By continuing to browse this site you agree to us using cookies as described in About Cookies
Wiley Online Library is migrating to a new platform powered by Atypon, the leading provider of scholarly publishing platforms. The new Wiley Online Library will be migrated over the weekend of March 17 & 18. You should not experience any issues or loss of access during this time. For more information, please visit our migration page: http://www.wileyactual.com/WOLMigration/
Alloreactive T-cell memory is present in every transplant recipient and endangers graft survival. Even in the absence of known sensitizing exposures, heterologous immunity and homeostatic T-cell proliferation generate ‘endogenous' memory T cells with donor-reactivity. We have recently shown that endogenous donor-reactive CD8 memory T cells infiltrate murine cardiac allografts within hours of reperfusion and amplify early posttransplant inflammation by producing IFN-γ. Here, we have tested the role of ICOS co-stimulation in eliciting effector function from these memory T cells. ICOS is not expressed on the cell surface of circulating CD8 memory T cells but is rapidly upregulated during cell division within the allograft parenchyma. Donor-reactive CD8 memory T-cell infiltration, proliferation and ICOS expression are regulated by donor class I MHC molecule expression. ICOS blockade significantly reduced IFN-γ production and other proinflammatory functions of the activated CD8 memory T cells. Our data demonstrate that this induction of ICOS expression within peripheral tissues is an important feature of CD8 memory T-cell activation and identify ICOS as a specific target for neutralizing proinflammatory functions of endogenous CD8 memory T cells.
B7-related protein 1 (ICOS-L, LICOS, B7h, GL50, B7-H2, CD275)
cytotoxic T lymphocyte antigen 4 immunoglobulin
inducible costimulatory molecule (CD278)
Donor-reactive T cells mediate allograft rejection through multiple mechanisms. Recent studies of early posttransplant events have revealed a novel mechanism of allograft injury wherein graft infiltrating neutrophils and donor-reactive CD8 memory T cells synergize to increase intragraft inflammation that promotes the subsequent recruitment of donor antigen- primed effector T cells to the allograft (1–3). Inhibiting the function of these CD8 memory T cells in human transplant recipients is predicted to reduce early inflammation and improve long-term outcomes.
Because the immunosuppression currently used lacks efficacy and causes morbidity, efforts are underway to develop immunosuppression targeted to donor-reactive memory and effector T cells. The costimulatory molecules that modulate the signals delivered through the TCR/CD3 complex are attractive therapeutic targets (4). The CD28 homologue ICOS (inducible costimulatory molecule) is a costimulatory molecule that has been reported to influence T-cell activation, differentiation, trafficking and expression of effector functions including cytokine production and T-cell help for humoral immune responses (5–11). ICOS is not expressed on naïve T cells but is upregulated during activation and is constitutively expressed on some memory T-cell subsets. The ICOS ligand B7RP-1 (B7-related protein 1) is expressed by many nonhematopoietic cells within peripheral tissues and is upregulated by inflammatory cytokines (12,13). These patterns of ICOS and B7RP-1 expression suggest that ICOS costimulation may regulate the elicitation phase of effector and memory T-cell responses.
The effect of ICOS blockade has been tested in many transplant models. Graft biopsies from nonhuman primates as well as from human recipients show prominent ICOS expression on graft-infiltrating lymphocytes and upregulated expression of B7RP-1 during acute rejection episodes (14–16). Disruption of ICOS/B7RP-1 interactions modestly prolongs allograft survival in full MHC-mismatched rodent models of heart, liver and islet transplantation, and ICOS blockade combined with α-CD40L mAb, CTLA4-Ig, cyclosporine or rapamycin can promote long-term allograft survival (17–24). Combining ICOS- and CD40L-directed therapies has been particularly effective in reducing the vascular lesions associated with chronic graft injury (17,19,25). Collectively, these data suggest that ICOS is a potential therapeutic target in transplantation. Despite this potential, the effects of ICOS blockade are complicated. ICOS monotherapy has failed to prolong allograft survival in several models and accelerated allograft rejection in two unrelated studies of murine kidney and heart transplantation (26–29). Early administration of ICOS monotherapy had no effect on graft survival, while delayed administration significantly prolonged allograft survival in a murine heart model (28). Last, ICOS blockade reduced antigen presenting cell expression of CD86 required for the beneficial effects of CTLA4-Ig (26).
