The roles of CD8 central and effector memory T cell subsets in allograft rejection

Abstract

The contribution of secondary lymphoid tissue homing central memory T cells (T_CM) and peripheral tissue homing effector memory T cells (T_EM) to allograft rejection is not known. We tested whether T_EM is the principal subset responsible for allograft rejection due to the non-lymphoid location of target antigens. Skin allograft rejection was studied after transferring either CD8 T_CM or T_EM to wild type mice and to mice that lack secondary lymphoid tissues. We found that CD8 T_CM and T_EM were equally effective at rejecting allografts in wild type hosts. However, CD8 T_EM were significantly better than T_CM at rejecting allografts in the absence of secondary lymphoid tissues. CD8 T_CM were dependent upon secondary lymphoid tissues more than T_EM for optimal differentiation into effectors that migrate into the allograft. Recall of either CD8 T_CM or T_EM led to accumulation of T_EM after allograft rejection. These findings indicate that either CD8 T_CM or T_EM mediate allograft rejection but T_EM have an advantage over T_CM in immune surveillance of peripheral tissues, including transplanted organs.

Introduction

The alloimmune response is a T-cell dependent process that culminates in the rejection of transplanted organs. The immune repertoire of adult humans contains high frequencies of alloreactive memory T cells that play an important role in allograft rejection (1, 2). The presence of donor-specific memory T cells correlates with acute and chronic rejection in humans and hinders tolerance induction in experimental animals (1, 3–7). Immunosuppressive strategies that inhibit naïve T cells are ineffective against memory T cells because of several inherent advantages over their naïve counterparts such as markedly extended lifespan, lower activation threshold, increased proliferative capacity, rapid secretion of multiple effector cytokines, and homing to both lymphoid and non-lymphoid (peripheral) tissues (8–10).

Two subsets of memory T cells have been described in mice and humans based on their homing patterns and effector function: secondary lymphoid tissue-homing CD62L^hi CCR7^hi central memory (T_CM) T cells with delayed effector function, and peripheral tissue-homing CD62L^lo CCR7^lo effector memory (T_EM) T cells with immediate effector function (11–14). The contribution of CD8 T_CM and T_EM subsets to protective immunity against bacterial and viral infections is dependent upon the type and location of pathogen. CD8 T_CM are superior to T_EM at clearing systemic lymphocytic choriomeningitis and vesicular stomatitis virus infections; CD8 T_CM are equivalent to T_EM at clearing Listeria monocytogenes infection; while CD8 T_CM are inferior to T_EM at clearing Sendai and vaccinia virus infections, in mice (12, 15–18). However, the relative importance of memory T cell subsets in allograft rejection is less clear (7, 19, 20). Understanding which memory T cell subset is the principal mediator in allograft rejection has biomedical significance because agents that inhibit lymphocyte trafficking are potential candidates for use in clinical transplantation (21).

We hypothesized that peripheral tissue-homing T_EM is the principal memory subset responsible for allograft rejection since a transplanted organ represents not only a foreign antigen, but also a complex non-lymphoid peripheral tissue harboring antigen presenting cells (APCs) and endothelial cells that can modulate memory T cell behavior. To test this hypothesis, polyclonal alloreactive memory T cells generated in wild type (wt) hosts were sorted for CD8 T_CM or T_EM and transferred into wt and splenectomized alymphoplastic (aly/aly) recipients of skin allografts. Splenectomized aly/aly (aly/aly-spleen) mice lack all secondary lymphoid tissues and are therefore suitable for testing the function of transferred CD8 T_CM and T_EM in non-lymphoid tissues (22). We report that CD8 T_CM and T_EM were equally effective at mediating allograft rejection in wt hosts that have secondary lymphoid tissues. In contrast, CD8 T_EM were significantly more effective than T_CM at rejecting skin allografts in aly/aly-spleen hosts that lack all secondary lymphoid tissues. CD8 T_EM infiltration into the allograft was independent of secondary lymphoid tissues whereas T_CM were dependent upon secondary lymphoid tissues for optimal differentiation into effectors that migrate into the allograft. These results show that either CD8 T_CM or T_EM reject allografts effectively and CD8 T_EM is the principal subpopulation responsible for immune surveillance of transplanted organs.

Materials and Methods

Mice

C57BL/6 (Thy1.2, H-2^b; hereafter wt), B6.PL-Thy1^a/Cy (Thy1.1, H-2^b; hereafter Thy1.1) and BALB/c (H-2^d) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Alymphoplasia mice (Map3k14^−/−, Thy1.2, H-2^b, hereafter aly/aly) were purchased from CLEA (Osaka, Japan) (23). All animals were maintained under SPF conditions and procedures were performed as per IACUC guidelines.

Splenectomy, skin transplantation and co-stimulation blockade

Splenectomy and partial thickness skin transplantation were performed using established techniques (24). Skin grafts were monitored daily and rejection was defined as > 90% graft necrosis. Wt recipients of skin allografts were treated after transplantation with CTLA4-Ig (0.25mg, i.p.) and anti-CD40L (MR1) (0.25mg, i.p.) (Bioexpress Inc, MA) (days 0, 2, 4 and 6). Kaplan-Meier survival analysis was used to assess differences in allograft survival and p value of < 0.05 was considered significant.

