ T cells are found in the valve lesions in humans and in animal models, and there is much interest in whether the self-Ags they recognize are structural mimics of bacterial peptides (3, 9). Whether CD4⁺ T cells are necessary effectors in the pathogenesis of autoimmune carditis, however, has not been clearly defined. The K/BxN mouse model has allowed us to test directly whether CD4⁺ T cells are required for the development of autoimmune carditis.

Inflammation entails the recruitment of leukocytes from the circulation into tissues. Cell surface adhesion molecules mediate leukocyte attachment to the vascular endothelium, a critical step in the inflammatory cascade. The β₂ integrins are a major family of adhesion molecules expressed by cells of the hematopoietic lineage. The common β-chain of β₂ integrins, CD18, heterodimerizes with one of four α subunits (CD11a, b, c, or d) to form functional receptors capable of binding a number of endothelial ligands, including ICAM-1 (CD54), as well as molecules of the extracellular matrix. In addition to their role in leukocyte adhesion to vascular endothelium, the β₂ integrins participate in hemostasis, the formation of stable synapses between leukocytes, and other processes (10–12). Highlighting the importance of these molecules during an inflammatory response, deficiency of CD18 in humans and mice causes leukocyte adhesion deficiency syndrome, characterized by leukocytosis, increased susceptibility to infections, and impaired wound healing (12, 13). Likewise, the β₂ integrins participate in inflammation in the context of autoimmune diseases. Studies using different animal models of autoimmunity have indicated a role for one or more of the β₂ integrins in promoting the development of type I diabetes, lupus-like disease, collagen-induced arthritis, experimental autoimmune encephalomyelitis (EAE), colitis, and psoriasis (14–21). Most relevant to the current study, mice lacking CD11a or CD18 are protected against the development of K/BxN serum-transferred arthritis (22). Collectively, these studies suggest that differential usage of β₂ integrin heterodimers directs leukocytes from the blood to different target tissues, the specificity of which might depend both on the anatomic site as well as the nature of the inflammatory stimulus.

Leukocyte adhesion to the vascular endothelium is not, however, the only mechanism by which β₂ integrins contribute to autoimmune disease pathogenesis. Interestingly, the development and function of regulatory T cells (Tregs) is compromised in mice lacking β₂ integrins, leading to impaired peripheral immunological tolerance. In a model of autoimmune colitis, deficiency of CD18 led to reduced numbers of Tregs in the secondary lymphoid organs and increased disease severity (23), a phenotype that others have suggested is due to absence of LFA-1 (CD11a/CD18) (24). Furthermore, in a model of psoriasis, Tregs derived from mice expressing a hypomorphic variant of CD18 demonstrated decreased proliferation and impaired suppressive function, attributed to reduced cell-to-cell contact (20). Understanding how β₂ integrins contribute to the pathogenesis of autoimmune diseases therefore requires consideration of not only their role in recruiting leukocytes to inflamed tissues, but also their capacity to help maintain immunological tolerance. In this study, we explored the contribution of CD4⁺ T cells and β₂ integrins to the development of arthritis versus endocarditis in the K/BxN mouse model.

KRN TCR transgenic mice on the C57BL/6 background (7) were a gift from Drs. Diane Mathis and Christophe Benoist (Harvard Medical School, Boston, MA) and the Institut de Génétique et de Biologie Moléculaire et Cellulaire (Strasbourg, France); C57BL/6 mice congenic for H2^g7 (B6.g7) (6) were also a gift from Drs. Mathis and Benoist. CD18-null mice on the C57BL/6 background (Itgb2^tm2Bay, stock no. 003329) (13), originally obtained from The Jackson Laboratory (Bar Harbor, ME), were a gift from Dr. Yoji Shimizu at the University of Minnesota; CD11a (Itgal^tm1Bll, stock no. 005257) (25), CD11b (Itgam^tm1Myd, stock no. 003991) (26), and Rag1-deficient mice (Rag1^tm1Mom, stock no. 002216) on the C57BL/6 background were purchased from The Jackson Laboratory and bred in our specific-pathogen-free colonies. The β₂ integrin-deficient K/BxN lines were created by breeding mice bearing the appropriate β₂ integrin knockout (KO) allele on the C57BL/6 background to KRN/B6 mice and also to C57BL/6 mice congenic for the NOD-derived MHC, A^g7. The MHC is the only NOD-derived genetic region the mice retain; to simplify nomenclature, however, we refer to the mice as K/BxN throughout this study as we have previously (6). Genotyping was performed by PCR and confirmed by flow cytometry. All studies were conducted in accordance with Institutional Animal Care and Use Committee-approved protocols at the University of Minnesota (protocol nos. 0611A96106 and 0909A72086).

