|Home | About | Journals | Submit | Contact Us | Français|
Acquired dysfunction of tumor-reactive T cells is one mechanism by which tumors can evade the immune system. Identifying and correcting pathways that contribute to such dysfunction should enable novel anti-cancer therapy design. During cancer growth, T cells show reduced NF-κB activity, which is required for tumor rejection. Impaired T cell-NF-κB may create a vicious cycle conducive to tumor progression and further T cell dysfunction. We hypothesized that forcing T cell-NF-κB activation might break this cycle and induce tumor elimination. NF-κB was activated in T cells by inducing the expression of a constitutively active form of the upstream activator IKKβ. T cell-restricted caIKKβ augmented the frequency of functional tumor-specific CD8+ T cells and improved tumor control. Transfer of caIKKβ-transduced T cells also boosted endogenous T cell responses that controlled pre-established tumors. Our results demonstrate that driving T cell-NF-κB can result in tumor control, thus identifying a pathway with potential clinical applicability.
T cells that recognize tumor-associated antigens (TAA) have the capacity to eliminate tumors (1). Tumor-reactive T cells can often be identified in cancer-bearing patients, both in the circulation and infiltrating tumor masses, and the presence of tumor-infiltrating lymphocytes (TIL) can serve as a powerful positive prognostic and predictive biomarker (2–5). Yet, these tumors usually progress nonetheless, suggesting that the functional properties of TILs are likely suppressed over time (6–9). Direct ex vivo analysis of tumor antigen-specific TILs has indeed revealed defective cytokine production or cytolytic activity in patients (6–8). Approaches to interfere with negative regulatory pathways in order to augment or restore T cell function have shown promising clinical activity (10–12). However, even patients who experience clinical benefit from these new agents often achieve only partial responses, such that additional work is necessary to fully understand the mechanisms that drive T cell dysfunction in cancer, to improve clinical efficacy further.
One signal transduction pathway critical for T cell function involves activation of the IκB kinase β (IKKβ), downstream of TCR/CD28 ligation, which activates the transcription factor NF-κB. The tumor context can result in inhibition of T cell-NF-κB (13, 14), and T cells isolated from cancer patients have been reported to have reduced NF-κB activity (15, 16). Using mice engineered to have impaired NF-κB downstream of the TCR, we have recently shown that T cell-NF-κB activation is required for cytokine secretion, antigen-specific cytotoxicity, and the elimination of immunogenic tumors in vivo (17). Collectively, these studies indicate that growing tumors can induce reduced T cell-NF-κB activity, which in turn results in impaired anti-tumor T cell immunity, thus creating a vicious cycle favoring tumor growth. Hence, it is of therapeutic interest to examine whether forcing T cell-intrinsic NF-κB activity can help improve anti-tumor immunity.
To test this hypothesis, we utilized novel genetic mouse models in which constitutively active IKKβ (caIKKβ) was expressed conditionally in select T cell populations in a constitutive or inducible manner. In addition, we used retroviral vectors to express caIKKβ in wild-type (WT) or TCR transgenic T cells. Our results demonstrate that T cell-restricted expression of caIKKβ markedly improved tumor control even for pre-established tumors. Thus, T cell-intrinsic NF-κB plays a critical role in the immune response against a growing cancer, and the IKKβ/NF-κB axis can be exploited therapeutically to enhance anti-tumor immunity.
