Immune checkpoint blockade, exemplified by antibodies targeting the PD-1 receptor, can induce durable tumor regressions in some patients. To enhance the efficacy of existing immunotherapies, we screened for small molecules capable of increasing the activity of T cells suppressed by PD-1. Here, we show that short-term exposure to small-molecule inhibitors of cyclin-dependent kinases 4 and 6 (CDK4/6) significantly enhances T-cell activation, contributing to antitumor effects in vivo, due in part to the derepression of NFAT family proteins and their target genes, critical regulators of T-cell function. Although CDK4/6 inhibitors decrease T-cell proliferation, they increase tumor infiltration and activation of effector T cells. Moreover, CDK4/6 inhibition augments the response to PD-1 blockade in a novel ex vivo organotypic tumor spheroid culture system and in multiple in vivo murine syngeneic models, thereby providing a rationale for combining CDK4/6 inhibitors and immunotherapies.
Significance: Our results define previously unrecognized immunomodulatory functions of CDK4/6 and suggest that combining CDK4/6 inhibitors with immune checkpoint blockade may increase treatment efficacy in patients. Furthermore, our study highlights the critical importance of identifying complementary strategies to improve the efficacy of immunotherapy for patients with cancer. Cancer Discov; 8(2); 216–33. ©2017 AACR.
See related commentary by Balko and Sosman, p. 143.
See related article by Jenkins et al., p. 196.
This article is highlighted in the In This Issue feature, p. 127
Immunotherapies that harness or enhance a patient's immune system to target their tumors have recently been developed (1–4). The discovery of immune checkpoint receptors, such as CTLA4 and PD-1 (5–9), that repress the activity of antitumor T cells, led to the development of blocking antibodies directed against these coinhibitory receptors or their ligands, including ipilimumab (anti-CTLA4), pembrolizumab (anti–PD-1), nivolumab (anti–PD-1), atezolizumab (anti–PD-L1), and durvalumab (anti–PD-L1). Strikingly, some patients treated with checkpoint inhibitors experience durable tumor regression, in contrast to targeted small-molecule therapies where tumor relapse is a common occurrence (10). This remarkable response has led to the rapid approval of these therapies for patients and a tremendous amount of optimism in the field (4, 10, 11).
However, despite promising clinical results, checkpoint blockade therapies are successful in only a subset of patients, and certain tumor types respond more favorably than others (11). Furthermore, it is increasingly appreciated that, as in the case of targeted therapies, tumors can acquire resistance against immunotherapies (12–14). Thus, it is crucial to more fully understand the mechanisms behind immunotherapies to develop complementary treatments that will broaden the types of tumors that respond to immunotherapy, and to further enhance the specificity and efficacy of antitumor activity of existing approaches.
One potential strategy is to pair immunotherapies with small-molecule inhibitors. There are many advantages to small-molecule drugs in comparison with antibody-based biologics, including greater exposure in the tumor microenvironment and access to intracellular targets (15). Moreover, increasing evidence indicates that signaling from oncogenic driver kinases, such as mutant BRAF in melanoma (16) or mutant EGFR in lung cancer (17), in addition to their role in promoting cellular transformation, can alter the tumor microenvironment to promote immunosuppression, and inhibition of these oncogenes using small-molecule drugs can contribute to immune reactivation. Furthermore, small-molecule inhibitors can directly alter immune cell function and contribute to antitumor immunity, as was reported in a recent study where inhibition of MAPK kinase (MEK) increased levels of effector CD8+ T cells in tumors and synergized with anti–PD-L1 blockade (18). As such, combining small-molecule kinase inhibitors, which induce dramatic but short-lived tumor regression, with immunotherapies, which have slower but potentially more durable responses, is an attractive potential treatment strategy that can be applied to different cancer types with distinct oncogenic driver genes.
Cyclin-dependent kinases (CDK) are a family of proline-directed serine/threonine kinases that are conserved across eukaryotes (15). The classical cell-cycle CDKs (1, 2, 4, and 6) regulate checkpoints to ensure proper progression through the cell cycle (19), and thus have long been attractive targets for pharmacologic inhibition for treating cancers. In particular, the CDK4/6 inhibitors palbociclib, ribociclib, and abemaciclib recently received FDA approval for the treatment of patients with advanced or metastatic breast cancers, and are in clinical trials for other indications, including non–small cell lung cancer. Although best known for their function in phosphorylating retinoblastoma (Rb) to promote cell-cycle progression, there is increasing evidence that CDK4/6 regulate lymphocytes in other ways (19). For example, CDK6 was recently reported to control hematopoietic stem cell activation via transcriptional repression of the transcription factor EGR1 (20, 21).
In this study, we identified a role for CDK4/6 in modulating T-cell activation via regulation of the activity of NFAT proteins, a family of transcription factors crucial for T-cell activation (22). Furthermore, we found that pharmacologic inhibition of CDK4/6 increased levels of tumor-infiltrating T cells in vivo, and synergized with anti–PD-1 blocking antibodies in multiple syngeneic tumor models. Taken together, our results uncover a mode of action for CDK4/6 inhibitors and suggest that treatment with CDK4/6 inhibitors may increase the effectiveness of immune checkpoint blockade therapies in patients.
Small-Molecule Screen Identifies CDK4/6 Inhibitors as Compounds that Enhance T-cell Activity
To identify small molecules capable of enhancing T-cell activation in the setting of PD-1 engagement, we screened for compounds that activate PD-1–overexpressing Jurkat T cells (23) by measuring IL2 secretion following α-CD3/CD28/IgG (“TCR/IgG”) or α-CD3/CD28/PD-1 (“TCR/PD-1”) stimulation (Fig. 1A; Supplementary Fig. S1A). In addition to known negative regulators of IL2 production [e.g., glycogen synthase kinase-3α/β (GSK3α/β); refs. 24, 25], this screen identified cyclin-dependent kinase 4 and 6 inhibitors (CDK4/6i) as top hits (Supplementary Table S1). As immunostimulatory properties have not been previously ascribed to CDK4/6i, we tested several optimized inhibitors, including the three FDA-approved compounds palbociclib, ribociclib, and abemaciclib, as well as trilaciclib (G1T28), a recently reported selective CDK4/6 inhibitor (26, 27), and found that three of the four tested compounds potently enhanced IL2 secretion, even when suppressed by PD-1 signaling (Fig. 1B). Although abemaciclib had the greatest stimulatory activity, kinome profiling revealed that it potently inhibits many other kinases, including GSK3α/β (Supplementary Fig. S1B–S1D and published data; ref. 26). As palbociclib and trilaciclib are significantly more selective for CDK4/6, we sought to minimize potential confounding effects due to off-target activity and focused our studies on these two compounds.
To further investigate this phenomenon, we stimulated primary human CD4+ T cells with α-CD3/CD28 and either recombinant PD-L1 or control IgG, and found that both palbociclib and trilaciclib treatment enhanced IL2 secretion (Fig. 1C). This recapitulated the effect we observed in Jurkat cells, confirming that CDK4/6i have potent immunostimulatory activity. Importantly, transfection of CDK4- or CDK6-specific siRNAs (Fig. 1D) revealed that knockdown of CDK6, but not CDK4, enhanced IL2 secretion (Fig. 1E), supporting on-target specificity of small-molecule CDK4/6 inhibitors and a predominant role for CDK6 inhibition.