mAbs used for flow cytometry included anti-Vβ6 (clone RR4-7), CD4 (RM4-5), CXCR5 (2G8), CD138 (281-2), CD16/32 (2.4G2), CD162 (2PH1), and isotype control Abs IgG1 (R3-34), IgG2a (R35-95), and IgG2b (A95-1) purchased from BD Pharmingen, and CD44 (IM7), CD45R (RA3-6B2), CD45.1 (A20), CD45.2 (104), CD62L (MEL-14), CD278 (7E.17G9), and isotype control Abs IgG2a (eBR2a) and IgG2b (eB149/10HS) from eBioscience. Intracellular staining for CD25⁺ (PC61.5) Foxp3⁺ (FJK-16s) cells was performed per the manufacturer’s protocol (eBioscience). Flow cytometry was performed using a FACSCalibur and an LSRII (BD Biosciences), and cells were analyzed using FlowJo v7.6 software (Tree Star).

Histological sections of cardiac valves were prepared using hearts snap-frozen in optimal cutting temperature compound, and 10-μm cryosections were prepared using a Leica CM305 S cryostat. Tissues were stained with H&E using standard protocols and imaged with an Olympus BX51 microscope or (where noted) a Leica DM5500 B stitching microscope. Mitral valve thickness was determined for each heart by measuring the thickest point of the valve in serial sections, as described (6). Aortic valve, pericardial, and coronary artery inflammation were assessed by examining serial sections containing each structure and scored for the presence or absence of inflammation.

Following Fc receptor blocking with anti-CD16/32 (clone 93; eBioscience) and anti-CD64 (clone N-19; Santa Cruz Biotechnology) Abs, heart sections were stained with biotinylated Abs recognizing CD4 (L3T4), CD11a (M17/4), CD54 (ICAM-1, clone 3E2), and isotype control Abs IgG2a (R35-95), IgG2b (A95-1), and IgG1 (A19-3) purchased from BD Pharmingen; anti-CD11c (N418), CD18 (M18/2), Foxp3 (FJK-16s), and IFN-γ (R4-6A2) were from eBioscience. Unconjugated anti–IL-17 (E-19) and secondary biotinylated IgG from Santa Cruz Biotechnology were used in combination. Biotinylated Abs were detected using the Vectastain ABC-AP kit with either the Vector Red substrate I or the ImmPACT diaminobenzidine substrate kit (Vector Laboratories) and counterstained with DAPI or hematoxylin to visualize nuclei. Imaging was performed using an Olympus BX51 fluorescent microscope equipped with a digital camera and DP-BSW software (Olympus).

Arthritis was measured and total serum IgG and anti-GPI titers as well as IgG subtypes were determined as described (27, 28).

Within 2 d onset of arthritis, K/BxN mice were treated with 30 or 60 μg purified monoclonal anti-mouse CD4 (GK1.5) or IgG2b isotype control Abs (eBioscience) via i.p. injection for 3 consecutive days. Mice were then aged for 4 wk and assessed for arthritis and carditis as described above. The different doses reflect two different experiments; no difference in response was observed between the two doses, so the results were combined.

Pooled serum (100 μl/dose) from CD18-sufficient or -deficient K/BxN mice was injected i.p. into 6-wk-old C57BL/6 recipient mice on days 0 and 2. Recipient mice were allowed to develop arthritis for 3 wk, after which time their hearts were harvested and assessed for carditis as described above.