Generation, isolation, and adoptive transfer of memory T cells

Thy1.1 mice were immunized with 3 × 10⁷ BALB/c splenocytes (i.p.). 6 – 12 weeks later, spleen and lymph node (LN) cells were enriched for T cells by negative selection via MACS (Miltenyi Biotec, Auburn, CA), and sorted for CD8⁺CD44^highCD62L^high (CD8 T_CM) and CD8⁺CD44^highCD62L^low (CD8 T_EM) populations (> 95% purity) on BD FACS Aria. 1.5 – 2 × 10⁶ CD8 T_CM or 4 – 6 × 10⁵ CD8 T_EM containing similar numbers of alloreactive IFNγ⁺ T cells were adoptively transferred in each experiment.

Cell harvest and enumeration after adoptive transfer

Adoptive hosts were sacrificed either at 8 days after transplantation (effector phase) or at 6 – 10 weeks after allograft rejection (secondary memory phase). Cells from spleen, lymph nodes, blood, liver, lungs, bone marrow and skin grafts were harvested as described (24–26). Harvested cells from tissues were counted and analyzed by flow cytometry after gating on CD8⁺Thy1.1⁺ population. Antigen-specific cells were analyzed by measuring BALB/c-reactive IFNγ⁺ T cells within the harvested CD8⁺Thy1.1⁺ population by flow cytometry after intracellular cytokine staining (24). Statistical analyses was performed using unpaired Student’s t test and differences with p < 0.05 were considered significant.

Flow cytometry, intracellular cytokine staining and in vivo cytotoxicity

Fluorochrome-tagged antibodies for flow cytometry were purchased from BD Pharmingen (San Diego, CA), eBioscience (San Diego, CA) and R&D systems (Minneapolis, MN). Intracellular IFNγ producing cells were detected after ex-vivo stimulation with BALB/c splenocytes (H-2^d) for 6-hrs in the presence of Brefeldin A. In-vivo cytotoxicity of CD8 T_CM and T_EM was assessed in wt B6 mice pre-treated with anti-NK1.1 (PK136, 300µg) to deplete NK cells. 3-days later, sorted CD8 T_CM (1.9 × 10⁶) or T_EM (0.5 × 10⁶) containing similar numbers of BALB/c-reactive IFNγ⁺ T cells were transferred. Equal numbers of CFSE labeled syngeneic H-2^b (5 × 10⁶, 2µM, B6) and allogeneic H-2^d (5 × 10⁶, 0.2µM, BALB/c) splenocytes were injected. 12-hrs later, in-vivo target cell killing was measured as loss of allogeneic vs syngeneic cells in adoptive hosts of T_CM or T_EM compared to naïve controls using the formula: 100-[(%H-2^d cells in memory T cell recipients/%H-2^b cells in memory T cell recipients÷%H-2^d cells in naïve mice/%H-2^b cells in naïve mice) × 100] (27). Flow acquisition was performed on LSRII analyzers (BD Biosciences, San Diego, CA), and data analyzed using Flowjo software (Treestar, Ashland, OR).

Results

Characterization of polyclonal alloreactive CD8 memory T cell subsets

Alloreactive memory T cells were generated in wt hosts to study polyclonal CD8 T_CM and T_EM subsets. Briefly, B6 Thy1.1 mice were immunized with BALB/c splenocytes. 6 – 12 weeks later, spleen and LN cells were harvested and sorted for CD8 memory T cell subsets by gating on CD8⁺CD44^highCD62L^high cells (CD8 T_CM) and CD8⁺CD44^highCD62L^low cells (CD8 T_EM) (Fig. 1A). Phenotypic analysis showed that CD8 T_EM contained a sub-population that is CD69^high, 1B11^high, CD27^low and CD127^low, suggesting a more activated phenotype and less dependence on co-stimulation and/or cytokines than T_CM (Fig. 1B) (17, 28, 29). Ex-vivo alloreactive IFNγ production was used to detect antigen-specific cells within the polyclonal CD8 T_CM and T_EM subsets since antigen-specific MHC-tetramer⁺ cells in either subset were IFNγ⁺ upon ex-vivo recall in viral studies (12). BALB/c-reactive IFNγ⁺ T cells constituted 1.6 ± 0.3% of unfractionated CD8 memory T cells, 1.3 ± 0.3% of CD8 T_CM, and 4.4 ± 1.3% of CD8 T_EM subset (Figs. 1C–1D). Intracellular TNFα correlated with IFNγ but Granzyme B and IL-2 were not detected above background (data not shown) (12). BALB/c-reactive, IFNγ⁺ cells within CD8 T_CM and T_EM increased over time upon ex-vivo re-stimulation and were approximately 3 – 4 fold more in T_EM than T_CM subset (Fig. 1E). These findings suggest that CD8 T_EM contained 3 – 4 fold more alloreactive memory T cells than T_CM.

Fig. 1. Characterization of polyclonal alloreactive CD8 memory T cell subsets.

