Inciting the cellular arm of adaptive immunity has been the fundamental goal of cancer immunotherapy strategies, specifically focusing on inducing tumor antigen–specific responses by CD8+ cytotoxic T lymphocytes (CTL). However, there is an emerging appreciation that the cytotoxic function of CD4+ T cells can be effective in a clinical setting. Harnessing this potential will require an understanding of how such cells arise. In this study, we use an IL12-transduced variant of the 70Z/3 leukemia cell line in a B6D2F1 (BDF1) murine model system to reveal a novel cascade of cells and soluble factors that activate anticancer CD4+ killer cells. We show that natural killer T cells play a pivotal role by activating dendritic cells in a contact-dependent manner; soluble products of this interaction, including MCP-1, propagate the activation signal, culminating in the development of CD4+ CTLs that directly mediate an antileukemia response while also orchestrating a multipronged attack by other effector cells. A more complete picture of the conditions that induce such a robust response will allow us to capitalize on CD4+ T-cell plasticity for maximum therapeutic effect. Cancer Immunol Res; 2(11); 1113–24. ©2014 AACR.
Recently, evidence has emerged that definitively identifies a subset of CD4+ T cells with cytotoxic capacity (1, 2), a subset that was previously underappreciated because of the classic designation of CD4+ T cells as helper cells. This understanding of their extended plasticity makes CD4+ T cells an ideal target for immune induction efforts as they are capable of direct and efficient cytotoxicity as well as orchestrating immune responses by other effector populations. Although CD4+ T cells with the capacity to use the same cytotoxic mechanisms as CD8+ cytotoxic T lymphocytes (CTL) have been reported, the conditions under which CD4+ CTLs arise are only beginning to emerge. We describe how natural killer T (NKT) cells and appropriately activated dendritic cells (DC) cooperate with CD4+ α/β T cells to eliminate leukemia. We show that direct contact between NKT cells and DCs, mediated by CD40, is required for activation, as is direct contact between DCs and CD4+ CTLs. At the cytokine level, we demonstrate that IL12, IFNγ, and MCP-1 each has a critical role in guiding the cellular cascade that leads to CD4+ CTL activation and leukemia cell killing.
CD4+ CTLs were often considered as a type of overactivated Th1 cell until two companion articles revealed a unique transcription factor signature for CD4+ CTLs: Runx3 expression resulted in diminished ThPOK expression and de-repression of the CD8+ T cell–associated cytotoxic program in mature, stimulated CD4+ T cells (1, 2). These studies used the gut as their model tissue and suggested that it is the inflammatory milieu of this unique environment that gives rise to such cells. Our data indicate that CD4+ CTLs play a role outside of the gut microenvironment.
An increased understanding of the mechanisms by which immunomodulatory elements influence CD4+ T cells provides a rational approach to design therapeutic regimens to activate this effector population and capitalize on the diverse roles of CD4+ T cells. One possibility our study highlights is the potential in manipulating the cytokine environment to create the right conditions for the generation of CD4+ CTLs.
Materials and Methods
Female BDF1 (C57BL/6 × DBA/2)F1 mice were purchased from The Jackson Laboratory. Mice were housed in micro-isolator cages under specific pathogen-free conditions in the animal facility of the Ontario Cancer Institute (Toronto, ON, Canada). Cells were harvested from 4- to 6-week-old mice for in vitro cultures. In vivo experiments used 8- to 14-week-old mice, unless noted otherwise. All experimental procedures were approved by The Animal Care Committee of the Ontario Cancer Institute.
70Z/3 is a μ+ pre-B leukemia line derived from BDF1 mice (3). 70Z/3 was stably transduced with a fusion form of the IL12 cDNA using a lentivirus vector as previously reported and authenticated by sequence matching (4). The IL12-transduced clone LV12.2, the parent 70Z/3 line, and the NKT cell line C1498 (obtained from the ATCC) were maintained in RPMI-1640 with 5% heat-inactivated FBS (Gibco), 10 mmol/L HEPES, 100 μg/mL penicillin–streptomycin or kanamycin, and 5.5 × 10−5 mol/L β-mercaptoethanol in a humidified atmosphere at 37°C in 5% CO2. The IFNγ-resistant cell line LV12.2-R (“R” denotes the resistant line) was produced by culturing cells in increasing amounts of IFNγ for a period of nearly 2 months until the concentration of IFNγ was equal to 10 ng/mL. This concentration is approximately 10 times the amount detected in the supernatant of coculture wells undergoing leukemia cell clearance. After an initial period of reduced proliferation, LV12.2-R cells grew at the same rate as the parent LV12.2 cell line. All cell lines were tested and validated to be Mycoplasma free; no other authentication assay was performed.