Six-week-old Rag1-deficient mice were sublethally irradiated with 300 rad. Four hours after irradiation, 10 × 10⁶ bone marrow cells from either K/BxN, CD18-deficient K/BxN, or a mixture thereof were injected retro-orbitally. Bone marrow chimeric mice were maintained on sulfamethoxazole and trimethoprim administered in drinking water for the duration of the study. Two months following bone marrow transplantation, the lymphoid organs and hearts were harvested for analysis.

Statistical differences between mean values for groups were calculated using the Student two-tailed t test. One-way ANOVA followed by a Tukey multiple comparison test was used to assess differences in the extent of cardiac pathology between integrin-sufficient and -deficient mice.

The predominant inflammatory cell types differ in the synovial joints versus the cardiac valves in K/BxN TCR transgenic mice. Whereas neutrophils are the main cell type found in the arthritic joints (7), the valve-infiltrating cells comprise primarily mononuclear cells including CD4⁺ T lymphocytes (6). This apparent difference in effector cell types led us to explore an additional requirement for CD4⁺ T cells in the pathogenesis of endocarditis beyond the initiation of the autoimmune response. Previous studies have shown that depletion of CD4⁺ T cells in K/BxN mice before the onset of arthritis prevented its development, whereas depletion of CD4⁺ T cells after arthritis onset had no effect on its severity (7) (Fig. 1A). In stark contrast, we found that depletion of CD4⁺ T cells after the onset of arthritis in K/BxN mice significantly decreased the severity of cardiac valve inflammation (Fig. 1B), demonstrating a critical requirement for a sustained CD4⁺ T cell presence in the pathogenesis of endocarditis. The valve lesions were characterized by a prominence of the Th1 effector cytokine IFN-γ, along with some IL-17 (Supplemental Fig. 1). We did not detect IL-4 in the valve lesions (data not shown). Thus, whereas autoantibodies are sufficient to provoke arthritis in this model, our earlier results (6) and these, considered together, suggest that both autoantibodies and CD4⁺ T cells are required to provoke the cardiac pathology.

We were interested in determining how members of the β₂ family of integrins contribute to cardiac pathology in K/BxN mice. Immunohistochemical analysis of inflamed mitral valves revealed the presence of CD18, CD11a, and CD11c (Fig. 2A), as well as CD11b expression (6). Increased endothelial cell surface expression of specific ligands for β₂ integrins occurs in the context of inflammation to facilitate leukocyte migration into the tissues. We next investigated whether the endothelium of K/BxN cardiac valves had an activated phenotype. Indeed, we observed increased expression of ICAM-1, the primary counterreceptor for the β₂ integrins, on inflamed K/BxN mitral valves compared with noninflamed valves from TCR transgene-negative control mice (Fig. 2B). Thus, β₂ integrins are expressed by the tissue-infiltrating leukocytes of K/BxN cardiac valves, and under inflammatory conditions the endothelial cells lining the cardiac valves upregulate a main β₂ integrin ligand. To determine the role of β₂ integrins in the development of autoimmune endocarditis, we generated K/BxN TCR transgenic mice lacking CD18, CD11a, or CD11b.

Based on the strict requirement for LFA-1 (CD11a/CD18) in the development of K/BxN serum-transferred arthritis, we anticipated that deficiency of CD18 in K/BxN TCR transgenic mice would similarly result in decreased arthritis severity (13, 22). However, the absence of CD18, CD11a, or CD11b had no observable effect on the development of arthritis in the TCR transgenic mice (Fig. 3A–F). Moreover, we observed no differences in the pathogenic anti-GPI autoantibody titers (Fig. 3G–I). These findings suggest that deficiency of β₂ integrins impairs neither the priming of KRN T cells nor their ability to provide help to GPI-specific B cells.

