|Home | About | Journals | Submit | Contact Us | Français|
The clinical use of adoptive immunotherapy with tumor-reactive T cells to treat established cancers is limited in part by the poor in vivo survival and function of the transferred T cells. Although administration of exogenous cytokines such as IL-2 can promote T cell survival, such strategies have many nonspecific activities and are often associated with toxicity. We show here that abrogating expression of Casitas B-lineage lymphoma b (Cbl-b), a negative regulator of lymphocyte activation, in tumor-reactive CD8+ T cells expanded ex vivo increased the efficacy of adoptive immunotherapy of disseminated leukemia in mice. Mechanistically, Cbl-b abrogation bypassed the requirement for exogenous IL-2 administration for tumor eradication in vivo. In addition, CD8+ T cells lacking Cbl-b demonstrated a lower threshold for activation, better survival following target recognition and stimulation, and enhanced proliferative responses as a result of both IL-2–dependent and –independent pathways. Importantly, siRNA knockdown of Cbl-b in human CD8+CD28– effector T cell clones similarly restored IL-2 production and proliferation following target recognition independent of exogenous IL-2, enhanced IFN-γ production, and increased target avidity. Thus, abrogating Cbl-b expression in effector T cells may improve the efficacy of adoptive therapy of some human malignancies.
Adoptive transfer of ex vivo expanded tumor-specific CD8+ T cells has exhibited efficacy in the treatment of some malignant diseases (1–6). However, substantive obstacles to making T cell therapy reproducibly effective remain, including isolating high-avidity tumor-reactive T cells and sustaining the in vivo survival and function of transferred cells. T cell persistence can be transiently maintained by the administration of IL-2 (2, 7–9), but this is often associated with significant toxicity in humans and may also undesirably promote expansion of Tregs (10). The poor T cell function often observed in therapy has been more difficult to address, as it, in part, reflects deficient expression of costimulatory proteins and/or increased expression of inhibitory ligands or suppressive cytokines by tumor cells (4, 11). The low avidity of most potentially tumor-reactive T cells, which can lead to inefficient recognition and elimination of tumor cells, appears to be intrinsic to targeting tumor antigens, as most tumor antigens are “self” proteins aberrantly expressed or overexpressed by the tumor. Thus, strategies to improve T cell survival, function, and avidity are needed, and one approach is to manipulate the intrinsic properties of T cells by genetic modification prior to transfer.
The E3 ubiquitin ligase, Casitas B-lineage lymphoma b (Cbl-b), is a negative regulator of lymphocyte function (12–15). Cbl-b increases the threshold for naive T cell activation by regulating TCR and CD28 signaling, in part by inhibiting PKCθ- and PI3K/Akt-dependent pathways, respectively (16–20). Abrogating Cbl-b expression rescues IL-2 production and proliferation by naive CD28–/– T cells (16, 17), which has therapeutic implications, since CD28 expression is often lost on T cells that have differentiated to effector cells during repetitive stimulations in vitro. Our laboratory has demonstrated that reconstitution of CD28 expression in human effector T cells restores autocrine IL-2 production and proliferation (21), indicating that the intracellular pathways required for IL-2 production remain accessible and functional in effector cells. However, often tumors lack expression of the costimulatory proteins that bind CD28, which may preclude this approach. Thus, disrupting Cbl-b expression in tumor-reactive T cells may provide a unique strategy for increasing autocrine IL-2 production for promoting proliferative responses and survival following target recognition in the absence of costimulation and may also reduce the threshold for T cell activation to allow low-avidity T cells to respond to tumor targets expressing limiting amounts of antigen.
Cblb–/– mice are resistant to outgrowth of transplantable and spontaneous tumors (12, 22). Tumor resistance has been attributed to the increased responsiveness of the small number of naive tumor-reactive Cblb–/– CD8+ T cells present in the polyclonal repertoire, since adoptive transfer of naive Cblb–/– CD8+ T cells into tumor-bearing wild-type mice resulted in detectable therapeutic activity (12, 22). However, these studies are not directly applicable to human therapeutic settings, as current adoptive therapy protocols require expanding the small numbers of tumor-reactive T cells isolated from the host in vitro via multiple cycles of stimulation to produce adequate numbers of differentiated effector cells for infusion (23). In the disseminated Friend murine leukemia virus–induced (FMuLV-induced) FBL leukemia therapy mouse model (24), adoptive transfer of large numbers of in vitro expanded CD8+ TCR transgenic T cells (TCRgag cells) specific for an epitope derived from the gag protein of FMuLV that is expressed by FBL tumor cells is insufficient to mediate tumor regression, unless — similar to the requirement for efficacy in human clinical trials — injections of exogenous IL-2 are provided to promote persistence of transferred T cells (14, 25–27). The requirement for a prolonged T cell response in this model provides an ideal setting for testing the efficacy of strategies designed to improve T cell activation, proliferation, and persistence.
We have used this mouse model of disseminated leukemia as well as in vitro cultured human effector T cell clones to examine whether ablating Cbl-b expression in in vitro expanded effector T cells might enhance the efficacy of adoptive immunotherapy. Naive and in vitro–derived effector Cblb+/+ and Cblb–/– cells were evaluated to determine whether observed functional differences were dependent on autocrine IL-2, paracrine IL-2, and/or CD28 signaling. The results demonstrate that ablating Cbl-b–enhanced mouse and human effector T cell function via both IL-2–dependent and IL-2–independent pathways, restored some CD28-mediated functions in T cells and increased the efficacy of adoptive immunotherapy in a clinically relevant murine model of disseminated leukemia.
