CD4 T cells, lymphopenia, and IL-7 in a multistep pathway to autoimmunity

Abstract

There are many inhibitory mechanisms that function at the cellular and molecular levels to maintain tolerance. Despite this, self-reactive clones escape regulatory mechanisms and cause autoimmunity in certain circumstances. We hypothesized that the same mechanisms that permit T cells to expand during homeostatic proliferation may inadvertently promote autoimmunity under certain conditions. One major homeostatic cytokine is IL-7, and studies have linked it or its receptor to the development of multiple sclerosis and other autoimmune diseases. We show in a model of β-islet cell self-reactivity that the transfer of activated autoreactive CD4 T cells can prime and expand endogenous autoreactive CD8 T cells in a CD28- and CD40-dependent manner through the licensing of dendritic cells. Despite this, mice do not develop diabetes. However, the provision of exogenous IL-7 or the physiological production of IL-7 associated with lymphopenia was able to profoundly promote the expansion of self-reactive clones even in the presence of regulatory T cells. Autoimmune diabetes rapidly ensued with CD4 help and the subsequent activation of CD8 T cells, which contributed to disease progression. With the advent of many biologicals targeting TNFα, IL-6, and IL-1 and their effective use in the treatment of autoimmune diseases, we propose that IL-7 and its receptor may be promising targets for biological agents in the treatment of autoimmunity.

The diversity of the T cell repertoire is required to allow recognition of all pathogens. A consequence of such diversity is the potential for pathological self-reactivity (1). Many mouse models of autoimmunity have been developed that rely on the transgenic expression of self-reactive T cell receptors, vaccination using self-antigens, viral infection, or adoptive transfer of activated self-reactive T cell clones. All these models have clearly defined a role for CD4 T cells in disease pathogenesis (2–9). The heavy bias toward CD4 T cell involvement in these models could indicate that T helper cells are required for full activation and expansion of effector CD8 T cells (10), that CD4 T cells are directly pathogenic (7), or that self-reactive CD4 T cell clones are more prone to spontaneous activation than CD8 because of their TCR repertoire, class II MHC restriction, or ability to respond to cytokines or inflammatory signals.

The complex nature of autoimmune disease and its incidence suggests that there are many checkpoints and cell types that influence the outcome of immune pathology. The exact mechanisms that promote activation and expansion of self-reactive clones are not clear (1). It is intriguing that, despite the many cellular and molecular inhibitory networks, self-reactive clones may persist in an activated state and proliferate (1). Although highly speculative, it is plausible that the mechanisms that permit homeostatic activation and proliferation of T cells also may function in an aberrant manner to allow the expansion and activation of self-reactive T cell clones (1). Interestingly, many studies have linked the development or exacerbation of autoimmunity to relative lymphodepletion (2, 4, 11, 12). Several mechanisms have been proposed for this increased propensity to develop autoimmune reactivity (13). One hypothesis is that the cytokine milieu associated with lymphopenia may support and augment immune responses, including autoimmunity (14). However, some of the models are potentially complicated by the preferential loss of regulatory T cells (Treg) (13).

If self-reactive CD4 T cells do become activated and expand in the presence of a permissive cytokine environment, how do they then participate in the pathogenesis of autoimmune disease (15–17)? The exact mechanism as to how CD4 T cells promote and sustain an autoimmune response is still unclear. CD4 T cells may by themselves be pathogenic by direct or indirect release of tissue toxic cytokines or by inducing Fas ligand-mediated apoptosis of target tissues (7). It also has been postulated that CD4 cells may promote or license dendritic cells, through CD40 interactions, which in turn initiate and promote CD8 T cell autoreactivity (17). Alternatively, CD4 T cells may directly promote CD8 autoimmune responses (18). Several reports have suggested that CD4 T cell help is essential to facilitate CD8 T cell immunity at least in cases where dendritic cells (DCs) are inefficiently conditioned by pathogen-derived signals (19). Such conditioning would be very limited in autoimmune responses. This hypothesis is supported by the adoptive transfer of transgenic CD4 T cells in nonobese diabetic (NOD).scid mice (15) and the induction of diabetes in rat insulin promoter (RIP)-OVA-transgenic mice coinjected with CD4 and CD8 OVA-specific T cells (16, 17).

