Purpose: Tumor-infiltrating lymphocytes (TILs) become hypofunctional, although the mechanisms are not clear. Our goal was to generate a model of human tumor-induced TIL hypofunction to study mechanisms and to test anti-human therapeutics.
Experimental Design: We transduced human T cells with a published, optimized T-cell receptor (TCR) that is directed to a peptide within the cancer testis antigen, NY-ESO-1. After demonstrating antigen-specific in vitro activity, these cells were used to target a human lung cancer line that expressed NY-ESO-1 in the appropriate HLA context growing in immunodeficient mice. The ability of anti-PD1 antibody to augment efficacy was tested.
Results: Injection of transgenic T cells had some antitumor activity, but did not eliminate the tumors. The injected T cells became profoundly hypofunctional accompanied by upregulation of PD1, Tim3, and Lag3 with coexpression of multiple inhibitory receptors in a high percentage of cells. This model allowed us to test reagents targeted specifically to human T cells. We found that injections of an anti-PD1 antibody in combination with T cells led to decreased TIL hypofunction and augmented the efficacy of the adoptively transferred T cells.
Conclusions: This model offers a platform for preclinical testing of adjuvant immunotherapeutics targeted to human T cells prior to transition to the bedside. Because the model employs engineering of human T cells with a TCR clone instead of a CAR, it allows for study of the biology of tumor-reactive TILs that signal through an endogenous TCR. The lessons learned from TCR-engineered TILs can thus be applied to tumor-reactive TILs. Clin Cancer Res; 22(2); 436–47. ©2015 AACR.
See related commentary by Yang, p. 275
Adoptive T-cell transfer (ATC) holds promise for cancer therapeutics; however, solid tumors pose significant hurdles for ATC, including the upregulation of inhibitory receptors on T cells after successful infiltration into tumor. Robust preclinical models are needed to study the inhibitory receptor expression pattern on tumor-infiltrating lymphocytes (TIL) and test the effects of tailored blockade of inhibitory receptors on antitumor activity of tumor-reactive T cells. We present such a model, where tumor-reactive (via TCR engineering) human T cells targeting human solid tumor successfully infiltrate into established tumors, but demonstrate hypofunction due, at least in part, to the upregulation of inhibitory receptors (e.g., PD1, Tim3, and Lag3), with PD1 being significantly upregulated on TCR-engineered TILs, especially those engineered with enhanced tumor reactivity. We further describe the ability to augment the activity of transferred engineered T cells by combining with injected anti-PD1 antibody.
The field of adoptive T-cell transfer (ATC) has made impressive progress over the last decade. Expanding from early experiences using ex vivo–expanded tumor-infiltrating lymphocytes in metastatic melanoma (1), the field is now exploring the use of autologous peripheral blood T cells that are genetically modified to express chimeric antigen receptors (CAR) or modified T-cell receptors (TCR) to redirect them toward tumor-associated antigens (TAA; ref. 2). The most impressive successes have involved the use of CARs directed against the B-cell antigen CD19, where a high percentage of complete clinical responses have been observed in both adults and children with chronic and acute leukemias (3).
In contrast, the treatment of solid tumors with ATC (or any other form of immunotherapy) has so far proven to be more challenging, though not without some success. There have been some promising results with ATC using TCR-engineered T cells derived from peripheral blood T cells. These T cells are genetically altered to express optimized TCRs that operate under restriction of the TCR–peptide–MHC complex, but offer the benefits of the full spectrum of activation signals that are exhibited by the wild-type receptor (4, 5), show more physiologically relevant levels of affinity for their cognate antigen (6), and can also be directed against antigen-expressing stromal cells and intracellular TAAs (7). Antigens that have been targeted by TCR engineering, to date, have included relatively immunogenic antigens derived from spontaneously occurring tumor-specific T cells in patients, such as the melanocyte differentiation antigens MART-1 (8), glycoprotein 100 (gp100; ref. 9), the melanoma-associated antigen (MAGE; ref. 10), and New York esophageal squamous cell carcinoma antigen (NYESO1; ref. 11). Using these targets, antitumor responses have been reported in a subset of patients enrolled in early-phase clinical trials of TCR-engineered T cells targeting melanoma (1, 9), colon cancer (12), and synovial cell sarcoma (13).
