C57BL/6 (CD45.2⁺), B6SJL-PTPRC (CD45.1⁺), OT-I, and Rag1^−/− mice were purchased from The Jackson Laboratory. All mice were kept in the animal facility at Fundación Ciencia y Vida and maintained according to the “Guide to Care and Use of Experimental Animals, Canadian Council on Animal Care.” Animal work was carried out under institutional regulations of Fundacion Ciencia & Vida and Facultad de Ciencias, Universidad de Chile and was approved by the local ethics review committees.

Naive CD8⁺ T cells were purified from spleens and lymph nodes of OT-I mice. Briefly, the spleen was perfused with RPMI 1640 supplemented with 10% FCS, and CD8⁺ T cells were negatively selected using MACS (Miltenyi Biotec) following the manufacturer’s instructions. Following enrichment of CD8⁺ T cells, naive CD8⁺ T cells (CD8⁺/CD44lo/CD62Lhi/CD25⁻) were obtained by cell sorting (FACS Aria III, BDBiosciences). Naive CD8⁺ T cells were cultured in a 96-well round bottom microplate (10⁵ CD8⁺ T cells/well) and were activated with soluble α-CD3 (1 µg/ml; clone 145-2C11, eBioscience) and α-CD28 (1 µg/ml; clone 37.51) for 4 days in the presence of different cytokine cocktails. To generate Tc17 cells, CD8⁺ T cells were differentiated for 4 days in the presence of 2 ng/ml recombinant human TGF-β1 (eBioscience), 20 ng/ml recombinant mouse IL-6 (eBioscience), and 5 µg/ml of anti-IFN-γ (clone XMG1.2, Biolegend). Tc1 cells were differentiated for 4 days in the presence of 10 ng/ml recombinant mouse IL-2 (eBioscience) using a protocol modified from the literature which replicates the clinical protocol of expanding human tumor infiltrating lymphocytes (13). Cells were then harvested for in vitro experiments, adoptive transfer experiments, RNA extraction, intracellular cytokine staining, and flow cytometry.

To assess cytokine production by adoptively transferred Tc1 and Tc17 cells, cells obtained from lymph nodes, spleen, and peritoneal cavity of recepient mice were re-stimulated with 0.25 µM PMA (Sigma-Aldrich) and 1 µg/ml ionomycin (Sigma-Aldrich) in the presence of GolgiPlug (BD Biosciences) for 4 h. Cells were stained with antibodies against the cell surface markers (congenic markers) and then resuspended in a fixation/permeabilization solution (Cytofix/Cytoperm; BD Pharmingen). Following fixation and permeabilization, the cells were incubated with antibodies against IFN-γ, IL-2, IL-17, and TNF-α for 30 min at 4°C. The cells were then washed with permeabilization buffer and resuspended in PBS supplemented with 2% FCS for FACS analysis (FACSCanto II; BD Bioscience). In some cases, Fixable Viability Dye (eBioscience) was used to discard dead cells from the analysis.

Intracellular staining for Ki67 was conducted by incubating surface stained cells in a fixation/permeabilization solution (Foxp3/transcription factor permeabilization and fixation buffer, eBioscience). Following fixation and permeabilization, the cells were incubated with the Ki-67 antibody (clone 11F6, Biolegend) in a permeabilization buffer (Foxp3/transcription factor permeabilization buffer, eBioscience). After staining, cells were washed and analyzed by FACS. FACS data analysis was performed using the FLOWJO software (Tree Star Inc., Ashland, OR, USA).

Tc1 and Tc17 cells were differentiated in vitro, and total RNA was extracted. Naive CD8⁺ T cells were isolated by cell sorting (CD8⁺/CD44lo/CD62Lhi/CD25⁻) before RNA extraction. Total RNA was obtained using EZNA Total RNA Kit I (Ω Bio-Tek). RNA (1 µg) was reverse-transcribed using M-MLV reverse transcriptase (Promega). The PCR reaction was performed using Brilliant II SYBR Green QPCR Master Mix (Agilent Technologies) in a Stratagene Mx3000P real-time PCR machine. For relative quantitation, the amplified fragments were normalized according to constitutive transcription of the housekeeping gene HPRT. The sequences of the primers used for quantification of each measured transcript were the following:

tcf7 forward 5′- CAA TCT GCT CAT GCC CTA CC -3′, reverse 5′- CTT GCT TCT GGC TGA TGT CC -3′; lef1 forward 5′- TGA GTG CAC GCT AAA GGA GA -3′, reverse 5′- CTG ACC AGC CTG GAT AAA GC -3′; bcl6 forward 5′- CTG CAG ATG GAG CAT GTT GT -3′, reverse 5′- CAC CCG GGA GTA TTT CTC AG -3′; klf2 forward 5′- TGT GAG AAA TGC CTT TGA GTT TAC TG -3′, reverse 5′- CCC TTA TAG AAA TAC AAT CGG TCA TAG TC-3′; tbx21 forward 5′- CCT GTT GTG GTC CAA GTT CA AC-3′, reverse 5′- CAC AAA CAT CCT GTA ATG GCT TGT -3′; eomes forward 5′- GCG CAT GTT TCC TTT CTT GAG -3′, reverse 5′- GGT CGG CCA GAA CCA CTT C -3′; prdm1 forward 5′- GAC GGG GGT ACT TCT GTT CA -3′, reverse 5′- GGC ATT CTT GGG AAC TGT GT -3′; grzb forward 5′- ATC AAG GAT CAG CAG CCT GA-3′, reverse 5′- TGA TGT CAT TGG AGA ATG TCT -3′; pfn1 forward 5′- GAT GTG AAC CCT AGG CCA GA -3′, reverse 5′- TGA TGT CAT TGG AGA ATG TCT -3′; hprt forward 5′- CTC CTC AGA CCG CTT TTT GC -3′, reverse 5′- TAA CCT GGT TCA TCA TCG CTA ATC -3′.

Cell lysis was performed using Cytobuster protein extraction reagent (Novagen) supplemented with protease and phosphatase inhibitors (complete PhosSTOP, Roche). Protein extracts were separated in 15% SDS-polyacrylamide gels and transferred to PDVF membranes (Millipore). Blocking was at room temperature for 1 h in 5% fat-free milk, and membranes were incubated overnight at 4°C with 1:2,000 dilution of Total OXPHOS Human WB Antibody Cocktail (ab110411), and then for 1 h at room temperature with a secondary antibody conjugated to horseradish peroxidase. Chemiluminescence detection used SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) and series of timed exposure images were acquired with a Chemidocs imaging system (BioRad) to ensure densitometric analyses were performed at exposures within the linear range. To ensure equal protein loading across gels, membranes were submerged in stripping buffer (Restore Western blot stripping buffer; Pierce), incubated at 37°C for 30 min, and re-probed with a loading control antibody. Image J was used for densitometric analysis.

To assess secondary expansion of transferred cells, Tc1 and Tc17 lymphocytes were generated in vitro from OT-I/CD45.2⁺ and OT-I CD45.1⁺/CD45.2⁺ mice, respectively. On the day of the transfer, the Tc1 and Tc17 lymphocytes were counted and mixed at a 1:1 ratio. The initial ratio (input) on the cytometer was then evaluated by staining against CD45.1 and CD45.2. A total of 2 × 10⁶ cells from this mixture were transferred intravenously into B6.SJL (CD45.1⁺) congenic mice. At day 23 post-adoptive transfer, the mice were immunized with 500 µg of OVA protein (Sigma-Aldrich) and 100 µl of Freund’s Complete Adjuvant (CFA, Sigma) intraperitoneally. On days 10, 20, and 27 post-adoptive transfer blood samples were obtained and analyzed by flow cytometry. Finally, on day 27 after the adoptive transfer, the mice were sacrificed, and cells obtained from the peritoneal cavity, secondary lymphoid organs, and spleen were analyzed by FACS.

To examine the ability of Tc17 cells to be retained in effector tissues, we performed intradermal injections of in vitro-differentiated CD45.2⁺ Tc1 and CD45.1⁺/CD45.2⁺ Tc17 cells (1.5 × 10⁶ cells per mice, at a 1:1 ratio) into CD45.1⁺ congenic mice, and 30 days later, we analyzed skin, secondary lymphoid organs, spleen, bone marrow, lung, and liver for the transferred cells. The intradermal transfer was performed by injecting 100 µl of cellular suspension in PBS divided in five spots of 20 µl each in the lower back skin of each mouse.

For in vitro homeostatic survival analysis, Tc1 and Tc17 cells were cultured in the presence of IL-15 (25 ng/ml, eBioscience) or IL-7 (10 ng/ml, eBioscience). Three days later, cells were harvested, counted, and analyzed by FACS.

For in vivo homeostatic survival analysis, Tc1 and Tc17 lymphocytes were generated in vitro from CD45.2⁺ and CD45.1⁺/CD45.2⁺ mice, respectively, and co-transferred (0.5 × 10⁶ cells of each type) in Rag1^−/− mice. The transferred cells were evaluated 10, 25, 75, and 115 days after in blood samples using counting beads (eBioscience).