DC generation and isolation
DCs were established according to methods described previously (5). Briefly, bone marrow was flushed from the femur and tibia of BDF1 mice, incubated in ammonia chloride potassium red blood cell lysis solution, washed with PBS, and resuspended at 5 × 105 cells/mL in RPMI supplemented with 20 ng/mL recombinant GM-CSF (Biosource, Invitrogen) and IL4-conditioned media (6). Cultures were fed on day 3 with 1 mL of RPMI containing growth factors. After 7 days, cells were harvested, and selection of the DC population was achieved using a CD11c Positive Selection Kit (STEMCELL Technologies) according to the manufacturer's instructions.
Single-cell suspensions were obtained by passing spleens from BDF1 mice through 40 μm cell strainers (BD, Falcon). After RBC lysis, CD4+ cells were obtained by magnetic selection using a CD4+ T-Cell Enrichment Kit (STEMCELL Technologies) according to the manufacturer's instructions. A negative selection kit was used to avoid nonspecific activation of the cells during the selection process.
Following selection, the CD4+ cells were stained with Abs recognizing CD4(-PE) and CD8(-FITC) and with PBS-57–loaded CD1d tetramer conjugated to allophycocyanin (APC; NIH Tetramer Core Facility) and incubated for 30 minutes on ice. Cells were washed twice with PBS containing 3.5 mmol/L EDTA and 10% FBS and sorted by FACS to deplete the CD4+ population of NKT cells. The NKT cell–depleted CD4+ population is defined as CD8−, CD4+, tetramer−, and the NKT cell population as CD8−, CD4+, tetramer+.
Cultures (2 mL) were established with combinations of the following: 5 × 104 leukemia cells, 20 ng/mL rIL12 (R&D Systems)—a similar concentration to that measured in the supernatant of LV12.2 cells cultured alone under the same conditions—5 × 105 CD11c+ cells, 106 CD4+ cells, and 105 NKT cells. DCs and leukemia cells were incubated together for 2 hours before the addition of NKT or CD4+ cells.
Transwell cultures were set up as described above, except that the lower chamber was plated in a total volume of 1 mL media and the upper chamber in 200 μL. The lower chamber contained either DCs, LV12.2, and CD4+ cells; DCs, LV12.2, and NKT-depleted CD4+ cells; or LV12.2 and NKT-depleted CD4+ cells without DCs. The upper chamber contained media alone; NKT cells; NKT cells and DCs; or NKT cells, DCs, and LV12.2 cells.
Enumeration of leukemia cells
After 3 days of coculture, cells were harvested and counted with a hemacytometer to determine the total live-cell count per well. As 70Z/3 are the only cells in the assay that express μ, flow cytometry was used to determine the proportion of μ+ cells within the live-cell gate for each well. The absolute number of μ+ cells was calculated and divided by the input number to arrive at a measure of the number of target cells remaining in cocultures. This value is reported as “fold increase in μ+ cells over input” and compared between test wells and control wells.
Limiting dilution assays
On days 1, 3, 5, 7, 10, and 15 of the coculture, samples were taken from wells containing all the necessary components for leukemia cell clearance. Cells were counted and resuspended at concentrations of 1,000, 500, 100, 50, 25, 12.5, 6.25, and 3.125 cells per well in 200 μL complete OptiMEM media (GIBCO) without the addition of growth factors. Forty-eight wells were plated at each dilution. On day 5, plates were scored for colony growth, and the fraction of nonresponsive wells was plotted against cell concentration. KaleidaGraph software was used to create a line of best fit and determine the frequency of leukemia cells.
Supernatants collected from cocultures at 4, 10, 12, and 14 hours after initiation were analyzed using the Mouse Inflammation Cytometric Bead Array (CBA; BD Biosciences). Standards and samples were processed according to the protocol provided, and flow cytometry was carried out using a FACScan and analyzed using CellQuest software version 3.1. IFNγ results are reported for hours 4 and 10. Granzyme B (GzmB) was detected by ELISA (eBioscience) used according to the manufacturer's instructions with supernatants collected from cocultures on days 3, 6, 11, 13, and 17. MCP-1 was detected using a plate-based multiplex ELISA (Mouse Cytokine Magnetic 20-Plex Panel; Invitrogen) from day 2–4 supernatants collected from cocultures or Transwell assays.