It is likely that the lack of consistent success of ATC for solid tumors seen so far is due to the same set of obstacles encountered by cancer vaccines and other forms of immunotherapy in general, that include: (i) inadequate T-cell trafficking (14), (ii) intra-tumoral metabolic and hypoxic challenges (15), and (iii) an immune-inhibitory tumor microenvironment (TME) milieu that includes stroma, suppressive immune cells, and soluble factors (16). In addition, there is expression of a set of surface inhibitory receptors (IRs) (i.e., CTLA4, PD1, Tim3, and Lag3) and intracellular checkpoints (i.e., SHP-1, diacylglycerol kinase, and the transcription factor Ikaros) that are naturally upregulated after TCR engagement to block continued activation and thus prevent autoimmunity (17–19). Although elucidating these factors in mouse models of immunotherapy has been the basis of much of our understanding of tumor immunology, it is clear that there are many important differences between mouse and human T-cell and tumor biology. It would therefore be of great value to be able to study human T cells targeted to human tumors in experimentally manipulatable animal models.
Recently, our lab published the details of such an animal model of human solid tumor where injection of “second-generation” CAR-engineered human T cells exerted some antitumor effects, but became reversibly hypofunctional after infiltration into tumors due to a variety of mechanisms (20). This hypofunction phenomenon was similar to that described in TILs isolated from human tumors (21–23). The goal of this study was to generate a similar model, however, using human T cells expressing a specific TCR, rather than a CAR. We reasoned that this model would not only provide a tool to study ATC using transgenic T cells, but, as all signaling occurs through a defined and “authentic” TCR, we could more feasibly extrapolate our findings to TILs observed in the majority of human cancers.
To accomplish this goal, we transduced human T cells with a previously published, optimized TCR (called Ly95), that is directed to a peptide within the cancer testis antigen (CTA) NY-ESO-1 (24). These cells were then used to target a human lung cancer line that expresses NY-ESO-1 in the appropriate HLA context growing in immunodeficient mice. Injection of these transgenic T cells had some antitumor activity, but did not eliminate the tumors. We observed that the injected T cells become hypofunctional and noted upregulation of PD1 and other IRs. These findings allowed us to take advantage of a major strength of this model, the ability to test reagents targeted specifically to human T cells. We found that PD1 blockade using an anti-human PD1 antibody led to decreased T-cell hypofunction and augmented the efficacy of the adoptively transferred T cells.
Materials and Methods
Cell culture conditions
See Supplementary Materials and Methods for details.
The NY-ESO1–reactive Ly95 TCR construct is an affinity-enhanced variant of the wild-type IG4 TCR identified from T cells recognizing the HLA-A2 restricted NY-ESO-1:157-165 peptide antigen. In the mutant form, the threonine at residue position 95 is substituted by leucine and the serine at residue position 96 is substituted by tyrosine. It was constructed using an overlapping PCR method (25) based on the description and sequences published previously (24) and incorporated into the lentiviral expression vector pELNS bearing the EF1α promoter (provided by Dr. Carl June at the University of Pennsylvania, Philadelphia, PA). Packaging of each plasmid into lentivirus has been previously described (26). Titering of lentiviral concentration was performed by transduction of Sup-T1 cells (ATCC CRL-1942) at different virus dilutions and measurement of transgenic TCR expression by flow cytometric analysis using an anti-human Vβ13.1 TCR chain antibody;
Isolation, bead activation, transduction, and expansion of primary human T lymphocytes
See Supplementary Materials and Methods for details.
Generation of the target lung cancer cell line
See Supplementary Materials and Methods for details.
See Supplementary Materials and Methods for details.
Flow cytometric T-cell activation assay
See Supplementary Materials and Methods for details.
In vitro testing of tumor cell killing by Ly95 TCR T cells
Control tumor cells and A549-A2-ESO cells were plated in a flat-bottom 96-well plate at 5,000 cells per well in triplicates. After overnight incubation, Ly95 T cells were cocultured at different effector:target (E:T) ratios. After 18 hours of incubation at 37°C and 5%CO2, supernatant from the wells were aspirated for cytokine analysis by ELISA, wells were washed, the remaining tumor cells were lysed, and luminescence was read in a Modulus II Microplate Multimode Plate Reader after addition of 100 μL of luciferin reagent (Promega E1501). The same assay was used to examine the tumor killing ability of TILs obtained from our in vivo experiments (see below).