Tc1 and Tc17 lymphocytes were generated in vitro from CD45.2⁺ mice and stained with eFluor 670 (Thermofisher Scientific, 5 µM) and CMTMR (Thermofisher Scientific, 10 µM), respectively. Then, the cells were co-transferred (0.5 × 10⁶ cells of each type) in CD45.1⁺ mice. Homing was evaluated 24 h later. Homing index in each organ was calculated as (Tc17_organ/Tc1_organ)/(Tc17_input/Tc1_input).

Oxygen consumption rates (OCRs) and extracellular acidification rates (ECARs) from in vitro generated Tc1 and Tc17 cells (0.5 × 10⁶ cells/well) were measured in non-buffered DMEM without phenol red (containing 25 mM glucose, 2 mM l-glutamine, and 1 mM sodium pyruvate) under basal conditions and in response to 1 µM oligomycin, 0.5 µM fluoro-carbonyl cyanide phenylhydrazone, and 1 µM rotenone + 1 µM antimycin A (Sigma). The cells were analyzed with the XFe-96 Extracellular Flux Analyzer (Seahorse Bioscience).

Data are presented as mean ± SEM. Differences between two groups were determined using two-tailed Mann–Whitney test. Differences between more than two groups were determined using one-way ANOVA with Tukey or Bonferroni post-test. A two-way ANOVA with Bonferroni post-test was used to compare secondary expansion rates. Statistical analysis and graphs were obtained with GRAPHPAD PRISM (GraphPad Software Inc).

In a previous study, we have demonstrated that Tc17 cells are endowed with high antitumor capacity and persistence, despite displaying a reduced cytotoxic potential in vitro (12). These observations lead us to hypothesize that Tc17 cells present a memory phenotype. To test this hypothesis, we co-transferred in vitro-generated Tc1 and Tc17 cells obtained from OT-I mice into wild type mice and then 3 weeks later, the mice were immunized i.p. with OVA-CFA. We analyzed the persistence and expansion of Tc1 and Tc17 cells before and after OVA immunization. As shown in Figure 1A, 10 and 20 days after the co-transfer, we could only detect the presence of Tc17 cells in blood, and these cells suffered a two-fold expansion following immunization with OVA, whereas Tc1 cells remained undetected. Accordingly, Tc17 cells were mainly found in the peritoneal cavity while Tc1 cells were present in low frequencies, ruling out the possibility that Tc1 cells were absent in blood because they were more efficiently recruited to the peritoneal cavity (Figures 1B,C). Interestingly, besides being detected in the peritoneal cavity, Tc17 cells were also present in spleen and secondary lymph nodes such as peripheral lymph nodes (PLN), mesenteric lymph node (MLN), and parathymic lymph nodes (pTLN) (Figures 1B,C). This observation suggests that Tc17 cells may be recruited to effector tissues but also, that a portion of Tc17 cells may remain circulating within secondary lymph nodes as a reservoir for future encounter with their cognate antigen. Next, we tested whether Tc17 cells continued to produce IL-17 or could give rise to IFN-γ-producing cells. As shown in Figure 1D, a high percentage of transferred Tc17 cells differentiated into IFN-γ-producing cells in spleen and peritoneal cavity, whereas a small fraction of Tc17 cells continued to produce IL-17 in spleen, confirming their high plasticity in this setting. Tc17 cells also expanded more rapidly and outnumbered Tc1 cells following in vitro polyclonal activation with anti-CD3 and anti-CD28 antibodies (data not shown). Taken together, these results demonstrate the ability of Tc17 cells to persist in the long term and rapidly expand upon a challenge.

Next, we evaluated whether Tc17 cells could differentiate to memory cells in vivo which could expand rapidly following a re-challenge with their cognate antigen. For this, we co-transferred in vitro-generated Tc1 and Tc17 cells obtained from OT-I mice into wild-type mice, and 24 h later, the mice were challenged with OVA. Four weeks later, mice were re-challenged with OVA and the frequencies of transferred cells in blood, PLN and spleen were analyzed. In this setting, the composition of Tc1 and Tc17 cells remained similar to the input before the re-challenge with OVA (dotted line, Figure S1C in Supplementary Material). However, following re-challenge with OVA protein (day 34), we found that Tc17 cells rapidly expanded (Figures S1A,B in Supplementary Material) and became almost 90% of transferred cells (Figure S1C in Supplementary Material), confirming that Tc17 cells are superior than Tc1 cells in proliferating in response to their cognate antigen. Similar results were obtained when analyzing the composition of transferred cells in spleen and PLN (Figure S1C in Supplementary Material). Interestingly, 1 week after the re-challenge, Tc17 cells found in the spleen and PLN gave rise to IFN-γ-producing cells (54.6 ± 7.3%) that do not express KLRG1 but express high levels of CD44 and CD127 (Figure S1D in Supplementary Material). Moreover, we found that a proportion of these Tc17 cells (29.6 ± 3.7%) express CD62L, which explains their potential to migrate within secondary lymph nodes (Figure S1D in Supplementary Material).