We generated Cblb–/– TCRgag transgenic mice (TCRgagCblb–/– mice), in which CD8+ T cells express an H-2Db–restricted TCR specific to the gag epitope derived from FBL leukemia (26, 28). Although Cblb–/– mice have increased susceptibility to autoimmunity (18, 29), TCRgagCblb–/– mice, which express a defined nonautoreactive TCR, did not exhibit symptoms of autoimmunity and remained healthy in our colony. Nonmanipulated TCRgagCblb+/+ and TCRgagCblb–/– mice had, based on binding an H-2Db tetramer with the gag peptide, similar absolute numbers of transgenic cells (average number of tetramer cells per spleen, 2.9 × 107 vs. 2.4 × 107, respectively). Transgenic T cells from TCRgagCblb+/+ and TCRgagCblb–/– mice also expressed comparable levels of TCR (Figure (Figure1A).1A). Tetramer+ cells from both TCRgagCblb+/+ and TCRgagCblb–/– mice were CD44lo, CD62Lhi, CD25–, and CD69–, indicating that Cbl-b deficiency did not modify the phenotype of naive TCRgag cells (Figure (Figure1B).1B). TCRgagCblb–/– cells produced significantly more IL-2 compared with TCRgag cells but did not differ in production of IFN-γ or TNF-α (Figure (Figure1C).1C). Higher levels of IL-2 were due to both a significantly higher percentage of TCRgagCblb–/– cells secreting IL-2 compared with TCRgagCblb+/+ cells (Figure (Figure1D)1D) as well as more IL-2 production on a per cell basis, as reflected by the MFI of IL-2 staining (Figure (Figure1E).1E). To evaluate whether Cbl-b regulates the threshold for naive CD8+ T cell proliferation, we incubated CFSE-labeled TCRgagCblb+/+ and TCRgagCblb–/– cells with splenocytes pulsed with titrating concentrations of peptide. TCRgagCblb–/– cells proliferated in response to 10-fold lower concentrations of peptide compared with TCRgagCblb+/+ cells (Figure (Figure1F).1F). In identical cell cultures supplemented with recombinant human IL-2 (r-h-IL-2) (25 U/ml), TCRgagCblb–/– cells still exhibited greater proliferation at all concentrations of peptide tested compared with TCRgagCblb+/+ cells (Figure (Figure1F,1F, bottom row). The increased proliferation of TCRgagCblb–/– cells reflected both a higher proportion of cells entering cell cycle as well as a greater number of cell divisions. Although the observed greater IL-2 production likely contributed to the enhanced division, the failure of excess exogenous IL-2 to restore proliferative responses of Cblb+/+ cells to the level of Cblb–/– cells suggests that IL-2–independent mechanisms may also contribute to the increased proliferative response in the absence of Cbl-b.
Adoptive immunotherapy of cancer patients requires repeated in vitro expansion of antigen-specific T cells, which generally are derived from the nontolerized, naive repertoire and have been induced to differentiate to effector CD8+ T cells during in vitro manipulation (2, 3). Therefore, to model human adoptive immunotherapy protocols, we stimulated naive TCRgagCblb+/+ and TCRgagCblb–/– cells in vitro with antigen in the presence of IL-2 and expanded the cells with 3 cycles of stimulation (denoted in vitro–derived effector cells). Gating on tetramer+ cells on day 5 after the third cycle of in vitro expansion revealed that the absence of Cbl-b did not influence CD8 or TCR expression, although TCRgagCblb–/– effectors were larger and more granular than wild-type cells (Figure (Figure2A).2A). Both TCRgagCblb+/+ and TCRgagCblb–/– cells expressed cell surface markers predominantly consistent with an effector phenotype (CD44hi, CD62Llo) (Figure (Figure2B),2B), but TCRgagCblb–/– effector cells expressed higher levels of CD44, CD25, and CD69. The higher expression of CD25 could be attributable to enhanced TCR signaling and/or more persistent activation from autocrine IL-2 usage by TCRgagCblb–/– T cells. To examine whether the absence of Cbl-b resulted in enhanced IL-2 signaling, TCRgagCblb+/+ and TCRgagCblb–/– effector cells were activated with antigen in the same wells, so that the cells would be exposed to a similar concentration of IL-2. After 1 day, effector cells were stained with antibodies to CD8 and Thy1.2 (a congenic marker that distinguished wild-type and Cblb–/– cells) and to phosphorylated-Stat5, which is downstream of the IL-2 receptor signal. TCRgagCblb–/– cells exhibited higher expression of phosphorylated-Stat5, suggesting the higher CD25 expression in Cblb–/– effector cells is associated with augmented IL-2R signaling (Figure (Figure2C).2C). The higher IL-2Rα (CD25) expression on Cblb–/– effector cells most likely reflects consequences of Cbl-b–mediated regulation of TCR-dependent activation, since IL-2 signaling does not appear to be directly regulated by Cbl-b (30), but CD25 as well as CD69 are induced in part by signals from the TCR activating NF-κB (31, 32).
To determine whether Cbl-b regulates effector T cell function, we stimulated TCRgagCblb+/+ and TCRgagCblb–/– effector cells with titrating doses of antigen and compared cytokine production and expression of cytolytic molecules. The percentage of TCRgagCblb–/– effector cells that produced IL-2 was consistently slightly higher compared with that of TCRgagCblb+/+ effector cells (Figure (Figure2D),2D), resulting in a small, but significant (3-fold) increase in the amount of IL-2 detected in the supernatant (Figure (Figure2E).2E). In contrast, we did not detect differences in IFN-γ production, indicating avidity was not changed (Figure (Figure2D).2D). The expression of granzyme B, perforin, and CD107a/b between TCRgagCblb+/+ and TCRgagCblb–/– cells was also similar, suggesting that Cbl-b does not play a role in regulating cytolytic function of effector T cells (data not shown). To determine whether Cbl-b regulates helper-independent proliferation of effector T cells, CFSE-labeled TCRgagCblb+/+ and TCRgagCblb–/– effector cells were stimulated with antigen for 4 days in the presence or absence of exogenous IL-2. In the absence of exogenous IL-2, a higher percentage of TCRgagCblb–/– effectors proliferated compared with Cblb+/+ effectors (Figure (Figure2F),2F), while addition of exogenous IL-2 minimized proliferative differences (Figure (Figure2F,2F, bottom row). The enhanced proliferation of TCRgagCblb–/– effector cells also resulted in an increase in absolute numbers of cells, as the number of TCRgagCblb–/– cells expanded more than 20 fold, even in the absence of exogenous IL-2 (Figure (Figure2G).2G). Thus, TCRgagCblb–/– effectors cells have similar cytolytic function but increased proliferation and expansion in vitro in the absence of exogenous IL-2, characteristics that might translate into improved in vivo survival and efficacy in adoptive immunotherapy.