Thy1.1 mice were immunized with BALB/c splenocytes (3 × 10⁷, i.p.) and spleen and LN cells were harvested 6 – 12 weeks later to sort for CD8 T_CM and T_EM. (A) Sorting of CD8 memory T cell subsets. CD8 memory T cell subsets were sorted after gating on CD8⁺CD44^highCD62L^high cells (CD8 T_CM) and CD8⁺CD44^highCD62L^low cells (CD8 T_EM). Sorted cells were tested for purity prior to adoptive transfer. (B) Phenotype of CD8 T_CM and T_EM. Expression of CD69, 1B11, CD25, CD27, CD127 and CD122 on CD8 T_CM and T_EM is shown (representative FACS plots of 5 – 6 experiments). (C) Ex-vivo IFNγ production by CD8 T_CM and T_EM. Sorted CD8 T_CM and T_EM were assessed for IFNγ production by intracellular cytokine staining after 5-hr ex-vivo stimulation with BALB/c splenocytes. IFNγ production within the Thy1.1⁺ population is shown after gating on CD8⁺ T cells. Naïve T cells from unimmunized Thy1.1 mice (Naïve) and unfractionated sorted CD8⁺ CD44^high memory T cells (CD8 memory) from immunized Thy1.1 mice that contained both CD62L^high and CD62L^low cells were used as controls. Representative FACS plots of 4 – 5 experiments are shown. (D) Quantitation of alloreactive IFNγ⁺ T cells within CD8 T_CM and T_EM subsets after 5-hr ex-vivo stimulation with BALB/c splenocytes. Percent of IFNγ⁺ T cells within naïve, unfractionated CD8 memory (memory) and sorted CD8 T_CM and T_EM populations. Mean ± SD of 4 – 5 experiments. (E) Alloreactive IFNγ⁺ T cells within CD8 T_CM and T_EM subsets upon ex-vivo stimulation with BALB/c splenocytes over time. Percent of IFNγ⁺ T cells within naïve, sorted CD8 T_CM and T_EM populations after ex-vivo stimulation with BALB/c splenocytes at 6hrs, 12hrs and 24hrs. (F) In-vivo cytotoxicity of alloreactive CD8 T_CM and T_EM subsets in wt adoptive hosts. Wt mice were treated with anti-NK1.1 (PK136, 300µg, i.p.) to deplete NK cells. 3-days later, sorted CD8 T_CM (1.9 × 10⁶) or CD8 T_EM (5 × 10⁵) containing similar numbers of BALB/c-reactive IFNγ⁺ T cells were transferred to wt adoptive hosts followed by injection of equal numbers of CFSE labeled H-2^b (5 × 10⁶, 2µM B6, syngeneic) and H-2^d (5 × 10⁶, 0.2µM, BALB/c, allogeneic) splenocytes on the same day. 12-hrs later, in vivo killing of target cells was measured in harvested spleen and LN cells as loss of H-2^d target cells compared to loss of H-2^b syngeneic cells in adoptive hosts of memory T cells (T_CM or T_EM) versus naïve B6 control mice (as described in Materials and Methods) (representative FACS plots of n = 3 mice/grp are shown). (G) Quantitation of in-vivo specific lysis of allogeneic cells by alloreactive CD8 T_CM and T_EM subsets in wt adoptive hosts. Percent of BALB/c-cell lysis in harvested spleen and LNs from wt adoptive hosts of CD8 T_CM or T_EM. Mean ± SD of n = 3 mice/grp. (H) Expression of chemokine and adhesion receptors on CD8 T_CM and T_EM. Expression of CCR7, CXCR3, CCR5, CD49d (α₄-integrin), β₁-integrin, α₄β₇ and CD11a (LFA-1) on CD8 T_CM and T_EM subsets shown in comparison to CD8⁺ CD44^low naïve T cells (representative FACS plots of 3 – 4 experiments).

In-vivo cytotoxicity of CD8 T_CM and T_EM was assessed after transfer to wt adoptive hosts pre-treated with anti-NK1.1 (PK136) antibody to inhibit NK cell mediated killing of allogeneic cells (30). Since CD8 T_EM contained 3 – 4 fold more alloreactive IFNγ⁺ cells than T_CM (Fig. 1D), 3 – 4 fold more total T_CM (1.9 × 10⁶) than T_EM (0.5 × 10⁶) were transferred to wt hosts to compare similar numbers of antigen-specific cells. Spleen and LN from adoptive hosts were analyzed to assess killing since splenocyte target cells were found mainly in these tissues and not in peripheral tissues (data not shown). BALB/c-cell killing by CD8 T_EM was significantly better than T_CM in the spleen (27 ± 3% vs 15 ± 3%, p = 0.04, n = 3, Figs. 1F–1G), and comparable to T_CM in the LN (21 ± 3% vs 18 ± 7%, p = 0.1, NS, n = 3, Figs. 1F–1G). Therefore, both CD8 T_CM and T_EM killed allogeneic cells in-vivo but T_EM were better cytotoxic cells than T_CM.

Alloreactive CD8 T_CM and T_EM were assessed for the expression of chemokine and adhesion receptors that guide their migration and recruitment into inflammatory sites. As reported, CCR7 expression was higher on CD8 T_CM than T_EM cells (Fig. 1H) (12, 31). CXCR3, CCR5 and LFA-1 were upregulated on both subsets whereas VLA-4 (α₄β₁) expression was restricted to CD8 T_EM (Fig. 1H). These findings suggest that both subsets can migrate in response to chemokines such as MIG, IP-10, I-TAC, MIP-1α, MIP-1β, and RANTES produced at inflammatory sites, including transplanted organs (32, 33). However, following migration, CD8 T_EM that express VLA-4 (α₄β₁) can extravasate readily into allografts (34–38) and hence, might have an advantage over T_CM in causing allograft rejection.