Understanding the function of ICOS on individual alloreactive T-cell populations will aid in assessing the risks and benefits of ICOS-based therapy in transplantation and in choosing appropriate adjunctive reagents. We have tested the hypothesis that ICOS costimulation is important for the expression of effector function by endogenous donor-reactive memory T cells that infiltrate allografts at early times posttransplant. We define donor-reactive endogenous CD8 memory T cells as the population of donor-reactive memory phenotype CD8 T cells constitutively present in recipients that have had no prior exposure to donor antigen. We demonstrate that the early infiltration of endogenous donor-reactive CD8 memory T cells into cardiac allografts is tightly regulated by TCR specificity and that CD8 memory T cells proliferate within transplanted tissues and upregulate ICOS during cell division. ICOS blockade does not inhibit the recruitment or proliferation of these CD8 memory T cells, but does inhibit expression of effector function. The findings indicate that the pattern of ICOS expression and consequences of ICOS ligation on CD8 memory T cells are distinct from those for primary effector T cells or memory CD4 T cells. These results identify ICOS as a target for neutralizing endogenous CD8 memory T cells early posttransplant in situations where newly activated T cells participating in the primary immune response are inhibited by other means.
Materials and Methods
The following inbred mice were used: C57BL/6 (H-2b) and A/J (H-2a) (Charles River Laboratories, Wilmington, MA); CD90.1-BALB/c (H-2d) and CD90.2-BALB/c-H-2dm2 (H-2d) (The Jackson Laboratory, Bar Harbor, ME); B6.CD8−/−, B6.Rag1−/− and B6.2C-TCR-Tg (colonies maintained at our facility). In all experiments 8–12-week-old male mice were used. All animal procedures were approved by the Cleveland Clinic IACUC.
Cardiac transplantation and recovery
Heterotopic intraabdominal cardiac transplantation was performed with modifications of the method of Corry and coworkers (30). Retransplantation was accomplished by joining the donor aorta and pulmonary artery with the abdominal great vessels of a new recipient after transecting these donor vessels between the original anastomoses and the heterotopic graft. Total operative times averaged 45 min and hearts resumed spontaneous contraction immediately upon reperfusion. Prior to retransplantation or during recovery, the heterotopic graft was thoroughly perfused with saline to remove blood from the circulation. Recovered grafts were immediately snap-frozen in liquid nitrogen or placed in media for digestion and subsequent analysis of graft-infiltrating cell populations.
Bone marrow transplantation
Following an established protocol, CD90.1-BALB/c and CD90.2-BALB/c-H-2dm2 mice received two 3.5 Gy doses of γ-irradiation with a 4-h interval between the doses (31). Twenty-four hours later 1 × 107 bone marrow cells were given intravenously. CD90.1 and CD90.2 expression was measured on peripheral blood leukocytes 8 weeks after bone marrow reconstitution to assess chimerism, which was ≥85% in all animals used.
Memory T-cell priming, purification and transfer
To generate donor-reactive polyclonal memory T-cell populations, groups of mice received full thickness trunk skin allografts. Recipient spleen cell suspensions prepared 8–10 weeks later were passed through CD3 negative selection columns (R&D Systems, Minneapolis, MN), and CD44lo and CD44hi CD8 T-cell populations were then flow sorted to greater than 98% purity using a FACSAria (BD Biosciences, San Jose, CA). Purified cell populations were transferred intravenously to syngeneic B6.CD8−/− recipients. To generate 2C-TCR-Tg memory cells, flow-sort purified 2C cells were adoptively transferred into Rag1−/− recipients following an established protocol (32).
5 × 105 flow-sort purified CD44lo or CD44hi CD8 T cells were cultured for 3 days in complete RPMI (Sigma, St. Louis, MO) in 48-well culture dishes coated with 3 μg/mL control antibody, α-CD3 mAb or α-CD3 mAb and α-CD28 mAb.
RNA purification and qRT-PCR
Snap-frozen grafts were crushed, homogenized, and RNA was isolated using fibrous tissue RNA recovery kits (Qiagen, Valencia, CA). Reverse transcription and Real-Time PCR were performed using commercially available reagents, probes and a 7500 Fast Real-Time thermocycler, all from Applied Biosystems (Foster City, CA).
Flow cytometric detection of graft-infiltrating cells was performed using a modification of the method published by Afanasyev and colleagues (33). Briefly, recovered tissues were weighed prior to incubation at 37°C in RPMI with type II collagenase (Sigma, St. Louis, MO) for 1 h without the addition of proteases. After incubation, samples were passed through 40 μm filters, washed twice in RPMI, counted using a hemocytometer and stained for common phenotypic surface markers using commercially available antibodies (BD Biosciences; eBioscience, San Diego, CA). The 7E.17G9 mAb (eBioscience) was used for all ICOS staining.