Previous studies have shown that naive T cells can proliferate via both IL-2–dependent and IL-2–independent mechanisms (33, 34). In contrast to naive T cells, in vitro expanded effector T cells fail to proliferate in the absence of exogenous cytokines, such as IL-2. Therefore, we addressed whether the loss of Cbl-b influenced T cell proliferation via IL-2–dependent and/or IL-2–independent pathways. To determine whether autocrine IL-2 was required for the enhanced proliferation of TCRgagCblb–/– cells, we blocked IL-2 signaling in vitro. CFSE-labeled naive or in vitro–derived effector TCRgagCblb–/– or TCRgagCblb+/+ cells were incubated with a combination of mAbs to both neutralize IL-2 in solution and to block binding to CD25 during in vitro activation with antigen. Four days later, we analyzed CFSE dilution in the presence or absence of IL-2 blockade to quantify cell proliferation. Consistent with previous studies (35, 36), IL-2 blockade inhibited some but not all antigen-induced proliferation of naive wild-type TCRgag cells (Figure (Figure3A).3A). In contrast, IL-2 blockade did not affect the proliferation of naive TCRgagCblb–/– cells (Figure (Figure3A)3A) or cell survival, as we observed significant expansion of Cblb–/– effector cells (data not shown). IL-2 blockade completely suppressed the small amount of proliferation that was observed in wild-type in vitro–derived effector cells but did not inhibit the proliferation of effector TCRgagCblb–/– cells (Figure (Figure3B).3B). The results suggest that Cbl-b regulates not only IL-2 production and the resulting IL-2–induced proliferation, but also regulates IL-2–independent T cell proliferation of both naive and effector T cells.
To determine whether the concentrations of the combined IL-2 blocking antibodies used in these studies were efficiently blocking physiological levels of IL-2 in these experiments, we cultured CFSE-labeled CTLL-2 cells, an IL-2–responsive cell line (37, 38), with 2 ng/ml recombinant mouse IL-2 (r-mIL-2), and with either blocking antibody alone or with r-mIL-2 and a combination of both blocking antibodies. Three days later, CFSE dilution was measured to quantify CTLL-2 cell proliferation. The combination of both blocking antibodies completely suppressed CTLL-2 cell proliferation in response to 2 ng/ml r-mIL-2 (Figure (Figure3C)3C) and inhibited over 90% of CTLL-2 cell proliferation in response to 1 log higher concentration of r-mIL-2 (20 ng/ml), which greatly exceeds the amount detected in activated T cell supernatants (data not shown).
To confirm that Cblb–/– effector T cell proliferation could be fully independent of autocrine IL-2, we crossed our mice to knockout mice lacking the IL-2 gene (TCRgagCblb–/–Il2–/– mice and control TCRgagIl2–/– mice). As predicted, naive (CD44lo, CD62Lhi, CD8+, Vα3+) TCRgagCblb+/+, TCRgagCblb–/–, TCRgagCblb+/+Il2+/+, and TCRgagCblb–/–Il2–/– cells proliferated in response to antigen stimulation independent of exogenous IL-2 (data not shown). After 3 cycles of stimulation with antigen and exogenous IL-2 in vitro to generate effector cells, only TCRgagCblb–/– and TCRgagCblb–/–Il2–/– effector cells proliferated in response to antigen in the absence of exogenous IL-2 (Figure (Figure3D).3D). At day 5, in response to stimulation with antigen, the TCRgagCblb–/–Il2–/– effector cells proliferated somewhat less than TCRgagCblb–/– cells capable of producing autocrine IL-2 (Figure (Figure3D),3D), but by day 7, the numbers of TCRgagCblb–/– and TCRgagCblb–/–Il2–/– effector cells that had expanded and survived were similar (Figure (Figure3E).3E). Addition of exogenous IL-2 and antigen resulted in significantly more, but essentially identical, expansion of TCRgagCblb+/+, TCRgagCblb–/–, TCRgagCblb+/+Il2+/+ and TCRgagCblb–/–Il2–/– cells (Figure (Figure3E).3E). Thus, abrogating Cbl-b expression permits proliferation of effector CD8+ T cells in the absence of exogenous or autocrine IL-2.
CD28 costimulatory signals can enhance T cell expansion via both IL-2–dependent and IL-2–independent mechanisms (35, 39). Previous studies have demonstrated that Cbl-b dampens CD28 signals in naive T cells and that Cbl-b deficiency results in reduced requirements for CD28 signaling for naive T cell proliferation (16, 17, 22). Therefore, we wanted to address whether Cbl-b deficiency reduces the requirement for CD28 signaling and/or IL-2 production for proliferation of effector T cells. CFSE-labeled in vitro–derived effector TCRgagCD28–/–Cblb+/+ and TCRgagCD28–/–Cblb–/– cells were activated in vitro with antigen for 4 days in the presence or absence of IL-2 blocking antibodies. While both TCRgagCD28–/–Cblb+/+ and TCRgagCD28–/–Cblb–/– effector cells proliferated in the presence of antigen and exogenous IL-2 (Figure (Figure3F),3F), only CD28–/– cells that were also deficient in Cbl-b proliferated in response to antigen in the absence of exogenous IL-2 (Figure (Figure3F).3F). Furthermore, IL-2 blockade inhibited some but not all of the proliferation of TCRgagCD28–/–Cblb–/– cells (Figure (Figure3F).3F). Thus, abrogation of Cbl-b appears to overcome the requirements for IL-2 and/or CD28 signaling for proliferation of effector T cells.