In the current study, we investigate the factors that promote CD4-mediated autoimmune disease. Specifically, we attempt to identify factors associated with lymphopenic states that augment and promote autoimmunity. We focused on IL-7 because this cytokine and its receptor have been linked to multiple autoimmune disorders in humans (20, 21). IL-7 is a major homeostatic cytokine, and transgenic overexpression results in severe autoimmune disease (22, 23), but its physiological relevance in the development of autoimmunity remains unclear. To address these questions, we used the RIP-glycoprotein (GP)-transgenic mouse model, in which the RIP drives the expression of the lymphocytic choriomeningitis virus (LCMV) GP in pancreatic β-islet cells (24). Constitutive LCMV GP expression in the pancreas of RIP-GP mice is not associated with significant central or peripheral deletion of GP-reactive CD8 T cells, nor does it result in spontaneous GP-specific T cell immunity and diabetes (24). To study the role of CD4 T cells in promoting autoimmunity, we crossed the RIP-GP mice with Smarta-transgenic mice bearing CD4 T cells expressing a T cell receptor (TCR) specific for the I-A^b-restricted LCMV GP61–80 (p13) epitope. Alternatively, these cells were adoptively transferred into RIP-GP mice (25). Using these tools, we were able to study the multiple mechanisms that are required to induce diabetes in this transgenic model.

Results

Transfer of Activated Smarta CD4 T Cells in RIP-GP Mice Promotes the Priming of Endogenous P14 TCR Transgenic CD8 T Cells Specifically in Draining Lymph Nodes and Causes Autoimmunity.

To investigate the potential role of CD4 T cells in initiating autoimmunity, in vitro-activated Smarta CD4 T cells were transferred into P14/RIP-GP double-transgenic mice. P14 mice have CD8 T cells that express a transgenic TCR recognizing the LCMV-GP33/H-2D^b-restricted epitope (26). Twenty days after Smarta transfer, ≈40% of P14/RIP-GP mice developed diabetes, whereas mice that did not receive the activated CD4 help remained euglycemic (Fig. 1A). The potential for Smarta to induce diabetes was increased to 100% with coinjection of the immunostimulatory unmethylated CpG oligonucleotide ODN 1826, which acts as a TLR9 ligand. The control GpC oligonucleotide ODN 1928 had no enhancing effect. The P14/RIP-GP model allowed us to dissect the effect of activated self-reactive CD4 T cells on the P14 GP-specific T cells, which are mostly Vα2⁺ and, hence, easily identified. We found that transferred Smarta T cells caused an up-regulation of the activation marker CD44 on P14 CD8 T cells specifically in the pancreatic draining lymph nodes (PDLN) (Fig. 1B). Therefore, the transfer of activated self-antigen-specific CD4 T cells is sufficient to initiate a CD8-mediated autoimmune response confined to the regional LN, which can be further augmented by TLR stimuli.

Fig. 1.

Activated GP-specific CD4 T cells promote the priming of GP-specific CD8 T cells in pancreatic draining LN of P14/RIP-GP mice, resulting in autoimmunity. (A) Kinetics of hyperglycemia development in P14/RIP-GP mice after adoptive transfer of activated Smarta CD4 T cells (n = 12 per group). Experiments were repeated three times. The P values were generated by using a time-to-event analysis and a log-rank test. The P value for the difference between Smarta+ODN1826 and Smarta+ODN1928 is 0.0012. (B) GP-specific P14 CD8 T cells in PDLN up-regulate the activation marker CD44 specifically in LN draining the pancreas and in the presence of GP-specific CD4 T cells. CD44 levels from nondraining iLN and PDLN are compared for each condition tested on day 13 after Smarta or ODN1826 injection. Dot plots are gated on CD8⁺Vα2⁺ cells (n = 4 per group). The experiment was repeated three times.

GP-Specific CD4 T Cells Must Be Activated to Promote Localized Priming and Expansion of GP-Specific CD8 T Cells.

