T Cells Primed by Leishmania major Infection Cross-React with Alloantigens and Alter the Course of Allograft Rejection

Materials and Methods

Results

In an effort to identify a model of cross-reactive priming to alloantigens, we initially treated mice of several different MHC haplotypes with a variety of immunologic stimuli, including immunization with several Ags, HEL, chicken OVA, and rabbit RTA mixed in CFA or in IFA. In addition, we infected mice with the intracellular pathogen L. major. Six weeks after the initial treatment, spleen cells from these animals or from naive controls were isolated and tested in cytokine ELISPOTs for their ability to respond to a panel of allogeneic stimulator cells. In this assay, Ag-reactive T cells produce cytokines in <24 h, a time period too short for proliferation and differentiation of the T cells in vitro (3). We have previously demonstrated that the ability to produce IFN-γ by ELISPOT in a <24-h assay is a marker of a primed T cell and directly correlates with down-regulation of L-selectin on the T cell surface (3).

Spleen cells from BALB/c (H-2^d), C3H (H-2^k), and B6 (H-2^b) mice immunized with either HEL, OVA, or RTA mixed in IFA or CFA did not respond stronger than naive controls to any of the tested alloantigenic stimulator cells (data not shown). However, spleen cells obtained from B6 mice 6 wk following infection with L. major (a time at which the animals had clinically recovered from the infection) exhibited an intriguing pattern suggesting cross-reactivity. Spleen cells from the L. major-infected animals responded by producing IFN-γ at a significantly higher frequency to P/J (H-2^p) stimulator cells compared with spleen cells from naive mice (∼90/million spleen cells vs <5/million, respectively). In contrast, the spleen cells from naive mice and from L. major-infected mice responded equally, and at low frequency, to SJL (H-2^s) stimulator cells (Fig. 1⇓, <20 per million IFN-γ producers for each group), and at a similar low frequency to a variety of other stimulator cells (H-2^{d, k, q, r} haplotypes, data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Spleen cells from B6 mice infected with L. major respond to P/J alloantigens. Pooled spleen cells were isolated from naive B6 mice (□) or from B6 mice 8 wk following infection with L. major (▪) and tested in IFN-γ ELISPOT assays for responses to a panel of allogeneic splenic stimulator cells (mitomycin C treated). Visible inflammation and foot swelling had resolved within 4 wk of the initial L. major infection. No IL-4 or IL-5 was detected over background (data not shown). The depicted results represent the mean values of duplicate wells (<10% variability between wells) as counted by the computer-assisted image analyzer. The difference between the anti-P/J immune response in naive vs L. major-primed mice reached statistical significance (p < 0.05). The experiment was repeated twice with similar results.