To assess whether Cbl-b regulates T cell survival, we stimulated TCRgagCblb+/+ and TCRgagCblb–/– in vitro–derived effector cells with antigen and 4 days later, stained transgenic cells with 7AAD and Annexin-V to determine the fraction of cells undergoing apoptosis. A higher percentage of TCRgagCblb+/+ cells were Annexin-V+ compared with TCRgagCblb–/– cells (Figure (Figure4A,4A, top row). Addition of exogenous IL-2 (r-h-IL-2 at 25 U/ml) resulted in a significant reduction in the percentage of Annexin-V+ TCRgag cells, confirming its role as a T cell survival factor, but slightly increased the percentage of Annexin-V+ Cblb–/– cells. This minor increase in cell death likely reflects the much greater overall proliferation and increased numbers of Cblb–/– cells that occurs in the presence of exogenous IL-2 (Figure (Figure4A,4A, bottom row). The antiapoptotic molecules Bcl-2, which is induced after IL-2 signaling, and Bcl-xL, which is CD28 responsive, both promote T cell survival. On day 4 after T cell activation, effector cells that had been stimulated with antigen in the absence of exogenous IL-2 were stained for intracellular Bcl-2 and Bcl-xL expression. Although Bcl-2 expression levels were not sufficient for detection in both TCRgagCblb+/+ or TCRgagCblb–/– effector cells, TCRgagCblb–/– effector cells did express higher levels of Bcl-xL than wild-type cells (Figure (Figure4B).4B). Thus, increased expression of Bcl-xL in Cblb–/– effector cells likely contributes to the increased cell survival.
To assess whether Cbl-b deficiency bypassed CD28 signaling for enhanced survival, we performed similar experiments with TCRgagCD28–/–Cblb+/+ and TCRgagCD28–/–Cblb–/– effector cells. TCRgagCD28–/–Cblb–/– cells exhibited a lower percentage of Annexin-V+ cells compared with TCRgagCD28–/–Cblb+/+ cells, which was again associated with higher levels of Bcl-xL expression (Figure (Figure4,4, C and D). Addition of exogenous IL-2 decreased the percentage of Cblb+/+CD28–/– and Cblb–/–CD28+/+ Annexin-V+ effector cells. In the presence of exogenous IL-2, a slightly higher percentage of CD28–/–Cblb–/– cells were Annexin-V+ compared with that of CD28–/–Cblb+/+ cells, which may again reflect the greater proliferation of Cblb–/– cells. Thus, ablating Cbl-b expression in either CD28+ or CD28– effector cells enhances Bcl-xL expression and survival.
To best mimic human adoptive immunotherapy protocols, we investigated whether effector CD8+ T cells with abrogated Cbl-b expression would enhance therapeutic efficacy in a mouse therapy model of disseminated leukemia. B6 Thy1.1 mice were inoculated with a lethal dose of FBL leukemia (40), and 5 days later, when leukemia was disseminated, mice were treated with cyclophosphamide chemotherapy, followed in 6 hours with 1 × 105 Thy1.2+TCRgagCblb+/+ or Cblb–/– effector cells that had been expanded in vitro by 3 consecutive cycles of antigen stimulation in the presence of exogenous IL-2 (25 U/ml). A cohort of mice from each group also received low-dose intraperitoneal IL-2 (104 U/day) for 10 days, which we have previously shown enhances therapeutic efficacy by promoting the survival of wild-type TCRgag cells and is required for complete tumor eradication (40). In the absence of exogenous IL-2, T cell therapy prolonged survival in only 1 out of 16 recipients of TCRgagCblb+/+ cells, whereas 13 out of 14 recipients of TCRgagCblb–/– cells appeared healthy and disease free at day 60 (P < 0.0001; Figure Figure5A).5A). By contrast, TCRgagCblb+/+ recipients or TCRgagCblb–/– recipients that also received IL-2 injections nearly uniformly achieved prolonged survival (9 out of 10 or 10 out of 10, respectively). Control in vitro–activated non–antigen-specific P14 Cblb–/– did not extend survival or cure any mice of tumor (data not shown), indicating, as extensively previously demonstrated, that T cell specificity is required in this model (25).
To evaluate whether ablating Cbl-b regulated T cell expansion and persistence during adoptive immunotherapy, we measured the percentage of donor T cells (Thy1.2+) in the blood and spleen at various time points after T cell transfer. Consistent with the enhanced therapeutic efficacy in the absence of exogenous IL-2 injections, TCRgagCblb–/– cells were detected at a 5-fold higher frequency than TCRgagCblb+/+ cells at 10 days after T cell transfer (Figure (Figure5,5, B and C; P < 0.05). Exogenous IL-2 injections promoted the expansion of both wild-type and Cblb–/– effectors during therapy (Figure (Figure5,5, B and C). However, neither transferred effector cell population formed substantial numbers of long-term memory cells regardless of IL-2 administration, as reflected by the failure to detect persisting transferred cells at day 60 (Figure (Figure5C).5C). These results are consistent with clinical data, indicating that effector cells expanded in vitro with IL-2 become terminally differentiated and fail to generate memory T cells with a self-renewal capacity (2). Thus, ablating Cbl-b does not appear to enhance the ability of in vitro–derived effector T cells to form long-term memory but rather acts to enhance the early proliferation and sustain the initial survival of T cells during the period required for tumor eradication.
We also evaluated whether abrogating Cbl-b in naive or minimally expanded T cells also imparts the enhanced efficacy of adoptive immunotherapy of disseminated leukemia, as improved techniques are being developed that may allow the generation of cells for therapy with the requirement for less in vitro expansion. Therefore, we transferred titrating doses of naive or minimally expanded (stimulated 1 cycle in vitro) TCRgagCblb+/+ and TCRgagCblb–/– cells into tumor-bearing mice. At the lowest T cell dose tested (1 × 104 T cells), only naive Cblb–/– T cells exhibited any therapeutic activity, but this was modest, with only 33% surviving in the absence of exogenous IL-2 and with 67% and 33% surviving in the presence of exogenous IL-2 following treatment with naive Cblb–/– and Cblb–/– T cells expanded by a single in vitro cycle, respectively (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI41991DS1). At a dose of 1 × 105 T cells, Cblb–/– T cells stimulated for a single cycle were equally as effective as naive Cblb–/– T cells, similar to what we observed with effector cells generated with 3 cycles of stimulation (Figure (Figure5A),5A), and again were therapeutically effective without requiring IL-2. By contrast, naive Cblb+/+ T cells were not effective at any tested dose in the absence of exogenous IL-2 and only exhibited any therapeutic activity at a dose of 1 × 105 T cells. Thus, targeting Cbl-b expression in CD8+ T cells that have undergone even minimal expansion in vitro still provides therapeutic benefit.