Antibodies and cytokines
Cells were harvested from cocultures, resuspended in FACS buffer, incubated at room temperature for 10 minutes with 0.5 μg/test α-CD16/CD32 Fc block (with the addition of rat serum for intracellular staining) and for 20 minutes with α-mouse–specific antibodies: μ (clone 33.60), CD4 (GK1.5; BioLegend), CD8 (53-6.7; BD Biosciences), CD11c (N418; eBioscience), CD80 (16-10A1; BD Biosciences), CD86 (GL1; eBioscience), TCRγδ (GL3; BD Biosciences), pan-TCRβ8 (F23.1; BD Pharmingen), Vβ (H57-597; BioLegend), Vβ8.1/8.2 (MR5-2; BD Pharmingen), Vβ811, γ/δTCR (RR3-15; BD Pharmingen), GzmB (GB11; BioLegend), and ThPOK (D9V5T; Cell Signaling Technology). Viability was determined with the Zombie UV Fixable Viability Kit (BioLegend). Cells stained with biotin-labeled antibodies were washed twice with FACS buffer, incubated for 15 minutes with an appropriate streptavidin secondary fluorochrome, and analyzed by flow cytometry using a FACSCalibur, or LSRFortessa (Becton Dickinson) and FlowJo software (TreeStar). Blocking Abs against MCP-1 (2H5; BD Pharmingen), CD40-L (MR1; BioLegend), and IL15 (polyclonal goat IgG; R&D Systems) were added at 40 μg/mL, 25 μg/mL, and 3 μg/mL, respectively. Hamster (BD Bioscience) or goat (Santa Cruz Biotechnology) IgG were used as isotype controls. To test the effectiveness of IL15 blocking, replicate wells of 105 splenocytes were plated with a titration of IL15, blocking antibody added at 3 μg/mL, and control wells containing no Ab. After 4 days, the plates were assayed by thymidine incorporation. Cytokines used to activate the DC population included IFNγ (R&D Systems) and MCP-1 (R&D Systems) and were used at 1 μg/mL, and 200 μg/mL, respectively. Anti-CD40 Ab (BioLegend) was used at 1 μg/mL.
RT-PCR detection of the iTCR and cytotoxic molecules
mRNA was prepared from cocultures sampled at various time points. Expression of invariant T-cell receptor (iTCR) mRNA was detected by reverse transcriptase PCR (RT-PCR) using two previously published primer sets: Vα14Jα18: Vα14-CACAGCCACCCTGCTGGAT and Jα18-CCAAAATGCAGCCTCCCTAA (7); Vα14Jα281: Vα14-CTAAGCACAGCACGCTGCACA, Jα281-CAGGTATGACAATCAGCTGAGTCC and α-constant CAGGCAGAGGGTGCTGTCCTG (8). The results for both primer sets agreed in all experiments. Fas-L: sense-GGAATGGGAAGACACATATGG, α-sense-CATATCTGGCCAGTAGTGCAG (9); perforin: sense-GAGAAGACCTATCAGGACCA, α-sense-AGCCTGTGGTAAGCATG (10); TRAIL: sense-AGTCCTCTCGGAAAGGGCAT, α-sense-TGGCTTCTTGATCCAGGTCC (11).
Ex vivo detection of ThPOK
Splenocytes harvested from mice 7 days after injection of 106 LV12.2 cells, or from control mice (untreated or injected with 106 70Z/3 cells), were cultured with 70Z/3 cells at a ratio of 500:1 for 4 hours in the presence of GolgiPlug (BD Biosiences). Cells were fixed and permeabilized using the FoxP3 Buffer Kit (eBioscience) and stained for CD4, ThPOK, and μ to exclude background staining by the tumor cells.
Labeling of target cells for specificity experiments
Target 70Z/3 or C1498 cells were labeled by incubation in media for 30 minutes at 37°C, with the fluorescent cytoplasmic dye TFL4 (Oncoimmunin, Inc.). Cells were washed once with media and plated with effector cells as described.
Labeled target cells were cultured with effector CD4+ cells derived from day 8 in vitro cultures that had completely cleared the original leukemia cells. These populations were cocultured for 2 hours, with or without the addition of the GzmB inhibitor Serpin A3N (SA3N; R&D Systems). Cell death was quantified by flow cytometry using Annexin V and propidium iodide (PI; Roche). Target cells were plated alone as a control, and the percentage of cytotoxicity was determined as follows: (experimental death − spontaneous death)/(maximum death − spontaneous death).