C57BL/6 (B6) mice were obtained from Envigo (Indianapolis, IN). R26StopFLikk2ca mice (C57BL/6-Gt(ROSA)26Sortm1(Ikbkb)Rsky/J) and CD4Cre (B6.Cg-Tg(Cd4-cre)1Cwi/BfluJ) were obtained from Jackson Laboratories. LckCreER mice were generated by cloning the cDNA encoding a tamoxifen-inducible Cre recombinase, into a cassette containing the Lck proximal promoter and a human CD2 enhancer (18). The tamoxifen-inducible Cre recombinase was generated by fusing Cre with a mutated form of the estrogen receptor (Cre-ERT2 fusion protein, from Addgene plasmid 14797) (19), such that tamoxifen administration (but not endogenous estrogen) results in Cre recombinase activity. 2C/RAG-KO (Thy1.1+/Thy1.2+) and OTI/RAG-KO (Thy1.1+) transgenic mice were maintained in the laboratory following crossing of 2C and OTI transgenic mice to RAG2-KO mice (Jackson Laboratories).
The B16.F10 spontaneous melanoma cell line was purchased from American Type Culture Collection. B16.SIY was engineered to express the model antigen SIYRYYGL, which can be recognized by CD8+ T cells in the context of H2-Kb (20).
Tumor cells were washed, resuspended in PBS and injected subcutaneously (s.c.). Tumors were measured with calipers and tumor area was calculated as the product of the greatest tumor diameter length and its perpendicular width.
Flow cytometric analyses were performed on single-cell suspensions stained in FACS buffer (PBS, 1% BSA and 0.01% NaN3). Biotinylated H2-Kb:SIY (SIYRYYGL) pentamers (Proimmune, Oxford, U.K.) were used according to the manufacturer’s protocol and revealed with streptavidin-PE. Cells were labeled with fluorochrome-conjugated antibodies and were obtained from BD Biosciences (San Jose, CA), eBioscience (San Diego, CA) or Biolegend (San Diego, CA). Following transfer of transduced TCR transgenic T cells, tumor-infiltrating lymphocytes cells were restimulated with SIY peptide (0.3 µM) for 5h in the presence of Brefeldin A for the last 4h and the percentage of cells positive for intracellular IFNγ was determined after subtracting the background staining in the absence of stimulation. Samples were acquired using Accuri, FACSCanto or LSR Fortessa (BD Biosciences) flow cytometers. Data were analyzed using FlowJo software (TreeStar).
250µg/mouse of anti-CD8 (2.43.1) and/or anti-CD4 (GK1.5), or anti-IFN-γ (XMG1.2.20) (generated by the Fitch Monoclonal Antibody Facility, University of Chicago), or of anti-Thy1.1 (19E12, BioXcell) were injected i.p.
The mouse IFN-γ ELISpot assay was conducted using the BD Bioscience (San Jose, CA) kit according to the manufacturer’s protocol. Splenocytes were plated at 106 cells/well. Stimulation was performed either with irradiated B16.SIY tumor cells (20,000 rads) at 5x104 cells/well, or with 160nM SIY peptide (Proimmune, Oxford, U.K.), as indicated, or PMA and ionomycin as a positive control. Developed plates were read using an ImmunoSpot Series 3 Analyzer, and analyzed with ImmunoSpot software.
Mice were treated with 7.5 mg Tamoxifen (Sigma) by oral gavage on day 0, 1, 3, 7, 14 and 21, and GFP expression in T cells was assessed by flow cytometry at day 7 as a readout of CreER recombinase activity.
Plasmids used to generate the retroviral vectors were previously described (21). Retroviral transduction of T cells was performed as described (22). Prior to adoptive transfer, greater than 98% of live cells were CD3+. Transduction efficacy was determined by flow cytometry and the number of T cells transferred was adjusted for the number of transduced (GFP+) T cells indicated. Cells were injected i.v. into C57BL/6 mice 7 days after tumor inoculation or into naïve mice. The latter animals were sacrificed 7 weeks later and hematoxylin/eosin staining was performed on paraffin-embedded sections from the lung, liver and kidney.
Comparisons of means were performed with GraphPad Prism (GraphPad Software) using the Mann-Whitney test or two-way ANOVA where appropriate with Bonferroni’s correction for multiple comparisons. Differences were considered significant for p values <0.05.
