Naïve CD8+ T cell derived tumor-specific cytotoxic effectors as a potential remedy for overcoming TGF-β immunosuppression in the tumor microenvironment

Despite of the potential implications for cancer immunotherapy, conventional approaches using in vitro expanded CD8+ T cells have suboptimal outcomes, mostly due to loss of functionality from cellular exhaustion. We therefore investigated the phenotypic and functional differences among in vitro activated CD8+ T cells of three different sources, namely naïve (NTeff), memory (MTeff) and tumor-infiltrating lymphocytes (TILeff) from human and mice, to better understand mechanisms behind potent effector functions and potential for overcoming current limitations. In line with the greater proliferation activity and longer telomere lengths of NTeff populations, cells of naïve origin exhibited significantly less amounts of T cell exhaustion markers than those of MTeff and TILeff, and moreover, acquired distinct expression patterns of memory-promoting transcription factors, T-bet and Eomes, induced in a rapid and sustainable manner. NTeff cells appeared to have lower expression of Foxp1 and were refractory to apoptosis upon TGF-β conditioning, implying better survival potential and resistance to tumor-induced immune suppression. Of CD8+ T cell pools activated to tumor-specific CTLs, naïve cell generated effectors possessed the most potent cytotoxic activity, validating implications for use in rational design of adoptive immunotherapy.


Results
Successful expansion of human naïve, memory, and TIL CD8 + T cells in vitro. CD8 + T cell subsets could be defined by the surface markers CD62L, CD45RO, CCR7, and CD45RA to naïve (CCR7 + CD45RA + or CD62L + CD45RO − ) and memory (CD45RA − or CD45RO + ) cells. Tumor-infiltrating CD8 + T cells were CD44 + , mostly with loss of homing receptors CD62L and CCR7 expression, and composed of two main sub-populations CD45RA − and CD45RA + , suggesting a relatively differentiated status. After MACS isolation and stimulation by CD3/CD28 Dynabeads on anti-CD2 coated plates, successful expansion was demonstrated in all three groups by either CFSE peak division of NT eff and MT eff subsets, or enlarged populations of CD44 + CD8 + TIL eff cells (Fig. 1A). CFSE staining of TIL eff proliferation was observed but peak division was indistinct due to the small number of initial TILs (data not shown). Activation kinetics differed for cell subsets where exponential expansion and accumulation was observed starting at day 3 for NT eff and later at day 5 for MT eff and TIL eff cells, and thus further analyses of all effector cells were performed on day 5 post-stimulation.
Phenotypic characteristics of effector cell populations were assessed by surface expression of activation markers CD62L, CD25, CD44, and OX40. Effector cells from all progenitors (naïve, memory, and TILs) exhibited an effector phenotype with significant up-regulation of activation markers in comparison to their progenitors, and were considered activated as CD62L − CD25 + CD44 + OX40 + populations (Fig. 1B). Despite variability in the percentage of effector cells produced among donors, a significantly higher percentage of OX40 expressing cells was observed for NT eff in comparison to MT eff (p < 0.05) or TIL eff (p < 0.005) cells (Fig. 1C). To further compare the proliferative potential among cell populations, relative telomere length (RTL), which correlate with replicative capacity, of NT eff , MT eff , and TIL eff cells were investigated against CCRF-CEM control cell line. Telomere length was greatest in NT eff , shorter in MT eff , and shortest in TIL eff cell subsets (Fig. 1D). This result was consistent with previously published reports of longer telomeres in human naive T cells, leading to increased proliferation of NT eff cells in comparison to MT eff cells after in vitro stimulation 9 .