Previous studies in our laboratory demonstrated that adoptive therapy with TCRgag cells in mice that express the gag tumor antigen also as a self antigen selectively in the liver does not result in detectable autoimmune injury (25). Since Cbl-b deficiency lowers the threshold for T cell activation and renders cells resistant to regulatory mechanisms such as TGF-β–mediated suppression and anergy induction (12, 22), we evaluated whether using T cells with abrogated Cbl-b expression in adoptive therapy influences autoimmune toxicity by transferring TCRgagCblb–/– cells in a therapeutic setting in which these cells recognize an antigen not only expressed at high levels by the tumor but also at lower levels in normal tissues, similar to many human tumor antigens. Control B6 mice and mice that express the gag transgene selectively in the liver (Alb/gag mice) were used as recipients for therapy of FBL leukemia. Transfer of 1 × 106 TCRgag or TCRgagCblb–/– in vitro–derived effector T cells, in combination with exogenous IL-2, cured Alb/gag mice of disseminated FBL (Figure (Figure6A)6A) but did not result in detectable autoimmunity, with all recipients appearing healthy through day 60. Neither Alb/gag mice nor B6 mice exhibited persistent elevations of the liver enzymes after T cell transfer, and there was no evidence of increased CD8+ T cell infiltration into the liver of Alb/Gag mice (Figure (Figure6,6, B and C).
Loss of CD28 expression commonly occurs in vivo during differentiation of human antigen-specific CD8+ T cells to effector/memory cells, resulting in an inability to both produce IL-2 and to support autocrine IL-2–mediated proliferation (41). CD28 is also generally not detected on effector T cells used for adoptive immunotherapy, as a result of differentiation during the repetitive in vitro stimulations with antigen necessary for T cell expansion. We have previously shown that restoration of CD28 expression in this population restores IL-2 production, indicating that CD28– effector cells retain the intracellular signaling pathways required to produce IL-2 (21). Based on these observations, as well as the enhanced function of murine Cblb–/– effector cells, we investigated whether targeting Cbl-b expression has functional consequences in human effector CD8 cells. We transduced the human CD8+CD28– cytomegalovirus-specific (CMVpp65-specific) effector T cell clone (MT-3) with retroviral vectors expressing Cbl-b–specific siRNAs designed to specifically target 3 different sequences in Cblb mRNA (the 3 different siRNA sequences are referred to as Cbl-b1, Cbl-b2, and Cbl-b3, as described in the Methods). The siRNA Cbl-b2 reduced Cbl-b protein expression by approximately 100 fold, while the siRNA Cbl-b3 reduced protein levels by approximately 10 fold (Figure (Figure7A).7A). To determine whether decreasing Cbl-b expression influenced functional avidity of CMV-specific clones, we cultured parental (unmodified) and Cbl-b siRNA-transduced MT-3 cells with APCs pulsed with decreasing concentrations of CMVpp65495–503 peptide, and IFN-γ secretion was measured by intracellular cytokine staining. Cbl-b2 siRNA-transduced MT-3 cells responded to a log lower concentration of peptide compared with nontransduced cells, indicating that reducing Cbl-b expression in differentiated human effector T cells improves the functional avidity of these cells by decreasing the amount of antigen required for T cell activation (Figure (Figure7B).7B). To test whether decreasing Cbl-b expression by specific siRNA resulted in a difference in the ability of CD8+ T cell clones to produce IL-2 in response to antigen, control or Cbl-b2 siRNA-transduced CTLs were stimulated with peptide-pulsed T2 cells, and IL-2 production in the supernatant was measured. While unmodified CD28–CD8+ clones failed to produce IL-2 after antigen stimulation, decreasing Cbl-b via the Cbl-b2 siRNA restored IL-2 secretion in MT-3 cells (Figure (Figure7C).7C). Furthermore, stimulation with APCs pulsed with the CMVpp65495–503 peptide induced the CD28–CD8+ clone with decreased Cbl-b expression to produce approximately 67% of the IL-2 produced by the same parental clone that had been transduced to re-express CD28 and therefore have an intact costimulatory signal (Figure (Figure7D). 7D).
To further evaluate the effect of reducing Cbl-b expression in human effector T cell clones of other specificities, particularly tumor-specific T cells, we transduced the MART-1/MelanA26–35–specific CD28–CD8+ T cell clone (SK-111) (2) using retroviral vectors encoding Cbl-b siRNA (Cbl-b2). Similar to the results with CMV-specific T cell clones, siRNA reduced Cbl-b protein expression approximately 100 fold in SK-111 tumor-specific CTLs (Figure (Figure7E).7E). To test whether reducing Cbl-b expression by siRNA resulted in differences in the ability to recognize tumor cells, parental and Cbl-b2 siRNA-transduced CD8+ T cell clones were incubated with melanoma tumor lines that either express or lack the MART-1 antigen. Cbl-b2 siRNA-transduced CD8+ T cell clones produced increased amounts of IFN-γ compared with control CD8+ T cell clones following recognition of the MART-1 expressing 526 melanoma cell line, which expresses HLA-A2 and the MelanA antigen (Figure (Figure7F).7F). Cbl-b2 siRNA-transduced clones did not respond to the melanoma tumor line that did not express the MART-1 antigen (HLA-A2+ 375), indicating that decreasing Cbl-b expression does not result in spontaneous recognition of tumors without TCR specificity. To test whether decreasing Cbl-b expression by specific siRNA also resulted in a difference in the ability of CD8+ T cell clones to produce IL-2 in response to antigen, control or Cbl-b2 siRNA-transduced CTLs were stimulated with peptide-pulsed T2 cells, and IL-2 production in the supernatant was measured. Decreasing Cbl-b via the Cbl-b2 siRNA restored IL-2 secretion in MelanA26–35–specific CD28–CD8+ T cell clones (Figure (Figure7G).7G). Furthermore, Cbl-b2 siRNA-transduced cells proliferated in response to antigen without the requirement for supplemental exogenous IL-2, whereas control CD8+ T cell clones required exogenous IL-2 for proliferation (Figure (Figure7E).7E). Thus, targeting Cbl-b in human T cells may provide a means to promote autocrine IL-2 production and T cell proliferation in effector CD8+ T cells and to reduce the threshold for activation as a means to enhance efficacy of adoptive therapy.