We further characterized this model in several ways. Surprisingly, we found that Smarta-transgenic mice intercrossed with RIP-GP mice rarely developed spontaneous diabetes (<1% of animals followed over 6 months). Furthermore, Smarta/RIP-GP mice develop diabetes infrequently (only 35%) even when infected with LCMV, which should activate the CD4 Smarta T cells in vivo. This low incidence of diabetes was due to a severely limited CD8 TCR repertoire in Smarta-transgenic animals and consequently a very skewed and limited LCMV-induced CD8 T cell response, resulting in poor pancreatic infiltration [supporting information (SI) Fig. 7]. These findings suggest that activated CD4 self-reactive T cells are required, together with CD8 T cells, to initiate autoimmunity in this model.

Using the adoptive transfer model, we next investigated whether the autoreactive GP-specific CD4 T cells were preferentially priming and expanding GP-specific CD8 T cells, rather than causing nonspecific activation of all CD8 in the PDLN. Interestingly, activated Smarta CD4 T cells transferred into RIP-GP mice with a polyclonal TCR repertoire resulted in the activation and expansion of the endogenous GP33-specific CD8 T cells in the PDLN but not in nondraining inguinal LNs (iLNs) (Fig. 2 A and B). To confirm whether prior activation of Smarta CD4 T cells was required for these effects, we transferred naïve GP-specific CD4 T cells into RIP-GP hosts. Naïve Smarta CD4 cells were not able to induce priming and expansion of GP-specific CD8 T cells (Fig. 2C). Furthermore, activated but not naïve Smarta T cells were able to induce infiltration of GP-specific T cells into the pancreas (Fig. 2C). Therefore, activated autoreactive CD4 T cells initiate autoimmunity by activating and expanding autoreactive CD8 T cells that encounter cross-presented tissue antigens in the draining LN.

Fig. 2.

Activated but not naïve GP-specific CD4 T cells promote the localized priming and pancreatic accumulation of GP-specific CD8 T cells from a polyclonal T cell repertoire. (A) Representative CD8 versus TetGP33 staining of iLN and PDLN from RIP-GP mice at 9 or 12 days after adoptive transfer of GP-specific Smarta CD4 T cells. Dot plots are gated on CD8⁺ cells (n = 4 per group). The experiment was repeated three times. (B) Time course analysis of the percentage of TetGP33⁺ cells among CD8⁺ cells in PDLN and pancreas after adoptive transfer of Smarta CD4 T cells. Graph shows average ± SEM (n = 4 per group). The experiment was repeated three times. *, P < 0.0001. (C) Requirement of Smarta CD4 T cell activation for the priming of GP-specific CD8 T cells in PDLN of RIP-GP mice. Activated or naïve Smarta CD4 T cells were injected in RIP-GP mice, and the expansion of TetGP33⁺ cells among CD8 T cells in PDLN or the pancreas was evaluated 12 days later. Dot plots are gated on CD8+ cells (n = 4 per group). The experiment was repeated three times.

Activated Smarta CD4 Cells Induce Functional Maturation of DCs.

DC licensing has been postulated as a mechanism that allows CD4 T cells to condition DCs, which in turn are able to activate CD8 CTL (27–29). We found that activated Smarta CD4 or naïve Smarta CD4, in the presence of their cognate antigen, p13, could induce the expression of DC activation markers on bone marrow- (BM) or splenic-derived DCs after overnight coculture. This was not observed if Smarta CD4 T cells were cultured with an irrelevant OVA peptide (Fig. 3A). Supernatants from the cocultures were analyzed for cytokine levels. Importantly, only DC cocultured with activated Smarta CD4 cells and cognate antigen produced large amounts of proinflammatory cytokines (Fig. 3B). These in vitro effects correlated with the in vivo maturation of APC induced by the transfer of activated Smarta CD4 cells and p13 injected in C57BL/6 mice (Fig. 3C).

Fig. 3.

Activated Smarta CD4 T cells promote the phenotypic and functional maturation of DCs both in vitro and in vivo. (A) Modulation of several DC activation markers after an overnight coculture of naïve or activated Smarta with BMDCs. The experiment was repeated three times. (B) The levels of multiple proinflammatory cytokines were quantitated by ELISA in the coculture supernatant from the experiment described in A. (C) Activated Smarta induce APC maturation in vivo. Activated Smarta and p13 were coinjected i.v. in a C57BL/6 mouse, and 44 h later the expression of multiple activation markers was assessed on CD3⁻B220⁻CD11c^HIGH cells or CD3⁻B220⁻CD11b^HIGH cells from iLN. The same cell populations from an uninjected mouse were used as a reference for basal levels of all markers tested. The experiment was repeated three times.