Exhaustion phenotypes differ among in vitro generated human and murine effector cells NT eff , MT eff , and TIL eff . Inhibitory receptors on T cell surfaces such as PD-1, CTLA-4, and KLRG-1, have been shown to facilitate T cell exhaustion by interaction with ligands on antigen presenting cells or tumor cells [3][4][5] . We therefore compared the expression of these inhibitory receptors on the three effector cell subtypes. All effector cells showed significant increase of exhaustion phenotypes during proliferation, but NT eff cells showed significantly less expression of PD-1 and CTLA4 compared to MT eff and TIL eff subsets ( Fig. 2A). Check point inhibitors showed varying levels of expression dependent on the time elapsed from activation with peak expression on days 4-5 for PD-1, days 5-7 for CTLA-4, and days 4-7 for KLRG-1 (data not shown). To investigate functional CFSE assay showed positive proliferation results after 3 to 5 days of culture. (B) Activation markers of effector cell subsets. Effector naïve and memory CD8 + cells (NT eff and MT eff , respectively) are characterized by CD62L -CD25 + CD44 + OX40 + expression. Re-stimulated CD8 + TIL (TIL eff ) populations also showed similar phenotypes. (C) OX40 + cell percentage of effector cell subsets. NT eff showed significantly higher values compared to MT eff and TIL eff (* P < 0.05 vs TIL eff , * * P < 0.005 vs MT eff ). (D) Telomere length comparison showed the longest telomeres in NT eff and shortest in TIL eff cells (* P < 0.05). RTLs of effector cells were measured against CCRF-CEM control cells (1:1 ratio, total 5.0 × 10 5 cells) with a DAKO Telomere PNA Kit. Data are representative of three to four independent experiments and are presented as mean ± SD. Effector cells (OX40 + ) derived from naïve cells (NT eff ) expressed lower levels of exhaustion markers PD1, CTLA4, KLRG1, and LAG3 compared to effectors from memory cells or TILs (MT eff or TIL eff ) (A) in human (* P < 0.05 for PD-1, * * P < 0.005 for CTLA4) and (B) in mice at 5 days post-stimulation with anti-CD3/CD28 and anti-CD2. MFI levels are the average result of three experiments and are presented as mean ± SD. (C) In vitro functional analysis of human effector cells by cytokine production. An equal number of progenitors (2 × 10 5 cells per subtype) activated to OX40 + effectors were characterized by flow cytometry for direct comparison among subtypes. NT eff cells demonstrated a greater increase in perforin + granzyme B + populations compared to MT eff and TIL eff subsets at two time points (day 3 and 5 post-stimulation). Data are representative of three independent experiments. relevance of exhaustion phenotypes, we then evaluated the secretory function of cytotoxic cytokines such as granzyme B, perforin, and IFN-ɤ from different human effectors. During days 3 to 5 post-stimulation, the secretion of granzyme B and perforin gradually increased in all effector cells derived from naïve, memory, and TIL populations (Fig. 2C). Among fully activated day 5 effector cells, NT eff possessed the highest expression level of perforin and granzyme B, and similar levels of IFN-ɤ (data not shown) compared to MT eff and TIL eff cells. In the murine model, effectors derived from CD8 + naïve and memory phenotype (MP) T cells, and TILs collected from OVAp-expressing EL4-EG7 tumors were successfully proliferated with CD44 + OX40 + phenotypes. Expression levels of inhibitory receptors PD-1, KLRG1, and LAG3 in murine CD8 + effectors showed similar patterns to those of human cells (Fig. 2B).

Expression kinetics of transcription factors T-bet and Eomes differ among CD8 + T cell effector populations.
For further characterization of the CD8 + T cell subpopulations, the expression levels of two T-box transcription factors, T-bet and Eomes, were examined using intracellular staining for FACS analysis and confirmed with Western blot. Although naïve CD8 + T cells expressed lower levels of T-bet and Eomes in comparison to memory cells before stimulation, activated effectors NT eff , MT eff , and TIL eff showed relatively similar MFI levels of T-bet and Eomes, demonstrating the capacity of naïve or memory cells to induce similar effector populations (Fig. 3A). However, considering the lower MFI for pre-stimulated naïve cells, this cell population displays a greater increase in T-bet and Eomes expression levels post-stimulation compared to memory T cells. To clarify and confirm the expression profile of these T-box transcription factors in the effector subpopulations, Western blot analysis was performed using whole cell lysates of naïve and memory T cells, and TILs on days 0, 3, 5 and 8 post TCR activation. Our experiments revealed clear differences in the kinetics of T-bet and Eomes expression during CD8 + T cell activation (Fig. 3B). Effectors derived from naïve cells increased in T-bet expression appreciably starting on day 3, reaching maximum on day 8, whereas memory cell derived effectors gained the highest expression of T-bet earlier on day 3 and were quickly down-regulated on days 5 through 8. TIL eff cells demonstrated poor expression of T-bet throughout stimulation time. While naïve and memory cells showed increasing Eomes expression with stimulation, TILs exhibited decreasing intensity during day 5 through 8 post-activation. Such kinetics suggest that NT eff cells express higher levels of both T-bet and Eomes, and expression is sustained for longer periods during activation, in comparison to other effector groups MT eff and TIL eff . Collectively, naïve T cell derived effectors appear to differentiate relatively rapidly, proliferate actively, and have potential to be converted into long-lived memory cells.