Measurement of Ly95 T-cell IFNγ secretion by ELISA
See Supplementary Materials and Methods for details.
In vivo xenograft experiments
A total of 5 × 106 A549-A2-ESO tumor cells were injected in the flanks of NSG mice in a solution of X-Vivo media (Lonza) and Matrigel (BD Biosciences). After tumors were established (100–200 mm3), the mice were randomly assigned to one of three intravenous (tail-vein) treatment groups: (i) saline, (ii) 10 × 106 mock-transduced and expanded (mock) T cells, and (iii) 10 × 106 Ly95-expressing T cells. In the experiments combining anti-PD-1 antibody with T cells, two additional groups were included: (iv) every 5-day intraperitoneal (i.p.) injection of 10 mg/kg anti-PD1 antibody (Ultra-LEAF, Biolegend), and (v) 10 × 106 Ly95 T cells intravenous plus every 5-day intraperitoneal injection of 10 mg/kg anti-PD1 antibody. Tumors were measured using calipers and tumor volumes were calculated using the formula (π/6) (length) × (width)2. When predefined protocol endpoints were reached, tumors were harvested, microdissected, and digested in a solution of 1:2 DNase:collagenase in a shaker incubator at 37°C for 2 hours. Digested tumors were then filtered through 70 μm nylon mesh cell strainers, and red blood cells were lysed if needed (BD Pharm Lyse; BD Biosciences). Spleens harvested from the same mice were also filtered through 70 μm nylon mesh cells trainers with red blood cell lysis. A total of 1 × 106 cells from single-cell suspensions were placed in standard FACS tubes and were stained with anti-human CD45, CD8, CD4, and TCRVβ13.1 antibodies to assess degree of infiltration of adoptively transferred T cells. In addition, we also stained cells with anti-PD1, anti-Tim3, and anti-Lag3 antibodies to measure expression of IRs on TILs. The in vivo experiments were repeated three times in an independent fashion. Groups contained 5 to 10 mice each.
Ex vivo TIL analysis
After digestion of harvested tumors, necrotic debris was first removed by processing the single-cell suspension using a Dead Cell Removal Kit (Miltenyi Biotech). TILs were subsequently isolated using an anti-human CD45-PE antibody (BD Biosciences) with the EasySEP PE Selection Kit (STEMCELL Technologies). Once isolated, functional analyses of TILs were performed in two different ways: (i) luciferase-based killing assays, and (ii) measurement of antigen-induced T cell IFNγ secretion by ELISA (see above). Pooling of samples was required in order to isolate sufficient numbers of viable TILs after processing (e.g., harvest, digestion, single-cell preparation via multiple filter and wash steps, dead cell removal, CD45 magnetic separation) to perform in vitro coculture killing experiments.
See Supplementary Materials and Methods for details.
See Supplementary Materials and Methods for details.
An engineered TCR can be efficiently expressed on the surface of human T cells
Transduction of human CD4 and CD8 T cells undergoing anti-CD3/CD28 bead activation with high titer lentivirus that encodes the Ly95 TCR recognizing NY-ESO-1 resulted in approximately 50% expression as measured by FACS analysis of T cells stained with an anti-human TCRVβ13.1 antibody. At the time of analysis, approximately 70% of the T cells were CD8+ and 29% were CD4+ (Fig. 1A).
Ly95-engineered human T cells demonstrate reactivity to A549-A2-ESO tumor cells in an antigen-specific fashion
After the Ly95 T cells were “rested down”, they were cocultured with A549-A2-ESO (A549 expressing NY-ESO-1 in the context of HLA-A2) or A549-A2 tumor cells (A549 expressing HLA-A2 but not NY-ESO-1) for 24 hours, and stained with different markers of activation and for intracellular cytokine secretion. Compared with T cells that were not cocultured with tumor cells, or T cells cocultured with control cells not expressing NY-ESO-1, T cells cocultured with A549-A2-ESO cells demonstrated increases in the percent of cells expressing: (i) intracellular IFNγ (∼2% to 15%), (ii) granzyme B (∼2% to 25%), (iii) CD107a (∼3% to 22%), and (iv) CD25 (∼10% to 47%; Fig. 1B).