Complete cocultures were initiated in a Transwell system as described. On day 3, the bottom chamber was harvested, washed in media, and resuspended in 1 mL of new media. One hundred microliters of the cell suspension was added to precoated ELISPOT wells specific for mouse GzmB (R&D Systems) and incubated at 37°C in 5% CO2 for 24 hours. GzmB was then detected according to the manufacturer's instructions.
In vitro specificity
LV12.2-primed CD4+ cells were generated from day 5 cocultures and plated at a ratio of 5:1 with labeled target cells in media supplemented with IL2. Target cells alone in media + IL2 were plated to establish fold change. Wells were harvested after 46 hours, and the number of viable target cells was assessed by flow cytometry. Dead cells were excluded by PI staining. Alternately, ratios of 10:1, 20:1, and 50:1 were plated, and these cultures were analyzed after 2 hours.
Ex vivo specificity
Mice received 3 (LV12.2, LV12.2, 70Z/3; n = 2) or 4 (LV12.2, LV12.2, 70Z/3, LV12.2; n = 3) i.p. injections of 106 cells at 7-day intervals. CD4+ cells were selected from spleens of primed or untreated mice and plated at 5 × 105 cells/mL in media supplemented with IL2. Labeled target cells were added at 2.5 × 105 or 5 × 104 cells per well. Cultures were harvested 2 and 3 days after initiation, and the number of viable target cells was assessed by flow cytometry. Dead cells were excluded by PI staining. Fold change was calculated by dividing the number of viable cells from primed cultures over naïve cultures for each target cell.
Determining the cellular mechanism of long-term immunity
Mice were injected with 106 70Z/3 cells 110 days after the primary challenge, but without further IL12 therapy. Groups of mice were depleted of CD4+ cells, CD8+ cells, both T-cell subsets, or IFNγ using specific Abs, as previously described (4). The depletion potential of each Ab was confirmed in vivo before use. A control group of naïve mice was injected with 70Z/3 cells alone.
Statistical analyses were performed using Prism GraphPad software.
CD4+ cytotoxic cells require IL12 and the presence of DCs during the primary response
We previously reported that treatment of 70Z/3-challenged mice with IL12 leads to a T cell–mediated immune response that eradicates the leukemia, resulting in survival of the challenged animals. Using the IL12-secreting leukemia cell clone LV12.2 in a cell-based approach, we showed that survival was dependent on a CD4+ cellular fraction (12). To study this result more carefully, we established a syngeneic in vitro assay using 70Z/3 leukemia target cells and CD4+ splenic cells.
CD11c+ DCs were derived from bone marrow cultured in GM-CSF and IL4 for 7 days, and selected on the basis of CD11c expression. After selection, DCs expressed both CD80 and CD86 when added to cultures. CD4+ cells were enriched from the spleens, and cultures were established containing CD4+ splenocytes and leukemia cells (parent 70Z/3 or LV12.2) with or without DCs. After 3 days of coculture, target cell clearance was observed only in wells that included DCs. Cumulatively, the data reveal that the establishment of a CD4+ cytotoxic population capable of clearing 70Z/3 to a measurable degree requires both a DC population and a source of IL12 (Fig. 1A). Flow cytometric analysis of the DC population after 2 days in coculture showed evidence of DC activation: Expression of CD80 was maintained on the CD11c+ population, and CD86 expression was substantially increased over the baseline level (Fig. 1B).
Quantification of leukemia cell clearance
Clearance of the leukemia cell targets was measured using two methods. First, because LV12.2 cells are the only μ+ cells in the cultures, we could enumerate them and compare with input numbers. A second method was to use limiting dilution to enumerate leukemic clones. This revealed an initial expansion of LV12.2 cells, followed by complete clearance (Fig. 1C).
The cytotoxic population contains αβ T cells and NKT cells
Enriching CD4+ cells before addition to cocultures excludes NK cells but not NKT cells. To determine the composition of this population, input CD4+ cells were stained with either a pan-TCRβ Ab or the γ/δTCR. Although nearly all cells were positive for TCRβ, γ/δTCR staining was undetectable. In addition, we stained for three of the more commonly expressed TCR Vβ chains—Vβ8, Vβ8.1/8.2, and Vβ11—and for NKT cells using a CD1d tetramer loaded with the αGalCer analogue PBS-57. A portion of the population was positive for either Vβ8, Vβ11, or the tetramer, but staining for Vβ8.1/8.2 was negative (Fig. 2A). This indicated that our CD4+ input cells represent a heterogeneous population of α/β T cells with approximately 6% NKT cells.