Allograft rejection in adoptive hosts of CD8 T_CM and T_EM

To test which alloreactive CD8 memory T cell subset is more effective at allograft rejection, skin allograft survival was compared in adoptive hosts of CD8 T_CM and T_EM. Sorted CD8 T_CM and T_EM (Fig. 1A) were transferred into wt recipients of BALB/c-skin allografts. Since CD8 T_EM contained 3 – 4 fold more alloreactive IFNγ⁺ T cells than T_CM (Fig. 1D), 3 – 4 fold more total T_CM (1.5 × 10⁶ – 2 × 10⁶) than T_EM (0.5 × 10⁶ – 0.6 × 10⁶) were transferred in each experiment to compare similar numbers of antigen-specific cells. Wt hosts of CD8 T_CM or T_EM rejected skin allografts comparable to sensitized recipients (MST = 15, 15, and 16 days, respectively, n = 3 – 4/grp; p = 0.4, NS), and not significantly different from naïve mice (MST = 18 days, n = 4; p = 0.06 and 0.4, respectively, NS) (Fig. 2A). Since endogenous naïve T cells also contributed to observed skin allograft rejection in wt hosts, recipients were treated with CTLA4-Ig and anti-CD40L to inhibit naïve T cell activation to better assess rejection mediated by transferred CD8 T_CM and T_EM (5, 6, 39). Recall of either CD8 T_CM or T_EM subset in wt recipients treated with co-stimulation blockade led to skin allograft rejection that was comparable to sensitized recipients (MST = 14, 16 and 15 days, respectively, n = 4 – 6/grp; p = 0.06 and 0.5, respectively, NS) and significantly accelerated than naïve recipients (MST = 25 days, n = 4; p = 0.02) (Fig. 2B). Therefore, in wt hosts, both CD8 T_CM and T_EM subsets were equally effective in mediating allograft rejection.

Fig. 2. Allograft rejection in adoptive hosts of CD8 T_CM and T_EM.

Sorted CD8 T_CM (1.5 × 10⁶ – 2 × 10⁶) or T_EM (0.4 × 10⁶ – 0.6 × 10⁶) containing similar numbers of BALB/c-reactive IFNγ⁺ T cells were transferred into wt and aly/aly-spleen adoptive hosts followed by BALB/c-skin transplantation 2-days later. (A) Skin allograft rejection mediated by CD8 T_CM and T_EM in wt hosts. BALB/c-skin allograft survival was tested in wt recipients of CD8 T_CM or T_EM and compared to unimmunized naïve mice (Naïve, no cells) and sensitized mice (Sensitized, no cells) harboring endogenous memory T cells that had rejected BALB/c-skin allografts 8 – 12 weeks earlier (n = 3 – 4 mice/grp). (B) Skin allograft rejection mediated by CD8 T_CM and T_EM in wt hosts treated with co-stimulation blockade. Wt recipients of CD8 T_CM or CD8 T_EM, unimmunized naïve mice (Naïve, no cells) and sensitized mice (Sensitized, no cells) were treated with CTLA4-Ig and anti-CD40L, 0.25mg each (i.p.), on day 0, 2, 4 and 6 after BALB/c-skin transplantation and allograft survival was compared (n = 4 – 6 mice/grp). (C) Skin allograft rejection mediated by CD8 T_CM and T_EM in aly/aly-spleen hosts. Aly/aly-spleen recipients of CD8 T_CM or CD8 T_EM or unfractionated CD8 memory T cells (1 × 10⁶) (CD8 memory) and aly/aly-spleen naïve mice without memory T cell transfer (Naïve, no cells) underwent BALB/c-skin transplantation and allograft survival was compared (n = 4 – 7 mice/grp).

CD8 T_CM have a resting phenotype and do not express VLA-4, suggesting that further differentiation is required for their extravasation and function in non-lymphoid tissues (Figs. 1B and 1H). It is not clear if the preferential location of CD8 T_CM within secondary lymphoid tissues suggests dependence on these tissues for differentiation into effectors analogous to naïve T cells (22). To address this question, sorted CD8 T_CM (1.5 × 10⁶ – 2 × 10⁶) or T_EM (0.4 × 10⁶ – 0.6 × 10⁶) containing similar numbers of alloreactive IFNγ⁺ T cells were transferred to aly/aly-spleen mice and BALB/c-skin allograft rejection was tested. Skin allograft rejection in aly/aly-spleen recipients of CD8 T_CM was significantly delayed than in recipients of T_EM (MST = 57 days, n = 7 vs MST = 36 days, n = 5, respectively; p = 0.008, Fig. 2C). Allografts in aly/aly-spleen mice that did not receive memory T cells survived > 100 days compared to recipients of memory T cells (n = 4; p = 0.0002, Fig. 2C). These data show that alloreactive CD8 T_EM were significantly more effective than T_CM in causing allograft rejection in the absence of secondary lymphoid tissues. However, in adoptive hosts of CD8 T_EM, skin allograft rejection was accelerated in wt than in aly/aly-spleen recipients (MST 16 and 36 days, respectively, n = 5 – 6/grp; p = 0.003) (Figs. 2B–2C) suggesting that secondary lymphoid organs support recall of both subsets.