In agreement with our previous observations in tumor bearing mice (12), we demonstrated that transferred Tc17 cells were always found in the spleen and secondary lymphoid organs (Figure 1B; Figure S1C in Supplementary Material), suggesting that they have a circulation pattern similar to central memory (T_CM) cells. To study the migration pattern of Tc17 cells, in vitro-generated Tc17 and Tc1 cells were stained and transferred into congenic mice. 24 h later, we analyzed spleen, PLN, MLN, bone marrow, liver, lung, and blood for the presence of the transferred cells. As shown in Figure 2A, Tc17 cells migrated preferentially to secondary lymphoid organs such as the spleen, PLN, and MLN compared to Tc1 cells. In contrast, Tc1 cells preferentially migrated to the bone-marrow compared to Tc17 cells (Figure 2A).

Next, we tested whether Tc17 cells injected in peripheral tissues can become established as memory cells in secondary lymphoid organs. For this, we intradermally co-transferred Tc1 and Tc17 cells in a 1:1 ratio into CD45.1⁺ mice and 30 days later, the presence of memory cells was analyzed in skin, spleen, PLN, MLN, bone marrow, liver, lung, and blood. As shown in Figure 2B, we did not find Tc1 or Tc17 cells in the skin, however, Tc17 cells were found in blood and in all the organs analyzed. Interestingly, when analyzing the absolute numbers, we observed that Tc17 cells were preferentially found in secondary lymphoid organs (Figure 2B). These results demonstrate that Tc17 cells are not retained in the skin but migrate into secondary lymphoid organs.

To analyze the homeostatic response of Tc17 cells, we cultured Tc17 and Tc1 cells with the homeostatic cytokines IL-15 and IL-7, and the proliferation and absolute numbers of cells were analyzed 3 days later. As shown in Figure 3A, Tc17 cells responded to both homeostatic cytokines while Tc1 cells proliferated only in response to IL-15, which is in agreement with the fact that only Tc17 cells express CD127. In spite of their rapid proliferation, Tc17 cells retained the ability to produce IL-17 and only a small fraction of cells produced IFN-γ (data not shown).

To further analyze the in vivo homeostatic proliferation of Tc17 cells, Tc1 and Tc17 cells were transferred into Rag1^−/− mice at a 1:1 ratio and several days later, the absolute number of transferred cells was analyzed. As shown in Figures 3B–D, both Tc1 and Tc17 cells were present in Rag1^−/− mice 11 and 25 days following the transfer, but the absolute number of Tc17 cells in blood always doubled the absolute number of Tc1 cells. Moreover, a higher percentage of Tc17 cells were positive for Ki67 staining at day 25 compared to Tc1 cells (Figure 3E). Thus, Tc17 cells are much more responsive to homeostatic stimulation than Tc1 cells.

We have previously shown that Tc17 cells express higher levels of tcf-7 and lef1 and lower levels of GzmB compared to Tc1 cells (12). To further characterize the phenotype of Tc17 cells, in vitro-generated Tc1 and Tc17 cells were analyzed by real-time PCR and by FACS in search of additional markers related to memory cells and effector cells. As shown in Figure 4A, we observed that Tc17 cells express markers related to memory T cells such as bcl-6 and lack markers of terminal differentiation such as tbx-21, eomes, and prdm1. On the other hand, Tc1 and Tc17 cells did not express Klf-2 transcriptional factor, which commands CD62L expression. Interestingly, in contrast to Tc1 cells, Tc17 cells present low levels of CD25, granzyme B, and TNF-α, while they express RORγt and produce high levels of IL-2 (Figure 4B).

Since Tc17 cells express high levels of bcl-6 and low levels of prdm1 which suggests that Tc17 cells present a metabolic signature related to memory T cells, we sought to determine the metabolic profile of Tc17 cells. A distinctive characteristic of memory CD8⁺ T cells is that they rely on an oxidative metabolism and present a high SRC (19). Thus, we determined the OCR and ECAR of in vitro-generated Tc1 and Tc17 cells and found that Tc17 cells present a higher OCR and SRC than Tc1 cells (Figures 5A,B). Moreover, Tc17 cells exhibit higher levels of complex IV proteins compared to Tc1 cells (Figures 5C,D and Figure S2 in Supplementary Material), suggesting that they present an increased efficiency in respiration. All these data demonstrate that Tc17 cells rely mainly on an oxidative metabolism which has been already related to a memory CD8⁺ T cell phenotype.