To more fully characterize the induced immune responses, we purified CD3⁺ T cells (by negative selection) from spleens of naive mice, L. major-infected mice, or B6 mice that previously rejected a P/J skin graft and tested them for reactivity to allogeneic stimulators and Leishmania-derived, or control, Ags (Fig. 2⇓). The T cells isolated from naive mice produced predominantly IL-2 in response to P/J and SJL stimulators (at a frequency of ∼30–40 per 50,000 T cells) with few IFN-γ, IL-4, or IL-5 producers (<10 per 50,000). This low frequency response is consistent with our previous studies demonstrating that naive animals have largely naive T cell repertoires that do not contain primed IFN-γ-producing T cells specific for alloantigens (3). As anticipated for a naive T cell repertoire, T cells isolated from naive B6 mice did not respond to SLA or to a control RTA (Fig. 2⇓B). In contrast, T cells obtained from animals that resolved L. major infections responded vigorously to SLA (∼250 IFN-γ ELISPOTs per 50,000 T cells) showing that specific anti-L. major immunity was induced by the infection. Consistent with the spleen cell results shown in Fig. 1⇑, the T cells from L. major-primed mice also produced IFN-γ in response to P/J stimulator cells (∼40–70 IFN-γ ELISPOTs per 50,000 T cells), confirming cross-reactive T cell priming. This cross-reactive immunity was type 1 in character, as no IL-4 or IL-5 was produced in response to P/J alloantigens (Fig. 2⇓A). Importantly, T cells isolated from L. major-infected mice did not produce IFN-γ in response to SJL stimulators, showing that the effect was not simply unspecific activation of a large proportion of the T cell repertoire. As one specificity control to determine whether T cell priming to any Ag would induce a cross-reactive immune response, we placed B6 male skin grafts on B6 female recipients, and 8 wk later tested splenic T cells for responses to P/J stimulators. As shown in Fig. 2⇓B, T cells from these male Ag-primed mice did not cross-react with P/J stimulators, but did respond at high frequency to donor male stimulator cells and to both MHC I- and MHC II-restricted, HY-derived peptides. The relative specificity of the L. major:P/J cross-reactive T cell immune response was confirmed by additional experiments in which spleen cells obtained from B6 mice 2 mo following pulmonary infection with Mycobacterium bovis (Bacillus of Calmette and Guerin) responded at high frequency (>400 per 50,000 T cells) to self APCs, as well as to a panel of stimulator cells (H-2^{d, k, q, r, s, p} haplotypes, data not shown). Similar results were noted using spleen cells isolated from B6 mice 4 wk following oral infection with Salmonella typhimurium. Thus, some chronic intracellular infections, such as Mycobacterium and Salmonella, seem to nonspecifically activate a wide repertoire of alloreactive T cells, while infection with L. major seems to result in a relatively specific cross-reactive T cell alloimmune response.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

T cells from L. major-infected B6 mice respond to P/J alloantigens. A, Pooled splenic T cells were isolated by negative selection (<92% CD3⁺ by flow cytometry) from naive B6 mice and from B6 mice 8 wk following infection with L. major, and tested in recall IFN-γ, IL-2, IL-4, and IL-5 ELISPOT assays. B, In a second set of specificity control experiments, purified splenic T cells isolated from naive and L. major-infected B6 mice as well as from B6 mice 3 wk following rejection of P/J skin grafts (∼5 wk after graft placement) and female B6 mice 4 wk following rejection of male B6 skin grafts (∼8 wk after graft placement) and tested in recall IFN-γ ELISPOT assays. P/J or third party SJL spleen cells were treated with mitomycin C and used as stimulator cells. Responses to SLA, RTA, and the male Ag-derived peptide determinants HY1pb (MHC I-restricted) and HY2pb (MHC II-restricted) were performed in the presence of mitomycin C-treated syngeneic B6 spleen cells isolated from naive animals. The depicted results represent the mean values of duplicate wells (<10% variability between wells) as counted by the computer-assisted image analyzer. The results are representative of three independently performed experiments. ∗, p < 0.05 vs naive controls.

We next purified CD4⁺ and CD8⁺ T cells from L. major-infected animals (3–4 mo after the initial infection) and tested these T cell subsets for reactivity to P/J and SJL Ags (Fig. 3⇓). CD4⁺ T cells from L. major-primed mice, but not from naive B6 mice, responded specifically to P/J (but not SJL) stimulator cells (∼200 per 100,000 CD4 T cells) by producing IFN-γ. No IL-4 or IL-5 was detected (data not shown). Consistent with the development of activated or memory T cells as suggested by IFN-γ ELISPOT, the percentage of CD4⁺ T cells from L. major-infected mice expressing cell surface markers associated with memory phenotype (either CD44^high or CD62 L-selectin (CD62L)^low) increased 2- to 3-fold compared with naive mice (Table I⇓). CD8⁺ T cells isolated from naive or L. major-primed mice did not respond strongly to either P/J or SJL stimulators (<10 per million IFN-γ ELISPOT), and L. major infection had only minimal effects on altering the cell surface expression of memory cell markers in this population of cells (Table I⇓). The data demonstrate that the induced cross-reactivity was found within the CD4⁺ T cell population, consistent with the fact that CD4⁺ T cells are the dominant immune mediators in response to L. major in this mouse strain (35, 36).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

