DNT cells infiltrate patient lung adenocarcinoma and have cytotoxic function
To explore the role of DNT cells in human lung cancer, we analyzed treatment naïve resected lung adenocarcinoma tissue (Additional file 1: Table S1). Single cell suspensions were generated from resected tumor tissue as well as from matching adjacent and grossly normal appearing tissue and analyzed for T cell presence. Flow cytometric analysis detected a population of DNT cells and conventional CD4+ and CD8+ T cells (Fig. 1a). Interestingly, whereas comparable levels of CD4+ and CD8+ T cells were observed in normal (NOR), adjacent (ADJ) or tumor tissue (CA), DNT cell frequency was significantly reduced in tumor tissues compared to adjacent or normal lung tissues (CA:4.2 ± 0.2% vs ADJ:6.5 ± 0.6% and NOR:7.0 ± 0.7%, respectively; Fig. 1b). Based on co-staining of CD45RA and CD27, human T cells can be broadly categorized as effector memory (CD45RA−CD27−) or central memory (CD45RA−CD27+) subsets [29,30,31]. We found that tumor infiltrating DNT cells were predominantly central memory cells with no significant differences observed between different tissues (Fig. 1c). However, we did observe a significantly higher frequency of central memory phenotype amongst CD4 and CD8 T cells within cancer tissue relative to adjacent and normal lung tissue (Fig. 1d and e).
Given the presence of DNT cells in tumor tissue, we determined whether DNT cells derived from lung cancer patients have anti-tumor function. Using our well-established DNT cell expansion protocol, by which we have previously expanded DNT cells from peripheral blood of both leukemia patients [22] and healthy donors [24], we attempted to selectively expand DNT cells from tumor samples, but failed, possibly due to the low frequency of DNT cells obtained from tumors and/or exhaustion of the obtained DNT cells. However, DNT cells expanded from peripheral blood of lung cancer patients resulted in high purity (> 90%) but with a lower yield than those derived from healthy donors (Additional file 2: Figure S1). Importantly, DNT cells expanded from both lung cancer patients and healthy donors displayed potent and comparable cytotoxicity against established lung cancer cell line, NCI-H460, and patient xenograft-derived cell line XDC137 (Fig. 1f). Next, we compared the potency of anti-tumor activity mediated by CD4, CD8 and DN T cells expanded from the same donor in in vitro killing assays against the two cell lines and found that while all expanded T cell subsets showed cytotoxicity towards lung cancer cell lines, DNT cells induced the highest degree of cytotoxicity (Additional file 2: Figure S2A).
Ex vivo expanded DNT cells from healthy donors can target advanced late-stage lung cancer xenografts
To determine whether DNT cells can target late-stage lung cancer in vivo, we generated two late-stage xenograft models. An NSCLC established cell line NCI-H460 and a patient-derived adenocarcinoma xenograft cell line XDC137 were inoculated subcutaneously (s.c.) into the flanks of sublethally irradiated NSG mice and let grown to ~100mm3. Tumor-bearing mice were then treated subcutaneously with 3 peritumoral injections of ex vivo expanded DNT cells or CD8 T cells in 3–4 days intervals. For the more aggressive NCI-H460 model, the PBS treated control tumor reached end-point by 20 days post treatment (Fig. 2a). However, DNT cell treatment resulted in a significant reduction of tumor growth as early as 6 days post 1st DNT cell injection. At 20 days post DNT cell treatment NCI-H460 tumor volume was reduced by 43.3 ± 15.9%, from 834.2 ± 234.8 mm3 in the control group to 473.2 ± 132.9 mm3 in the DNT cell treated group (Fig. 2a). In contrast, injection of an equal number of CD8 T cells was not able to reduce tumor growth during this observation period (Additional file 2: Figure S2B). Additionally, DNT cell-mediated inhibition of tumor growth led to a significant increase in the survival of NCI-H460 tumor-bearing mice, with a humane endpoint extending from median 24 days to 38 days (Fig. 2b). Though patient-derived xenograft model XDC137 grew much slower than NCI-H460, with humane endpoint not being reached by 71 days of observation, DNT cell treatment significantly reduced XDC137 xenograft volume from 160.8 ± 39.5mm3 in the PBS control group to 86.2 ± 34.8mm3 in the DNT cell treated group (Fig. 2c), resulting in a 46.4 ± 21.6% reduction in tumor volume. These results show that adoptive transfer of healthy donor-derived DNT cells can significantly inhibit the growth of both aggressive and slow growing lung cancer xenografts. As DNT cells were found in lung cancer patient TILs, we next determined whether DNT cells would be detectable within tumor xenografts at experimental endpoints. Using immunohistochemical staining for human CD3+ cells, we detected DNT cells infiltrating both aggressive xenograft, NCI-H460 (Fig. 2d) and slower growing xenograft, XDC137 (Fig. 2e), at days 21 and day 71, respectively.