Unexpectedly, we observed more severe cardiovascular disease in CD18-deficient K/BxN mice compared with CD18-sufficient littermates. The cardiac involvement in K/BxN mice is largely limited to the mitral valve with occasional aortic valve inflammation (6), consistent with what we observed in CD18-heterozygous K/BxN mice (Fig. 4). Although the severity of mitral valve inflammation was not affected by CD18 deficiency (Supplemental Fig. 2), additional cardiac structures were inflamed. Specifically, we found that the absence of CD18 resulted in inflammation not only of the mitral valve (n = 8 of 8 mice) but also the aortic valve (n = 8 of 8), coronary arteries (n = 5 of 8), and pericardium (n = 5 of 8) (Fig. 4). K/BxN mice lacking CD11a exhibited a similar yet less penetrant increase in the scope of carditis compared with CD11a-heterozygous littermates, whereas deficiency of CD11b did not increase the extent of the cardiac phenotype (Fig. 4I). Thus, although LFA-1 is essential for arthritis induced by the transfer of K/BxN serum (22), deficiency of β₂ integrins in K/BxN TCR transgenic mice had no apparent effect on the course of arthritis and actually increased the extent of cardiovascular pathology, an effect attributable primarily to LFA-1. We next investigated the possible mechanisms by which the absence of CD18 augmented autoimmune endocarditis.

Given the critical role of CD4⁺ T cells in the pathogenesis of K/BxN carditis, we first explored whether differences in T cell activation might underlie the more extensive cardiac involvement in β₂ integrin-deficient K/BxN mice. Autoreactive CD4⁺ T lymphocytes isolated from arthritic K/BxN mice have an activated phenotype consistent with their disease state (29). We found that the expression of the T cell activation marker CD44 was further increased in the absence of CD18 (Fig. 5A). A similar hyperactivated state was noted among T cells from K/BxN mice lacking CD11a; however, this phenotype was less apparent in CD11b-deficient mice (Fig. 5B, 5C). This increased expression of CD44 in CD18- and CD11a-deficient K/BxN mice echoed the exacerbated cardiovascular pathology noted in the same animals, potentially indicating a defect in T cell regulation in the absence of β₂ integrins.

We also asked whether dysregulation of B cells could underlie the increased severity of the cardiovascular pathology seen in CD18-deficient K/BxN mice, as we have previously shown that activating Fcγ receptors (that bind IgG) are required specifically for the development of K/BxN endocarditis (6) and both K/BxN and CD18-deficient C57BL/6 mice are reported to have hypergammaglobulinemia (7, 13). Indeed, the genetic combination of the KRN transgene and CD18 deficiency resulted in even more exaggerated hypergammaglobulinemia with levels of circulating total IgG 4- to 5-fold greater in CD18^−/− K/BxN mice compared with CD18-sufficient K/BxN mice or with CD18-deficient KRN transgene-negative littermates (Fig. 5D). In contrast, mice lacking CD11a or CD11b had total serum IgG titers equivalent to control mice (Fig. 5E, 5F). Consistent with the fact that IgG1 is the predominant IgG subtype generated in K/BxN mice, we found that the elevated total IgG concentration in CD18-deficient K/BxN mice was due largely to increased levels of IgG1, although levels of IgG2b and IgG2c (the C57BL/6 variant of IgG2a; see Ref. 30) were also increased relative to control CD18-sufficient K/BxN mice (Supplemental Fig. 3A). Furthermore, although the titers of anti-GPI IgG were equivalent in CD18-deficient and -sufficient K/BxN mice (Fig. 3G), the distribution of IgG subtypes within the GPI-reactive Ab pool was altered in the CD18-deficient mice, with a shift toward IgG2b and IgG2c and away from IgG1 (Supplemental Fig. 3B).

To address the possibility that these observed changes in IgG subtype distribution are responsible for the increased extent of cardiac disease in CD18-deficient K/BxN mice, we performed serum-transferred arthritis experiments using serum from CD18-sufficient or -deficient K/BxN mice. Mice that received serum from CD18^−/− K/BxN mice developed slightly more severe arthritis than those given the control K/BxN serum (Supplemental Fig. 4A, 4B). Importantly, none of the mice developed cardiac pathology (Supplemental Fig. 4C), suggesting that alterations in Ab production and/or subclass usage alone are not responsible for the increased extent of cardiac disease in CD18^−/− K/BxN mice. These findings are consistent with the notion that autoreactive CD4⁺ T cells are required in addition to autoantibodies for the pathogenesis of carditis (see Fig. 1).