CD4⁺ but not CD8⁺ T cells from L. major-infected B6 mice cross-react with P/J alloantigens. Pooled splenic CD4⁺ or CD8⁺ T cells were isolated by negative selection (>90% CD4⁺ or CD4⁺ by flow cytometry, respectively) from naive B6 mice or B6 mice 12 wk following infection with L. major tested in recall IFN-γ ELISPOT assays. P/J or third party SJL spleen cells were treated with mitomycin C and used as stimulator cells. Responses to SLA and RTA were performed in the presence of mitomycin C-treated syngeneic B6 spleen cells isolated from naive animals. The depicted results represent the mean values of duplicate wells (<10% variability between wells) as counted by the computer-assisted image analyzer. The results depict one of two independently performed experiments with similar results. The differences between the anti-P/J and the anti-SLA CD4 immune responses in naive vs L. major-primed mice reached statistical significance (p < 0.05).

T cells from B6 mice that rejected P/J skin grafts responded strongly to P/J stimulator cells (∼100 IFN-γ ELISPOTs per 50,000 T cells, Fig. 2⇑). However, these CD4⁺ T cells did not react with SLA, suggesting that the anti-P/J cross-reactivity primed by L. major infection was directed toward Leishmania Ags not found in the SLA preparation (which only contains a portion of the total antigenic repertoire of the L. major organism; Ref. 29).

To determine whether the induced anti-L. major-immune response also exhibited B cell cross-reactivity to P/J alloantigens, we tested serum from naive mice, L. major-primed mice, and P/J skin graft-primed B6 mice for binding to P/J (or control SJL) thymocytes by flow cytometry. No anti-P/J Abs were detected in naive mice or in mice that cleared L. major infections (n = 6, data not shown), suggesting that the anti-P/J cross-reactivity induced by L. major infection was restricted to T cells and did not involve the humoral immune response (as a positive control, anti-P/J-specific alloantibodies were detected in B6 mice that rejected P/J skin grafts).

To determine the in vivo relevance of the detected cross-reactivity, we next placed P/J or control SJL skin grafts on L. major-infected B6 mice. We reasoned that if the L. major infection-primed T cells with cross-reactivity to P/J alloantigens, then these animals might specifically reject P/J skin grafts with accelerated kinetics compared with naive animals. As shown in Fig. 4⇓, mice previously infected with L. major rejected P/J skin grafts 2 days faster than naive mice, an acceleration that was both statistically and biologically significant (p < 0.05 by Kaplan Meier survival analysis). In addition, the accelerated rejection rate was not statistically different from second set rejection of P/J skin (median survival time of 11.5 days, n = 4, data not shown), which is considered the maximum rate of rejection in this model. In control experiments, naive mice and mice that previously cleared L. major infections both rejected third party SJL skin grafts with similar kinetics (median survival time of 14.8 days per group, Fig. 4⇓), confirming that the L. major-induced cross-reactive immune response was specific for P/J.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Accelerated rejection of P/J skin grafts in mice previously infected with L. major. Naive B6 mice (open symbols) or B6 mice infected 8–12 wk earlier with L. major (filled symbols) were transplanted with either P/J (circles) or third party control SJL (squares) skin grafts. Graft survival is depicted for each group. The difference in P/J skin graft survival between naive and L. major-primed mice reached statistical significance (p < 0.05). There was no difference in SJL skin graft survival between the two groups.

One of the goals of current transplantation immunology research is to prolong graft survival through specific immunomodulatory interventions, with the ultimate goal of inducing specific immune tolerance. Several approaches, including anti-CD40L-based costimulatory blockade with concomitant donor-specific transfusion (DST) are promising in this regard (26, 27, 37, 38). The influence of previously primed T cells (and more specifically, T cells that were primed by exposure to Ags that cross-react with allogeneic tissues) on the effectiveness of these protocols remains poorly understood. As primed T cells have lower costimulatory requirements and lower activation thresholds when compared with naive T cells, they may not be as susceptible to the tolerogenic effects of costimulatory blockade. Therefore, we next sought to determine whether P/J-reactive T cells by an L. major infection affected the ability to prolong graft survival through costimulatory blockade.