To verify this discovery in a more physiologic setting, patient-derived organotypic tumor spheroids (PDOTS) were treated with CDK4/6i (Supplementary Fig. S1E) in a novel ex vivo three-dimensional (3-D) microfluidic culture system (28). PDOTS contain autologous tumor-infiltrating immune cells (Supplementary Fig. S1F and S1G), and bead-based cytokine profiling of conditioned media from spheroids loaded into 3-D microfluidic devices revealed increased levels of TH1 cytokines (e.g., CXCL9, CXCL10, IFNγ, IL16, and CXCL16; refs. 29, 30) following treatment with palbociclib or trilaciclib (Fig. 1F and G). Although the concentration of IL2 was below the detection range in this system, these findings suggest that CDK4/6i may activate CTL/TH1 responses to elicit antitumor immunity.
CDK6 Regulates NFAT Activity
NFAT family proteins are crucial for T-cell activation and transcriptional regulation of IL2 (22). To investigate the link between CDK4/6 and NFAT in regulating IL2 production, we measured IL2 secretion from PD-1–overexpressing Jurkat cells stimulated in the presence of palbociclib and cyclosporine A (CsA), a calcineurin inhibitor that prevents activation of the NFAT pathway (Fig. 2A). Addition of CsA ablated the production of IL2, even in the presence of palbociclib, suggesting that CDK4/6 inhibitors increase IL2 secretion through heightened NFAT signaling and not via an alternative pathway. Interestingly, a recent biochemical screen suggested that NFAT4 (NFATc3) is a substrate of CDK4/6 (31). To assess phosphorylation of NFAT4 by CDK4/6, we performed two-dimensional (2-D) 15N heteronuclear single quantum correlation (2-D–HSQC) experiments to analyze changes in chemical shifts of the regulatory domain of NFAT4 after incubation with either recombinant CDK4/Cyclin D1 or CDK6/Cyclin D3. The 15N HSQC spectrum of NFATc3 (1–400) has narrow dispersion (∼1 ppm) in the 1H dimension centered on the random coil chemical shift of 8.0 ppm, consistent with an unstructured protein (Supplementary Fig. S2A). The few resonances around the 1H frequency of 7.5 ppm indicate that a minor part of this protein harbors structured elements, which is in accordance with the disorder prediction from primary sequence information (Supplementary Fig. S2C). When NFAT4 was incubated with CDK6, we observed the appearance of resonances corresponding to phospho-serine residues, upfield of 8.5 ppm in the 1H dimension, and a number of distinct chemical shift perturbations for the residues neighboring the phosphorylation sites (Supplementary Fig. S2A; refs. 32, 33). However, this did not occur when NFAT4 was incubated with CDK4 (Supplementary Fig. S2A), consistent with our previous knockdown data (Fig. 1D and E), although we observed some nonspecific peak broadening due to the presence of glycerol in the enzyme mixture. Importantly, CDK6-induced phosphorylation of NFAT4 was inhibited when the kinase was preincubated with palbociclib (Supplementary Fig. S2B), where we observed neither phospho-serine resonances nor distinct chemical shift perturbations associated with phosphorylation. Although we observed the broadening of a few resonances after the addition of palbociclib, we confirmed that this was a nonspecific effect due to the addition of DMSO (Supplementary Fig. S2B).
As our nuclear magnetic resonance (NMR) results indicated that CDK6 is an upstream NFAT kinase, we hypothesized that CDK4/6i would result in decreased phospho-NFAT, which could lead to increased nuclear translocation and enhanced NFAT transcriptional activity (22, 26). As we were unable to assign the residues of NFAT4 that were phosphorylated by CDK6, we instead examined levels of phospho-Ser172-NFAT2, a site reported to regulate the nuclear localization of NFAT2 (34). Although we do not have evidence that CDK4/6 directly phosphorylates NFAT2, we found that treatment of PD-1–Jurkat cells with palbociclib reduced levels of phospho-Ser172-NFAT2 (Fig. 2B), suggesting that multiple members of the NFAT family may be regulated by CDK4/6.
As phosphorylation of the regulatory domains of NFATs is a key regulator of their nuclear localization, we isolated nuclear and cytoplasmic fractions from unstimulated or stimulated PD-1–Jurkat cells treated with palbociclib or vehicle control, and found that CDK4/6 inhibition increased nuclear levels of NFAT4 (Fig. 2C). Consistent with increased levels of NFATs in the nucleus, we also found that exposure to palbociclib increased NFAT transcriptional activity (Fig. 2D) and mRNA expression of IL2, IL3, and GM-CSF (Fig. 2E), three previously reported NFAT targets (35). Taken together, these results reveal a novel role for CDK6 as an upstream regulator of NFAT activity, and demonstrate that pharmacologic CDK4/6 inhibition can enhance T-cell activation in vitro.
CDK4/6 Inhibition Enhances T-cell Infiltration into Lung Tumors
To determine the impact of CDK4/6 inhibition on tumor-infiltrating immune cells in vivo, we treated KrasLSL-G12DTrp53fl/fl (KP) mice, representing an immunocompetent genetically engineered mouse model (GEMM) of human non–small cell lung cancer (36), with either palbociclib or trilaciclib. Both agents increased infiltration of CD4+ T cells and CD8+ cells, to a lesser degree, into lung tumors among total lung-infiltrating leukocytes (Fig. 3A). This increase of CD4+ cells was confirmed in two additional GEMMs, including the KrasLSL-G12D (K) and the KrasLSL-G12DLkb1fl/fl (KL) model (36), in which CDK4/6i also increased infiltration of tumor-infiltrating leukocytes (TIL) into lung tumors (Supplementary Fig. S3A).
Although CDK6 plays a critical role in T-cell proliferation (26, 37), transient inhibition of CDK4/6 did not decrease total number of TILs in these lung tumors, whereas absolute numbers of CD4+ and CD8+ cells only mildly changed (Fig. 3B). This finding suggests that CDK4/6 inhibition can either induce intratumoral T-cell expansion, which is unlikely given the requirement for CDK4/6 for cell proliferation (19), or can lead to increased homing of effector T cells to the tumor. To explore the impact of CDK4/6i on TIL proliferation, we evaluated bromodeoxyuridine (BrdUrd) incorporation in vivo. CDK4/6i did not alter BrdUrd incorporation in CD4+ or CD8+ cells from naïve mice without tumors (Fig. 3C, top; Supplementary Fig. S4A), but did diminish BrdUrd incorporation in both CD4+FOXP3− conventional T cells (Tconv) and CD4+FOXP3+ regulatory T cells (Treg), but not CD8+ cells, isolated from mice bearing KrasLSL-G12DTrp53fl/fl allografts (Fig. 3C, bottom, Supplementary Fig. S4A). Similarly, CDK4/6i more potently reduced proliferation of T cells from tumor-bearing mice than naïve mice after stimulation ex vivo (Supplementary Fig. S4A and S4B), possibly because proliferation of naïve T cells relies on CDK1 and other transcriptional factors such as T-bet (19, 38), whereas tumor-infiltrating CD4+ lymphocytes are more susceptible to CDK4/6i. However, the percentage of Tregs did not show significant changes among CD4+ TILs after CDK4/6i treatment (Supplementary Fig. S3B and S3C).