Adoptive T cell immunotherapy of cancer provides an opportunity to not only use tumor-reactive T cells for therapy, but also to genetically engineer the T cells prior to infusion as a means to enhance function and overcome therapeutic obstacles such as the poor immunogenicity of many tumor cells and the limited in vivo survival and function of transferred cells. This study provides the first evidence to our knowledge that abrogating expression of a negative regulator of lymphocyte function, Cbl-b, in effector CD8+ T cells generated by several cycles of in vitro stimulation and expansion can significantly enhance the efficacy of adoptive therapy of a disseminated malignancy. Previous studies of the regulatory role of Cbl-b in T cell function have largely focused on the activity in naive T cells during the primary response to antigen (16–18, 20, 42, 43). Transfer of naive Cblb–/– CD8+ T cells into tumor-bearing hosts can also lead to efficient eradication of some established tumors (12, 22, 44). In addition, transfer of a mixture of polyclonal naive and antigen-specific CD4+ and CD8+ Cblb–/– T cells that had been acutely activated in vivo transiently mediated tumor destruction in a spontaneous insulinoma model (45). However, these studies, which suggest a potential benefit to ablating Cbl-b in T cells targeting a tumor, are not directly applicable to clinical settings that require extensive in vitro expansion of a small number of tumor-specific CD8+ T cells. In vitro expansion generally induces differentiation to effector cells, which, in contrast to naive T cells, usually have lost the ability to produce IL-2 following target recognition, have reduced proliferative capacity compared with naive T cells, and become dependent on CD4+ T helper cells or exogenous cytokines for proliferation and sustained function (46–51). Our observation that Cblb–/– effector CD8+ T cells retain proliferative capacity suggests that Cbl-b may be an important regulatory checkpoint in the differentiation of CD8+ T cells to become helper dependent.
Exogenous IL-2 is required for proliferation of in vitro expanded effector T cells and is commonly administered during therapy in humans to promote the survival and expansion of transferred CD8+ T cells. However, therapeutic administration of this cytokine, often at high doses, can lead to significant toxicities, including a potentially life-threatening capillary leak syndrome (1, 2). In the model used in this study, eradication of FBL leukemia requires the transferred T cells to survive for a minimum of 30 days (24), and, similar to clinical settings (3), in vivo IL-2 administration following cell transfer is essential for T cell survival and tumor eradication (52). Our findings that in vitro expanded Cblb–/– effector cells cured mice of disseminated FBL leukemia in the absence of exogenous IL-2 administration suggests that targeting Cbl-b expression in effector T cells might facilitate the design of therapy protocols that would no longer require inclusion of potentially toxic IL-2 regimens.
Naive and effector Cblb–/– T cells were much less dependent than wild-type T cells on both autocrine and exogenous IL-2 for proliferation, and Cbl-b was found to not only regulate IL-2 production and the subsequent IL-2–induced proliferation, but also to regulate an IL-2–independent proliferative pathway. Proliferation of resting T lymphocytes depends upon the engagement of TCR and costimulatory pathways, such as CD28. In naive T cells, initiation of T cell cycling can occur in the absence of IL-2 (33, 35, 53) but appears to require prolonged TCR/CD28 engagement and the downstream activation of PI3K/PKB, Akt, and mammalian target of rapamycin (36). Cbl-b antagonizes TCR- and CD28-dependent proximal signaling of PI3K/PKB, Akt, and NF-κB (18–20, 54), and the enhanced proliferation of Cblb–/– T cells has been shown to be dependent on PI3K (19). Thus, Cbl-b deficiency may be amplifying the IL-2–independent proliferation pathway in effector T cells by prolonging TCR and CD28 signaling pathways.
Our studies demonstrate that Cbl-b deficiency enhanced both proliferation and survival of effector T cells. In fact, Cblb–/– effector CD8+ T cells did not require CD28 signaling or exogenous IL-2 for proliferation, which is consistent with previous studies that have shown naive Cblb–/– CD4+ T cells are also independent of CD28 costimulation for proliferation (16, 17). As far as we know, this is the first study to demonstrate increased survival in Cblb–/– CD8+ effector cells, suggesting one potential mechanism for the increased efficacy observed in tumor therapy. CD28 was also not required for the enhanced of survival of Cblb–/– effector T cells, which was independent of exogenous IL-2 and associated with increased expression of the CD28-responsive antiapoptotic molecule Bcl-xL (39, 55). The enhanced survival of Cblb–/– T cells was most evident at day 4 after activation, rather than earlier time points, and could reflect Cbl-b–mediated regulation of IL-2 secretion, as addition of exogenous IL-2 rescued wild-type T cells from increased apoptosis.
One concern about transferring tumor/self-reactive T cells with enhanced function into patients is autoimmune toxicity, since most tumor antigens are also self-antigens. While increased autoimmunity has been reported in Cblb–/– mice in which the whole repertoire is deficient in Cbl-b (16, 17), no evidence of acute or persistent autoimmune toxicity was detected following transfer of TCRgagCblb–/– cells into hosts in which the targeted gag tumor antigen was also expressed as a self-antigen in the liver. The safety of Cblb–/– T cells in therapy targeting self-proteins may be enhanced by the inability of transferred Cblb–/– effector CD8+ T cells to form long-term memory cells with self-renewing capacity in the host. Cbl-b has been shown to regulate anergy induction of naive T cells in vitro and in vivo (54, 56–58), and thus long-term persistence of transferred self-reactive Cblb–/– effector CD8+ T cells might pose risks of reactivation and autoimmunity. In our studies, congenically marked transferred Cblb–/– effector cells were rarely detectable by flow cytometry 60 days after transfer. Furthermore, neither immunization of cured mice with Listeria monocytogenes that expressed the gag epitope, nor in vitro expansion of splenocytes from cured mice with antigen and IL-2, resulted in a detectable population of Cblb–/– cells (data not shown). Thus, ablating Cbl-b did not appear to enhance the ability of in vitro–derived effector T cells cultured with antigen and IL-2 to form long-term self-renewing memory cells but rather did act transiently to enhance the proliferation and sustain the survival of the transferred CD8+ T cells through the period required for tumor eradication without causing autoimmunity. Moreover, ablating Cbl-b in tumor-reactive T cells may provide additional benefits for cellular therapy, particularly if targeting tumors of other histologies in which TGF-β– and Treg-mediated suppression are known to be important. Naive Cblb–/– T cells have been shown to be resistant to both TGF-β– and Treg-mediated suppression and to better infiltrate solid tumors (12, 22, 42, 44, 45, 59), and studies in relevant solid tumor therapy models will need to be performed to determine whether this is also true for effector CD8+ T cells.