Foxp1 expression and apoptosis studies under TGF-β conditioning show favorable results for CD8 + NT eff cells. To determine the immunosuppressive resistance potential of CD8 + NT eff cells in the tumor microenvironment, we examined expression of Foxp1 among effector cells derived from naïve, memory, and TIL progenitors in the presence of TGF-β . After 5 days of proliferation, effector cells were exposed to TGF-β (10 ng/mL) for 48 hours and Foxp1 expression (day 7 post-stimulation) was evaluated by intracellular FACS staining. MT eff and TIL eff populations showed increased Foxp1 expression with TGF-β conditioning, whereas NT eff cells maintained their lower expression levels in three independent experiments (Fig. 3C). To confirm the increased expression of Foxp1 in CD8 + effector cells in the tumor microenvironment, Western blot analysis was performed for effector T cell subsets under TGF-β exposure using non-stimulated naïve cells as the control. As expected, NT eff populations expressed the lowed amount of Foxp1 protein expression relative to other effector groups, consistent with intracellular staining results (Fig. 3D). After conditioning with TGF-β , flow cytometry detection of apoptosis with Annexin V displayed limited induction of apoptotic cell populations for NT eff cells relative to MT eff and TIL eff subgroups (Fig. 3E). Taken together, these results suggest that CD8 + effector subgroups differ in Foxp1 expression and apoptotic response to TGF-β exposure, and that NT eff cells maintain resistance to inhibitory factors induced from the tumor microenvironment.

Cytotoxic activity against targeted tumor cells is superior for effector cells of naïve T cell origin in both human and mouse models.
To investigate the cytotoxic function of tumor-specific human CTLs, antigen (U266 MM cell line) loaded autologous mature dendritic cells (mDCs) were co-cultured for 7 days with IL-2 primed naïve and memory CD8 + T cells and the generated CTLs were assessed via IFN-ɤ ELISPOT assay and LDH Cytotoxicity Detection Kit. A total of five cell groups were harvested for cytotoxic analyses including CD8 + (Naïve D14, Naïve D7, Naïve D0, and Memory) and CD3 + CTLs. Briefly, Naïve D14 and Naïve D7 cells refer to CD8 + naïve cells co-cultured with high-dose IL-2 (1 μ g/mL) for 14 and 7 days respectively, whereas Naïve D0 and Memory cells are co-cultured with conventional dose IL-2 (10 ng/mL) and are of naïve and memory T cell origin, respectively (refer to Methods). IFN-ɤ ELISPOT was measured in the presence and absence of MHC class I specific monoclonal antibody (W6/32, 20 µg/mL) with CTL alone as the control. High dose IL-2 primed Naïve D14 effector cells showed the highest secretion of IFN-ɤ against target U266 cells when co-cultured for one day at a 10:1 ratio of effector (E) to target (T) cells (Fig. 4A.). Although CD3 + cells also secreted higher levels of IFN-ɤ relative to Naïve D7, Naïve D0, and Memory CD8 + cells, they also displayed non-specific secretion as observed in the control CTL alone group. Direct cytotoxicity testing against U266 target cells using the LDH cytotoxic assay revealed enhanced CTL activity for Naïve D14 cells that was significantly greater than other effector CTLs when E:T ratios were at least 10:1 (Fig. 4B.). In the murine model, in vivo CTL activity against inoculated OVAp-expressing EL4-EG7 lymphoma cells in B6 mice was examined for NT eff , MT eff , and TIL eff CD8 + effector subsets obtained from OT-1 Thy1.1 mice. Tumor growth was negative for at least 35 days after tumor inoculation in mice injected with NT eff cells, whereas MT eff , and TIL eff injected mice showed positive tumor growth within the same time frame (Fig. 4C). Tumor growth became positive in one NT eff injected mouse at 42 days post inoculation (data not shown).