We next evaluated the impact of CDK4/6i on the immune microenvironment beyond T-cell proliferation and IL2 secretion by investigating chemokines, expression of exhaustion markers, and the proliferation of other stromal cells. Levels of the Th1 chemokines CXCL9 and CXCL10, which govern the trafficking of effector T cells to tumor sites (30, 39), were increased in the lung after CDK4/6 inhibition (Supplementary Fig. S4C and S4D). Levels of coinhibitory molecules, including PD-1 and CTLA4, were reduced in both CD4+ and CD8+ T cells after palbociclib or trilaciclib treatment, albeit to different extents (Fig. 3D; Supplementary Fig. S3D and S3E). CDK4/6i also reduced the abundance of CD11c+ myeloid cells (Fig. 3E), which may be due to decreased proliferation of bone marrow hematopoietic progenitors (26). We also observed reduced levels of IL6, IL10, and IL23 after CDK4/6i (Supplementary Fig. S4D), three cytokines produced by myeloid cells that suppress the TH1 response in cancer (40, 41). Taken together, these data indicate that despite effects on T-cell proliferation, CDK4/6 inhibition results in an increased percentage of effector cells within the tumor microenvironment, correlated to chemokine secretion, with apparent downregulation of coinhibitory molecules in some of the models tested. Moreover, the antiproliferative effect of CDK4/6i does not result in an increase of Tregs among TILs, but does result in a reduced number of the myeloid subpopulation.
Tumor Antigen–Experienced T Cells More Sensitive to CDK4/6 Inhibition than Naïve T Cells
As a recent report demonstrated that lymphocyte proliferation inhibition by CDK4/6i is transient and reversible (27), it is possible that properly timed and sequenced doses of CDK4/6i can activate effector T cells without adversely suppressing their proliferation. To evaluate the impact of CDK4/6i on T-cell activation, IFNγ secretion was evaluated. Total splenocytes isolated from tumor-bearing mice, but not naïve mice, treated with trilaciclib in vivo demonstrated increased IFNγ secretion (Supplementary Fig. S5A and S5B). This finding was further confirmed by treatment with trilaciclib ex vivo, which increased IL2 production only in cells from tumor-bearing mice (Fig. 4A; Supplementary Fig. S5C). Moreover, although CDK4/6i did not significantly alter IFNγ secretion by CD8+ cytotoxic T cells alone (Supplementary Fig. S5D and S5E), coculture of splenic CD8+ T cells from tumor-bearing mice with Tregs in the presence of trilaciclib relieved Treg-mediated suppression, as IFNγ production increased by approximately 10-fold. In contrast, the effect of CDK4/6i on IFNγ production from naïve CD8+ T cells cocultured with Tregs was minimal (Fig. 4B; Supplementary Fig. S5F). These data suggest that CDK4/6i can augment effector T-cell function even in the presence of Tregs.
To further investigate the effects of CDK4/6 inhibition on tumor-infiltrating T cells in vivo, we performed single-cell RNA sequencing (RNA-seq) on CD3+ T cells isolated from KP GEMM lung tumors. Gene ontology (GO) analysis revealed enrichment for processes related to lymphocyte activation and proliferation (Supplementary Table S2). Several reported NFAT targets were upregulated by trilaciclib, including Ccr2, Clcf1, Cx3cr1, and Cd86 (Fig. 4C; Supplementary Table S3), consistent with our in vitro findings (Fig. 2). Conversely, we observed downregulation of Ccna2 and Cd22, which are negatively regulated by NFATs (Fig. 4C; Supplementary Table S3). Thus, single-cell RNA-seq data was consistent with our in vitro findings, indicating that inhibition of CDK4/6 derepresses NFAT activity.
We further analyzed the T-cell RNA-seq data by unsupervised density-based clustering on t-Distributed Stochastic Neighbor Embedding (t-SNE) analysis to separate cells into three different groups (clusters) according to gene expression signatures (Fig. 4D). One group was comprised almost exclusively of cells from trilaciclib-treated mice (group 3). A second group contained cells predominantly from trilaciclib-treated mice, but also from vehicle-treated animals (group 1). The final group (group 2) represented a mixture of cells from vehicle and trilaciclib-treated mice (Fig. 4D). Trilaciclib treatment significantly increased IL2 signaling activation in group 3, as well as in group 1, to a lesser extent. This activation includes upregulation of the IL2 receptors IL2Rα, IL2Rβ, and IL2Rγ (Fig. 4E). Treatment with trilaciclib increased the proportion of T cells in the G1 phase in groups 1 and 3 (Supplementary Fig. S6A), confirming on-target pharmacodynamic effects in these cells. Compared with cells in group 2, cells from groups 1 and 3 showed evidence of highly activated NFAT signaling (Supplementary Fig. S6E), along with heightened upregulation of activation markers, including 4-1BB (Tnfrsf9), Icosl, GITR (Tnfrsf18), Cd40, and Cd86 (Fig. 4E; Supplementary Fig. S7A). Compared with group 1, cells in group 3 showed greater downregulation of inhibitory markers (42), including Pd-l1 (Cd274), Pd-l2 (Pdcd1lg2), Tim3 (Havcr2), and Cd200 and its receptors (Fig. 4E; Supplementary Fig. S7B). In addition, these cells demonstrated greater TCR signaling, manifested by upregulation of Zap70, Lat, Skap1, and Cd6, which are important for continued T-cell activation after TCR engagement, as well as for the effector T-cell function (Fig. 4E). Interestingly, these hyperactive cells were primarily in the G1 phase (Supplementary Fig. S6B), consistent with the effect of CDK4/6 inhibition on traversal from G1–S in both cancer cells and immune cells (Supplementary Fig. S6). Therefore, despite inhibitory effects on cell-cycle progression, short-term exposure to a CDK4/6i resulted in a gene transcription signature consistent with enhanced T-cell function.
CDK4/6 Inhibition Augments Anti–PD-1 Antibody–Induced Antitumor Immunity
We next examined effects of CDK4/6i on tumor burden. CDK4/6i alone was not sufficient to eradicate tumors despite reduced tumor proliferation and increased T-cell activation and infiltration in the KP GEMM model (Fig. 5A), consistent with a previous report that palbociclib reduces the growth of Kras-driven murine lung tumors (43). We therefore evaluated the ability of CDK4/6i to complement PD-1 blockade. As Kras-mutant GEMMs are not responsive to checkpoint blockade (3, 44), in part due to the low levels of somatic mutations (45), we utilized the murine syngeneic colon adenocarcinoma model MC38. We first demonstrated that anti–PD-1 combined with CDK4/6i synergistically induced cell death ex vivo in MC38 murine-derived organotypic tumor spheroids (MDOTS; ref. 28; Fig. 5B; Supplementary Fig. S8A). Furthermore, combination treatment of CDK4/6i with PD-1 blockade downregulated levels of CCL2, CXCL1, and CCL3, which negatively regulate the TH1 response (Fig. 5C). However, when MDOTS were generated from tumors grown in Rag1−/− immunodeficient mice, which lack both B and T lymphocytes, or when the TH1 response was blocked by addition of an anti-IFNγ neutralizing antibody, we no longer observed synergistic effects of CDK4/6i combined with anti–PD-1 treatment (Supplementary Fig. S8B and S8C). In contrast, addition of a neutralizing antibody against CCL5 had no such rescue effect (Supplementary Fig. S8C), suggesting that the T cells are the key cellular mediators of the antitumor activity of CDK4/6i.