The hypergammaglobulinemia and increased T cell activation observed in CD18-deficient K/BxN mice prompted us to explore whether a more generalized defect in immune regulation existed. Accordingly, we found that CD18-deficient K/BxN mice had significant splenomegaly with a corresponding increase in total splenocyte numbers relative to CD18-sufficient K/BxN mice (Fig. 6A, 6B). We investigated whether a deficiency of Tregs might underlie this phenotype. Indeed, both the percentage and absolute number of Tregs in the lymph nodes of K/BxN mice lacking CD18 were less than half of those found in CD18-sufficient mice (Fig. 6C–E). The percentage of Tregs was also decreased in the spleens of CD18-deficient animals, indicating a deficit in the total numbers of Tregs per mouse (Fig. 6D, 6E). Furthermore, although Foxp3⁺ cells were easily identified in the cardiac valves of CD18-sufficient mice, we observed fewer of these cells in the valves of CD18-deficient K/BxN mice (Fig. 7).

CD18-deficient K/BxN mice shared many phenotypic characteristics reported recently in K/BxN mice lacking all Tregs, due to the scurfy mutation (31). In particular, as reported in K/BxN scurfy mice (31), we found that CD18-deficient K/BxN mice had increased numbers of follicular Th cells and extrafollicular Th cells relative to K/BxN mice (Fig. 8A), along with an exaggerated accumulation of plasma cells in the secondary lymphoid organs and decreased expression of the inhibitory Fc receptor FcγRIIB on the plasma cells (Fig. 8B–D). These data are consistent with an intrinsic Treg defect in CD18-deficient K/BxN mice.

To explore this possibility further, we generated mixed bone marrow chimeric mice. Specifically, we sought to determine whether wild-type (WT) K/BxN Tregs could exert dominant immunological tolerance and thereby constrain the increased cardiac pathology engendered by CD18 deficiency. Congruent with our genetic model, transplantation of WT K/BxN bone marrow alone resulted in less severe cardiovascular disease than did transplantation of CD18-deficient (KO) K/BxN marrow (Fig. 9A). Importantly, both the 50:50 (WT/KO) and 10:90 (WT/KO) mixed chimeras were protected from severe cardiac pathology relative to those mice that received only CD18 KO marrow (Fig. 9A), demonstrating a dominant tolerance-inducing effect of the WT bone marrow. These experiments also revealed a proliferative or survival disadvantage of the CD18^−/− K/BxN T cells, as just a few hundred of the KO T cells were detected in the lymph nodes and spleens of the 50:50 chimeras and comprised only ∼10% of the T cells in the 10:90 (WT/KO) chimeras (Fig. 9B, 9C and data not shown). Additionally, the presence of WT Tregs resulted in both the WT and KO effector T cell populations displaying a phenotype similar to WT K/BxN mice (higher CD62L, lower CD44) rather than the hyperactivated phenotype observed in CD18-deficient K/BxN mice (Fig. 9D, 9E), further supporting the dominant suppressive effect of the WT Tregs. These findings suggest that deficiency of CD18 impairs Tregs numerically and functionally, leading to increased activation of pathogenic effector CD4⁺ T cells as well as hypergammaglobulinemia, and culminating in more extensive autoimmune carditis.

A better understanding of the disparate means by which systemic inflammation generates tissue-specific destruction is essential for the development of more directed therapeutic strategies for multiorgan autoimmune diseases. The genetic manipulability of the K/BxN TCR transgenic mouse model of spontaneous coexisting autoimmune endocarditis and arthritis affords the opportunity to dissect such mechanistic requirements. In this study, we have shown that CD4⁺ T lymphocytes in addition to autoantibodies are required as effectors of autoimmune carditis. We have also shown that β₂ integrins play a critical and unexpected role in limiting the extent of cardiovascular pathology via their influence on Tregs.