CD28 is a critical costimulatory molecule required for the generation of functional T cell–mediated immunity (39). CD28 is expressed on naive T cells but is generally lost when the small number of naive, potentially tumor-reactive T cells detectable in a patient is expanded by multiple cycles of in vitro stimulation to obtain a high enough number of tumor-reactive effector T cells for adoptive transfer to achieve therapeutic efficacy. Effector T cells that no longer express CD28 exhibit very limited proliferative responses in the absence of exogenous cytokines (21). A previous study in our laboratory demonstrated that genetically modifying human CD28–CD8+ antigen-specific effector T cells to reconstitute CD28 expression restored their capacity to produce IL-2 following target recognition and to induce autocrine-regulated proliferation (21). In this study, we identified an alternative approach to restore the capacity to induce IL-2 and proliferation by human effector memory CD28– T cells following target recognition by transducing 2 different human effector CD28–CD8+ effector T cell clones with siRNAs that specifically target Cbl-b. Reducing Cbl-b expression in the relatively low-avidity MelanA-specific clones restored IL-2 production and autocrine proliferation in the absence of exogenous IL-2, functions that could likely render these genetically modified tumor-reactive cells more effective in adoptive immunotherapy. Reducing Cbl-b protein expression either approximately 10 or 100 fold in higher avidity CMV-specific T cell clones was sufficient to achieve a similar benefit as restoring CD28 expression on IL-2 production but had the benefit of not requiring that the target cell being recognized express a matching costimulatory ligand. Moreover, reducing Cbl-b expression with siRNA reduced the threshold for activation, making it possible for effector T cells to respond to targets expressing lower amounts of the selected antigen, which represents a common evasion strategy observed with tumors (60). Thus, targeting Cbl-b expression in human T cells with siRNA appears to represent a clinically feasible approach to improve the therapeutic activity of effector T cells that could be administered to treat human malignancies.
Alb/gag and TCRgag transgenic mice have been previously described (25, 26). Cblb–/– mice were originally received from Hua Gu (Columbia University) and crossed to TCRgag and TCRgagCD28–/– mice in our animal facility. Thy1.1 C57BL/6 and Il2+/– mice were obtained from The Jackson Laboratory. The Il2+/– mice were crossed to TCRgag and TCRgagCblb–/– mice in our animal facility to generate TCRgagIl2+/– and TCRgagCblb–/–Il2–/– mice. All mice were maintained under specific pathogen–free conditions. Our murine studies were executed according to our approved animal protocol and the policies of the Institute for Animal Care and Use Committee in the Department of Comparative Medicine, University of Washington.
FBL, a Friend virus-induced erythroleukemia of C57BL/6 (H-2b) origin that expresses MHC class I and FMuLV Gag-encoded products, has been previously described (61). The FMuLV Gag peptide (CCLCLTVFL) recognized by TCRgag CD8+ T cells was purchased from Pi Proteomics. The fluorochrome-conjugated antibodies CD8, Thy1.1, CD44, CD25, CD69, CD62L, Bcl-2, IL-2, and IFN-γ and the mouse Th1/Th2 cytobead arrays were purchased from BD Biosciences. The neutralizing mouse IL-2–specific antibodies, blocking antibodies to IL-2Rα (CD25), and recombinant murine IL-2 were purchased from R&D Systems. Rabbit anti-mouse Bcl-xL and antibodies to phospho-Stat 5 (Tyr694) were purchased from Cell Signaling Technologies, and intracellular staining was detected via secondary labeling with anti-rabbit 546 (Invitrogen). The Annexin-V Staining kit was obtained from BD Biosciences, and all intracellular staining assays were performed using the BD Biosciences Fix/Perm Kit. The gag/H-2Db tetramer was synthesized by the Immune Monitoring Core at the Fred Hutchinson Cancer Research Center. The IL-2–dependent cell line (CTLL-2 cells) has been previously described (37) and was maintained in vitro by culture with 50 U/ml IL-2.
CD8+ T cells were purified from the spleens of naive transgenic mice by negative selection using CD8+ isolation kit (Dynal). TCRgagCblb+/+, TCRgagCblb–/–, TCRgagIl2–/–, TCRgagCblb–/–Il2–/–, TCRgagCD28–/–, and TCRgagCblb–/–CD28–/– T cell lines were generated as previously described (25). Briefly, 1 × 106 transgenic T cells were cultured with 5 × 106 irradiated, syngeneic feeders cells, 3 × 106 irradiated FBL leukemia, and r-h-IL-2 (25 U/ml) in 10 ml of complete media (RPMI 1640 supplemented with 2 μM glutamine, 100 U/ml penicillin/streptomycin, 10% fetal calf serum, and 30 μM 2-mercaptoethanol). CD8+ T cells were stimulated every 7 days, using the same protocol. Five days after the third in vitro stimulation with antigen, in vitro–derived effector T cells were used in various assays or transferred for adoptive immunotherapy.
CD8+ transgenic T cells were labeled with 1 μg/ml CFSE for 30 minutes at 37°C. A total of 1 × 105 CFSE-labeled transgenic CD8+ T cells were incubated in round-bottom 96-well plates with syngeneic splenocytes (5 × 105) that had been pulsed with titrating concentrations of gag peptide. CFSE dilution of transgenic T cells was assessed using flow cytometry by gating on Thy1.2+ CD8+ cells. To block IL-2, neutralizing antibodies to mouse IL-2 and blocking antibodies to CD25 were included in the assay by adding the antibodies (20 μg/ml) to T cell cultures 1 hour prior to activation with antigen.