We chose to study the effects of DST plus anti-CD40L Ab MR1 treatment (DST/MR1) on graft survival using a previously published protocol in which animals are treated with DST (20 million spleen cells, i.v.) combined with repeat i.p. injections of MR1 (250 μg) on days −7, −4, 0, and +4 with respect to graft placement (26, 27, 38). Used in this manner, P/J DST/MR1 treatment of naive mice induced prolonged skin graft survival in naive mice (Fig. 5⇓, median survival time >45 days, vs 14 days in untreated mice). Despite the prolongation of graft survival, the animals were not truly tolerant to the skin grafts, as the grafts were all eventually rejected and second P/J skin grafts were not accepted (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Prior infection with L. major prevents the effects of costimulatory blockade on prolonging skin graft survival. Naive B6 mice were either engrafted with P/J skin (□) or treated with P/J DST plus MR1 and engrafted with P/J skin (○, statistically significant prolongation in graft survival vs nontreated naive mice, p < 0.05). Eight to 12 wk following infection with L. major, B6 mice were treated with P/J DST plus MR1 and engrafted with P/J skin (•). n = 4–8 per group. All of the skin grafts in the naive mice treated with DST plus MR1 eventually rejected (days 50–60) and second P/J skin grafts were rejected by these mice by day 15 demonstrating that true tolerance was not achieved in any animal.

In contrast to the findings in naive animals, the same DST/MR1 Ab treatment had minimal effect on prolonging P/J skin graft survival in mice that previously cleared L. major infections (Fig. 5⇑). All six of the P/J grafts were rejected within 15–20 days of graft placement.

In an effort to evaluate how the DST/MR1 treatment affected the induced T cell immune response in these mice following graft placement, an additional set of experiments was performed. Naive or L. major-infected B6 mice were treated with DST/MR1 and a P/J skin graft was placed as above. As positive controls, P/J skin grafts were placed onto naive B6 mice without any treatment. All animals were sacrificed on day 13 following placement of the skin graft. At this time point, the P/J skin grafts placed on naive, untreated recipients were fully rejected, the grafts placed on the L. major-infected mice treated with DST/MR1 were undergoing acute rejection (∼50% necrotic by visual inspection), and the grafts placed on uninfected recipients treated with DST/MR1 did not demonstrate any visual evidence of rejection. Immune cells were isolated from animals in each group and were tested in recall cytokine ELISPOT assays (Fig. 6⇓). Immune cells from uninfected, untreated recipients that rejected P/J skin grafts contained a high frequency of IFN-γ and IL-2 producers specific for P/J stimulators (∼1200 IFN-γ producers per million lymph node cells), consistent with the results in Fig. 2⇑. Immune cells from uninfected mice treated with DST/MR1 associated with prolonged skin graft survival responded only weakly to donor stimulator cells at frequencies no different from those detected in naive mice (Fig. 6⇓, ∼50 per million IFN-γ producers). Furthermore, there was no evidence for type 2 immune deviation as no donor-reactive IL-4 (or IL-5, data not shown) was detected. In contrast, immune cells from the L. major-infected mice treated with DST/MR1 that were undergoing acute rejection of P/J skin responded to donor P/J stimulator cells (∼700 per million IFN-γ producers), albeit at lower frequencies than that found in untreated mice (Fig. 6⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Prior L. major infection abrogates the depressed alloimmune response induced by DST plus MR1. Pooled lymph node cells were isolated on day 13 following skin graft placement from naive B6 mice (□), naive B6 mice treated with P/J DST plus MR1 followed by a P/J skin graft (▪), and L. major-infected B6 mice treated 12 wk later with P/J DST plus MR1 followed by a P/J skin graft (▦) and tested in recall cytokine ELISPOT assays against P/J and third party SJL stimulator cells. n = 3 per group. The depicted results represent the mean values of duplicate wells (<10% variability between wells) as counted by the computer-assisted image analyzer. The results are representative of two independently performed experiments with similar results. ∗, Reached statistical significance vs the DST/MR1-treated group, p < 0.05.