Statistical testing was performed using GraphPad software (San Diego, CA). Kruskal–Wallis, Mann–Whitney U, and Spearman tests were used. Error bars reflect SEM throughout.
B7RP-1 levels rapidly increase in iso- and allografts following transplantation
Because the expression levels and tissue distribution of B7RP-1 determine the role of ICOS costimulation during the effector phase of immune responses, we first measured B7RP-1 mRNA levels in cardiac iso- and fully MHC-mismatched A/J (H-2a) allografts retrieved from C57BL/6 (H-2b) mice (Figure 1A). Relative to native hearts, B7RP-1 mRNA expression increased 3-fold in both iso- and allografts as early as 12 h posttransplant and this increase was sustained throughout the first 72 h posttransplant.
Intragraft ICOS levels increase in CD8-replete allograft recipients and correlate with IFN-γ production
We have previously shown that endogenous CD8 memory T cells infiltrate cardiac allografts and are stimulated to produce IFN-γ as early as 24 h posttransplant, which is not observed in isografts (1). To begin to test the hypothesis that ICOS/B7RP-1 costimulation is required for the proinflammatory functions expressed by these graft-infiltrating CD8 memory T cells, we measured ICOS mRNA expression levels in iso- and allografts retrieved 24, 48 and 72 h posttransplant (Figure 1B). ICOS mRNA levels increased 6-fold between 24 and 72 h posttransplant in allo- but not isografts. Allografts transplanted into B6.CD8−/− recipients or wild-type recipients depleted of CD8 T cells had reduced levels of ICOS mRNA 72 h posttransplant when compared to allografts transplanted into CD8-replete recipients. Furthermore, we observed a direct correlation between ICOS and IFN-γ expression in a grouped analysis of iso- and allografts retrieved between 24 and 72 h posttransplant (Figure 1C).
CD8 memory T cells upregulate ICOS as they divide within the allograft parenchyma
We next determined the levels of ICOS expression on endogenous memory T cells purified from the spleens of C57BL/6 mice. Unlike CD4 memory T cells, resting CD8 memory T cells do not constitutively express ICOS (Figure 2A). To reconcile these data with our finding of CD8 T-cell-dependent increases in intragraft ICOS mRNA (Figure 1B), the expression of ICOS on endogenous CD8+ memory T cells infiltrating the allograft was directly tested and compared to the levels of expression on CD44hi CD8 T cells in the spleen (Figure 2B). It is important to note that the graft-infiltrating CD8 cells are exclusively CD44hi (1). Whereas ICOS expression was not observed on CD8 memory T cells in the spleen, 62% of the endogenous graft-infiltrating CD8 memory T cells expressed ICOS at 72 h posttransplant.
We then tested if donor-reactive CD8 memory T cells upregulate ICOS during infiltration into the graft (Figure 2C). A test population of BALB/c (H-2d) reactive CD8+CD44hi memory T cells was purified from the spleens of C57BL/6 mice 8 weeks after these mice had rejected BALB/c skin grafts. These cells were CFSE-labeled and transferred into B6.CD8−/− mice one day prior to transplantation of BALB/c cardiac allografts. Graft-infiltrating CD8 memory T cells diluted CFSE to levels consistent with more than seven cycles of cell division and the divided cells markedly upregulated ICOS 72 h posttransplant. In contrast, transferred donor-reactive CD8 memory T cells recovered from the spleen had not divided more than once and remained ICOS negative within the first 72 h posttransplant.
We next tested the possibility that CD8 memory T-cell division and ICOS expression occur in a peripheral location (e.g. the spleen or lymph nodes) before these T cells infiltrate the allograft. BALB/c cardiac allografts were first transplanted into B6.CD8−/− recipients containing adoptively transferred BALB/c-reactive CD8 memory T cells, then were retrieved 24 h after transplantation, and finally were retransplanted into B6.CD8−/− recipients that had not received adoptively transferred memory cells (Figure 2D). Prior to retransplantation, grafts were infiltrated with undivided ICOS-negative CD8 memory T cells. When retransplanted allografts were analyzed 72 h posttransplant (48 h post retransplantation), only CFSElowICOS+ graft-infiltrating CD8 T cells were observed. These results demonstrate that CD8 memory T cells divide within allografts after crossing the endothelial barrier, and that ICOS is upregulated during this cell division.