T Cells Are Required for Antitumor Immunity Induced by Combinational Treatment of CDK4/6 Inhibitor and Anti–PD-1 Antibody
As previously reported, in vivo PD-1 blockade induced partial tumor growth inhibition in the MC38 model (46); however, consistent with the results in MDOTS, the addition of intermittent exposure to trilaciclib nearly eliminated tumor growth (Fig. 6A, left). Furthermore, we found that treatment with palbociclib in combination with PD-1 blockade had a similar effect in mice bearing tumors derived from CT26 colon carcinoma cells (47), which are far less responsive to PD-1 blockade alone (Fig. 6A, right; Fig. 6B).
Profiling of TILs from MC38 tumors revealed that anti–PD-1 alone increased CD8+ IFNγ production but not CD4+ IL2 production (Fig. 6C). Thus, in this model, PD-1 blockade increased the cytotoxicity of CD8+ T cells, but did not increase T-cell proliferation through IL2. Addition of trilaciclib to PD-1 blockade resulted in an approximately 10-fold increase in the levels of IFNγ in CD8+ TILs and approximately 2-fold increase in CD4+ IL2 production (Fig. 6C). The increase in IL2 was also observed in the CT26 model treated with palbociclib alone or in combination with PD-1 (Supplementary Fig. S9A). Of note, in the MC38 model, IL2 production was also increased in the murine inguinal lymph nodes (Supplementary Fig. S9B), albeit to a lesser extent, compared with TILs, whereas IFNγ levels remained unchanged (Fig. 6D).
Importantly, we found that depletion of either CD4+ or CD8+ T cells in the CT26 model reversed the antitumor effect induced by combined treatment of palbociclib and anti–PD-1 (Fig. 6E). As T-cell depletion completely ablated the antitumor activity of the combination treatment, the predominant antitumor activity of palbociclib in this model could not be independent of T cells (i.e., a direct antiproliferative effect on tumor cells due to tumor cell CDK4/6 inhibition); instead, our results demonstrate that palbociclib amplifies the T cell–dependent antitumor effects of PD-1 blockade. Finally, combination treatment of CDK4/6i and PD-1 blockade was superior to single agents alone in treating established tumors, although eventual relapse was evident in all treatment groups (Supplementary Fig. S9C and S9D). Thus, CDK4/6 inhibitors greatly potentiate the effects of PD-1 blockade in vivo, and the major factors of the CDK4/6i-induced antitumor immune response are T cells.
Here, we show that pharmacologic inhibition of CDK4/6 promotes T-cell activation. Beginning with an unbiased small-molecule screen, we identified CDK4/6 inhibitors as a class of compounds that could enhance the production of IL2, a surrogate marker for T-cell activation, even when suppressive signaling from PD-1 was enforced. Mechanistically, we identified that CDK4/6 regulated the activity of NFAT family transcription factors, which are critically important for proper activation and function of T cells. Finally, ex vivo and in vivo studies revealed that small molecule–mediated inhibition of CDK4/6 resulted in increased antitumor activity, particularly in conjunction with PD-1 blockade, and this effect was largely dependent on T cells.
This finding was especially surprising because inhibiting proliferation should disrupt the clonal expansion of tumor antigen–specific T cells, thereby reducing the activity of the antitumor immune response. Instead, it is apparent that properly timed doses of CDK4/6i can promote T-cell activation and augment the effects of PD-1 blockade. Indeed, we found that short-term treatment with CDK4/6i led to heightened secretion of IFNγ from CD8+ T cells in the presence of Tregs, which often correlates with enhanced antitumor cytotoxicity (3, 30). Moreover, we found in both murine models and human patient samples that treatment with CDK4/6i resulted in increased levels of TH1 cytokines/chemokines, including CXCL9 and CXCL10. As CXCL9/10 are known to be strongly induced by IFNγ, we speculate that the heightened levels of TH1 cytokines/chemokines are partly due to increased levels of IFNγ, resulting from enhanced T-cell activity after CDK4/6 inhibition.
Interestingly, our study suggests that certain types of T cells, especially T regulatory cells, are more susceptible to CDK4/6 inhibition, which may be due to differing expression levels of CDK4/6. A recent study performed transcriptional analysis of human tissue lymphocytes (including Tregs, TH1, and TH17) located either within tumors or in normal tissue (48). Interestingly, they reported that Tregs in general had higher expression of CDK6 than other T-cell subtypes, including tissue-resident TH1 and TH17 cells, as well circulating naïve, central memory, and effector memory CD8+ T cells, suggesting that higher levels of CDK6 and potentially greater dependence on CDK6 in Tregs could account for their increased sensitivity to CDK4/6 inhibitors, which in turn releases suppression of IFNγ production from CD8+ T cells. Future studies will be necessary to determine whether similar expression patterns occur in murine tumor models and in patients.
Although CDK4/6 are best known for their role in regulating cell-cycle progression through phosphorylation of the tumor suppressor protein RB, novel functions for these kinases have recently emerged. In particular, CDK6 has been reported to have a critical role in regulating transcription. For example, Kollmann and colleagues reported that CDK6 is a component of a transcription complex that induces the expression of the tumor suppressor p16INK4a (49). However, as they reported that this activity is unrelated to its kinase activity, it seems unlikely that the same mechanism is in effect in our system. In addition, Handschick and colleagues reported that CDK6 is a chromatin-bound cofactor for NFκB-dependent gene expression (50). However, they had previously found that CDK6 phosphorylates Ser536 of p65 (51), but we found that, in the setting of Jurkat cells stimulated with anti-CD3/CD28, with or without treatment with palbociclib, levels of phospho-Ser536-p65 do not change (Supplementary Fig. S10). Thus, we again do not believe that this plays a significant role in our system. Nevertheless, there is accumulating evidence that CDK6 has a critical role in regulating transcription, potentially in both kinase-dependent and kinase-independent fashions. Future studies will be necessary to carefully interrogate the transcriptional functions of CDK6 and its interacting proteins during T-cell activation.
Here, we identified that NFAT4 is a novel substrate of CDK6, but not CDK4, and that CDK4/6 inhibitors enhance NFAT activity in activated T cells. Further studies will be necessary to elucidate which other NFAT family members are direct substrates of CDK4/6, and whether NFAT activity fluctuates as cells progress through the cell cycle. Moreover, recent studies have identified NFAT signaling, in particular NFAT1 in the absence of AP1, as contributors to T-cell exhaustion (52). Although we did not find evidence of increased T-cell exhaustion after treatment with CDK4/6i, further experiments will be necessary to carefully determine the effects of using small molecules to transiently increase NFAT activity on T-cell exhaustion.