Tumor infiltrating and ex vivo expanded DNT cells express PD-1
With the observation that significantly fewer DNT cells were found in the patient TILs than in adjacent or normal tissue (Fig. 1b), we hypothesized that the immunosuppressive tumor microenvironment may prevent DNT cell infiltration. Consistent with this hypothesis, PD-1 was expressed on DNT cells within resected lung tissue, similar to that seen for CD4+ and CD8+ T cells (Fig. 3a). Further, a significantly higher proportion of DNT cells expressed PD-1 within tumors compared to adjacent or normal tissue (CA: 55.5 ± 11.7% vs ADJ: 36.1 ± 14.5% and NOR: 35.5 ± 9.1%). Though tumor-infiltrating DNT cells expressed PD-1, they were the least frequent PD-1+ T cell subset and showed the most variability in PD-1 expression compared to CD4+ and CD8+ T cells (CD4: 65.8 ± 7.1%, CD8: 67.2 ± 7.2%, DNT: 55.5 ± 11.7%, Fig. 3b).
Since patient-derived DNT cells induced a similar level of cytotoxicity against lung cancer cells as those from healthy donors (Fig. 1f), and DNT cells expanded from healthy donors possess features allowing them to be used as an “off-the-shelf” ACT [32], we utilized healthy donor DNT cells to understand the role of PD-1 expression on DNT cells. Prior to expansion, PD-1 expression varied amongst donors (Fig. 3c, day 0 of expansion). Upon expansion, donor DNT cells followed a similar expression profile: sharply increasing expression of PD-1 at day 3 of culture then gradually returning to baselines by day 17 (Fig. 3c). We observed a similar trend of PD-1 expression for CD8 T cells expanded in this manner. In contrast, CD4 T cells maintained a significantly higher level of PD-1 expression than DNT and CD8 T cells from day 10 until the end of the expansion culture (Fig. 3d). Given that PD-1 expression was higher in tumor infiltrating DNT cells than those in adjacent or normal lung tissues (Fig. 3a), and lung cancer cell lines express different levels of PD-L1 (Additional file 2: Figure S3A), we determined if co-culture of DNT cells with lung cancer cells was sufficient to induce PD-1 expression. Consistent with the observations in patients, in vitro coculture with 4 different PD-L1+ lung cancer cell lines (A549, H460, H520 and XDC137, Additional file 2: Figure S3A), all resulted in a significant increase in PD-1+ DNT cells when compared with DNT cells cultured alone (Fig. 3e and Additional file 2: Figure S3B). PD-1 induction was not dependent on the level of PD-L1 expression on lung cancer cells as H520 expressed the lowest level of PD-L1 (Additional file 2: Figure S3A) but induced a similar level of PD1+DNT cells as H460 which showed a very high level of PD-L1 expression (Fig. 3e and Additional file 2: Figure S3A). Prolonged co-cultures with lung cancer cells did not further increase PD-1+ DNT cells for any given cell line (Additional file 2: Figure S3B). Co-culture with lung cancer cell lines also increased intracellular expression of IFNγ and TNFα in DNT cells (Additional file 2: Figure S4), suggesting the activation of these T cells by lung cancer cells.
Anti-PD-1 treatment enhances DNT cell-mediated anti-tumor activity
With the propensity of DNT to upregulate PD-1 and cytokines expression in the presence of lung cancer, we sought to determine if addition of anti-PD-1 may augment DNT cell-mediated anti-tumor activity in vivo. To observe whether anti-PD-1 can benefit adoptive DNT therapy in vivo, PD-L1 expressing NCI-H460 lung cancer cell line was subcutaneously implanted and established to ~ 100 mm3 and DNT cells, with or without anti-PD-1, were administered using two methods, either locally by s.c. peritumoral injection or systemically by intravenous (i.v.) tail vein injection as shown schematically in Fig. 4a and Additional file 2: Figure S5A, respectively. Anti-PD-1 treatment alone had no effect on tumor growth compared with PBS treated controls (Additional file 2: Figure S6) and consistent with Fig. 2a, peritumoral infusion of DNT cells significantly reduced NCI-H460 tumor volume from 922.1 ± 164.2 mm3 in the control group to 546.5 ± 125.7 mm3 in the DNT cell treated group, resulting in a 40.7 ± 13.6% reduction in tumor volume. Interestingly, the combination of DNT cell injection with anti-PD-1 resulted in an additional 43.1 ± 29.4% reduction of tumor volume (from 546.5 ± 125.7 mm3 in the DNT cell alone treated group to 310.7 ± 160.9 mm3 in the combination group) by day 20 (Fig. 4b). Similarly, systemic i.v. infusion of DNT cells also significantly reduced NCI-H460 tumor volume from 1017.49 ± 246.2 mm3 in the control group