Smarta CD4-Mediated in Vivo Priming of GP-Specific CD8 T Cells Requires CD28 and CD40 Expression on Host T Cells and APC.

CD4-mediated licensing of DC is thought, at least in part, to be mediated through CD40 signaling (27, 29). To investigate this requirement, we transferred activated Smarta CD4 T cells into RIP-GP hosts deficient in CD40 or CD28. In the absence of CD40 or CD28 on host cells, GP-specific CD8 cells were not primed and did not expand (Fig. 4A). To examine whether CD40 was required on host APC or the GP-specific CD8 cells, we adoptively transferred naïve CFSE-labeled P14 CD8 T cells deficient in CD40 into WT RIP-GP hosts and then 3 days later transferred activated Smarta CD4. The CD40-deficient P14 CD8 T cells expanded equally as well as WT P14 CD8 T cells, indicating that the host APC enforced the requirement for CD40 (Fig. 4B). It has been postulated that CD40 and/or TLR triggering on APC alone is sufficient to induce an immune response in the absence of CD4 help (27, 29). However, in our autoimmune model and consistent with other reports, we find that CD40 agonistic antibodies in the absence of CD4 Smarta transfer do not induce priming and expansion of endogenous GP-specific CD8 T cells in RIP-GP hosts (Fig. 4C) (17, 19). Therefore, activated self-reactive CD4⁺ T cells are able to promote a CD28-dependent proliferation of CD8⁺ autoreactive T cells via CD40-mediated interactions with BM-derived cells.

Fig. 4.

The Smarta-induced in vivo priming of GP-specific CD8 T cells depends on CD28 and CD40 triggering on host T cells and APC. (A) TetGP33 versus CD8 dot plots (gated on CD8⁺ cells) in PDLN and pancreas of WT, CD28^−/−, or CD40^−/− RIP-GP mice 9 days after adoptive transfer of activated Smarta cells. The experiment was repeated three times (n = 4 per group). (B) CD40 on responding CD8 T cells is not required for the priming of GP-specific CD8 T cells. CFSE-labeled GP-specific naïve WT or CD40^−/− P14 T cells were adoptively transferred into WT RIP-GP mice; 3 days after adoptive transfer of activated Smarta CD4 T cells, the amount of P14 T cell proliferation was evaluated by measuring the extent of CFSE dilution in the PDLN. Dot plots were gated on CFSE⁺CD8⁺Vα2⁺ cells from PDLN. The experiment was repeated three times. (C) CD40 triggering in the absence of GP-specific CD4 Smarta cells is insufficient to prime GP-specific CD8 T cells in PDLN of RIP-GP mice. RIP-GP mice were injected i.v. with 100 μg of a CD40 agonistic Ab; 9 days later, the modulation of CD44 on CD8⁺ T cells (Left), the number of lymph node cells (Center), and the percentage of TetGP33⁺ cells among CD8 cells of PDLN (Right) were measured. Dot plots and histograms are gated on CD8⁺ T cells. The experiment was repeated three times (n = 4 per group). Bar graphs show average + SEM.

Exogenous IL-7 or Endogenously Generated High IL-7 Levels Associated with Lymphopenia Are Required for Full Induction of Autoimmune Diabetes.

Despite the above observations (Figs. 2 and 4), few RIP-GP mice with a polyclonal CD8 TCR repertoire, as opposed to P14/RIP-GP mice, developed diabetes after the transfer of activated Smarta CD4 (Fig. 5A). This suggested that further steps or hits are required for the induction of diabetes. Previous work has suggested that lymphopenia and the associated availability of homeostatic cytokines contribute to the induction of autoimmunity (11, 14, 30). Cyclophosphamide (CTX) administration induces lymphopenia, and its use also has been linked to autoimmunity (12, 13). We found that CTX administration 2 days before the transfer of activated Smarta T cells induced diabetes in 100% of RIP-GP mice (Fig. 5B). Interestingly, several studies have found that tumor-adoptive immune therapy is augmented by CTX treatment (31, 32). We postulated that the most likely cytokine to be causing this effect was IL-7. Indeed, when we blocked IL-7 receptor with an IL-7Rα-blocking mAb (33), the incidence of diabetes returned to background levels (Fig. 5B). To ensure that the effects of CTX were mediated by an immune response and that CD8 T cells were contributing to the development of autoimmunity in our model, we depleted CD8 cells in host RIP-GP mice after Smarta CD4 transfer and again found that mice did not develop diabetes (Fig. 5B). It may be possible that our CD8-depleting antibodies also depleted CD8⁺DC, which may be important in our model. However, the results of the CD8-depleting experiment, together with the low rates of diabetes in Smarta/RIP-GP mice that have a skewed CD8 repertoire, indicate that CD8 T cells may be important in disease progression.