We next measured specific lysis of tumor cells by Ly95 T cells via coculture killing assays (Fig. 1C). Ly95 T cells demonstrated high efficiency killing of A549-A2-ESO tumor cells in a dose-dependent fashion, but did not kill a variety of control cells that did not express HLA-2, NY-ESO-1, or both (Fig. 1C, top). We observed the same pattern of high level of dose-dependent IFNγ secretion in response to A549-A2-ESO tumor cells with no IFNγ secretion after exposure to the control cell lines (Fig. 1C, bottom).
Ly95 T cells are able to slow progression but not induce regression of A549-A2-ESO flank tumors in NSG mice
To test the in vivo activity of Ly95 T cells, 10 × 106 Ly95-expressing T cells were injected once intravenously in NSG mice bearing large, established A549-A2-ESO flank tumors (approximately 200 mm3 in size). We observed slowing of tumor growth after 11 days post-adoptive transfer. At day 30, the tumors in the Ly95 T cell–treated group were approximately 50% smaller than those of the untreated group (P < 0.01). Administration of mock-transduced T cells (to control for allogenic responses) had no significant effect on flank tumor growth (Fig. 2A).
Ly95 T cells successfully infiltrate A549-A2-ESO flank tumors after intravenous administration
At 30 days post-adoptive transfer, flank tumors and spleens were harvested, digested into single-cell suspension, and analyzed by flow cytometry. There was some infiltration detected in the spleens of adoptively transferred mice (2.0% of total cells in Ly95-treated mice, 1.5% of total cells in mock-transduced T-cell–treated mice; Fig. 2B); however, this was increased in the tumors, where we observed a distinct population of human CD45+ T cells comprising 5.8% of the single-cell tumor suspension in the “Ly95 tumors” versus 0.3% of that in “mock tumors” (P < 0.05; Fig. 2C). The majority of TILs were CD8+ (76%) and Ly95+ (64%; Fig. 2C).
Ly95 TILs are reversibly hypofunctional
Although a significant number of live Ly95 TILs with confirmed surface expression of the engineered TCR were detected within A549-A2-ESO flank tumors, their ability to kill freshly cultured A549-A2-ESO cells (Fig. 2D, top) and secrete IFNγ (Fig. 2D, bottom) at a 20:1 E:T ratio compared to the infusion product [i.e. cryopreserved Ly95 T cells (“infused T”)] was dampened after isolation from tumors (“fresh TIL”). However, when the isolated Ly95 “fresh TILs” were cultured in cell culture medium (“rested”) overnight in 37°C and 5% CO2 (without IL2), and then cocultured with tumor cells the following day, they exhibited significant improvements in effector function and IFNγ secretion (Fig. 2D, right columns).
Hypofunctional Ly95 TILs have increased expression of inhibitory receptors compared with cryopreserved Ly95 T cells
We sought to compare the expression of three IRs on freshly isolated Ly95 TILs versus that on infused T cells (the latter serving as baseline) to determine whether the hypofunctionality of freshly isolated Ly95 TILs could at least be partially attributed to increased IR expression. In the infused T cells, CD8 and CD4 cells showed similar expression of PD1 (both 3%), Tim3 (2% and 3%, respectively), and Lag3 (1% and 3%, respectively; Fig. 3A, top). In comparison with infused T cells, both CD4 and CD8 freshly isolated Ly95 TILs exhibited increased expression of all three IRs (Fig. 3A, bottom and Fig. 3B). When repeated independently two additional times, upregulation of all three IRs was consistently seen on both CD8 and CD4 TILs compared with CD8 and CD4 infused T cells. (Supplementary Fig. S1).
We were also interested in comparing IR expression on the Ly95-expressing versus non-Ly95–expressing T cells from the same tumors. We first examined the phenotype of the infused T cells. As shown in Fig. 3C (top, small boxes), the percent of cells expressing PD1, Tim3, and Lag 3 was low, but very similar between the TCR− and the TCR+ (Vβ13.1–positive) cells. When we examined the isolated TILs (Fig. 3C, bottom and 3D), we noted increased expression of all of the IRs. However, there were differences in the expression levels of T cells with the transgenic TCR compared to the T cells not expressing the TCR. Specifically, TCR+ TILs, as compared to TCR− TILs, showed greater upregulation of levels of PD1 (74% vs. 36%, P < 0.01) and Tim3 (49% vs. 41%, P < 0.05), but no increases in Lag3 (Fig. 3B, bottom and bar graph).