Discussion

Our studies provide clear evidence that T cell immunity induced in response to an infectious agent can cross-react with alloantigens and thereby affect the course of allograft rejection. CD4⁺ T cells (but not CD8⁺ T cells) from mice that recovered from L. major infection responded to P/J alloantigens by producing IFN-γ at increased frequency in comparison to naive animals ( Figs. 1–3⇑⇑⇑). The induced T cell immunity exhibited a type 1 cytokine phenotype (IFN-γ, IL-2 without IL-4 or IL-5) consistent with what is known about the anti-L. major immune response in B6 mice (35, 36, 39) and further consistent with the dominant cytokine phenotype induced following transplantation (3, 24). Moreover, the cross-reactive immune response was confined to the cellular arm of the immune repertoire, as no alloantibodies were detected in L. major-primed mice.

The detection of cross-reactive immunity between an infectious agent and alloantigens is reminiscent of the concept of “molecular mimicry” between infectious agents and autoantigens. This phenomenon is relevant to initiation of certain autoimmune diseases both in mice and in humans (40, 41, 42). Our findings in a model of transplant rejection provide the first experimental evidence that an analogous molecular mimicry phenomenon can alter the course of transplant rejection.

The results have important implications for human recipients of organ transplants. It has been noted for many years that peripheral blood of normal human volunteers contains alloreactive T cells from within both the primed and the naive T cell populations (9, 10). Although some of these human alloreactive T cells were likely primed directly to alloantigens through exposure to blood transfusion and/or pregnancies, it has been hypothesized that infections and/or vaccinations could prime Ag-specific T cells that happen to cross-react with alloantigens. In vitro studies using human CTL clones have in fact documented cross-reactivity between immunodominant epitopes of the EBV and specific HLA molecules (43). Our data provide proof of principal that immunity to an infectious agent can cross-react with some alloantigens with relative specificity, and provide the first animal model with which to carefully study this phenomenon and its consequences on transplantation.

Although we have presented strong evidence for cross-reactivity, we do not know the specific L. major-derived determinants that cross-react with P/J alloantigens. T cells from B6 mice sensitized to P/J skin grafts did not respond to SLA, an Ag preparation produced from the promastigote stage of the L. major life cycle. The Ags found in this soluble preparation contain well-defined immunodominant Ags, such as LACK (44), and it is known that soluble Ag is comparable to irradiated whole parasite for induction of in vivo resistance or susceptibility to subsequent infection (29, 45). Nonetheless, the SLA preparation excludes insoluble promastigote Ags as well as some determinants that are unique to the tissue-phase, amastigote form of L. major (29, 45). Thus, it seems likely that the cross-reactive epitopes derive from one of these latter sources. Alternatively, it is possible that Leishmania infection triggers anti-P/J Ag T cells specificities not typically induced by P/J allograft rejection, such as a normally sequestered Ag.

Regardless of the initial antigenic stimulus, it is important to note that primed T cells exhibit immunologically relevant properties that differ from naive T cells. Recent studies have revealed that priming of naive T cells leads to differentiation into effector cells, massive expansion, and alteration of cell surface marker expression, thus enabling the primed cells to traffic into peripheral organs (16, 20). Following resolution of the immune response, there is widespread apoptosis of this Ag-reactive T cell repertoire leaving a small number of residual, specific memory cells (46, 47, 48). In comparison to naive T cells that are largely confined to the primary lymphoid organs, these memory T cells maintain their ability to circulate into peripheral organs (16, 20). In addition, compared with naive cells, memory T cells are resistant to apoptotic signals (49, 50, 51), and have lower costimulatory requirements (17, 18), lower activation thresholds (19), higher functional avidities (14), and more rapid onset of effector functions upon re-encounter of their Ag (16, 20, 21). These are key features for protection from reinfection as they permit effective immune surveillance and consequent rapid control of infectious agents previously encountered by the host.