Fig. 5.

Endogenous or exogenous provision of IL-7 dramatically enhances the incidence of Smarta-induced autoimmune diabetes in RIP-GP mice. (A) Incidence of diabetes in RIP-GP mice after injection of activated Smarta (n = 21). Control PBS-treated animals without Smarta transfer did not develop diabetes (data not shown). P values = log-rank test comparing Smarta+PBS versus control animals (n = 12 per group). The experiment was repeated three times. (B) Diabetes induction after injection of activated Smarta in lymphodepleted RIP-GP mice is IL-7- and CD8-dependent. Lymphodepletion was induced by CTX treatment 2 days before the Smarta transfer. Mice were injected with 100 μg of a blocking IL-7Rα Ab (n = 5), a CD8-depleting Ab (n = 3), or an isotype control Ab (n = 13) on days 0, 2, 4, 6, 8, and 10. The experiment was repeated three times. (C) Exogenous IL-7 therapy increased the incidence of Smarta-induced autoimmune diabetes in RIP-GP mice even in the absence of lymphodepletion (n = 9). The P value for the difference between Smarta+IL-7 (C) and Smarta+PBS (A) group is shown.

If the effects of CTX and lymphodepletion were mediated via IL-7 rather than other mechanisms (13, 34), we hypothesized that the exogenous administration of IL-7 in our models, together with the transfer of Smarta CD4 T cells, should recapitulate the high rate of diabetes. Indeed, we found that recombinant human IL-7 administration, together with Smarta transfer, caused diabetes in nearly all RIP-GP mice with the same kinetics as CTX treatment (Fig. 5C). Therefore, homeostatic cytokines such as IL-7 can promote CD8⁺-dependent autoimmunity by activated CD4⁺ T cells.

Effects of IL-7 in the RIP-GP and Smarta CD4 Transfer Model of Autoimmune Diabetes.

Further studies were done to explore the mechanism of action of IL-7 in this model. The adoptive transfer of activated Smarta T cells, together with IL-7, increased the total number of CD4 and CD8 T cells and also the proportions of GP-specific CD8 cells, identified by using GP33 and GP276 tetramer staining in the pancreas of RIP-GP mice (Fig. 6 A and B). The clear accumulation of GP276-specific CD8 T cells in IL-7-treated mice indicates that this cytokine promotes the expansion of T cell clones recognizing subdominant epitopes (35).

Fig. 6.

Smarta-adoptive transfer combined with IL-7 treatment promotes the expansion of CD8 T cells specific for subdominant GP epitopes and Foxp3⁺CD4⁺ T cell accumulation in the pancreas. (A) Numbers of CD4 and CD8 T cells in PILs 9 days after activated Smarta-adoptive transfer ± IL-7 treatment. Bar graph shows average + SEM (n = 4 per group). The experiment was repeated three times. (B) Dot plots showing GP33- and GP276-specific CD8 T cells in PIL from PBS- or IL-7-treated mice 9 days after activated Smarta-adoptive transfer (n = 4 per group). The experiment was repeated three times. (C) Accumulation of Foxp3⁺ CD4⁺ T cells in the pancreas of PBS- or IL-7-treated RIP-GP mice 9 days after activated Smarta injection. Bar graphs show average + SEM (n = 4 per group). The experiment was repeated three times. A representative dot plot from each group is shown.