In addition, we analyzed the frequency of cells that expressed more than one IR (Fig. 4 and Supplementary Fig. S2). A clear increase in cells expressing multiple IRs was seen in the TILs versus the infused T cells (Fig. 4, top). Interestingly, the percentage of cells expressing two or three inhibitory receptors was consistently higher on TCR+ TILs compared with TCR− TILs (Fig. 4, lower panels); comparing TCR+ TILs and TCR− TILs, 23% and 12% expressed all three IRs, whereas of the infused TCR+ and TCR− cryopreserved T cells, only 0.6% and 0.2% expressed all three IRs, respectively.
Repeated intraperitoneal injections of anti-human PD1 antibody augments the efficiency of Ly95 T cells in controlling the growth of A549-A2-ESO tumors, and preserves Ly95 TIL function
Given the increased levels of PD1 that we observed in hypofunctional TILs, we hypothesized that this upregulation had functional significance, and that blocking PD1 with an anti-human PD1 antibody might augment T-cell function and antitumor efficacy. Accordingly, NSG with established A549-A2-ESO flank tumors were treated with both Ly95 T cells and repeated doses of anti-PD1 antibody. As shown in Fig. 5, the anti-PD1 antibody alone had no effect on tumor growth compared with control, while Ly95 T cells significantly reduced tumor growth, as above. However, tumor-bearing mice that received repeated intraperitoneal injections of PD1 antibody in addition to a single intravenous injection of 10 million Ly95 T cells, demonstrated the slowest rate of tumor growth for 6 weeks from the start of treatment. At the end of the experiment, the Ly95 T-cell + PD1 antibody-treated group had tumors that were, on average, 35% smaller than those in the Ly95 T-cell–treated group (P < 0.05). This experiment was repeated with similar results.
At this time point, the tumors were harvested/digested and processed into single-cell suspensions. After dead cell removal, FACS analysis revealed that the degree of T-cell infiltration into the tumors and enrichment of TCR-engineered T cells was similar between Ly95 T cell and Ly95 T cell + PD1 antibody-treated mice (Fig. 6A). However, when Ly95 TILs were isolated and co-cultured with A549-A2-ESO tumor cells at 20:1 E:T ratio for 18 hours, TILs from Ly95 T-cell + PD1 antibody-treated mice exhibited less hypofunction than those treated with Ly95 T cells in terms of improved killing (39% vs. 17%; P < 0.01; Fig. 6B, left) and increased IFNγ secretion (4,180 pg/mL vs. 720 pg/mL; P < 0.01; Fig. 6B, right). TILs from Ly95 T-cell–treated mice were also cocultured ex vivo with A549-A2-ESO in the presence of 10 μg/mL of PD1 antibody. The addition of PD1 antibody during the ex vivo killing assay led to partial recovery of Ly95 TIL function as measured by killing (33% vs. 17%, P < 0.01) and IFNγ secretion (1,510 pg/mL vs. 720 pg/mL, P < 0.05; Fig. 6B).
Interestingly, the A549-A2-ESO tumor cells that upregulate PDL1 in response to IFNγ in vitro (Supplementary Fig. S3A) also increased PDL1 expression in vivo when Ly95 T cells were injected with PD1 antibody (Supplementary Fig. S3B).
Dysfunction of antigen-specific T cells infiltrating murine tumors is now well established (27). A number of investigators have documented that although TILs are often found in high numbers within tumors, they are usually hypofunctional upon isolation, with an inability to kill tumor cells or release cytokines. Key features of this hypofunction phenomenon in most models include a proximal TCR signaling defect and reversible dysfunction when they are removed from the tumor microenvironment (27). Thus, it appears that tumor-induced hypofunction is distinct from what has been described in anergy, tolerance, senescence, or exhaustion where dysfunction is usually not reversible (28, 29).