Recall and differentiation of CD8 T_CM and T_EM in adoptive hosts

To understand why allograft rejection was delayed in aly/aly-spleen hosts of CD8 T_CM, differentiation of CD8 T_CM and T_EM into effectors and migration into the allograft was compared in adoptive hosts. Sorted CD8 T_CM (2 × 10⁶) or T_EM (0.5 × 10⁶) containing similar numbers of alloreactive IFNγ⁺ T cells were transferred into wt and aly/aly-spleen hosts that underwent BALB/c-skin transplantation. 8 days later, cells were harvested from lymphoid and non-lymphoid tissues in adoptive hosts, and analyzed after gating on CD8⁺Thy1.1⁺ T cells. A large population of CD8 T_CM had acquired a CD49d^high CD62L^low effector phenotype in wt hosts (50 ± 10%, n = 3) compared to aly/aly-spleen hosts (15 ± 4%, n = 4) suggesting that differentiation of CD8 T_CM into effectors was impaired in the absence of secondary lymphoid tissues (Fig. 3A). CD8 T_EM retained a CD49d^high CD62L^low effector phenotype in harvested tissues of both adoptive hosts (Fig. 3A). BALB/c-reactive IFNγ⁺ T cells derived from CD8 T_CM and T_EM were found mainly in the CD49d^high population (data not shown). At 8 days after skin transplantation, BALB/c-reactive IFNγ⁺ T cells present in the harvested CD8⁺Thy1.1⁺ T cells were found exclusively within the proliferated CFSE^low population in adoptive hosts of CD8 T_CM and T_EM suggesting differentiation into alloreactive effectors (Fig. 3B). Transferred CD8 T_CM and T_EM contained CD44^hi memory phenotype T cells of irrelevant specificities that possibly underwent bystander proliferation in adoptive hosts contributing to CFSE^low cells within the CD8⁺Thy1.1⁺ population that were not IFNγ⁺ (Fig. 3B) (40–42). CFSE^low IFNγ⁺ effector T cells from CD8 T_CM in wt constituted a larger population of harvested CD8⁺Thy1.1⁺ cells than in aly/aly-spleen hosts (15 ± 5% vs 1.5 ± 0.5%, n = 3, respectively) compared to CFSE^low IFNγ⁺ effectors from T_EM (6 ± 3% vs 3 ± 2%, n = 3, respectively) (Fig. 3B). The greater proliferative advantage of alloreactive CD8 T_CM in wt hosts contributed to 100-fold more alloreactive IFNγ⁺ effectors from T_CM compared to a 4-fold increase in alloreactive effectors from T_EM in wt than in aly/aly-spleen hosts (p = 0.008 and p = 0.06, NS, respectively, n = 3 – 4, Figs. 3C–3D). Alloreactive effector T cells from CD8 T_CM compared to T_EM were 13-fold more in wt hosts (p = 0.03, n = 3 – 4) and 3-fold less in aly/aly-spleen hosts (p = 0.07, NS, n = 3 – 4) at 8 days (Figs. 3C–3D). Fewer alloreactive IFNγ⁺ effector T cells from CD8 T_CM in aly/aly-spleen hosts was not due to impaired survival since T_CM were recovered from naïve adoptive hosts in comparable numbers (Fig. 3E) (24), and ratio of recovered T_CM to T_EM (3.5:1) in allograft recipients was similar to cells transferred (Figs. 3F–3G). At 30-days after transplantation, more CD8 T_CM in aly/aly-spleen hosts had acquired a CD49d^high CD62L^low effector phenotype and differentiated into 3-fold more alloreactive IFNγ⁺ effector T cells than T_EM (Figs. 3H–3I). These findings suggested that differentiation of CD8 T_CM into effectors was significantly delayed in the absence of secondary lymphoid tissues.

Fig. 3. Differentiation of CD8 T_CM and T_EM in adoptive hosts.