It has been postulated that one mechanism of action of CTX is the preferential reduction of Treg (13, 36). However, other studies have shown that, even in the genetic absence of Treg, homeostatic cytokines associated with lymphopenia promoted antitumor responses against transplanted melanoma cell lines in mouse-adoptive immune therapy models (37). Interestingly, the number of Foxp3⁺ Treg in pancreatic-infiltrating lymphocytes (PILs) in IL-7-treated RIP-GP mice receiving Smarta CD4 was higher than controls (Fig. 6C). Given the high frequency of diabetes in the IL-7 group, this finding would indicate that this cytokine is in some way interfering with suppression, a phenomenon alluded to in previous studies (38). We are currently attempting to identify the mechanism. We were surprised to see the large accumulation of Treg even in control mice and questioned whether conventional CD4 cells were converting to Treg. Using the EGFP-Foxp3 gene-targeted mice, we found that CD4⁺EGFP⁻Foxp3⁻ congenically marked cells did not convert to CD4⁺EGFP⁺Foxp3⁺ cells after Smarta CD4 transfer in our models, consistent with previous reports (SI Fig. 8) (39).

Discussion

We specifically addressed the role of CD4 cells in the initiation of autoimmunity. We showed that activated autoreactive CD4 T cells are critically required for the induction of autoimmune diabetes in our model. Several reports have suggested that CD4 T cells alone are capable of inducing autoimmune diabetes through cytokine- or Fas-mediated killing (7). We show in our model that CD8 T cells, in response to Smarta CD4, contribute to the progression of autoimmune diabetes.

We were able to dissect the mechanisms promoting diabetes induction in our model and showed that activated Smarta CD4 T cells were able to promote priming and expansion of GP-specific CD8 in the PDLN, with accumulation of these cells in the pancreas. Importantly, we showed that the generation of a GP-specific CD8 response may be secondary to CD4 licensing of DCs through CD40, contrary to other reports (18). It has been postulated that TLR or CD40 triggering independent of CD4 T cell help is sufficient for the induction of immune responses (27, 29). We show that in our autoimmune model, the provision of CD40 triggering in the absence of Smarta CD4 help was not sufficient for the induction of diabetes. It is possible that Smarta CD4 T cells also may have a role in the pancreas by directly or indirectly increasing tissue antigen-release and, hence, making it available for presentation.

The relevance of CD4 T cells in the initiation of immune responses has been undermined recently by reports that immunity can be T helper-independent, and the sole role of CD4 cells is in the maintenance of immunological memory (40–43). Our data along with others from the literature suggest that the initiation of autoimmune diseases has different requirements for CD4 T cells, compared with the initiation of particular antiviral responses. This could be due in part to the inability of self-antigens to be presented in the context of a strong proinflammatory environment, whereas viral antigens would be accompanied by potent TLR-mediated activation of the innate immune cells. In this regard, one would predict that potent APC maturation signals such as anti-CD40 or CpGs would play a key role (Figs. 1 and 4A). In support of this dichotomy, the induction of diabetes in RIP-GP mice by viral infection is CD4-independent (24).

If CD4 help is the sole requisite for the initiation of autoimmune responses, we would expect the incidence of autoimmune disease to be very high. Studies have suggested that simply providing CD4 help induces autoimmune diabetes at a high frequency (16). In our models, we find that transfer of activated CD4 Smarta is critical, but not sufficient, for the induction of diabetes. Many inhibitory pathways, at both the cellular and molecular levels, are likely to stop CD4 and CD8 autoreactivity (1). We postulated that factors that permit homeostatic proliferation of T cells may in aberrant circumstances promote CD4 and CD8 autoreactive responses by releasing them from inhibitory networks (1). We found that the combination of Smarta CD4 help and lymphodepletion caused a dramatic increase in the incidence of diabetes in our model. We postulated that the most likely cytokine mediating this enhancement would be the homeostatic cytokine IL-7. Indeed, in our models, blocking IL-7 signaling abrogated the effects of lymphodepletion, and the rates of diabetes fell dramatically. This suggests that autoimmunity is a multistep process requiring CD4 help in the setting of proinflammatory factors, perhaps associated with infection and release of endogenous self-antigens, combined with a cytokine milieu that promotes expansion and perhaps overcomes Treg and other forms of inhibition. The provision of cytokines like IL-7, perhaps secondary to transient lymphopenia induced by infection or other mechanisms, would then promote overt autoimmunity. Recent studies have shown that, in murine models, IL-7 is critical for the development and maintenance of chronic colitis (9). We show that IL-7 expands both autoreactive CD4 and CD8 T cells and even facilitates expansion of CD8 T cells specific for subdominant epitopes despite the presence of Treg. We are currently attempting to identify the IL-7-mediated mechanisms that permit activation and expansion of T cells in the face of what appear to be overwhelming inhibitory networks.