Much less is known about human tumor–induced T-cell dysfunction. A limited number of studies have functionally analyzed human TIL (30) and, in general, the findings of reversible TIL hypofunction seem similar (see Supplementary Table S1 for a summary). Our data are consistent with these findings. However, it should be noted that the mechanisms may be somewhat variable and perhaps tumor-specific. For example, in an examination of renal cell cancers, TILs were reversibly hypofunctional, but had normal proximal TCR signaling and faulty distal signaling, with high DGK levels and decreased MAPK activation (31), whereas in an examination of melanoma tumors, TILs were irreversibly hypofunctional and demonstrated abnormal proximal TCR signaling, specifically downregulation of CD3ζ and Lck (32). In addition, PD1 signaling has been shown to be a significant contributing factor to TIL hypofunction in patients (including those with lung cancer and human TILs with NY-ESO-1 reactivity), similar to our model (33, 34).
Despite extensive investigation, the mechanisms responsible for tumor-induced T-cell dysfunction have yet to be fully elucidated, although many factors have been implicated in murine models (see Introduction). Given the complexity of this process and the importance of the tumor microenvironment, TIL hypofunction must be studied using intact tumor models, rather than cell culture or in vitro studies. Although experiments using murine tumor models have been important, it is clear that human and mouse T cells and tumors are very different, requiring tractable models of human TIL. To date, analysis of human TIL have been largely restricted to correlation analyses (35–38) or very short-term ex vivo studies using digested tumor extracts (21, 31), thus significantly limiting possible experimental manipulations.
In this article, we describe an experimental system that allows the long-term study and manipulation of antigen-specific human T cells reacting with human tumor cells in an intact (although murine, see below) tumor microenvironment. To do so, we transduced activated human T cells with a modified transgenic TCR (already being used in clinical trials (39)) that targets the tumor antigen NY-ESO-1 in the context of HLA-A2 restriction on tumor cells. These effector T cells were highly active in vitro. After injection into immunosuppressed mice bearing large human tumors, the T cells infiltrated the tumors, persisted, and proliferated. They were able to significantly slow tumor growth; however, they did not induce actual tumor regression. Importantly, after the transgenic TILs were isolated and interrogated for functional analyses ex vivo, we found them to be very hypofunctional. That is, when compared with infusion product (cryopreserved) T cells, the TILs were significantly inhibited in their ability to kill target tumor cells and to secrete IFNγ. Importantly, this hypofunction was reversible by isolating the TILs away from tumor and “resting” them overnight. This phenotype is very similar to that of actual human TIL as described above.
We believe this model offers unique opportunities to study human TIL biology. First, it provides the ability to isolate large numbers of TILs in a controlled, reproducible laboratory setting. Many clinical studies have encountered difficulties in isolating sufficient TILs for analysis due to difficulty in obtaining sufficient quantities of freshly-harvested tumor from the operating room. In addition, there is obviously a large amount of patient heterogeneity. Because of these issues, many previous patient studies have been limited to analyses of tumor-reactive T cells isolated from peripheral blood instead of from actual tumors (40, 41), an obviously suboptimal approach given the importance of the tumor microenvironment.
Second, as we are using T cells with a defined transgenic TCR, we can conduct ex vivo studies examining the reaction of the T cells to their known tumor antigen and look at their actual tumor cell killing ability, as well as stimulation of the T cells (i.e., cytokine release and CD107a upregulation). This is in contrast to the vast majority of patient-derived TIL studies that rely on nonphysiologic stimulation of TIL using agents such as CD3/CD28 antibodies or phorbol ester/ionomycin and can only examine surrogate markers of T-cell killing such as IFNγ secretion or CD107a upreguation by TILs.
Third, we can isolate tumor-reactive versus tumor-non-reactive T cells from within the same tumor microenvironment. As patient TILs are polyclonal with varying degrees of tumor reactivity, further analysis of the specific tumor-reactive TIL population is very difficult. In our model, we have just two distinct populations, tumor-reactive T cells assessed by TCRVβ13.1 positivity (the majority of the TILs due to enrichment in the tumor) or tumor nonreactive T cells (TCRVβ13.1–negative cells). Using flow cytometry, within one tumor sample, we can thus compare the phenotype of TCR-engineered TILs and compare them with TILs that do not bear the TCR transgene by flow cytometry. Isolation and genetic analysis of these populations is also possible.