Discussion
Current adoptive immunotherapy involves mainly the use of CD3-collected T cells or autologous genetically engineered TILs, with increasing prospect for memory CD8 + cells 18,19 . Autologous CD8 + subpopulations hold the advantage of avoiding contamination of immunosuppressive regulatory T cells and decrease in terminal differentiation with repeated clonal proliferation in cancer patients. In order to demonstrate the superior background of tumor-specific CTLs derived from naïve CD8 + populations we focused on clarifying the different exhaustion characteristics, including expression check point inhibitors and transcription factors, and divergence of cytotoxic function among different effector groups NT eff , MT eff , and TIL eff . We found that naïve CD8 + effectors have relatively favorable resistance patterns to immunosuppression and strong CTL potency against tumor targets. This may be explained by the "fresh" nature of these cell groups that display lower levels of inhibitory immune checkpoint receptors such as PD-1, CTLA-4, and KLRG1, leading to a greater activation and replicative potential 20 . Our results were consistent with reports by Hinrichs et al. 21 indicating effectors derived from naïve CD8 + T cells may be the preferable candidate for T-cell adoptive immunotherapy based on evaluation of KLRG1, CD57, CD27 expression and telomere length. Such markers represent tendency of MT eff or TIL eff cells to quickly enter into a T cell "exhaustion" phase with elevated levels of inhibitory receptors following long-term exposure to tumor antigen and ultimately acquire a non-functional state 4,22,23 .
Another point of consideration is the expression of two T-box transcriptional factors, T bet and Eomes, as well as their effect on CTL differentiation. Among various transcription factors that function in pairs to regulate effector and memory CD8 + T cell development, short-and long-term effects of T-bet and Eomes have been relatively well identified in the murine model, but are still limited to the context of infection and disease in humans 16,[24][25][26][27] . Previous studies have emphasized on the role of high T-bet expression, which is associated with long-term resilience, low expression of inhibitory receptors, and protection from CD8 + T-cell exhaustion 27,28 , as well as Eomes induced inhibition of cell death leading to generation or survival of memory cells 17,29 . In our experiments, data showed that NT eff cells gradually increase and maintain high expression levels of both T-bet and Eomes throughout activation, whereas other effectors quickly down-regulated these factors. It has been suggested that high expression of T-bet correlates with a granzyme B + perforin + phenotype where high perforin secretion was only detected in T-bet hi Eomes hi/lo cells 15 . Stable expression of Eomes levels in NT eff and MT eff subsets show their ability to become memory-precursor effector cells (MPECs), whereas TIL eff cells enter a terminal differentiation phase via down regulation of Eomes. Collectively, our data indicates that NT eff populations co-express sustained, high levels of T-beta and Eomes, and a granzyme B + perforin + phenotype representative of full cytotoxic function.
To clarify the relationship between patterns of Foxp1 expression and TGF-β , we activated T cells by in vitro TCR stimulation and evaluated Foxp1 levels of these effectors before and after TGF-β exposure with intention to mimic the immunosuppressive TME. As shown in a recent article 7 , overexpression of Foxp1 is thought to be mediated by TGF-β signaling through Smad protein interaction, which in turn mediates c-Myc and c-Jun transcriptional repression, and is expected to be upregulated after homing to the TME. However, in this study, only whole mouse CD8 + cells were investigated and did not address subsets of human CD8 + T cells. As inferred from our results, CD8 + NT eff are less influenced by tumor-derived inhibitory factor TGF-β , i.e., effectors derived from naïve cells are less affected by tumor suppression mechanisms driven by Foxp1 in response to TGF-β . Importantly, 48 hours post treatment of TGF-β , a smaller percentage of NT eff cells entered apoptosis compared to MT eff and TIL eff subsets, which further offers evidence for the superior features of naïve cell derived effectors in resistance to the TGF-β induced apoptosis effect. These findings provide indication for the use of TGF-β inhibitors 29 , NT eff engineered to express a dominant-negative mutant of the TGF-β receptor 30 , and Foxp1-deficient NT eff , with objectives to enhance the immunosuppressive resistance of lymphocytes in the TME in various clinical applications.