Our studies have important implications for the induction of autoimmunity and provide insights into potential therapeutic interventions. We are currently investigating whether cytokines like IL-7 are simply required for the initiation of autoimmunity or whether they also play a role in the maintenance of disease, which would have major therapeutic implications. Biological agents targeting TNFα and IL-1 have had a tremendous impact on the management of autoimmune diseases (44). Studies are now addressing the usefulness of biologicals that target IL-6 (44). Strangely, no efforts have been made to target IL-7. Our studies strongly support a role for IL-7 in the induction and perhaps maintenance of at least some autoimmune diseases.

Methods

Mice and Viral Infections.

P14 (26), Smarta (25), or RIP-GP (24) mice were interbred to obtain the indicated combinations of transgenes. See SI Methods for details of transgenic TCRs and other mice. An i.v. dose of 3,000 pfu LCMV Arm was used to infect RIP-GP mice.

T Cell Separation, Adoptive Transfers, CFSE Labeling, and CD40 Treatment.

For the adoptive transfer of naïve P14 CD8 T cells or naïve Smarta CD4 T cells, CD8⁺ or CD4⁺ T cells were purified by negative magnetic separation (Miltenyi Biotec). CD8 T cells were labeled with CFSE as described (45) by using 5 μM CFSE (Molecular Probes). RIP-GP mice were injected i.v. in the tail vein with 4 × 10⁶ cells, and the extent of T cell proliferation in the LN was assessed 3 days later. Where indicated, 100 μg of agonistic anti-CD40 Ab (FGK45; purified rat anti-mouse) was injected i.v. into RIP-GP animals.

In Vitro Activation and Expansion of Smarta CD4 T Cells.

Smarta CD4 T cells were activated in vitro in the presence of p13 peptide, IL-2, and APC as detailed in SI Methods.

Lymphodepletion, Antibody, ODN, and IL-7 Treatments.

Where indicated, RIP-GP mice were partially lymphodepleted by injection of 200 mg/kg CTX (Sigma–Aldrich) 2 days before adoptive transfer of in vitro-activated Smarta. Blocking and depleting antibodies are detailed in SI Methods. Where indicated, 50 μg of the TLR9 ODN agonist or control were injected i.v. into P14/RIP-GP mice (see SI Methods for details). Recombinant human IL-7 was provided by the Cytheris Corporation. Mice were injected s.c. with 10 μg of IL-7 every second day for 2 weeks.

Isolation of PILs and Immunohistochemistry.

PILs were isolated as described (45).

Flow-Cytometry Analysis and ELISA.

A detailed description of mAb and tetramers is given in SI Methods. Cells were stained with tetramers for 1 h at 4°C. All flow-cytometry data were acquired on a FACSCalibur and analyzed by using FlowJo. Cytokine levels in supernatant from Smarta/BMDC cocultures were measured by using ELISA kits from BD Pharmingen and e-Bioscience according to the manufacturer's recommendations.

BM-Derived DC and Smarta Cocultures.

BM cells harvested from the femur and tibia of mice were cultured with GM-CSF. After 8 days, DC were harvested and cultured with Smarta CD4 T cells in the presence of p13 peptide or control OVA peptide (see SI Methods for details). After 20 h, cells and supernatants were harvested for flow-cytometry analysis or cytokine ELISA, respectively.

In Vivo p13 Injection and DC Maturation by Activated Smarta CD4 T Cells.

In some experiments, 10⁷ activated Smarta and 400 μg of p13 peptide were coinjected i.v. into C57BL/6 animals, and 44 h later the phenotypic changes occurring on different subsets of antigen-presenting cells (APCs) in iLN were monitored by flow cytometry.

Statistical Analysis.

An unpaired t test was used to generate P values by using Prism (GraphPad). Tests were two-tailed, with a confidence interval of 95%. Time-to-event analysis of diabetes incidence was performed by using a log-rank test and SAS version 9.1 (SAS Institute).