We also tested the CD4+ cells for expression of the iTCR rearrangement that characterizes classic CD1d-restricted NKT cells (hereafter referred to as NKT cells). We used two sets of PCR primers that recognize the Vα14-Jα281 (8) and Vα14-Jα18 (7) regions of the iTCR, both of which gave identical results. C1498, an NKT cell line, was used as a positive control. The CD4+ population expressed iTCR, but the CD11c+ fraction and the LV12.2 target cells were negative (Fig. 2B). Thus, NKT cells are present within the starting CD4+ population.
NKT cells are required, but not sufficient, for leukemia cell killing
NKT cells are known to play a role in tumor immunosurveillance (7) and to produce a copious amount of IFNγ (13), a cytokine required for the initiation of a primary Th1 response (14). Indeed, in our culture system, NKT cells are the initial sources of IFNγ, producing it only in the presence of target cells (Fig. 2C). In addition to priming the immune response, IFNγ also has antiproliferative properties in cancer models (15), which could confound our results. To avoid confusion, we cultured LV12.2 cells with increasing concentrations of IFNγ to generate a resistant cell line (LV12.2-R) for our assays.
We isolated NKT cells from the CD4+ population by FACS, then repeated the assay with NKT-depleted CD4+ cells and NKT cells alone (Fig. 2D). We also included a condition in which NKT cells were removed from the population, and then recombined with the depleted CD4+ fraction to control for any effect of sorting on the NKT population. Clearance of leukemia cells was observed only under conditions in which both CD4+ and NKT cells were present (Fig. 2D). Neither the NKT-depleted CD4+ cells, nor the NKT cells alone, were capable of mediating target-cell killing. Furthermore, rIFNγ was not a sufficient substitute for NKT cells, indicating that the role of NKT cells in the cytotoxic response is not limited to IFNγ production. FACS-mediated depletion was confirmed by PCR detection of the iTCR (Fig. 2B).
NKT cells are required to activate the DC population
We used a Transwell culture system to determine which cell types need to be in contact for activation of the cytotoxic population and clearance of target cells. NKT cells, DCs, CD4+ T cells, and LV12.2 target cells were cocultured in various combinations in the upper and lower chambers separated by a Transwell membrane (Table 1). We found that leukemia cell killing requires that NKT cells directly interact with DCs but not with the other CD4+ cells or the target cells. DCs must also interact with both the cytotoxic cells and the target cells.
These observations suggest that NKT cells play a critical role during the priming phase, but they are not required during the effector phase of the antileukemia response. To test this possibility, DCs, NKT cells, and LV12.2 cells were combined in the top chamber, whereas DCs were seeded alone in the bottom. After 24 hours of culture, the preactivated DCs (aDC) in the bottom chamber were harvested and transferred into an NKT cell–deficient culture system that does not otherwise support target-cell clearance. Target-cell clearance was as effective with aDCs as was observed under complete coculture conditions (Fig. 3A). In this experiment, the control consisted of DCs harvested from preactivation cultures containing NKT-depleted CD4+ cells. Thus, cell–cell contact between DCs and NKT cells leads to the activation of DCs that are then capable of supporting the development of a CD4+ cytotoxic population.
MCP-1 plays a critical role in DC activation
Using the Transwell assay, we also observed that the particular DCs that interact with cytotoxic and target cells need not be the same DCs that were directly activated by the NKT cells (Table 1). This finding indicates that, in addition to a requirement for cell–cell contact between NKT cells and DCs during the activation stage, there must also be a soluble factor produced as a consequence of this contact that is responsible for activating DCs in the lower chamber. We collected supernatants from DC preactivation Transwell assays to compare cytokines produced when DCs were activated in the presence of NKT cells versus the control condition of culture with NKT-depleted CD4+ T cells. Supernatants were analyzed using multiplex ELISA. We found that MCP-1 production was increased consistently under conditions that led to DC activation (Fig. 3B). To test whether MCP-1 is capable of activating DCs, we conducted assays using undepleted CD4+ splenocytes and blocking Abs against MCP-1, and found that this abrogated the activation of cytotoxicity (Fig. 3C). These data suggest that the soluble products of IL12 signaling and DC–NKT interaction lead to the activation of DCs, allowing for induction of a cytolytic program in a CD4+ T-cell population. To test this possibility, cytokines MCP-1 and IFNγ were added to cultures containing DCs, LV12.2, and CD4+ cells depleted of NKT cells. Figure 3D shows that these conditions led to the clearance of target cells even in the absence of DC–NKT interactions. It has been reported that IL15 plays a role in the induction of CD4+ CTL (1). We examined the possibility that IL15 was also required for the clearance of leukemia cells in our assay by inclusion of an anti-IL15 antibody, which is capable of eliminating IL15-induced splenic cell proliferation (see the legend of Fig. 3C). This blocking antibody failed to prevent the clearance of leukemia cells, suggesting that IL15 does not play a role in this assay system (Fig. 3C). We also did not detect this cytokine using IL15-specific ELISA (data not shown).