TCR stimulation is sufficient to induce ICOS expression on endogenous CD8 memory T cells
To better understand the requirements for ICOS induction on CD8 memory T cells, we compared ICOS expression on naïve and endogenous memory CD8 T cells during stimulation in vitro. CD44lo and CD44hi CD8 T cells were flow-sort purified from the spleens of naïve wild-type C57BL/6 mice and aliquots were cultured for 72 h in wells coated with α-CD3 mAb, α-CD3 and α-CD28 mAbs or rat IgG (Figure 3). CD3 stimulation alone was sufficient to induce CD44hi memory CD8 T cells to undergo robust proliferation and upregulate ICOS. In contrast, only a minority of naïve T cells divided more than three times or expressed high levels of ICOS. Adding α-CD28 mAb to the cultures stimulated all viable naïve and memory T cells to divide and increased the intensity of ICOS staining on a per cell basis, demonstrating that additional costimulatory pathways can synergize with TCR ligation to enhance or sustain ICOS expression. These results indicate that ICOS expression on endogenous CD8 memory T cells is induced by TCR ligation and occurs coincidently with cell division.
Class I MHC expression on donor parenchymal cells activates CD8 memory T cells to upregulate ICOS
Our previous experiments demonstrated that the direct pathway of alloreactivity stimulates graft-infiltrating donor-reactive CD8 memory T cells to produce IFN-γ (1). Because TCR signaling alone is sufficient for endogenous CD8 memory T-cell activation in vitro (Figure 3), we investigated whether the activating class I MHC molecule leading to ICOS upregulation is expressed on donor hematopoietic or parenchymal cells. H-2Ld-reactive 2C TCR-Tg CD8 T cells were adoptively transferred into B6.Rag1−/− mice and allowed to divide until homeostatic equilibrium was reached (6 weeks). These mice then received: (1) syngeneic grafts; (2) Ld-positive BALB/c grafts; (3) grafts from H-2dm2 mice, a naturally occurring BALB/c mutant that has lost expression of Ld; (4) grafts from chimeric BALB/c→H-2dm2 mice that express Ld on hematopoietic cells only or (5) grafts from chimeric H-2dm2→BALB/c mice that express Ld on parenchymal cells only. CD8 T-cell infiltration and ICOS expression were measured 72 h after transplantation (Figure 4A). The 2C CD8 T cells failed to infiltrate syngeneic grafts and H-2dm2 allografts. Although Ld-expression on either hematopoietic or parenchymal cells restored CD8 T-cell infiltration into the allografts (Figure 4B), graft infiltrating cells upregulated ICOS only when donor parenchymal cells expressed Ld (Figure 4C). These results indicate that TCR specificity regulates CD8 memory T-cell infiltration into cardiac allografts and donor nonhematopoietic cells present allogeneic class I MHC molecules to activate graft-infiltrating CD8 memory T-cell expression of ICOS.
ICOS blockade reduces endogenous CD8 memory T-cell effector function
We next tested the ability of a nondepleting α-ICOS mAb to alter the effector functions expressed by endogenous donor-reactive CD8 memory T cells. C57BL/6 mice received A/J cardiac allografts and were treated with 0.5 mg doses of control IgG or α-ICOS mAb on the day before and the day after transplantation. Grafts were retrieved for analysis 72 h posttransplant. Markers associated with effector CD8 T-cell function including mRNA levels of IFN-γ, perforin, granzyme B and FasL were significantly decreased by α-ICOS mAb therapy (Figure 5A, B). Intragraft levels of IL-4 and IL-17 mRNA were also significantly reduced by ICOS blockade (Figure 5C).
ICOS blockade does not affect the infiltration or proliferation of donor-reactive CD8 memory T cells
To determine at what point ICOS blockade inhibited CD8 memory T-cell function in the allografts, we used an adoptive transfer model to test whether ICOS-blockade affected donor-reactive CD8 memory T-cell infiltration, proliferation or cytokine production. CD8 memory T cells were purified from skin allograft recipients, CFSE-labeled and adoptively transferred into CD8-deficient mice. B6.CD8−/− mice containing transferred A/J-reactive CD8 memory T cells received A/J cardiac allografts and were treated with 0.5 mg control IgG or anti-ICOS mAb on days 0 and +1 posttransplant (Figure 6). Infiltration of the adoptively transferred CD8 T cells, proliferation of the cells, and intragraft IFN-γ mRNA levels were evaluated 72 h after transplantation. Consistent with the previous experiments, IFN-γ mRNA was markedly reduced by anti-ICOS mAb treatment (Figure 6E). However, when compared to control antibody treatment, anti-ICOS mAb did not affect the infiltration or division of donor-reactive CD8 memory T cells (Figure 6B–D). These data indicate that ICOS blockade does not inhibit CD8 memory T-cell activation to infiltrate the graft and divide but does directly inhibit the costimulation required for expression of effector function by the graft-infiltrating CD8 memory T cells.