T cells from tumor-bearing hosts have been reported to have reduced NF-κB activity (15, 16, 23, 24). To determine whether forcing NF-κB activity in T cells could prevent T cell dysfunction and promote tumor elimination, R26StopFLikk2ca mice were crossed with CD4Cre transgenic (Tg) mice to produce CD4Cre x caIKKβ mice that express caIKKβ selectively in CD4+ and CD8+ T cells starting at the double positive stage of thymocyte development. We previously reported that peripheral T cells from these mice displayed increased levels of nuclear NF-κB (25). We confirmed that GFP expression was restricted to T cells and did not grossly affect thymocyte development, as CD4Cre x caIKKβ and control littermates had similar total numbers of CD4+ and CD8+ T cells in the thymus and spleen (Supplemental Figure 1). Following B16.SIY inoculation, tumors grew progressively in littermate controls, but were controlled in CD4Cre x caIKKβ mice (Figure 1A). Notably, expression of the model antigen SIY was not required for tumor rejection, as poorly immunogenic B16.F10 tumors that do not express SIY were still rejected by CD4Cre x caIKKβ mice (Figure 1B). These results demonstrate that caIKKβ expression in T cells can lead to markedly improved tumor control.
To confirm that tumor rejection in CD4Cre x caIKKβ mice was dependent on T cells, mice were injected with anti-CD8 and anti-CD4 depleting antibodies (Ab) prior to B16.SIY tumor inoculation. This resulted in greater than 95% deletion of T cells from the blood starting one week after Ab injection, as assessed by the ratio of CD8+/CD4+ or CD4+/CD8+ T cells for CD8+ or CD4+ T cell depletion, respectively (Figure 1F). As expected, depletion of T cells eliminated tumor control in CD4Cre x caIKKβ mice, demonstrating a clear T cell requirement (Figure 1C). In addition, single depletion of CD8+ T cells prior to tumor inoculation (Figure 1D), but not of CD4+ T cells (Figure 1E), eliminated tumor control in CD4Cre x caIKKβ mice. These results demonstrate that caIKKβ-expressing CD8+ T cells are necessary for increased anti-tumor immunity in CD4Cre x caIKKβ mice, and argue against the requirement for CD4+ T cells for tumor rejection when IKKβ is constitutively active in CD8+ cells.
An effective anti-tumor T cell response requires efficient T cell priming by activated APCs leading to expansion, differentiation and accumulation of TAA-specific T cells. To investigate if forced NF-κB activity enhanced T cell priming, we compared the expansion of tumor-specific T cells in CD4Cre x caIKKβ and littermate control mice. Seven days after the inoculation of B16.SIY cells, there was a marked increase in the frequency of SIY-specific CD8+ T cells in the spleens of CD4Cre x caIKKβ mice as compared to control mice (Figure 2A,B). IFNγ has been shown to play an important role in tumor rejection (26). Using ELISpot assays, we observed a significant increase in the frequency of IFNγ+ tumor-reactive splenocytes in CD4Cre x caIKKβ mice compared to littermate controls (Figure 2C). Furthermore, caIKKβ-expressing cells secreted higher levels of IFNγ on a per cell basis, as measured by the mean IFNγ ELISpot size (Figure 2D). Tumor control was dependent on IFNγ, as injection of Abs to neutralize this cytokine abolished tumor elimination (Figure 2E). This is similar to the role of IFNγ for rejection of B16.SIY in settings of Treg depletion and homeostatic proliferation (27). Taken together, these data indicate that caIKKβ expression increased both the frequency and the functional capacity of tumor-specific CD8+ T cells.