Finally, and perhaps most importantly, a major opportunity of this model is the ability to study and manipulate the adoptively transferred human T cells within the tumor microenvironment over time, using anti-human directed reagents. Unlike observational human TIL studies, it is possible to alter the T cells themselves via genetic manipulation at the time of transduction or to administer various agents that affect the tumor microenvironment and/or the T cells themselves. As an example, we describe studies using an anti-human antibody directed to the immune IR PD1 (see below). However, ongoing studies in our lab using shRNA, dominant-negative constructs, and other antibodies are examining the effects of other types of IRs, chemokine receptors, phosphatases (such as SHP-1), T-cell enzymes (DGK and protein kinase A), and transcription factors (Ikaros). Manipulations of the tumor microenvironment can also be done, either prior to T-cell transfer, during T-cell transfer, or after T-cell infiltration to tumor. For example, we are studying the effects of cyclooxygenase inhibition or targeting tumor-associated fibroblasts to breakdown tumor stroma (42) prior to transfer of tumor reactive T cells.
In this article, we utilized our model to study cell surface T cell IRs. IRs operate at multiple levels to ensure appropriate T-cell homeostasis, activation, and differentiation (43). Many studies have demonstrated significant upregulation of IRs on human TILs that negatively affect their antitumor function (44, 45). More recently, studies have shown that Lag3 and PD1 can be coexpressed on tolerized TILs, suggesting that they may contribute to tumor-induced immunosuppression (33, 46). Consistent with these data from human TILs, when we compared the phenotype of the hypofunctional Ly95 CD4 and CD8 TILs with the infused, cryopreserved Ly95 T cells, there was significant upregulation of surface expression of PD1, Tim3, and Lag3 (Fig. 3). In addition, a significant percentage of TILs, particularly TCR+ TILs, had upregulation of more than one IR, with 25% of TCR+ TILs and 12% of TCR− TILs expressing PD1, Tim3, and Lag3 (Fig. 4). Of direct relevance to our model, Matsuzaki and colleagues showed that NY-ESO-1–reactive TILs from patients with ovarian cancer had high expression of both Lag3 and PD1 and were blunted in their ability to secrete cytokines (33).
Our model also allowed us to extend these observations and pursue a number of more novel and interesting questions. For example, we could study if the increases in PD1 and Lag3 expression were due to chronic TCR activation or external microenvironmental influences. We were able to address this question by using flow cytometry to compare the IR expression on the TILs expressing the TCR transgene versus those that did not in the same tumor. As shown in Fig. 3, all of the IRs increased in both the TCR+ TILs and the TCR− T cells, especially for Lag3 and Tim3. This suggests that the overall tumor microenvironment plays an important role. In contrast, the increase in PD1 was more prominent in the TCR+ TILs, indicating a more important role for chronic antigen stimulation for this IR. These data thus suggest that IR upregulation is a complex process that involves intrinsic factors such as stimulation of the TCR on the T cell, as well as microenvironmental factors (i.e., cytokines).
A second question that we were able to address in our model was the functional importance of TIL PD1 expression. As shown in Fig. 6, the addition of an anti-PD1 antibody to our ex vivo hypofunctional TILs was able to partially restore function. More importantly, we were able to test the effects of repeated injections of an anti-human PD1 antibody in animals in combination with the Ly95 T cells. As we hypothesized, injections of anti-PD1 antibody significantly augmented tumor control by the single injection of 10 million Ly95 T cells by over 30%. This finding is similar to what was published recently by John and colleagues in a model of murine ATC (47), however, to our knowledge, this is the first description of TCR-engineered human T cells being augmented by IR antibody blockade against a human cancer model. Mechanistically, although we did not note increased numbers of TILs within the treated tumors, we did observe that when isolated, the TILs from the anti-PD1 antibody-treated mice had enhanced ex vivo killing and cytokine function (Fig. 6).