NKT cells and DCs likely interact through CD40–CD40-L
CD40–CD40-L interactions play an important role in the reciprocal activation of DCs and NKT cells. We used a blocking Ab for CD40-L to assess whether the requisite contact-dependent DC–NKT cell interaction is mediated by CD40–CD40-L. Figure 3C shows decreased target-cell clearance when CD40 ligation is inhibited, confirming that CD40–CD40L play a critical role in the development of CD4+ effector cells with cytotoxic potential. This result, along with results from the two-chamber Transwell experiment, suggests that NKT–DC interactions produce soluble factors that can propagate DC activation, leading to CD4 CTL activity and target-cell clearance. To determine whether the CD4+ cells activated in our in vitro cultures have cytotoxic potential, we applied the isolated CD4+ cells directly on target cells at different ratios. We observed direct cytotoxicity, as shown in Fig. 3E.
Next, we tested the specificity of CD4+ CTLs in this assay. We set up cultures and allowed CD4+ cells to clear LV12.2 cells. The CD4+ cells obtained from these cultures were tested for toxicity on non–IL12-producing 70Z/3 cells or an unrelated leukemia cell line, C1498. Figure 3F shows that only 70Z/3 cells were cleared.
GzmB release is one killing mechanism used by the CD4+ cytotoxic population
To determine whether contact is required for killing, we cultured LV12.2 cells in supernatants derived from cultures that had effectively cleared target cells. No killing was observed, suggesting that cytotoxicity requires cell–cell contact (data not shown). We next investigated membrane-bound cytotoxic ligands expressed by effector populations: Fas-L, TRAIL, perforin, and GzmB. RT-PCR for Fas-L and TRAIL showed that these two effector molecules are not differentially regulated at the transcriptional level between populations with cytotoxic capacity and the parental input populations (Fig. 4A). On the other hand, the detection of perforin transcripts by RT-PCR and of GzmB protein levels by ELISA (Fig. 4A and B) occurred only under killing conditions, and the expression of both factors occurred together. To test the role of GzmB in mediating killing, CD4+ cells that had already cleared targets in a coculture assay were rechallenged in the presence of a GzmB inhibitor, and there was decreased target-cell clearance (Fig. 4C). As noted previously (Table 1), we found that leukemia cell clearance required direct contact between DCs, CD4+, T cells, and target cells. NKT cells also played a key role in the antileukemia response but could do so at a distance; they must be in contact with DCs. A similar experiment was carried out to determine whether the CD4+ T cells (NKT-depleted) produce GzmB in these cultures. The Transwell assay was set up as described (NKT+DC in the top chamber), and DCs, CD4+T cells (NKT-depleted), and leukemia target cells in the bottom chamber. Figure 4D shows by ELISPOT that the CD4+ T cells in the bottom wells produced GzmB.
CD4+ effector cells decrease expression of ThPOK during culture
It was discovered recently that the transcription factor ThPOK, which suppresses the cytotoxic fate of CD4+ T cells and which is lost during CD8+ T-cell maturation, is also lost in CD4+ cells that demonstrate cytotoxic ability (1, 2). We therefore analyzed ThPOK expression in our culture system. We found a progressive loss of ThPOK during the 5-day culture period. Low ThPOK expression was maintained in CD4 cells undergoing rechallenge in vitro (Fig. 4E). We analyzed rechallenged cocultures by flow cytometry to determine whether CD4 cells expressing a low level of ThPOK also produced GzmB. Gating on TCRβ+ CD1d tetramer− cells, we observed a GzmB+ ThPOKlow population that was absent in the naïve spleen.