Donor-reactive memory T cells generated as a result of direct sensitization or heterologous immunity are present at varying levels in all transplant recipients and threaten the long-term survival of transplanted organs (34–38). Direct prior exposure to donor antigen accounts for severe cases of recipient sensitization, but memory T cells with cross-reactivity to donor antigens are continuously generated during protective host–defense responses and during physiologic and lymphopenia-driven homeostatic proliferation (39–41). This frequently observed cross-reactivity to allogeneic class I MHC molecules is the result of the low TCR engagement thresholds and reduced costimulatory requirements needed to activate memory T cells (42,43). Similar to the resistance of alloreactive CD4 and CD8 memory T cells to CD40L and CD28 costimulatory blockade, the endogenous CD8 memory T cells that initiate adaptive alloimmunity to MHC-mismatched heart grafts are resistant to such ‘conventional’ costimulatory blockade (1,35,36).
We have recently shown that endogenous donor-reactive CD8 memory T cells infiltrate allografts within hours of reperfusion and amplify early posttransplant inflammation by producing IFN-γ (1). Here, we have tested the hypothesis that ICOS costimulation regulates early proinflammatory functions of graft-infiltrating CD8 memory T cells. Our data confirm the hypothesized costimulatory role for ICOS and reveal novel aspects of CD8 memory T-cell activation in response to the allografts. First, TCR specificity tightly controls the early posttransplant infiltration of CD8 memory T cells into allografts despite inflammation in both iso- and allografts. Second, CD8 memory T cells divide within the graft but not in the spleen during the first 72 h posttransplant despite the presence of donor antigen-presenting cells at each site. Third, donor-derived ‘nonprofessional’ antigen presenting cells activate graft-infiltrating CD8 memory T cells to express ICOS. Fourth, ICOS costimulates donor-reactive CD8 memory T cells to produce IFN-γ.
The infiltration of 2C TCR-Tg T cells into allografts strictly required donor expression of H-2Ld. A model in which endothelial cells regulate T lymphocyte transendothelial migration via TCR-MHC binding and subsequent activation is suggested although Ld expression on donor hematopoietic-derived cells also restored CD8 T-cell infiltration into chimeric grafts. Because we cannot exclude the possibility that stem cells from the Ld-positive bone marrow innoculum matured into graft endothelial cells (44), definite conclusions about the role of endothelial cells in T lymphocyte extravasation cannot be made. In a related experiment, adoptively transferred HY-specific TCR-Tg T cells were challenged to infiltrate male or female peritoneal membranes following intraperitoneal injection of IFN-γ (45). In that study, T-cell diapedesis, but not rolling or adhesion, was dependent upon cognate TCR-MHC/peptide binding. While TCR-based selection may serve as a mechanism to recruit high-avidity T-cell clones into inflamed tissues, a vulnerability in host–defense could be exploited by pathogens that do not share antigenic material with the gatekeeping cells. Nevertheless, TCR specificity tightly regulates endogenous CD8 memory T-cell infiltration into allogeneic cardiac allografts.
Organized lymphoid structures facilitate T-cell interaction with antigen presenting cells promoting clonal expansion of the T cells. Whether or not secondary lymphoid organs are strictly required for the activation of memory T cells during secondary immune responses remains unclear (46,47). In this study, adoptively transferred donor-reactive CD8 memory T cells were recovered from both the spleen and the allograft but only the allograft-infiltrating CD8 T cells diluted CFSE within the first 72 h posttransplant. Retransplantation was then used as a tool to suggest that the CD8 memory T-cell expansion occurred within the allograft itself. It is important to note that CD8 T cells do not undergo homeostatic proliferation following adoptive transfer into CD8−/− mice (data not shown), and therefore homeostatic proliferation was not a stimulus for cell division in our experiment. Although it is possible that graft-infiltrating cells exited the graft after retransplantion, divided in the periphery and then reinfiltrated, we consider this possibility unlikely. The robust proliferative capacity of graft-infiltrating CD8 memory T cells in the graft raises important questions. Are these cells long-lived within the allograft? Are these cells fully competent to mediate rejection or chronic vascular injury when the primary immune response is suppressed?