One of the chief challenges in the clinic is determining which types and subpopulations of patients will respond to any candidate immunotherapy treatment, either alone or in combination with other agents. Here, we tested a novel organotypic tumor spheroid system; as these tumor-derived spheroids are comprised of a complex mixture of tumor, immune, and stromal cells, they represent a more faithful representation of the actual tumor microenvironment. Importantly, we found that CDK4/6 inhibitors had similar effects in this ex vivo model as they did in vivo, suggesting that the MDOTS system could be an effective tool for determining whether small-molecule compounds or biologics, either as single agents or in combination, have antitumor activity without performing an in vivo experiment. As this technology matures, PDOTS could potentially be used as a predictive assay for determining optimal treatments in the clinic.
In summary, the effects of CDK4/6i on cell-cycle progression and T-cell proliferation are balanced favorably by increased T-cell recruitment and enhanced effector cell function in tumors, mediated in part by activation of NFAT family transcription factors. Despite the antiproliferative effects of CDK4/6i on T cells, the net result of short-term pharmacologic inhibition of CDK4/6 was augmentation of antitumor immunity, which translated to an improved antitumor response that was largely dependent on the activity of T cells. Whereas prolonged CDK4/6i treatment could be immunosuppressive due to adverse effects on lymphocyte proliferation, which has been observed in both murine models and in the clinic (27, 53, 54), properly timed and sequenced doses of CDK4/6i may potentiate the clinical impact of anti–PD-1/PD-L1 antibodies, thereby sensitizing tumors to immune checkpoint blockade. As palbociclib, ribociclib, and abemaciclib are FDA-approved and other agents are in clinical trials, we expect that this hypothesis will undergo rapid testing in humans.
PD-1–overexpressing Jurkat cells were plated at a concentration of 1 × 105 cells/well in a total volume of 80 μL. Compounds (100 nL each) from the Institute of Chemistry and Cell Biology (ICCB) EMD Kinase Inhibitor I collection (244 compounds total; ref. 55), consisting of three libraries sold by EMD as InhibitorSelect 96-Well Protein Kinase Inhibitor I (catalog no.: 539744, 80 compounds), InhibitorSelect 96-Well Protein Kinase Inhibitor II (catalog no.: 539745, 80 compounds), and InhibitorSelect 96-Well Protein Kinase Inhibitor III (catalog no.: 539746, 84 compounds), were transferred by stainless steel pin array from library plates to each assay plate. Dynabeads conjugated to α-CD3, α-CD28, and α-PD-1 antibodies were added in 20 μL, for a final assay volume of 100 μL, with a final compound concentration of 3.3 μmol/L and an 8:1 bead:cell ratio. Beads were conjugated to α-CD3/α-CD28/control IgG and added to wells containing DMSO-treated cells as a positive control, whereas beads conjugated to α-CD3/α-CD28/α-PD1 were added to wells containing DMSO-treated cells as a negative control. Supernatants from each well were analyzed for IL2 levels by AlphaLISA (PerkinElmer) according to the manufacturer's protocol. Average and SD values were calculated from the PD-1 controls (DMSO-treated cells stimulated with α-CD3/α-CD28/α-PD-1 beads); hits were defined as compounds scoring at least 3 SDs from the mean of the controls.
PD-1–overexpressing Jurkat cells as published previously (23) were stimulated with Dynabeads conjugated to α-CD3 (UCHT1), α-CD28 (28.2), and α-PD-1 (clone EH12 from Gordon J. Freeman) or control IgG at a 4:1 bead:cell ratio in the presence of 1 μmol/L CDK4/6 inhibitor for 18 hours. For primary human T cells, normal donor human blood was obtained through Dana-Farber Cancer Institute (DFCI) Institutional Review Board (IRB) Protocol 04-430. Peripheral blood mononuclear cells (PBMC) were isolated using a Ficoll-Paque density gradient, and purified populations of CD4+ T lymphocytes were obtained through a negative magnetic selection kit according to the manufacturer's instructions (Miltenyi Biotec). Primary human CD4+ T cells were stimulated with Dynabeads conjugated to α-CD3 (UCHT1), α-CD28 (28.2), and recombinant hPD-L1-IgG fusion protein (from Gordon J. Freeman) or control IgG at a 4:1 bead:cell ratio in the presence of 1 μmol/L CDK4/6 inhibitor for 18 hours. IL2 levels in the supernatant were analyzed by AlphaLISA (PerkinElmer) according to the manufacturer's protocol.
Palbociclib and abemaciclib were profiled by DiscoveRx using KINOMEscan (55). Briefly, the two compounds were tested at the concentration of 100 nmol/L and 1,000 nmol/L, respectively. Targeted kinases were visualized using the TREEspot compound profile visualization tool. Z'LYTE kinase assays were conducted for GSK3α and GSK3β at Life Technologies using Km ATP concentrations.
Expression and Purification of NFATc3 Regulatory Domain
The regulatory domain of human NFATc3 (residues 1–400) was cloned into a pET151/D-TOPO plasmid and expressed as a fusion protein with a N-terminal His-GB1 solubility tag cleavable with TEV protease. Escherichia coli strain BL21 (DE3) carrying the above plasmid was grown at 37°C in M9 media containing 6 g/L Na2HPO4, 3 g/L KH2PO4, 0.5 g/L NaCl, 1 mmol/L MgSO4, 0.1 mmol/L CaCl2 in H2O supplemented with 4 g/L 12C-glucose, and 1 g/L of 15NH4Cl isotopes. Protein expression was induced at an optical density of 0.7 by 1 mmol/L isopropyl β-D-1-thiogalactopyranoside (IPTG) at 20°C. Cells were grown for an additional 15 hours at 20°C before harvesting. The harvested cells were resuspended in 40 mL of 50 mmol/L Tris-HCl (pH 8.0), 350 mmol/L NaCl, 10 mmol/L imidazole, and 5 mmol/L β-mercaptoethanol (β-ME). The suspended cells were then disrupted by sonication, and the insoluble fraction was removed by centrifugation for 40 minutes at 16,000 rpm. The protein was initially purified by affinity chromatography using 5 mL of Ni-NTA resin (Qiagen). The supernatant from the cell lysate was incubated with the Ni-NTA resin for one hour. After washing the bound resin with 40 mL of 50 mmol/L Tris-HCl (pH 8.0), 350 mmol/L NaCl, 40 mmol/L imidazole, and 5 mmol/L β-ME, the protein was eluted in an identical buffer containing 350 mmol/L imidazole. The elution fraction was dialyzed against a buffer containing 30 mmol/L Na2HPO4 (pH 6.7), NaCl (150 mmol/L), and DTT (5 mmol/L), and the His-GB1 solubility tag was cleaved using TEV protease. The digested NFATc3 and His-GB1 were separated and further purified using size exclusion chromatography (GE Healthcare Life Sciences “Superdex 75 10/300 GL”).
In Vitro Phosphorylation of NFATc3
NMR experiments were performed on a Varian (Agilent DD2 700) spectrometer equipped with a cryogenically cooled probe, and the spectrum was recorded at 287 K. CDK4/cyclin D1 and CDK6/cyclin D3 kinases were purchased from Signalchem. The phosphorylation reaction was performed with a sample containing 0.1 mmol/L 15N-labeled NFATc3 with the addition of 10 μg CDK4 or 10 μg CDK6 in kinase reaction buffer [50 mmol/L MES (pH 6.7), 140 mmol/L NaCl, 10 mmol/L MgCl2, 0.1 mmol/L EDTA, 2 mmol/L ATP and 5 mmol/L DTT]. First, a control experiment of unphosphorylated NFATc3 in same kinase reaction buffer was recorded, followed by addition of kinases and phosphorylation was monitored by using 2-D 15N-HSQC experiments. In the inhibition assay, approximately 0.7 μmol/L of CKD6 was preincubated with 7 μmol/L inhibitor before addition to 15N-labeled NFATc3 sample. Here, 2.5 μL of a 1 mmol/L stock of the inhibitor was added to 350 μL of the NMR sample. In a control experiment, the same amount of DMSO (2.5 μL) was added. All spectra were processed using nmrPipe and analyzed with CcpNmr-Analysis (version 2.4.1).