In CD4Cre x caIKKβ mice, caIKKβ expression is induced at the CD4+CD8+ double-positive stage of thymocyte development (28), raising the possibility that improved anti-tumor immunity could be a consequence of altered T cell development or of skewed TCR repertoire. To address this, we developed a mouse model in which caIKKβ can be expressed in mature peripheral T cells in an inducible manner following oral gavage of tamoxifen (LckCreER x caIKKβ mice). In these animals, tamoxifen treatment resulted in expression of caIKKβ in 5–25% of peripheral CD4+ and CD8+ T cells on d7 (Figure 3A). When these mice were inoculated on d7 with B16.SIY tumor cells, tumor growth was significantly reduced compared to tamoxifen-treated littermate controls (LckCreER-negative-STOPfl/fl-caIKKβmice, Figure 3B). GFP+ CD4+ and CD8+ cells could still be detected at day 24 in the spleen and tumor, albeit at low frequency (Figure 3C). Thus, improved tumor control could be achieved when caIKKβ expression was induced in only a small fraction of peripheral T cells. These results also suggest that the anti-tumor effect of caIKKβ-expressing T cells depends on a gain-of-function by effector T cells rather than altered thymocyte development.
The use of the genetic models described above revealed the potential therapeutic benefit of inducing caIKKβ expression in T cells. However, a clinical approach would require caIKKβ transduction into primary T cells before adoptive transfer. To test the value of such an approach, we used retroviral vectors driving expression of either caIKKβ or GFP alone, which resulted in 10–40% transduction of anti-CD3/CD28-stimulated splenic T cells (Figure 4A). WT mice were inoculated with B16.SIY tumor cells 7 days before adoptive transfer of 0.5–1x106 transduced T cells. Injection of caIKKβ-transduced T cells resulted in a 3-fold increase in the percentage of splenic SIY-specific CD8+ T cells 7 days after T cell transfer (14 days post tumor inoculation) compared with GFP-transduced T cells (Figure 4B), and a sharp augmentation in the frequency of splenocytes that secreted IFNγ in response to restimulation with SIY peptide (Figure 4C). Interestingly, mice with caIKKβ-transduced T cells had a >100-fold increase in intra-tumoral SIY-specific CD8+ cells relative to mice with control-transduced T cells (Figure 4B), suggesting selective accumulation of TAA-specific T cells in the tumor. Furthermore, adoptive transfer of caIKKβ-transduced T cells into mice with pre-established tumors resulted in a significant reduction of tumor growth (Figure 4D). This rapid tumor control by caIKKβ-transduced T cells also demonstrates that there is no requirement for chronic inflammation in the host, as can be observed in aged CD4-CrexcaIKKβ mice (29).
To determine whether tumor control was mediated by antigen-specific caIKKβ-expressing T cells, 2C/RAG-KO T cells (specific for the tumor-expressed SIY model antigen) or OT-I/RAG-KO T cells (specific for the irrelevant antigen OVA) were transduced with control or caIKKβ retroviral vectors and transferred into congenic mice bearing 7 day-pre-established B16.SIY tumors. Tumor elimination was only induced by transfer of caIKKβ-expressing 2C cells (Figure 4E), indicating the dual requirement for tumor antigen specificity and active IKKβ. To analyze the function of tumor-infiltrating cells, a cohort of similar mice was sacrificed 7 days post-transfer of caIKKβ- or GFP-transduced 2C cells. Transduced T cells were not detectable at this time point, but the frequency of IFNγ-expressing endogenous CD8+ T cells upon SIY restimulation was greater in hosts of caIKKβ- than control-transduced 2C cells (Figure 4F), suggesting that caIKKβ-transduced T cells promote endogenous anti-tumor immunity. To investigate directly whether caIKKβ T cells are dispensable after they help the endogenous response, caIKKβ- or control-transduced polyclonal Thy1.1 T cells were transferred into Thy1.2 congenic hosts bearing 7-day pre-established tumors, and depleted 7 days later using anti-Thy1.1 mAb. Effective deletion of the caIKKβ-transduced T cells (Figure 4G) did not prevent tumor control (Figure 4H), further suggesting that caIKKβ-transduced T cells empower the endogenous anti-tumor immune response. Taken together, these results strongly support the notion that retroviral transduction of caIKKβ into primary polyclonal T cells can be a powerful approach for cancer immunotherapy. Importantly, adoptive transfer of caIKKβ-expressing T cells did not trigger overt autoimmunity, as determined by lack of histological inflammation in the lung, liver or kidneys, or signs of colitis 7 weeks later (Supplemental Figure 2). This is distinct from the autoimmunity that can be observed in aged CD4-CrexcaIKKβ mice (29) and may reflect the lack of long-term survival of transferred transduced T cells.