to 572.5 ± 186.5 mm3 in the DNT cell treated group, resulting in a 43.7 ± 18.3% reduction in tumor volume, and the combination therapy of i.v. inoculated DNT cells and anti-PD-1 treatment resulted in an additional 32.6 ± 20.0% reduction in tumor volume (from 572.5 ± 186.5 mm3 in the DNT cell alone treated group to 385.9 ± 114.3 mm3 when in combination) by day 20 (Additional file 2: Figure S5B). Importantly, combination therapy prolonged the survival of both s.c. peritumoral inoculated DNT cell treated mice from median 38 days to 48.5 days (Fig. 4c) and i.v. inoculated DNT cell treated mice from median 33 days to 38 days (Additional file 2: Figure S5C). Analysis of hematoxylin and eosin (H&E) stained tumor tissue shortly after DNT treatment revealed that though tumor size remained similar (181.0 ± 53.7 mm3 for DNT cell treated vs 152.2 ± 54.7 mm3 for DNT cell and anti-PD-1 treated), anti-PD-1 significantly increased the proportion of necrotic area detected within tumors from mice receiving combination treatment (64.9 ± 11.7% vs 41.3 ± 14.5%; Fig. 4d), with a similar result observed for i.v. inoculated DNT cells (42.1 ± 10.4% vs 22.4 ± 7.2%; Additional file 2: Figure S5D). These results suggest that DNT cells inhibit tumor growth by actively targeting tumor cells and causing tumor necrosis, and that this activity was enhanced by anti-PD1 therapy. Overall, these results show that addition of anti-PD-1 augments the ability of DNT cells to reduce tumor growth and increase survival of mice.
Anti-PD-1 treatment increases DNT cell infiltration into tumor xenografts
To understand how anti-PD-1 augmented DNT cell-mediated tumor growth inhibition, we first determined whether the presence of anti-PD-1 altered in vitro cytotoxicity of DNT cells to lung cancer cell lines expressing different levels of PD-L1 (Additional file 2: Figure S7A). We found that addition of anti-PD-1 to the cocultures did not alter DNT cell cytotoxicity towards lung cancer cell lines H460, XDC137 and A549 natively expressing PD-L1, but significantly increased killing of PD-L1 overexpressing cell line A549-PD-L1 (Additional file 2: Figure S7B). To analyze how anti-PD-1 enhanced DNT cell treatment towards lung cancer xenografts in vivo we analyzed tumor infiltrating DNT cells post treatment. Consistent with PD-1 induction on DNT cells by lung cancer in vitro (Fig. 3e), flow cytometric analysis of xenograft infiltrating DNT cells showed a 2-fold increase in PD-1 expression compared to DNT cells prior to infusion (Fig. 5a). Further, anti-PD-1 treatment abrogated PD-1 expression on xenograft infiltrating DNT cells as shown by the lack of staining using anti-PD1 clone EH12.2H7 that recognizes a Nivolumab shared epitope of PD-1 [33, 34] (Fig. 5a), suggesting that the Nivolumab treatment effectively blocked the PD-1 epitope on tumor infiltrating DNT cells.
To determine whether anti-PD-1 treatment affects tumor infiltration of DNT cells, we quantified DNT cell infiltration of tumor xenografts by histological analysis. Mice receiving combination treatment of DNT cells and anti-PD-1 antibody had a 5.9 ± 1.2-fold increase in the number of tumor infiltrating DNT cells relative to mice that received DNT cells alone (Fig. 5b). Similarly, i.v. infusion of DNT cells also resulted in a 1.7 ± 0.3-fold increase in DNT cells accumulating in tumor xenografts (Additional file 2: Figure S5E). These data indicate that anti-PD-1 treatment can increase the accumulation of DNT cells in tumor tissue. We next analyzed whether anti-PD-1 treatment could alter the phenotype of tumor infiltrating DNT cells. To this end, tumor infiltrating DNT cells were isolated from mice receiving different treatments and expression of cytolytic molecules known to be involved in DNT cell anti-tumor responses were analyzed by flow cytometry [24, 25, 35]. We found that DNT cells expressing NKG2D and DNAM1 were present in both control and anti-PD-1 treated mice but were more abundant in mice receiving combination therapy than those receiving DNT cells alone, though differences did not reach statistical significance (Fig. 5c). Similarly, mice that received anti-PD-1 showed a greater number of TNFα+ and IFNγ+ DNT cells in the tumor (Fig. 5d). Importantly, consistent with the cytotoxic nature of DNT cells, anti-PD-1 treatment significantly increased the frequency of CD107a+, perforin+, and granzyme B+ DNT cells within tumors (Fig. 5e). These data suggest that anti-PD-1 treatment increases the accumulation of DNT cells within tumors expressing molecules involved in anti-tumor responses.