Western Blots and Antibodies
Cells were lysed in M-PER buffer (Thermo Scientific) containing protease/phosphatase inhibitor cocktail (Roche). Protein concentration was measured using a BCA assay (Pierce). Equivalent amounts of each sample were loaded on 4%–12% Bis-Tris gels (Invitrogen), transferred to nitrocellulose membranes, and immunoblotted with antibodies against CDK4, CDK6, β-catenin, active β-catenin, phospho-S536-p65, total p65, and Actin (Cell Signaling Technology), pS172-NFAT2 (R&D Systems), and NFAT2 (Invitrogen). IRDye 800–labeled goat anti-rabbit IgG and IRDye 680–labeled goat anti-mouse IgG secondary antibodies were purchased from LI-COR Biosciences, and membranes were detected on an Odyssey detection system (LI-COR Biosciences).
All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at DFCI (Boston, MA). The GEMM harboring a conditional activating mutation of endogenous Kras (KrasLSL-G12D/+) crossed with p53 conditional knockout Trp53fl/fl has been described previously (36). CRE recombinase was induced through intranasal inhalation of 5 × 106 p.f.u. adeno-Cre (University of Iowa adenoviral core).
For drug treatment studies in GEMMs, mice were evaluated by MRI imaging to quantify lung tumor burden before and after drug treatment. Mice were treated with either vehicle, 100 mg/kg trilaciclib, or 100 mg/kg palbociclib daily by oral gavage.
For allograft studies, lung tumor nodules were isolated from KrasLSL-G12DTrp53fl/fl mice (C57BL/6 background), minced into small pieces, and plated onto tissue culture plates and passaged at least 5 times before implantation into mice.
For syngeneic models, MC38 and CT26 cells were injected into 6–8-week-old C57BL/6 or BALB/c female mice subcutaneously, respectively. Vehicle control, CDK4/6 inhibitors (trilaciclib or palbociclib) were treated alone or together with PD-1 antibody starting at the indicated time point, using an intermittent dosing schedule of 3 days on, 4 days off until experimental endpoint. PD-1 antibody was administered three times a week (Monday, Wednesday, and Friday) at 200 μg/mouse through an intraperitoneal injection.
Samples from human subjects treated at Massachusetts General Hospital and DFCI were assembled for PDOTS profiling and culture between June and October 2016. Studies were conducted according to the Declaration of Helsinki and informed consent was obtained from all subjects. Tumor samples were collected and analyzed according to Dana-Farber/Harvard Cancer Center IRB-approved protocols.
Lung-infiltrating immune cells were stained with different combinations of fluorochrome-coupled antibodies against mouse CD45 (clone 30-F11, BioLegend), CD3 (clone 17A2, BioLegend), CD4 (clone GK1.5, BioLegend), CD8 (clone 53-6.7, BioLegend), CD11b (clone M1/70, BioLegend), CD11c (clone N418, BioLegend), FOXP3 (clone FJK-16s, eBioscience), CD279 (PD-1, clone 29F.1A12, BioLegend), CD152 (CTLA4, clone UC10-4B9, eBioscience), TIM3 (clone RMT3-23, eBioscience), CD223 (LAG3, clone C9B7W, BioLegend), IL2 (clone JES6-5H4, BioLegend), IFNγ (clone XMG1.2, BioLegend), BrdUrd (clone Bu20a, BioLegend). Jurkat, PD-1–Jurkat, and human PBMCs were stained with fluorochrome-coupled antibodies against human CD3 (clone HIT3a, BioLegend), CD4 (clone OKT4, BioLegend), and CD279 (PD-1, clone EH12.2H7, BioLegend).
Animals were anesthetized with isoflurane to perform MRI of the lung field using BioSpec USR70/30 horizontal bore system (Bruker) to scan 24 consecutive sections. Tumor volumes within the whole lung were quantified using 3-D slicer software to reconstruct MRI volumetric measurements as described previously (36). Acquisition of the MRI signal was adapted according to cardiac and respiratory cycles to minimize motion effects during imaging.
Spheroid Preparation and Microfluidic Culture
Experiments were performed as described previously (28). Briefly, fresh tumor specimens from human patients were received in media (DMEM) on ice and minced in 10-cm dishes (on ice) in a sterile field. S2 fractions (40–100 μm) were used for ex vivo culture as described previously (28). An aliquot of the S2 fraction was pelleted and resuspended in type I rat tail collagen (Corning) and the spheroid–collagen mixture was then injected into the center gel region of the 3-D microfluidic culture device. After 30 minutes at 37°C, collagen hydrogels containing PDOTS/MDOTS were hydrated with media with indicated treatments. MDOTS were treated with IgG isotype control (10 μg/mL, clone 2A3) or rat-α-mouse anti–PD-1 (10 μg/mL, clone RMP1-14, BioXCell). Both MDOTS and PDOTS were treated with vehicle (DMSO), palbociclib (palb; 100 nmol/L), or trilaciclib (100 nmol/L).
Dual labeling was performed by loading microfluidic device with Nexcelom ViaStain AO/PI Staining Solution (Nexcelom, CS2-0106). Following incubation with the dyes (20 minutes at room temperature in the dark), images were captured on a Nikon Eclipse 80i fluorescence microscope equipped with Z-stack (Prior) and CoolSNAP CCD camera (Roper Scientific). Image capture and analysis was performed using NIS-Elements AR software package. Whole device images were achieved by stitching in multiple captures. Live and dead cell quantification was performed by measuring total cell area of each dye.
Cytokine Profiling Analysis of Murine BAL Fluid
Mouse lung bronchoalveolar lavage (BAL) was performed by intratracheal injection of 2 mL of sterile PBS followed by collection by aspiration. Cytokines were measured using 19-plex mouse magnetic Luminex kit (R&D Systems), Mouse Cytokine 23-plex Assay (Bio-Rad), or Human Cytokine 40-plex Assay (Bio-Rad) and measured on Bio-Plex 200 system (Bio-Rad). Concentrations (pg/mL) of each protein were derived from 5-parameter curve fitting models. Fold changes relative to the control were calculated and plotted as log2 fold change. Lower and upper limits of quantitation (LLOQ/ULOQ) were derived from standard curves for cytokines above or below detection. Mouse IL6 and IL10 concentrations were further confirmed by ELISA (BioLegend).