We have previously proposed that T cells contribute directly to the pathogenesis of endocarditis based on the finding that arthritis can be elicited by transfer of K/BxN serum, splenocytes, or bone marrow (7, 29), whereas endocarditis is only faithfully recapitulated with transplantation of bone marrow (6). The present study supports this model. Specifically, depletion of CD4⁺ T cells after the onset of arthritis protected K/BxN mice from endocarditis without affecting arthritis severity. Cumulatively, our findings suggest that autoantibodies alone are sufficient to induce arthritis, whereas the development of carditis in K/BxN mice depends on the additional sustained presence of effector CD4⁺ T cells.

Our finding that CD18 deficiency resulted in more severe cardiac pathology was somewhat unexpected given the requirement for CD18 in the K/BxN serum-transferred arthritis model (22).

Although initial reports of CD18-null mice pointed to a potential defect in T cell priming (13), this appears not to impair the development of spontaneous autoimmunity in several models, now including K/BxN TCR transgenic mice. Rather, the β₂ integrins seem to contribute critically to the development and/or function of Tregs in this and several other models of autoimmune or inflammatory diseases (20, 23, 24), resulting in exaggerated disease phenotypes. It is possible that deficiency of CD18 on effector T cells renders those cells less susceptible to Treg-mediated inhibition; however, the alteration in CD18^−/− K/BxN effector T cell activation phenotype in the 10:90 (WT/KO) bone marrow chimera and less severe carditis indicates functional suppression of those cells. Tregs are critically required to exert control over the development of K/BxN arthritis (32), and genetic ablation of Tregs in the K/BxN TCR transgenic model (via introduction of the scurfy allele) results in more aggressive inflammatory arthritis (27), with a corresponding increase in follicular and extrafollicular Th cells and an aberrant accumulation of plasma cells in the secondary lymphoid organs (31). Although we did not observe increased arthritis severity in the absence of CD18, we did see similar increases in the Th subsets as well as the noted plasma cell abnormalities. That we observed no difference in arthritis severity due to CD18 deficiency likely reflects the fact that the scurfy mutation eliminates all Foxp3⁺ Tregs, whereas the CD18-deficient mice retain some Tregs.

The requirement for B cells, CD40, and FcRγ in the development of endocarditis suggests that autoantibodies contribute to the cardiac pathology via engagement of activating Fc receptors (5). Although most anti-GPI Abs produced by K/BxN mice are of the IgG1 subtype (7), we were intrigued to find that deficiency of CD18 resulted in increased production of IgG2b and IgG2c along with more severe cardiac disease, potentially indicating a role for FcγRI and/or FcγRIV, the activating Fc receptors that bind these IgG subtypes preferentially, in the pathogenesis of endocarditis in this model (33). However, it seems unlikely that hypergammaglobulinemia and altered IgG subtype usage alone are responsible for the more severe cardiovascular disease in the CD18-deficient K/BxN mice, because the CD11a-deficient K/BxN mice had more severe carditis without hypergammaglobulinemia, and transfer of CD18^−/− K/BxN serum failed to provoke cardiac pathology. These findings underscore our interpretation that carditis in K/BxN mice depends on both autoantibodies and effector CD4⁺ T cells.