However, for T cells that recognize alloantigens, these same features could theoretically result in deleterious effects with regard to the outcome of a transplanted organ. In support of this concept, the data presented in this manuscript show that the pretransplant presence of primed alloreactive T cells induced through a cross-reactive infection were specifically associated with accelerated rejection of a skin graft (Fig. 4⇑). Mice that recovered from L. major infections and developed cross-reactive immune responses to P/J alloantigens rejected P/J skin grafts at rates commensurate with second set rejection (Fig. 4⇑). The L. major-infected animals rejected third party SJL skin grafts at the same rate as did naive mice, consistent with the lack of detectable cross-reactivity to SJL stimulators in vitro ( Figs. 1–4⇑⇑⇑⇑). Overall, the data demonstrate that the induced anti-P/J immune response following L. major infection was not simply an in vitro observation, but had specific, clinically significant in vivo consequences. It should be noted that live L. major parasites can be isolated from B6 mice for many months after they have clinically resolved their infections (52), thus making it possible that both effector and/or memory CD4⁺ T cells contributed to the L. major-specific (and cross-reactive P/J-specific) responses detected in our experiments.

As the T cell repertoire of previously nontransplanted humans often contains alloreactive memory T cells (9, 11, 23), some of which were presumably primed through cross-reactivity to infectious stimuli, the findings from our studies may have important implications for transplant patients. Indeed, others and we have provided evidence that the frequency of donor-specific memory T cells pretransplant may predict the posttransplant risk of acute rejection episodes (9, 11, 23).

Our data additionally highlight a heretofore under-appreciated observation, that an immune manipulation effective in prolonging allograft survival in naive mice (DST/MR1) is largely ineffective in animals with preexisting alloreactive T cells primed through a cross-reactive infectious stimulus. Although DST plus anti-CD40L mAb significantly prolonged P/J skin graft survival in naive animals, the same therapy had essentially no effect in mice that clinically controlled an L. major infection (Fig. 5⇑). Moreover, previous L. major infection prevented donor-specific T cell hyporesponsiveness induced by DST/MR1 treatment (Fig. 6⇑). It should be noted that recently published studies by Rossini and colleagues (53, 54), in which acute LCMV infection was shown to abrogate the effects of DST/MR1, addressed a related but clearly different question. Although certainly quite relevant to the transplant patient who develops an acute viral infection, such data do not address the question evaluated by our studies: how does the presence of primed T cells that cross-react with alloantigens affect the course of graft rejection?

CD40 binding to CD40L is known to be an important costimulatory signal for activation of naive T cells, and Ag recognition by naive T cells in the absence of such costimulation can actively induce a tolerant state (26, 27, 55, 56). However, in contrast to naive T cells, the lower costimulatory requirements of effector or memory T cells can allow activation in the total absence of costimulation (even in the absence of CD4 or CD8 interactions), particularly if the TCR:MHC:peptide interaction is of high affinity. In addition, the up-regulated expression of antiapoptotic genes such as bcl-2 renders memory/effector cells resistant to activation-induced cell death (51, 57). Therefore, it might be anticipated that costimulatory blockade would not have the same effects on primed vs naive T cells—a hypothesis supported by the present data. It remains to be determined whether higher doses of costimulatory blockade and/or more prolonged treatment would be effective in mice with primed T cell immune responses in this system. Alternate protocols of costimulatory blockade have indeed been effective in selected animals that were presensitized to alloantigens directly before graft placement (58). Nonetheless, our data strongly suggest that cross-reactive T cell immunity is sufficient to render standard DST/MR1 therapy ineffective—a finding that in itself has important implications for human transplantation.

It is further interesting to note that preliminary work in nonhuman primates suggests that some animals are relatively resistant to the effects of costimulatory blockade, particularly in the absence of concomitant immunosuppression (59, 60). Although there are many differences between rodent models and primate or human transplant recipients, one speculative explanation for the difference in efficacy is that the immune repertoires of outbred primates maintained under standard living conditions (but not naive laboratory rodents) may contain donor-reactive memory T cells that are resistant to costimulatory blockade.

In conclusion, these studies definitively demonstrate that cross-reactive priming of alloreactive T cells can occur in vivo and furthermore provide evidence that such primed T cells can have a significant impact on the outcome of a subsequent transplant. The results have important implications for human transplant recipients whose immune repertoires may contain cross-reactively primed allospecific T cells.