Tumor-Infiltrating Immune Cell Isolation and FACS Analysis
Mice were sacrificed, and lungs were perfused using sterile PBS through heart perfusion from the left ventricle after BAL fluid collection. The whole lung was minced into small pieces and digested in collagenase D (Sigma) and DNase I (Sigma) in Hank's Balanced Salt Solution (HBSS) at 37°C for 30 minutes. After incubation, the digested tissue was filtered through a 70-μm cell strainer (Thermo Fisher Scientific) to obtain single-cell suspensions. Separated cells were treated with 1× RBC lysis buffer (BioLegend) to lyse red blood cells. Live cells were determined by LIVE/DEAD fixable aqua dead cell stain kit (Molecular Probes). The cell pellets were resuspended in PBS with 2% FBS for FACS analysis. Cells were stained with cell surface markers as indicated followed by fixation/permeabilization using FOXP3 fixation/permeabilization kit (eBioscience). Cells were imaged on BD LSRFortessa (BD Biosciences) and analyzed using FlowJo software (Tree Star).
Library preparation and preprocessing.
Single-cell suspensions from KrasG12D/+Trp53fl/fl GEMM mice treated with trilaciclib were isolated as described for tumor-infiltrating immune cells, with modifications. After isolation, live cells were stained and sorted for the CD45+CD3+DAPI− population and plated at one cell/well of a skirted twin.tec 96-well plate (Eppendorf) containing 1× TCL buffer (Qiagen catalog no. 1031576) spiked with ERCC (Ambion, 1:2,000,000 dilution ratio). In total, four 96-well plates were generated, two plates with and two plates without ERCC spike-ins. After sorting, full-length RNA-seq from isolated single cells was performed according to SMART-seq2 protocol with modifications. Briefly, total RNA was purified using RNA-SPRI beads. Poly(A)+ mRNA was converted to cDNA for amplification. The converted cDNA transcript was subject to barcoding specific to each sample using transposon-based fragmentation that used dual indexing. For single-cell sequencing, each cell was given its own combination of barcodes. Barcoded cDNA fragments were then pooled prior to sequencing. Sequencing was carried out as paired-end (PE) 2 × 36 bp with an additional 8 cycles for each index on NexSeq 500 desktop sequencer (Illumina). To obtain quantitative mapping information, PE reads were mapped to the mouse genome (mm9), concatenated with ERCC sequences for spiked-in samples, by STAR (56). Estimated transcript counts and transcripts per million (TPM) for the mouse Gencode vM1 annotation, concatenated with ERCC sequence information for spiked-in samples, were obtained using the pseudo-aligner Kallisto (57). Aggregated and library scaled TPM values for genes were obtained according to the methods described (58) and were used in further downstream differential distribution, GO, and cell-cycle analysis. Only cells that had a minimum of 100,000 PE reads, and with at least 20% alignment to the transcriptome, were retained for further analysis. To further exclude cells that displayed low quality, we collected quality metrics for library size, library complexity, duplicate reads, and mitochondrial and ribosomal read fraction, and performed principal component analysis (PCA) combined with density-based clustering (dbscan; http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.71.1980) to identify and remove outlier cells relative to the largest and homogeneous group of single cells (27 cells removed). Genes were considered not expressed if TPM < 1 and were subsequently removed if not detected in at least 10% of remaining cells.
The single-cell RNA-seq results have been deposited in the National Center for Biotechnology Information's Gene Expression Omnibus and are accessible through GEO Series accession number GSE89477 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE89477).
Single-Cell RNA-seq Analysis
We used a multistep approach using the R package SCONE to identify the optimal normalization strategy and account for possible batch effects, observed and hidden technical covariates (59). First, we identified the most stable 200 genes between the two plates that contained spike-ins. Next, these genes were used in the SCONE framework as negative control genes for unwanted variation for all plates. We continued with highest scored normalization strategy according to SCONE metrics, which included adjusting for batch and biological effects, removal of observed technical variation based on previously identified quality metrics, scaling for library size with DESeq, and imputing drop-out events using a combined clustering and probabilistic scoring algorithm.
To assign cells to a cell-cycle stage, we applied the cyclone classification tool as described previously (60).
Feature selection and cell clustering.
To identify the most informative genes for clustering single cells, we continued only with the Gencode defined gene types, protein_coding and lincRNA, which contain most genes and displayed the highest coefficient of variation. Subsequently, we combined two approaches. First, we identified genes that displayed more than expected variance modeled by the relationship between variance and log expression with LOESS. Next, these genes were used to perform PCA, and the 100 most correlated and anticorrelated genes for the first five principal components were retained for reducing dimensionality and separating cells in gene expression space with t-SNE. Distinct groups were identified applying density-based clustering (dbscan) on the t-SNE–generated coordinates, resulting in 3 groups of cells.
Differential distribution and GO analysis.
Genes that display differential distribution between previously identified groups or between treatments were discovered by performing pairwise comparisons with the scDD (http://biorxiv.org/content/early/2015/12/29/035501) package in R. Enriched biological processes were identified using the online GOrilla tool (http://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-10-48).
C57BL/6 mice were subjected to tail-vein injection with the KrasG12D/+Trp53fl/fl (KP) tumor cell line (1 × 106 cells/mouse) to induce orthotopic tumor growth in the lung. Tumor-bearing mice or C57BL/6 background naïve mice were treated with vehicle (10 μL/g), palbociclib (100 mg/kg), or trilaciclib (100 mg/kg) by daily oral gavage for two consecutive days. At day 3, mice received an intraperitoneal injection of BrdUrd (BD Biosciences) at 2 mg/mouse in sterile PBS. Mice were sacrificed 24 hours after BrdUrd injection and splenocytes were isolated and stained for surface markers. Cells were fixed and permeabilized with FOXP3 fixation/permeabilization buffer (eBioscience), followed by DNase I digestion (0.3 mg/mL, Roche) at 37°C for 1 hour. Cells were stained with fluorochrome-conjugated anti-BrdUrd antibody (BioLegend) and analyzed on LSRFortessa (BD Biosciences).
Cell Coculture and Cytokine Production
Naïve or KP tumor–bearing C57BL/6 mice were sacrificed and total splenocytes were harvested. Spleens were digested with collagenase D (Roche) and DNase I (Roche) at 37°C for 30 minutes, followed by 1× ACS lysis buffer (BioLegend) incubation to lyse red blood cells. The collected total splenocytes were stained with the fluorochrome-conjugated cell-surface markers CD3, CD4, CD8, and CD25 to isolate different T-cell subpopulations, including conventional T cells Tconv (CD3+CD4+CD25−), Treg (CD3+CD4+CD25+), and CD8+ (CD3+CD8+), using BD FACSAria II SORP cell sorter (BD Biosciences). DAPI (4′,6-diamidino-2-phenylindole) staining was used to exclude dead cells. Sorted cells were cultured in 96-well plates precoated with CD3 antibody (eBioscience) and treated with trilaciclib in the presence of CD28 (eBioscience). Cells were collected 3 days after culturing and cytokine production of IFNγ and IL2 was determined by intracellular staining and analyzed on BD LSRFortessa (BD Biosciences).
siRNA targeting human CDK4 or CDK6 (GE Dharmacon) or constructs for NFAT-Firefly Luciferase or Renilla Luciferase-SV40 (Addgene) were electroporated into cells using the Neon transfection system (Invitrogen) according to the manufacturer's protocol.