The essential role of high-dose IL-2 in cancer has now been widely discussed, especially because IL-2 is suspected to optimize all stages of CD8 + T cell response, including primary expansion, contraction, memory

. Tumor-specific CTL function of in vitro human and in vivo murine CD8 + T cells. (A)
In vitro comparison of tumor-specific CTL function of IL-2 primed human CD8 + T cells against U266 MM cells by IFN-ɤ ELISPOT assay. CTL alone represents the control group and U266 refers to cytotoxic activity of effectors against U266 target cells (E:T ratio of 10:1) in the absence of MHC class I restriction, expressed as the number of IFN-ɤ spots observed. Naïve T cells activated under high-dose IL-2 for 14 days (Naïve D14) displayed highest activity against U266 target cells. (B) LDH cytotoxicity assay of IL-2 primed human CD8 + T cells. Specific cytotoxicity lysis percentages against U266 MM target cells were highest for Naïve D14 effectors compared to other groups when E:T ratios were at least 10:1 (* p < 0.05). (C) In vivo CTL activity in the murine model. Following injection of NT eff , MT eff , and TIL eff CD8 + T cells, OVAp-expressing EL4-EG7 cells (8 × 10 5 cells/200 µ L) were inoculated subcutaneously in B6 recipient mice in a 1:2 ratio of target tumor to effector cells. All B6 mice injected with TIL eff had positive mass (1 cm 3 ) formation after 28 days, whereas tumor growth was suppressed in NT eff harboring B6 mice for longer than 35 days. All data are representative of three independent experiments and shown as mean ± SD.
Scientific RepoRts | 6:28208 | DOI: 10.1038/srep28208 generation, and secondary expansion 31 . In addition, high-dose IL-2 may strongly induce the proliferation as well as stimulation of naïve CD8 + T cells both in vitro and in vivo 32 . In our research, high-dose IL-2 (1 μ g/mL) stimulated naïve CD8 + T cells, which were then co-cultured with autologous DCs, could successfully generate strong CTLs within a short duration. By evaluation of ELISPOT and LDH assay, Naïve D14 and Naïve D7 both generated CTLs with a high number of IFN-ɤ spots and better (Naïve D14) or similar (Naïve D7) cytotoxic effects compared to CD3 + CTLs. As mentioned, high concentrations of IL-2 (0.1-1 μ g/mL) are necessary for stimulation of CD8 + naïve T cells through the antigen-independent manner 13 . Although our antigen-loaded DCs could induce antigen-specific CTLs from CD8 + naive T cells, higher than conventional dose IL-2 was required not only for proliferation of the limited number of antigen-specific naïve CD8 + T cells but also for generation of potent CTLs. During cytokine stimulation, high-dose IL-2 plays an important role in both pre-activation (during 7 days co-culture with naïve T cells) and co-activation (during 7 days co-culture with IL-2 primed T cells and DCs). In the co-activation period, high-dose IL-2 could potentially be used with other ɤ c cytokines (IL-7, IL-15) for greater generation of CTLs. However, the optimal exposure times of high-dose IL-2 to tumor-specific naïve CD8 + T cells remains to be determined.
Taken together, our results provide a relatively complete framework for collecting effector cells derived from naïve origin and present a practical method to effectively generate strongly functional NT eff populations with the use of high-dose IL-2 (1 μ g/mL). Nowadays, human T lymphocytes, which can be genetically engineered prior to adoptive transfer, can express virtually any target gene, by application of techniques involving genes encoding T-cell receptors (TCRs) or chimeric antigen receptors (CARs) to provide the desired T-cell specificity 11 . CAR-redirected T lymphocytes (CAR-T cells) are mostly limited to memory and effector T cells [33][34][35] due to potent cytotoxic function. However, the appropriate substrate of effector T cells for clinical use is still controversial and the development of antigen escape variants may represent the limitation of monoclonal specificity. To obtain an adequate amount of effector cells with exceeding cytotoxic function, we suggest that naïve cells should be considered as a candidate for CAR-T therapy.