These findings have some interesting translational implications for human IR therapy. Most importantly, our data support clinical trials using IR antibodies in combination with human T cell transfer approaches. This could apply to T cells expressing the same TCR used in our study, as NY-ESO-1–targeted transgenic T cells have shown promise in trials of sarcomas (13) and NY-ESO-1 is also expressed in up to 30% of lung cancer specimens (48). IR therapy could also be of value in other transgenic TCR approaches, for TIL therapy, and for chimeric antigen receptor-based therapies. However, the fact that anti-PD1 antibody augmented the slowing of tumor growth, but did not induce tumor regression, supports the idea that the TME-induced immunosuppression is multifactorial, and blockade of other inhibitory pathways or other inhibitory factors in a combinatorial fashion will need to be tested. Finally, our data suggest that analysis of the TILs, rather than just the tumor cells, may be useful in determining the choice of adjuvant IR blocking therapy and predicting response to therapy. In this animal model, we would predict that blockade of Lag3 and Tim3 might show added benefits. We have preliminary data that support this approach. This is consistent with other studies that have demonstrated that hypofunctional TILs express more than one IR (33, 44) and blockade of multiple IRs is needed to achieve optimal antitumor effects (45, 49)
Limitations of the model must also be acknowledged. The NSG mice lack a complete immune system, thus limiting the ability to study the effects of important tumor immune cells such at T-regulatory cells, NK cells, dendritic cells, B cells, and some myeloid cells. The innate immune cells present in NSG mice are not isologous with the human T cells and tumor cells. This could result in some species noncompatibility issues (i.e., some ligands in the murine microenvironment might not react with human T-cell receptors). For example, murine inteferons are not able to affect human tumor or T cells. The adoptive transfer of other human cell types (i.e., Tregs) or the use of mice with more fully “humanized” immune systems may be advantageous. In addition, nontumor–derived signals (even if murine) could potentially influence adoptively transferred human T cells as supported by descriptions of xenoreactive/homeostatic signals that can affect adoptively transferred cells (50, 51). Although possible, data support the observation that the upregulation of IR expression on the TILs are due to the tumor microenvironment and not just to nontumor xenoreactive/homeostatic signals. One piece of data is that the TCR-positive TILs had much higher proportions of double and triple IR positive cells than the TCR-negative TILs (Fig. 4). Perhaps even stronger data are that the infiltrating human lymphocytes isolated from the mouse spleens (which were subject to the same xenoreactive and homeostatic expansion as the tumor TILs) showed much less IR upregulation than did the TILs (Supplementary Fig. S4).
An area that has not yet been explored is the issue of tumor heterogeneity. Response to adoptively transferred T-cell therapy appears to vary among tumor types, even when they express the same targeted TAA (13, 39). Response to checkpoint blockade also varies among different tumors (52). It will thus be of interest to compare T-cell efficacy, hypofunction, and IR upregulation among different tumors. To address this, we are currently transfecting other human tumor cell lines with HLA-A2 and NY-ESO-1. We will then compare the degree of TIL hypofunction, expression of various IRs, and responses to anti-PD1 and other antibodies.
In summary, despite these potential limitations, the human Ly95 TILs isolated from the human tumors do seem to strongly resemble patient TILs suggesting that many of the key factors that induce T-cell hypofunction are present, without these other cell types. We thus believe the model will be useful in studying human adoptive T-cell transfer and further understanding the biology of tumor-reactive TILs that signal through an endogenous TCR. The model should also serve as an efficient and straightforward platform for preclinical testing of new human-specific immunotherapeutics prior to transition to the bedside.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: E.K. Moon, Y. Zhao, S.M. Albelda
Development of methodology: E.K. Moon, R. Ranganathan, S. Kim, X. Liu, Y. Zhao
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.K. Moon, R. Ranganathan, E. Eruslanov, S. Kim, K. Newick, S. O'Brien, A. Lo, X. Liu, Y. Zhao, S.M. Albelda
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.K. Moon, R. Ranganathan, E. Eruslanov, S. Kim, K. Newick, Y. Zhao, S.M. Albelda
Writing, review, and/or revision of the manuscript: E.K. Moon, R. Ranganathan, K. Newick, S.M. Albelda
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.K. Moon, R. Ranganathan, S. Kim, A. Lo
Study supervision: E.K. Moon, Y. Zhao, S.M. Albelda
The authors thank Drs. Linda Snyder and Raluca Verona of Janssen R&D for providing the anti-human PD1 antibody used for in vivo experiments.
The research was supported by funding from the NCI (NIH, K08 CA163941-04).
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.