Total RNA was extracted from cells using TRIzol (Invitrogen), and cDNA was generated using the SuperScript II Reverse Transcriptase Kit (Invitrogen). Quantitative PCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems), and transcript levels were normalized to Actin. Samples were run in triplicate. Primer sequences are as follows: Forward primer (5′–3′): IL2, AACTCACCAGGATGCTCACA; IL3, CAACCTCAATGGGGAAGACCA; GM-CSF, TGCTGAGATGAATGAAACAGTAGA; Actin, CGCACCACTGGCATTGTCAT. Reverse primer (5′–3′): IL2, TTGCTGATTAAGTCCCTGGGT; IL3, TGGATTGGATGTCGCGTGG; GM-CSF, CTGGGTTGCACAGGAAGTT; Actin, TTCTCCTTGATGTCACGCAC.
Luminescence was measured using the Dual-Glo Luciferase Assay System (Promega) from cells transiently transfected with NFAT-Firefly Luciferase and Renilla Luciferase-SV40 on a Clariostar Microplate Reader (BMG Labtech). Samples were run in triplicate.
Data are presented as mean with SEM unless otherwise specified. Statistical comparisons were performed using unpaired Student t test for two-tailed P values unless otherwise specified (*, P < 0.05; **, P < 0.01; ***, P < 0.001).
Disclosure of Potential Conflicts of Interest
J. Deng has ownership interest in a patent application. E.S. Wang has ownership interest in a patent application. R.W. Jenkins has ownership interest in DFCI US patent WO20112172A1. C.P. Paweletz has received speakers’ bureau honoraria from AstraZeneca and Bio-Rad. J.E. Bisi has ownership interest (including patents) in G1 Therapeutics. P.J. Roberts has ownership interest (including patents) in G1 Therapeutics. J.C. Strum has ownership interest (including patents) in G1 Therapeutics. R. Bueno reports receiving commercial research grants from Exosome, Inc., Novartis Institutes for Biomedical Research, HTG Molecular Diagnostics, Inc., Gritstone Oncology, Inc., Genentech, Inc., Verastem, Inc., Siemens, Myriad Genetics, Inc., and Castle Bioscience; has ownership interest (including patents) in Navigation Sciences; is a consultant/advisory board member for BioMedical Insights, Inc.; and has provided expert testimony for Arthur Tuverson, LLC, Balick & Balick, LLC, David Weiss, LLC, Morrison Mahoney, Neil Leifer, Esq., Satterly & Kelly, LLC, Rice Dolan & Kershaw, Ferraro Law Firm, and AstraZeneca. G.J. Freeman has received honoraria from the speakers bureaus of MI Bioresearch, Expert Conn, and MPM Capital, and has ownership interest (including patents) in Novartis (all related to licensed patents in the PD-1/PD-L1 field), Roche, Bristol-Myers Squibb, Merck, EMD Serono, AstraZeneca, and Dako. N.E. Sharpless has ownership interest (including patents) in G1 Therapeutics and is a consultant/advisory board member for the same. G.I. Shapiro reports receiving commercial research grants from Pfizer, Lilly, and Merck/EMD Serono and is a consultant/advisory board member for Pfizer, Lilly, G1 Therapeutics, Roche, and Vertex Pharmaceuticals. D.A. Barbie has ownership interest in DFCI US Patent WO2016112172A1 and is a consultant/advisory board member for N of One. K.-K. Wong has ownership interest (including patents) in G1 Therapeutics and is a consultant/advisory board member for the same. No potential conflicts of interest were disclosed by the other authors.
Conception and design: J. Deng, E.S. Wang, R.W. Jenkins, A.R. Aref, J.A. Sorrentino, J.H. Lorch, R. Bueno, G.I. Shapiro, D.A. Barbie, N.S. Gray, K.-K. Wong
Development of methodology: J. Deng, E.S. Wang, R.W. Jenkins, K. Yates, A.R. Aref, E. Ivanova, C.P. Paweletz, M. Bowden, T. Chen, E. Haines, P.J. Roberts, J.H. Lorch, S. Parangi, G.J. Freeman, W.N. Haining, D.A. Barbie, K.-K. Wong
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Deng, E.S. Wang, R.W. Jenkins, S. Li, K. Yates, W. Huang, A.R. Aref, E. Ivanova, M. Bowden, C.W. Zhou, G.S. Herter-Sprie, J.A. Sorrentino, J.E. Bisi, P.H. Lizotte, A.A. Merlino, M.M. Quinn, L.E. Bufe, A. Yang, Y. Zhang, H. Zhang, P. Gao, T. Chen, M.E. Cavanaugh, A.J. Rode, E. Haines, P.J. Roberts, W.G. Richards, J.H. Lorch, V. Gunda, G.M. Boland, R. Bueno, S. Palakurthi, J. Ritz, H. Arthanari, G.I. Shapiro, K.-K. Wong
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Deng, E.S. Wang, R.W. Jenkins, S. Li, R. Dries, K. Yates, H. Liu, A.R. Aref, E. Ivanova, C.P. Paweletz, M. Bowden, C.W. Zhou, G.S. Herter-Sprie, J.A. Sorrentino, P.H. Lizotte, T. Chen, M.E. Cavanaugh, P.J. Roberts, J.C. Strum, J.H. Lorch, S. Palakurthi, J. Ritz, N.E. Sharpless, H. Arthanari, G.I. Shapiro, K.-K. Wong
Writing, review, and/or revision of the manuscript: J. Deng, E.S. Wang, R.W. Jenkins, R. Dries, S. Chhabra, A.R. Aref, E. Ivanova, M. Bowden, G.S. Herter-Sprie, J.A. Sorrentino, J.E. Bisi, M.E. Cavanaugh, P.J. Roberts, J.C. Strum, W.G. Richards, J.H. Lorch, S. Parangi, V. Gunda, G.M. Boland, G.J. Freeman, J. Ritz, W.N. Haining, N.E. Sharpless, G.I. Shapiro, D.A. Barbie, N.S. Gray, K.-K. Wong
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.S. Wang, A.R. Aref, P. Gao, G.I. Shapiro, N.S. Gray
Study supervision: J. Deng, E.S. Wang, T. Chen, J.H. Lorch, R. Bueno, G.I. Shapiro, D.A. Barbie, N.S. Gray, K.-K. Wong
Other (designed, performed, and analyzed NMR experiments on NFAT): S. Chhabra
We thank the Dana-Farber Cancer Institute Animal Resources Facility staff for their support of the animal studies. We would also like to acknowledge the Brigham and Women's Hospital, Dana-Farber Cancer Institute, and Massachusetts General Hospital for providing the patient specimens used in this study. We thank the Broad Institute for facilitating the single-cell sequencing experiment. And finally, we appreciate the generous gift of anti–PD-1 antibody from G.J. Freeman used for the mouse studies. This work was supported by a Damon Runyon Cancer Research Fellowship DRG-2270-16 (to E.S. Wang), CA154303-05 (to N.S. Gray), and CA179483-03 (to N.S. Gray), NIH/NCI P01CA120964 (to K.-K. Wong), 5R01CA163896-04 (to K.-K. Wong), 5R01CA140594-07 (to K.-K. Wong), 5R01CA122794-10 (to K.-K. Wong), and 5R01CA166480-04 (to K.-K. Wong), and funding from a Medical Oncology Discovery grant, Dana-Farber Cancer Institute.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.