That our results differ from the previous report that LFA-1 (CD11a/CD18) is required for the development of arthritis induced by injection of serum from K/BxN mice (22) highlights the inherent differences between the K/BxN TCR transgenic model and its derivative model, serum-transferred arthritis. We have previously reported a similar difference between the two models in mice lacking FcRγ, the signaling chain shared by activating Fc receptors: that is, complete protection from serum-transferred arthritis but no effect on arthritogenesis in the transgenic system (6). These differences likely reflect the fact that the serum-transferred arthritis model interrogates the effector mechanisms linking autoantibodies to the development of arthritis, in the absence of an ongoing adaptive autoimmune response. In contrast, K/BxN TCR transgenic mice have a sustained adaptive autoimmune response characterized by the presence of autoreactive T and B lymphocytes and substantially higher titers of anti-GPI autoantibodies than those achieved in the serum transfer system. Thus, in the K/BxN serum transfer arthritis system, it is likely that absence of LFA-1 impairs neutrophil migration to the joints in response to the injected autoantibodies (22). Indeed, LFA-1 expression by neutrophils was recently demonstrated to be crucial in the serum transfer model (34). In contrast, in the K/BxN TCR transgenic animals, deficiency of β₂ integrins results in impaired regulation of the ongoing adaptive autoimmune response, resulting in the hyperimmune state and more severe cardiovascular pathology we report. That arthritis develops in K/BxN TCR transgenic animals even in the absence of β₂ integrins suggests that the inflammatory drive in these mice engages alternative mechanisms to recruit cells to the joints (and cardiac valves).

A critical difference between the heart and the joints is that dendritic cells (DCs), which are potent APCs, are well-described residents of normal cardiac valves (35), but not of the healthy synovium (36). Thus, one plausible sequence of events leading to the development of endocarditis in this model is that circulating immune complexes bind to and activate the valve-resident DCs via Fcγ receptors, resulting in the recruitment and activation of pathogenic effector T cells (37, 38). Tregs may be similarly recruited and activated, helping to constrain the inflammatory response (39, 40). Interestingly, CD18- and CD11a-deficient Tregs have been reported to lose the ability to form aggregates around DCs, resulting in a functional inability of Tregs to produce suppressive cytokines (20, 41). This may then allow the inflammatory process in the heart of CD18-deficient K/BxN mice to go unchecked and spread to additional cardiac structures, as we observed. Note also that this working model leaves open the possibility that GPI might not be the crucial T or B cell autoantigen in the development of endocarditis—other antigenic specifities (or Ag nonspecific immune complexes) could be at play. Importantly, a similar model, including T cell epitope spreading, has been proposed for the pathogenesis of rheumatic heart disease (3).

How do our findings inform our understanding of human diseases? Cell adhesion molecules have been attractive targets of therapeutic intervention for a variety of diseases, based on the notion that such interference would impede trafficking of leukocytes to sites of inflammation (10–12). For instance, mAbs specific for integrins have been approved for the treatment of autoimmune diseases in humans, such as natalizumab (anti-α₄ integrin) for multiple sclerosis and efalizumab (anti-CD11a) for psoriasis. Similarly, a small molecule inhibitor of LFA-1 has recently been shown to reduce the severity of collagen-induced and anti-collagen Ab-induced arthritis in mice (42). However, our data suggest that one must also consider the effect that β₂ integrins exert on the maintenance of peripheral tolerance, and that interference with such pathways might sometimes provoke autoimmune pathology. This may explain the reported occurrence of immune-mediated hematologic cytopenias and other immune-mediated disorders in some patients taking efalizumab (43–46), as well as the seemingly paradoxical occurrence of juvenile arthritis in some patients with leukocyte adhesion deficiency syndrome (47).

The data presented in this study offer new insights into the complex orchestration of the cells and molecules involved in the two related, yet distinct, diseases of the K/BxN TCR transgenic mouse. Autoreactive T lymphocytes provide the requisite B cell help to stimulate the production of high-titer autoantibodies necessary for the induction of both endocarditis and arthritis (7, 8, 29); however the immune perpetrators responsible for pathogenic inflammation in the heart and joints differ. Once autoantibodies are formed, they are sufficient to provoke arthritis. In contrast, the cardiac pathology depends on an additional ongoing contribution of effector CD4⁺ T cells. Deficiency of CD18 impairs Tregs, resulting in hyperimmune activation and more severe cardiac pathology.

We thank Drs. Christophe Benoist and Diane Mathis for mice, Dr. Yoji Shimizu for helpful review of the manuscript and for mice, Sindhuja Rao, Tyler Potts, Nadia Rusinak, and Jonathan Dexter for technical assistance, and Dr. Phillipe Gaillard and Andrew Way for statistical analysis.

The authors have no financial conflicts of interest.