Sorted CD8 T_CM (2 × 10⁶) or T_EM (0.5 × 10⁶) containing similar numbers of BALB/c-reactive IFNγ⁺ T cells were transferred to wt and aly/aly-spleen mice followed by BALB/c-skin transplantation 2-days later. CD8 T_CM and T_EM were labeled with 2µM CFSE prior to adoptive transfer to assess proliferation in adoptive hosts at 8 days after BALB/c-skin transplantation. Differentiation and proliferation of CD8 T_CM and T_EM was assessed after skin transplantation. Cells were harvested from liver (LV), lungs (LG), bone marrow (BM), blood and skin allograft (SK) in both wt and aly/aly-spleen hosts. Spleen (SP), draining lymph node (dLN) and non-draining LN cells were also harvested in wt hosts. (A) Phenotype of activated CD8 T_CM and T_EM in wt and aly/aly-spleen hosts at 8-days. Expression of CD49d and CD62L is shown after gating on CD8⁺Thy1.1⁺ population within the harvested cells from spleen and lungs in wt mice, and lungs in aly/aly-spleen mice. Harvested cells from LN, BM and liver tissues were similar in phenotype (data not shown). Representative FACS plots of 3 – 4 experiments are shown. (B) Proliferation of alloreactive CD8 T_CM and T_EM after recall in adoptive hosts at 8-days. BALB/c-reactive IFNγ⁺ T cells within the harvested cells from all tissues were assessed by intracellular cytokine staining and analyzed after gating on CD8⁺Thy1.1⁺ population. CFSE dilution and IFNγ are shown after gating on CD8⁺Thy1.1⁺ T cells. Representative FACS plots of 3 experiments are shown. (C – D) Quantitation of BALB/c-reactive IFNγ⁺ T cells derived from CD8 T_CM (C) and CD8 T_EM (D) in wt and aly/aly-spleen adoptive hosts at 8-days after BALB/c-skin transplantation. BALB/c-reactive IFNγ⁺ T cells within the harvested cells from all tissues were assessed by intracellular cytokine staining and enumerated after gating on CD8⁺Thy1.1⁺ population. Mean ± SD of 3 – 4 mice/grp. (E – G) Quantitation of CD8⁺Thy1.1⁺ cells from harvested tissues in adoptive hosts of CD8 T_CM at 8 days after transfer (E) or at 8 days after transplantation in adoptive hosts of CD8 T_CM and T_EM (F – G) (Mean ± SD of 3 – 4 mice/grp). (H) Phenotype of activated CD8 T_CM and T_EM in aly/aly-spleen hosts at 30-days after BALB/c-skin transplantation (similar to 3A). (I) Quantitation of BALB/c-reactive IFNγ⁺ T cells derived from CD8 T_CM and T_EM in aly/aly-spleen adoptive hosts at 30-days after BALB/c-skin transplantation (similar to 3C-D).

Migration of alloreactive effectors derived from CD8 T_CM and T_EM and their infiltration into skin allografts was compared in adoptive hosts. Alloreactive IFNγ⁺ effector T cells from either CD8 T_CM or T_EM migrated broadly to non-lymphoid tissues and infiltrated into skin allografts to a comparable extent in wt hosts (Fig. 4A; p = 0.08, NS, n = 3 – 4, Fig. 4C). Despite greater numbers of total alloreactive effectors from CD8 T_CM than T_EM in wt hosts (Figs. 3B – 3D), alloreactive effectors from T_EM were comparable to T_CM in skin allografts suggesting that early migration and accumulation of T_EM in the allograft contributed to equally effective allograft rejection as T_CM (Fig. 2B). In aly/aly-spleen allograft recipients, alloreactive IFNγ⁺ effector T cells from CD8 T_CM were found mainly in lungs and liver but not in skin allograft, whereas alloreactive effectors derived from T_EM were found in lungs, bone marrow and skin allograft at 8 days (Fig. 4B; p = 0.01, n = 3 – 4, Fig. 4D). Alloreactive IFNγ⁺ effector T cells from both subsets were found in blood, lung, liver, skin allograft (Fig. 4E) and more alloreactive effectors from T_EM than T_CM had infiltrated skin allografts (Fig. 4E–4F) despite fewer total alloreactive IFNγ⁺ effectors derived from T_EM (Fig. 3I) in aly/aly-spleen recipients at 30-days. Thus, CD8 T_EM infiltrated skin allografts early after transplantation whereas differentiation of T_CM into alloreactive effectors that infiltrate the allograft was delayed in the absence of secondary lymphoid tissues contributing to significantly delayed allograft rejection.

Fig. 4. Migration of alloreactive effectors in adoptive hosts of CD8 T_CM and T_EM.

Sorted CD8 T_CM (2 × 10⁶) or T_EM (0.5 × 10⁶) containing similar numbers of BALB/c-reactive IFNγ⁺ T cells were transferred to wt and aly/aly-spleen mice followed by BALB/c-skin transplantation 2-days later. Alloreactive effectors from CD8 T_CM and T_EM were assessed after BALB/c-skin transplantation. Cells were harvested from liver (LV), lungs (LG), bone marrow (BM), blood and skin allograft (SK) in both wt and aly/aly-spleen hosts. Spleen (SP), draining lymph node (dLN) and non-draining LN cells were also harvested in wt hosts. BALB/c-reactive IFNγ⁺ T cells within the harvested cells from all tissues were assessed by intracellular cytokine staining after gating on CD8⁺Thy1.1⁺ population to identify alloreactive effectors derived from CD8 T_CM and T_EM. (A – B) Tissue distribution of alloreactive effector T cells generated from transferred CD8 T_CM and T_EM in wt (A) and aly/aly-spleen (B) hosts. CD8⁺Thy1.1⁺ IFNγ⁺ T cells harvested from each tissue is shown as % of total CD8⁺Thy1.1⁺ IFNγ⁺ T cells harvested from all tissues in that recipient (Mean ± SD of 3 – 4 mice/grp). CD8⁺ Thy1.1⁺ IFNγ⁺ T cells harvested from blood and non-draining LN were < 1% of total and are not shown. (C – D) Quantitation of BALB/c-reactive IFNγ⁺ T cells within the harvested CD8⁺ Thy1.1⁺ population from skin allografts in either wt (C) or aly/aly-spleen (D) adoptive hosts of CD8 T_CM and T_EM (Mean ± SD of 3 – 4 mice/grp). (E) Tissue distribution of alloreactive effector T cells generated from transferred CD8 T_CM and T_EM in aly/aly-spleen hosts at 30-days after BALB/c-skin transplantation (similar to 4B). (F) Quantitation of BALB/c-reactive IFNγ⁺ T cells within the harvested CD8⁺ Thy1.1⁺ population from skin allografts in aly/aly-spleen adoptive hosts of CD8 T_CM and T_EM at 30-days after BALB/c-skin transplantation.