In summary, primary effector cells derived from naïve CD8 + T cells showed the highest potential for adoptive cell transfer therapy, with lower expression of inhibitory surface markers, higher secretion of cytotoxic cytokines, and improved resistance to TGF-β induced suppression in the TME, in comparison to secondary effectors derived from memory or TIL CD8 + cells. High-dose IL-2 priming amplifies the number of tumor-specific naïve CD8 + T cells without clonal exhaustion and results in the generation of potent tumor-reactive CTLs with capacity to overcome tumor-derived immune suppression.

Methods
Ethical approval of studies and informed consent. All experimental protocols were approved by the institutional ethical committee at ChonnamNational University Hwasun Hospital (CNUHH2014-146) and methods were carried out in accordance with the approved guidelines. Human peripheral blood and tumor masses were donated from healthy donors or lung cancer patients after obtaining written informed consent from all subjects. Animal experiments were performed with ethical approval from the Chonnam National University Animal Research Committee (HCRL15001-2).

Isolation of human CD8 + T cell subsets and tumor-infiltrating lymphocytes (TILs).
Peripheral blood mononuclear cells (PBMCs) were obtained from healthy donors by density gradient centrifugation with Lymphoprep (AXIS-SHIED Rodelokka, Oslo, Norway). CD3 + and CD8 + lymphocytes were isolated by a magnetic-activated cell sorter (MACS) using CD3 Microbeads and a CD8 T cell isolation kit (MiltenyiBiotec), respectively. The cells were then labeled with anti-CD8, anti-CCR7, and anti-CD45RA (Ebiosciences) and sorted into naïve and memory CD8 + T cell subsets on a BD fluorescence activated cell sorting (FACS) Aria sorter (BD Biosciences). For extraction of TILs, tissue from lung tumor samples were minced and digested with collagenase (2.5 mg/ml collagenase I) at 37 °C for one hour. Cell suspension was then twice filtered through 100-μ m and 40-μ m cell strainers (BD Biosciences) to obtain single cells. TILs were isolated from single cells utilizing density gradient centrifugation with Lymphoprep and further purified using CD8 MicroBeads (MiltenyiBiotec). Surface and intracellular staining using FACS. For surface staining, cells were labeled with mAbs against various targets including CD8, CD62L, CCR7, CD45RA, CD45RO, CD25, CD44, OX40, PD-1, CTLA-4, and KLRG-1 (Ebiosciences), incubated at 4 °C for 15 minutes, then washed twice with PBS 1% FBS (FACS buffer), and finally fixed in PBS containing 1% paraformaldehyde (Fix buffer). For T-bet (4B10) and Eomes (Dan11mag), and Foxp1 intracellular staining, cells were first labeled with surface markers CD8, CD62L, OX40 (Ebiosciences) for T-bet and Eomes, or CD25 (BD Biosciences), CD8, CD27, and CD44 (Ebiosciences) for Foxp1 detection, respectively, and then fixed and permeabilized with Foxp3/Transcription Factor Fixation/Permeabilization Concentrate and Diluent (Ebiosciences) in a 96-round-well plate. Intracellular labeling for T-bet and Eomes (Ebiosciences), and Foxp1 (LifeSpan Technologies) was performed according to manufacturer's instruction. For evaluation of Perforin, Granzyme B, and IFN-ɤ cytokine secretion, effector cells were incubated with brefeldin A for 4 hours at 37 °C to disrupt Golgi-mediated transport and accumulate cytokines. Cells were then surface Generation of human tumor-specific CTLs. Mature dendritic cells (mDCs) were pulsed with apoptotic body (irradiated U266 cell line) at a 2:1 ratio (2 × 10 5 DC: 10 5 U266) from day 6 to day 8 and then presented to CD8 + T cells at a 1:10 ratio (2 × 10 5 DC: 2 × 10 6 CD8 + T cell). Naïve and memory T cells were isolated from the same donor and cultured in different conditions. To visualize cytokine-activated proliferation and determine the optimal co-culturing start time with antigen-loaded mDCs post high-dose IL-2 activation, activated naïve CD8 + T cells were labeled with CFSE and analyzed at various time points (Supplementary Fig. S1). Initiation of peak expansion at 7 days post high-dose IL-2 lead to the decision to start co-culture of cells with mDCs after 7 days of high-dose IL-2 pretreatment. Naïve cells were activated with high-dose IL-2 (1 µg/mL) in two conditions; 14 day exposure to high-dose IL-2 (Naïve D14; 7 days pretreatment and during 7 days of mDC co-culture) and 7 day exposure to high-dose IL-2 (Naïve D7; 7 days pretreatment only followed by 7 days of mDC co-culture without high-dose IL-2). For control, we cultured naïve and memory cells under equal conditions; 7 day mDC co-culture with IL-2 (10 ng/mL) and IL-7 (10 ng/mL), labeled Naïve D0 and Memory, respectively. After mDC co-culture, 4 groups of CD8 + CTLs (Naïve D14, Naïve D7, Naïve D0, and Memory) and CD3 + CTLs (cultured as previously described) were harvested for cytotoxic analysis.