For intracellular cytokine staining, naive or in vitro–derived effector transgenic T cells were incubated with peptide-pulsed congenic splenocytes for 5 hours in the presence of monensin. Cells were then labeled with antibodies to CD8 and Thy1.2, fixed and subsequently permeabilized, and stained with antibodies to mouse IL-2 and IFN-γ according to the manufacturer’s directions (BD Biosciences). Phospho-Stat5 staining was detected with anti-rabbit 546 on fixed and permeabilized cells 24 hours after T cell activation with antigen using BD Biosciences Fix/Perm buffers. Bcl-2 and Bcl-xL staining was performed on day 4 after T cell activation with antigen, also using BD Biosciences Fix/Perm buffers. Anti–Bcl-2 was directly conjugated to FITC, and an anti-rabbit 546 antibody was used to detect Bcl-xL. Annexin-V and 7AAD staining was performed on day 4 after in vitro activation according to the manufacture’s protocol (BD Biosciences). Stained cells were analyzed by flow cytometry.
Adoptive immunotherapy experiments were performed as previously described (25, 27). Briefly, C57BL/6 Thy1.1 or Alb/gag Thy1.1 recipient mice were inoculated with 5 × 106 live FBL leukemic cells by intraperitoneal injection. Five days later, after the leukemia was disseminated, mice received cyclophosphamide (180 mg/kg), followed 5 hours later — to permit clearance of the cyclophosphamide — by injection of the indicated number of in vitro–derived effector Thy1.2+ TCRgagCblb+/+, TCRgagCblb–/–, TCRgagCD28–/–, or TCRgagCblb–/–CD28–/– T cells. Some recipients also received IL-2 injections for 10 days (1 × 104 U/ml/mouse/d). Mice were monitored for ascites formation or nodal masses (as an indicator of tumor burden) and euthanized if the detectable tumor burden increased to a size that predictably led to mortality within 24–48 hours. T cell persistence and function was monitored by bleeding mice and sacrificing cohorts of mice at various time points and analyzing for Thy1.2+ CD8+ T cells by flow cytometry.
Seven μM frozen liver sections were stained with primary antibodies to rat anti-mouse CD8 (BD Biosciences) and with secondary anti-rat-488 antibodies and counterstained with DAPI (Invitrogen). A total of 12 sections from 3–4 mice per group were compared by analysis on a Leica fluorescence microscope. Photographic images were captured using a fluorescence microscope equipped with a digital camera (Orca-ER) and assembled into RGB images with Adobe Photoshop.
Putative Cbl-b siRNA sequences (21-nt long, initial aa motif, ~50% GC) homologous between human and mouse were predicted and synthesized by Genscript and cloned into the pQCXIH retroviral construct (Clontech). EGFP (Clontech) was subsequently cloned into the multiple cloning sites to generate pQEGFPXIH-siRNA vectors containing the following siRNA hairpin sequences: Cbl-b1, 5′-AAAGTGGTAAGACTGTGCCAA-(TTGATATCCG)-TTGGCACAGTCTTACCACTTT-3′; Cbl-b2, 5′-AACACAGACGCCATGATTTGC-(TTGATATCCG)-GCAAATCATGGCGTCTGTGTT-3′; and Cbl-b3, 5′-AAGATGTCAAGATTGAGCCTT-(TTGATATCCG)-AAGGCTCAATCTTGACATCTT-3′. The sequence in parenthesis is the hairpin loop that is included in all of the siRNA constructs. Retroviral constructs were transiently transfected into the Phoenix-GALV packaging cell line by calcium phosphate transfection as previously described (21). Supernatants from transfected Phoenix-GALV packaging cells were collected on days 2 and 3 after transfection and concentrated by ultracentrifugation. The human MT-3 CMVpp65-specific CD8+ T cell clone was produced in our laboratory by Max Topp, and the SK-111 CTL line was provided by Cassian Yee (Fred Hutchinson Cancer Research Center). MT-3 or SK-111 cells were stimulated with anti-CD3 (OKT-3) and IL-2 (50 U/ml) and transduced with concentrated retroviral supernatants on day 3 of expansion by spinfection (1,000 g, 1 hour, 32°C) in the presence of polybrene (8 μg/ml). EGFP+ CTLs were sorted on a FACSVantage flow cytometer and reexpanded with anti-CD3 and IL-2. Cbl-b protein expression levels were assessed by Western blot using anti–Cbl-b (clone G, Santa Cruz Biosciences Inc.) or anti-actin (C-2, Santa Cruz Biosciences Inc.) antibodies and detected with HRP-linked goat anti-mouse Ig secondary antibody.
To measure IFN-γ and IL-2 production, expanded control or Cbl-b2 siRNA-transduced CTLs were stimulated with T2 cells pulsed with the CMVpp65495–503 peptide (NLVPMVATV) or the MelanA26–35 peptide analog (ELAGIGILTV), which binds stably to HLA-A*02, at the indicated concentrations or with the HLA-A2+ melanoma cells A375 (MelanA26–35–) or Mel 526 (MelanA26–35+). Cytokine production was measured by standard ELISA or by cytokine bead Array (BD Biosciences) on a FACSCanto flow cytometer. To measure proliferation, expanded control or Cbl-b2 siRNA-transduced CTLs were stimulated with mitomycin-c–treated T2 cells pulsed with the indicated peptide concentrations, and proliferation was measured on day 3 by pulsing with 3H-thymidine.
Results are displayed as mean ± SEM. Statistical differences between groups were determined using a 2-tailed unpaired Student’s t test, where a P value less than 0.05 was considered significant, except for those in Figure Figure5A,5A, where analyses of survival curves were performed using a log-rank Mantel-Cox test.
We thank C. Chou, C. Fowler, A. Schietinger, and T.D. Kim for helpful discussion and C. Georgopolos and I. Roberts for technical support. This work was supported in part by grants K01 CA117985-01 and R37 CA33084 from the United States National Institute of Health and a grant from the Leukemia and Lymphoma Society. Support was also received from a Korean GRL program grant from KRIBB. I.M. Stromnes was supported by Training Basic and Physician Scientists in Immunology Training grant 5T32AI007411.
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2010;120(10):3722–3734. doi:10.1172/JCI41991.