Persistence of CD8 T_CM and T_EM in adoptive hosts after allograft rejection

Hallmark of memory T cell recall is the generation of secondary memory T cells that mediate better protective immunity (43, 44). To test which alloreactive CD8 memory subset persists as secondary memory T cells after allograft rejection, sorted CD8 T_CM (2 × 10⁶) or T_EM (0.5 × 10⁶) containing similar numbers of alloreactive memory T cells were transferred to wt and aly/aly-spleen skin allograft recipients. 6 – 10 weeks after rejection, cells were harvested from lymphoid and non-lymphoid tissues and analyzed after gating on CD8⁺Thy1.1⁺ T cells. BALB/c-reactive IFNγ⁺ T cells within the harvested CD8⁺Thy1.1⁺ population were measured to assess antigen-specific secondary memory T cells derived from T_CM and T_EM. Alloreactive IFNγ⁺ T cells derived from CD8 T_CM and T_EM were comparable in harvested tissues of adoptive hosts (n = 4 – 5/grp) (Figs. 5A and 5B) suggesting that both subsets persist as secondary memory T cells in either wt or aly/aly-spleen adoptive hosts after allograft rejection. Recall of CD8 T_CM in wt hosts led to both CD62L^high CD49d^low T_CM and CD62L^low CD49d^high T_EM that contained alloreactive IFNγ⁺ T cells compared to CD62L^low CD49d^high alloreactive T_EM alone in aly/aly-spleen hosts (Figs. 5C and 5D). CD8 T_EM persisted as CD62L^low CD49d^high, alloreactive T_EM in either wt or aly/aly-spleen hosts (Figs. 5C and 5D). Thus, CD8 T_CM alone differentiated into alloreactive T_CM secondary memory after allograft rejection in wt and not in aly/aly-spleen hosts suggesting that generation of T_CM secondary memory requires secondary lymphoid tissues and/or that non-lymphoid tissues promote differentiation of T_CM to T_EM (29). T_CM vs T_EM phenotype of alloreactive secondary memory T cells was confirmed by assessing their tissue distribution in adoptive hosts. In wt hosts, alloreactive IFNγ⁺ T cells from CD8 T_CM were found mainly in spleen and LNs than non-lymphoid tissues while those from T_EM were found in spleen, bone marrow and liver (n = 4) (Fig. 5E). Alloreactive IFNγ⁺ T cells from either CD8 T_CM or T_EM were comparable in bone marrow, lung and liver tissues in aly/aly-spleen hosts (n = 5) (Fig. 5F). Thus, either CD8 T_CM or T_EM persisted as alloreactive secondary memory in lymphoid and non-lymphoid tissues and both subsets led to T_EM after allograft rejection that can function independently of secondary lymphoid tissues.

Fig. 5. Persistence of CD8 T_CM and T_EM in adoptive hosts after allograft rejection.

Sorted CD8 T_CM (2 × 10⁶) or T_EM (0.5 × 10⁶) containing similar numbers of BALB/c-reactive IFNγ⁺ T cells were transferred to wt and aly/aly-spleen mice followed by BALB/c-skin transplantation 2-days later. BALB/c-reactive IFNγ⁺ T cells derived from transferred CD8 T_CM or CD8 T_EM were assessed in adoptive hosts 6 – 10 weeks after allograft rejection. Cells were harvested from liver (LV), lungs (LG), bone marrow (BM) and blood in both wt and aly/aly-spleen hosts. Spleen (SP) and LN cells were also harvested in wt hosts. (A – B) Quantitation of alloreactive secondary memory T cells generated from CD8 T_CM and T_EM in wt (A) and aly/aly-spleen (B) adoptive hosts. BALB/c-reactive IFNγ⁺ T cells within the harvested cells from all tissues were assessed by intracellular cytokine staining and enumerated after gating on CD8⁺Thy1.1⁺ population (Mean ± SD of 4 – 5 mice/grp). (C – D) Phenotype of secondary memory T cells derived from transferred CD8 T_CM or T_EM in wt and aly/aly-spleen hosts. Expression of CD49d and CD62L is shown after gating on CD8⁺Thy1.1⁺ population within the harvested cells from spleen and lungs in wt mice, and lungs in aly/aly-spleen mice (C). Expression of CD49d and IFNγ is shown after gating on CD8⁺Thy1.1⁺ population within the harvested cells from spleen in wt mice and lungs in aly/aly-spleen mice (D). Harvested cells from LNs, BM and liver tissues were comparable in phenotype (data not shown). Representative FACS plots of 4 – 5 experiments are shown. (E – F) Tissue distribution of BALB/c-reactive IFNγ⁺ T cells derived from transferred CD8 T_CM or T_EM in wt (E) and aly/aly-spleen (F) hosts. CD8⁺Thy1.1⁺ IFNγ⁺ T cells harvested from each tissue is shown as % of total CD8⁺Thy1.1⁺ IFNγ⁺ T cells harvested from all tissues in that recipient (Mean ± SD of 4 – 5 mice/grp). CD8⁺Thy1.1⁺ IFNγ⁺ T cells harvested from blood were < 1% of total and are not shown.