In Fig. 5, we present a model that summarizes our findings and describes the minimum requirements for CD4+ CTL–mediated leukemia cell clearance in our system.
The in vivo clearance of LV12.2 cells is primarily dependent on a CD4+ cell population with capacity for providing long-lasting immunity.
We reported previously that mice surviving an initial challenge with LV12.2 tumor cells are able to reject subsequent leukemia challenge with parental, non–IL12-secreting 70Z/3 months later, and that the primary response depends on the presence of CD4+ cells but not CD8+ cells (12). We therefore examined the memory response to determine whether it has the same mechanism as the primary. A new cohort of mice was injected with LV12.2 cells and allowed to clear the leukemia before being injected 110 days later with specific antibodies to deplete CD4+ and CD8+ cells during rechallenge with parent 70Z/3 leukemia cells. Figure 6A shows that long-lasting immunity resides in both the CD4+ and CD8+ T-cell compartments, and either population alone is sufficient to reject leukemia without a need for IL12 administration during the second challenge. In previous experiments, we had shown that mice immunized with LV12.2 cells were able to reject the parental 70Z/3 cells but were unable to reject unrelated leukemia cells, suggesting that the response was specific to the immunizing cells (12). To determine whether CD4+ cells from primed mice would show similar specificity when challenged in vitro, we isolated CD4+ cells from the spleens of mice that had rejected LV12.2 cells and tested their ability to clear 70Z/3 target cells compared with C1498 target cells. As shown in Fig. 6B (top), the primed CD4+ cells were unable to clear the unrelated C1498 cells but they had begun to clear the immunogenically related 70Z/3 cells. We also tested the prediction, suggested by our in vitro experiments, that immunized mice should have an increased number of CD4+ cells expressing decreased levels of ThPOK. As shown in Fig. 6B (middle and bottom), this was indeed the case.
There is growing interest in CD4+ effector T cells with anticancer activity (16, 17). In this report, we describe novel cellular interactions and resultant molecular mediators necessary for the development of a CD4+ CTL population, which requires support from both DCs and NKT cells. We showed that the leukemia cell–derived IL12 induces IFNγ production by NKT cells, leading to the activation of DCs through CD40 ligation and production of MCP-1. DC activation licenses them to provide all necessary signals for maturation of cytotoxic CD4+ T cells. Maturation is accompanied by decreased expression of the transcription factor ThPOK, which normally suppresses the cytotoxic program. These signals culminate in contact-dependent target-cell killing by CD4+ CTLs, mediated in part by perforin and GzmB.
CD4+ T cells have an important role in activating CD8+ CTLs. There are also reports of CD4+ T cells orchestrating anticancer responses mediated by other effector populations such as macrophages (18, 19) and NK cells (20). In addition, there are reports that CD4+ T cells can display cytotoxic activity directly (16, 17). This phenomenon has been observed mostly in chronic infections (21), but there is growing recognition of this population of cells in malignancy (22), including their detection in peripheral blood of patients with B-cell chronic lymphocytic leukemia (23). Two recent studies have demonstrated CD4+ T-cell plasticity (1, 2) under steady state in the gut. Our data showed that CD4+ T cells with the capacity to activate a cytotoxic program also reside in the spleen. Emerging data support the idea that CD4+ CTLs could have clinical effectiveness. A patient with metastatic melanoma, treated with an autologous CD4+ effector clone recognizing NY-ESO-1, experienced complete disease regression, which was attributed to the expansion of T cells reactive to other tumor-associated antigens, through Ag-spreading, and the initiation of a de novo CD8+ CTL response (24). This result underscores one of the key attractions of mobilizing CD4+ T cells: Not only can they support anticancer responses by other effector populations, but they can also perform cytotoxic functions. Results from another study showed the induction of CD4+ cytotoxic cells in patients treated with ipilimumab (25). Our data demonstrate that CD4+ anticancer CTLs can be generated when the right cells and soluble factors are present, and suggest the possibility of inciting this population for therapeutic benefit.