To better understand the role of ICOS costimulation on donor-reactive CD8 memory T cells, we tested the ability of α-ICOS mAb to disrupt the infiltration, proliferation, cytokine production and/or cytotoxic function of these cells. Whereas ICOS blockade had no effect on endogenous CD8 memory T-cell infiltration into the allografts, we previously reported that ICOS blockade decreased the infiltration of primary CD8 effector T cells into allograft tissue (48). These findings are consistent with the ICOS expression observed after activation and division and the subsequent regulation of antigen-stimulated T-cell function. Naïve donor-reactive T cells upregulate ICOS during activation and cell division in the spleen and therefore B7RP-1 can influence the infiltration of newly primed effector T cells as they reach the allograft. Memory T cells, in contrast, do not express ICOS until after they are activated within the allograft and thus ICOS blockade has no effect on the infiltration of these cells into allografts.
ICOS costimulation has been reported to enhance proliferation of CD8 memory T cells cultured with suboptimal doses of anti-CD3 mAb and exogenous IL-2 (49). However, anti-ICOS mAb had no effect on the proliferation of endogenous CD8 memory T cells in vivo. Intragraft cytokine production and markers of cytotolytic function were significantly reduced in allograft recipients treated with α-ICOS mAb, indicating that ICOS costimulation is required for expression of effector functions within the allograft. We have previously demonstrated that endogenous donor-reactive CD8 memory T cells directly produce IFN-γ in MHC class I mismatched cardiac allografts (1). The cells producing IL-4 and IL-17 within allografts early posttransplant are as yet unknown. Donor-reactive CD4 memory T cells expressing ICOS may produce these cytokines, or the population of graft-infiltrating CD8 memory T cells may be heterogeneous and produce multiple cytokines at this early stage of the adaptive immune response.
The importance of the direct pathway of allorecognition and antigen presentation by donor-derived ‘ nonprofessional’ antigen presenting cells to CD8 memory T cells has been illustrated in several models and is evident in this study (50–52). Allograft parenchymal cells stimulate the proliferation of endogenous CD8 memory T cells, induce ICOS on these cells and possibly serve as B7RP-1 expressing allogeneic target cells. Endothelial cells activate allogeneic human T cells in vitro and in vivo and show high levels of B7RP-1 expression in biopsies from human cardiac allografts experiencing rejection episodes (16). Consistent with this, treatment with the proinflammatory cytokines IL-1β, TNF-α and IFN-γ increased B7RP-1 expression by murine primary endothelial cell cultures (data not shown). Immunofluorescence studies, however, showed colocalization of endothelial cell markers and B7RP-1 staining late (day 7) but not early (<72 h) posttransplant (data not shown). These data suggest a model in which endothelial cells regulate the infiltration and proliferation of donor-reactive CD8 memory T cells before these T cells infiltrate the parenchymal tissue and receive the B7RP-1 costimulation that stimulates effector function. Future experiments will determine the exact cell types that: (1) stimulate ICOS induction; (2) stimulate CD8 memory T-cell proliferation; (3) provide B7RP-1 costimulation.
The mixture of improved and exacerbated outcomes following transplantation with ICOS blockade is mirrored in other disease models (23,53–55). ICOS is expressed during naïve T-cell activation, as well as constitutively or following TCR-ligation on memory and regulatory T cells. This widespread expression and diversity of ICOS-regulated functions suggests that indiscriminate ICOS blockade may be counterproductive. Therefore, an important treatment goal is to identify both unique and overlapping functions of costimulatory molecules on subsets of alloreactive T cells so that primary and memory immune responses can be inhibited with mutually compatible strategies. This study provides further evidence that donor-reactive CD8 memory T cells are critical mediators of early allograft inflammation and identifies ICOS blockade as an effective means of controlling this form of T-cell-mediated allograft injury.
The authors thank Jennifer Powers for excellent flow sorting. The authors have no conflicting financial interests. This work was supported by NIH AI58088 (A.V.), AI40459 (R.F.), AI51620 (R.F.), and by the Roche Organ Transplant Research Foundation grant 60495086 (R.F.). A.S. was supported in part by NIH T32 GM007250, an AHA Predoctoral Fellowship, and the Case Western Reserve University Medical Scientist Training Program.