Generation of human CD8
Cytotoxic functional evaluation of human tumor-specific CTLs. The U266 cell line was chosen as target cells for our cytotoxic assays based on its expression of E-Cadherin and PD-L1 ( Supplementary Fig. S2), which are ligands of KLRG-1 and PD-1, respectively. ELISPOT assay (BD Biosience) was performed to quantify antigen-specific IFN-ɤ releasing effector T cells. 2 × 10 5 of the effector T cells were co-cultured with 2 × 10 4 U266 target cells in a 96-well nitrocellulose flat-bottom plate for 24 hours at 37 °C. IFN-ɤ ELISPOT was measured in the presence or absence of MHC class I specific monoclonal antibody (W6/32) with CTLs alone as the control, and spots were counted with ImmunoSpot Reader (Cellular Technology Ltd, Ohio). Data is presented as the mean number of spots ± SD of IFN-ɤ secreting cells per well of triplicate samples.
Evaluation of CD8 + and CD3 + CTL functional activity was performed using the Cytotoxicity Detection Kit LDH (Roche Applied Science, Basel, Switzerland) according to the manufacturer's protocol 36,37 . Cytotoxicity of CTLs was calculated according to the following formula: % cell lysis = (experimental -effector spontaneouslow control) x 100/(high control -low control), where "experimental" corresponds to the experimental signal value, "effector spontaneous" to the spontaneous background signal value of the effector cells alone, "low control" to the spontaneous background signal value of target tumor cells alone, and "high control" to the maximum signal value of target cells in medium containing 1% Trixton X-100. Cytotoxicity assays were performed for varying effector cell (E) to target cell (T) ratios, and specific cytotoxicity lysis percentages are reported as the mean ± SD of triplicate samples.
T cell purification, in vitro stimulation, and in vivo CTL assay of the murine model. C57BL/6 (B6) mice were purchased from Orient Bio (Iksan, South Korea) and OT-1.Thy1.1 TCR transgenic mice on a B6 background were obtained from the Institute for Basic Science (IBS; Pohang, South Korea). All mice were used in experiments at 6 to 12 weeks of age according to protocol. Pooled lymph nodes from OT-1 transgenic mice were stained with fluorochrome-conjugated antibodies to CD4, CD8, CD25, CD44 and CD62L and sorted to obtain Scientific RepoRts | 6:28208 | DOI: 10.1038/srep28208 CD44 − CD62L + naïve or CD44 + memory-phenotype (MP) CD8 + T cells using FACS. CD8 + TILs were obtained from B6 mice that were inoculated subcutaneously with OVAp-expressing EL4-EG7 tumor cell lines. For in vitro stimulation, sorted naïve, MP, and TIL CD8 + T cells were cultured on plate-bound anti-CD3 and anti-CD28 with IL-2 (10 ng/mL) and IL-15 (10 ng/mL). Effector cell surface staining procedure for PD-1, KLRG-1, and LAG3 labeling was identical to human cell methods previously described. In vivo CTL functional assay was performed using OVAp-expressing EL4-EG7 tumor cell lines as target cells. EL4-EG7 tumor cells (0.8 × 10 5 cells/200 µL) were inoculated subcutaneously in recipient B6 mice and 8 hours later, NT eff , MT eff , and TIL eff were injected intravenously in a 2:1 ratio of effector to target cells. A tumor mass growth of greater than 1 cm 3 was considered as positive tumor formation.
Statistical analysis. The two-tailed student's t test was used to determine differences in the mean. P values of less than 0.05 were considered statistically significant.