Our initial in vitro results showed that optimal target-cell clearance required DC support and a source of IL12. We optimized our system by including CD11c+-selected DCs. This is not surprising because DCs are the most potent inducers of Th1 responses through their production of the polarizing cytokine IL12p70. However, DCs do not constitutively produce IL12 p70, as this requires IFNγ stimulation (15). Furthermore, DCs typically do not produce IFNγ, and T cells only produce IFNγ after activation by DCs. Because IL12 can promote its own production (26), IL12 produced by the leukemia cells could bypass the need for IFNγ stimulation. However, DCs do not constitutively express the β1 chain of the heterodimeric IL12 receptor (IL12R; ref. 26), which is also induced by IFNγ. Therefore, another cell type (27) must provide the IFNγ that stimulates DCs to increase expression of MHC class II and costimulatory molecules on their surface and produce IL12p70 so that they can stimulate maturation of naïve CD4+ T cells. We and others have shown that DC activation is characterized by increased CD86 expression while CD80 expression remained relatively stable (28, 29).
NKT cells constitutively express IL12R and produce copious amounts of IFNγ in response to IL12 signaling. They are also present in the CD4+ population, as evidenced by detection of iTCR transcripts. However, NKT cells are also capable of anticancer responses. By depleting the CD4+ population of NKT cells, we showed that they are critical for leukemia cell clearance, but are insufficient when administered alone. The Transwell assays showed that there must be initial contact between the DC and NKT populations to produce signals that could expand activated DCs. This result indicated that NKT cells are required for the activation of DCs, and the activation requires both a membrane-bound and a soluble factor. DCs and NKT cells can activate each other reciprocally through CD40–CD40-L binding, and IFNγ increases CD40 expression by DCs (30). Results from the CD40-L blockade experiment confirmed that this molecule plays a critical role in the activation phase of the immune response.
MCP-1 was produced, to a greater extent, under conditions in which DCs were activated. DCs in the lower chamber of the Transwell assay did not require contact with NKT cells but were activated as a consequence of the interaction that occurred in the top chamber. MCP-1 may be a product of DC–NKT cell interaction, through CD40 signaling, and mediates further DC activation. In fact, under some circumstances, CD40 ligation can stimulate the production of MCP-1 (31). However, contradictory reports exist in the literature regarding the role of MCP-1 in the activation and differentiation of monocytes, leading either to inhibition of tumor growth (32) or to cancer growth and invasion (33). The kinetics of DC exposure to MCP-1 appears to be pivotal to the resultant effect and may determine whether a Th1 or Th2 response is promoted (34). However, in our system, MCP-1 appears to be critical to optimal DC activation and is required for antileukemia activity.
Culturing LV12.2 target cells in supernatants collected during tumor clearance did not lead to leukemia killing if effector cells were not included, demonstrating that cell–cell contact is required for killing. In these assays, the only molecules that were consistently produced under clearance conditions were GzmB and perforin.
We observed in our in vivo system that a primary response, mediated predominantly by CD4+ cells, resulted in memory within this T-cell compartment that is sufficient to mediate a recall response, suggesting that the effector population consists of bona fide CD4+ CTLs. However, this may not be the only mechanism used as a result of IL12 treatment in this model; in fact, neutralization conducted during rechallenge confirms the establishment of CD8+ CTLs. Cumulatively, these results indicate that CD4+ CTLs constitute an important component of the response to cell-based IL12.
We believe that CD4+ effector cells function under a wider set of circumstances than currently are appreciated, but that their contribution is masked by the understanding of their role as helper cells. Our model unmasks this role because CD4+ CTLs are the prominent effector population in our experimental system and has allowed us to study how they arise. A greater understanding of the conditions that lead to CD4+ CTL development will allow for this population to be specifically activated and implemented in a clinical setting.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: M.E. Nelles, J.A. Medin, C.J. Paige
Development of methodology: M.E. Nelles, C. Furlonger, C.J. Paige
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.E. Nelles, J.M. Moreau, C. Furlonger
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Nelles, J.M. Moreau, C. Furlonger, A. Berger, C.J. Paige
Writing, review, and/or revision of the manuscript: M. Nelles, J.M. Moreau, A. Berger, C.J. Paige
This work was supported by funding from the Terry Fox Research Institute, the Leukemia and Lymphoma Society, the Canadian Institutes of Health Research, and the Princess Margaret Cancer Centre Foundation for facilities through grants held by C.J. Paige.
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.
The authors thank Dr. Pamela Ohashi for thoughtful reading of the article. The authors gratefully acknowledge the NIH Tetramer Core Facility for provision of the PBS-57–loaded CD1d tetramer. They also thank Selena Cen, Michael Mielnik, and Yuanfeng Liu of the Paige laboratory for helpful comments and suggestions.
- Received November 27, 2013.
- Revision received July 7, 2014.
- Accepted August 6, 2014.
- ©2014 American Association for Cancer Research.