Tumor-derived factors have been shown to reduce T cell-NFκB activity (13, 14) and T cell-NF-κB is required for effective T cell-mediated tumor elimination (17). This vicious cycle can be one of the mechanisms leading to T cell dysfunction and tumor progression. In our study, we have exploited caIKKβ expression to enhance NF-κB activity in T cells, though it should be noted that IKKβ may have other signaling targets than IkBα (30, 31). The enhanced capacity of caIKKβ T cells to control tumor growth correlated with increased frequency of IFN-γ-producing TAA-specific CD8+ T cells and was dependent on IFN-γ signaling.
Current strategies to increase anti-tumor T cell function in patients include the use of monoclonal Abs that block T cell inhibitory pathways, such as anti-CTLA-4 and anti-PD-1/PDL1, or agonists of costimulatory molecules, such as anti-CD137 and OX40 (32) which have shown important clinical activity in patients with melanoma and other cancers. We have previously shown that engagement of CTLA-4 and PD-1 inhibits the NF-κB pathway (33, 34), and CD137 activation has been shown to induce NF-κB (35) making it conceivable that the success of these immunotherapies relies, at least in part, on their ability to enhance T cell-NF-κB activity.
T cell adoptive transfer is also being explored clinically, with particular success reported in melanoma (36). However, the efficacy of these therapies seems to be dependent on lymphopenic conditioning of the host to support homeostatic proliferation and persistence of the transferred cells, and to deplete suppressive cell populations (36). Our data indicate that adoptive transfer of retrovirally caIKKβ-transduced polyclonal T cells was sufficient to enhance immunity to pre-established tumors, without having to create lymphopenic conditions or deplete regulatory cells. Moreover, this strategy does not rely on prior knowledge of TAAs and HLA genotypes. In addition, the use of a polyclonal TCR repertoire might prevent resurgence of escape variants as the population of T cells may recognize several rather than just one tumor antigen and empower endogenous anti-tumor T cells. Finally, measurable anti-tumor activity could be detected when only a small fraction of T cells displayed enhanced NF-κB activity, suggesting that even transduction of a small number of T cells may have clinical efficacy and arguing against a possible role for chronic autoimmunity for achieving successful tumor control by caIKKβ T cells. Future studies should investigate whether combining forced caIKKβ expression along with other therapeutic modalities such as checkpoint blockade might synergize for a more complete elimination of pre-established tumors. Though transduced T cells do not appear to persist long-term, engineering of a suicide gene into caIKKβ vectors may provide an additional safety level to prevent greater autoimmunity than that observed following checkpoint blockade agents. Our data suggest that translational strategies to improve NF-κB signaling in T cells should be considered for clinical development.
We would like to thank Jessalynn Holman for breeding and genotyping all the animals, Linda Degenstein in the Transgenic Core Facility of the University of Chicago and the University of Chicago Flow Cytometry Core Facility.
1This project was financially supported by the University of Chicago Comprehensive Cancer Research Center Pilot Grant 5 P30 CAO14599-35 and NIH/NIAID RO1 AI052352 both to M.L.A; C.E. was funded by a University of Chicago Committee on Cancer Biology Fellowship and a Fundação para a Ciência e Tecnologia fellowship SFRH/BPD/80353/2011, S.S. by a Cancer Research Institute post-doctoral fellowship and T.F.G. by R01 CA118153.