Discussion

We addressed the roles of alloreactive CD8 T_CM and T_EM subsets in allograft rejection using an in vivo polyclonal system that is analogous to physiologic immune responses in the transplant setting. Alloreactive memory T cells were generated in wt hosts, sorted into CD8 T_CM and T_EM subsets and transferred to wt or aly/aly-spleen hosts that lack secondary lymphoid tissues to compare skin allograft rejection. We find that CD8 T_CM and T_EM were equally effective at mediating skin allograft rejection in wt hosts. However, in hosts that lack secondary lymphoid tissues, CD8 T_EM were significantly more effective than T_CM in mediating skin allograft rejection. Optimal differentiation of CD8 T_CM into effectors that migrate into the allograft was impaired in the absence of secondary lymphoid tissues leading to delayed allograft rejection. Following allograft rejection, either CD8 T_CM or T_EM can persist as alloreactive T_EM secondary memory that function independently of secondary lymphoid tissues.

Allograft rejection by CD8 T_CM and T_EM was addressed in an in-vivo polyclonal system since (a) alloimmune responses target a diverse array of antigenic epitopes and are polyclonal (45); (b) function of a monoclonal population of cells such as TCR-tg T cells may not be representative of physiologic immune responses; and (c) TCR-tg T cells result in artificially high antigen specific precursor frequencies that skew the generation of T_CM vs T_EM and alter the outcome of immune responses (46–48). Antigen-specific cells present within the polyclonal CD8 T_CM and T_EM subsets were detected by identifying alloreactive IFNγ⁺ T cells. CD8 T_CM or T_EM containing similar numbers of alloreactive IFNγ⁺ cells were transferred to accurately compare function of CD8 T_CM vs T_EM that led to transfer of 3 – 4 fold more total CD8 T_CM than T_EM in each experiment. Using this approach might have lead to transfer of more alloreactive CD8 T_CM than T_EM and possibly underestimated the functional advantages of T_EM over T_CM in vivo. Inspite of transferring more total CD8 T_CM, T_EM were equally effective to T_CM in wt hosts and better than T_CM in aly/aly-spleen hosts in rejecting allografts. When total number of transferred CD8 T_CM was matched to T_EM (0.5 × 10⁶ each), skin allograft survival in aly/aly-spleen hosts was prolonged >100 days (data not shown).

Secondary lymphoid organs supported differentiation of both CD8 T_CM and T_EM resulting in increased alloreactive effectors and accelerated skin allograft rejection in wt compared to aly/aly-spleen hosts. It is possible that residual activation of endogenous T and B cells despite co-stimulation blockade also contributed to allograft rejection in wt recipients. CD8 T_CM and T_EM mediated equally effective skin allograft rejection in wt hosts whereas T_EM functioned significantly better than T_CM in aly/aly-spleen hosts. Differentiation of CD8 T_CM into effectors that upregulated VLA-4 and migrated into the allograft was significantly delayed in the absence of secondary lymphoid tissues. Alloreactive CD8 T_EM expressed VLA-4 constitutively that could have contributed to early infiltration into the allograft independent of differentiation in secondary lymphoid tissues. CD8 T_EM were also better than T_CM in lysis of allogeneic target cells in-vivo. These subtle advantages of CD8 T_EM over T_CM led to more rapid allograft rejection by T_EM in hosts that lacked secondary lymphoid tissues but were of lesser significance in wt hosts that supported differentiation of both subsets. It is possible that optimal activation of CD8 T_CM requires interaction with antigen-bearing professional APCs that is promoted in secondary lymphoid organs (49). Since skin allograft rejection is a lymph-node dependent response (22, 50) and more likely to be dependent on CD8 T_CM than T_EM, it remains to be determined whether CD8 T_CM and T_EM respond similarly to other organ allografts (such as heart and kidney) that are less dependent on LN and more dependent on spleen (51–53).

Migration to sites where antigen is located and extravasation into inflamed tissues is essential for immune surveillance (54, 55). Although either CD8 T_CM or T_EM subset can cause allograft rejection, ability of CD8 T_EM to extravasate directly into allografts is an advantage over T_CM in promoting allograft rejection when activation requirements are constrained. The findings reported here implicate alloreactive CD8 T_EM as the principal subset responsible for immune surveillance of allografts. Similarly, T_EM control viruses residing in peripheral tissues and limit metastatic invasion in colorectal cancer (55–57). Our findings along with these reports emphasize the role of T_EM in immune surveillance of non-lymphoid tissues.

Abbreviations used in this paper

aly

alymphoplasia

T_CM

central memory

T_EM

effector memory