An attractive treatment of cancer consists in inducing tumor-eradicating CD8+ CTL specific for tumor-associated Ags, such as NY-ESO-1 (ESO), a strongly immunogenic cancer germ line gene-encoded tumor-associated Ag, widely expressed on diverse tumors. To establish optimal priming of ESO-specific CTL and to define critical vaccine variables and mechanisms, we used HLA-A2/DR1 H-2−/− transgenic mice and sequential immunization with immunodominant DR1- and A2-restricted ESO peptides. Immunization of mice first with the DR1-restricted ESO123–137 peptide and subsequently with mature dendritic cells (DCs) presenting this and the A2-restriced ESO157–165 epitope generated abundant, circulating, high-avidity primary and memory CD8+ T cells that efficiently killed A2/ESO157–165+ tumor cells. This prime boost regimen was superior to other vaccine regimes and required strong Th1 cell responses, copresentation of MHC class I and MHC class II peptides by the same DC, and resulted in upregulation of sphingosine 1-phosphate receptor 1, and thus egress of freshly primed CD8+ T cells from the draining lymph nodes into circulation. This well-defined system allowed detailed mechanistic analysis, which revealed that 1) the Th1 cytokines IFN-γ and IL-2 played key roles in CTL priming, namely by upregulating on naive CD8+ T cells the chemokine receptor CCR5; 2) the inflammatory chemokines CCL4 (MIP-1β) and CCL3 (MIP-1α) chemoattracted primed CD4+ T cells to mature DCs and activated, naive CD8+ T cells to DC–CD4 conjugates, respectively; and 3) blockade of these chemokines or their common receptor CCR5 ablated priming of CD8+ T cells and upregulation of sphingosine 1-phosphate receptor 1. These findings provide new opportunities for improving T cell cancer vaccines.
From previous cancer vaccine trials, it has emerged that in order for tumor-associated Ag (TAA)-specific CD8+ T cell responses to be tumoricidal, they need to be persistent, have high titers and high avidity, and they must infiltrate tumors (1–5). Although, in some cases, tumor regression was achieved, correlation between objective tumor control and vaccine-induced T cell responses remained modest, stressing the need to vigorously investigate key variables that determine the efficacy of T cell-based cancer vaccines.
An efficient cancer vaccine must induce maturation of dendritic cells (DCs). Mature DCs highly express MHC class I (MHC I) and MHC class II (MHC II) molecules, CD40, and ligands for costimulatory molecules, such as CD80 (B7.1), CD86 (B7.2), and CD70 (1, 6). While triggering of CD28 on naive T cells by CD80 and CD86 promotes T cell differentiation and IL-2 production, triggering of CD27 by CD70 drives the generation and maintenance of CD8 memory (1, 6, 7). DCs naturally mature upon infections with bacteria or viruses in which inflammatory cytokines (e.g., IFN type I, TNF-α, IL-6, and IL-1β) act on DCs, in addition to triggering various receptors. In cancer vaccines, various adjuvants have been tested to promote adequate DC maturation. Good results have been obtained with the TLR9 agonists CpG combined with emulsion-forming agents such as IFA or Montanide (1–3, 8). DCs play diverse and preeminent roles in tumor immunology, and some tumors escape T cell immunity by inhibiting DC maturation, thus favoring the induction of regulatory T cell and T cell tolerance rather than of adaptive T cell immunity (1, 9). Moreover, there exist different types of DCs, and the CD8+ T cell responses they induce can be strikingly different (10).
Induction of effective and persistent anti-TAA CD8+ T cell responses typically requires adequate help from Th1 cells (2, 11–15). In mice, it has been shown that in the absence of Th1 cells (e.g., in MHC II or CD4 knockout mice), formation of protective and persistent CD8+ T cell responses is seriously compromised (13, 14). Upon interaction with mature, cognate DCs, Th1 cells secrete substantial amounts of Th1 cytokines (e.g., IFN-γ, TNF-α, IL-2) and stimulate DC cytokine production (1, 6, 16). Whereas IL-2 promotes the proliferation and differentiation of Ag-selected cytotoxic T cells and CD8 memory (17, 18), IFN-γ upregulates the expression of IL-12, adhesion molecules, and MHC I and MHC II molecules and inhibits Th2 cell activities (6, 16).
There are different views and models on the mechanism by which Th1 cells assist priming of naive CD8+ T cells. According to one concept, Th1 cells help CD8+ T cell priming by licensing DCs for efficient priming of naive CD8+ T cells (6, 15). This relies primarily on DC activation by CD40 triggering upon engagement with CD40L (CD154) expressed on activated CD4+ T cells, and it results in IL-12 production, a cytokine that supports the generation of CTL (1, 6, 16). In support of this model, it has been shown that CD40 agonists strongly boost CD8+ T cell priming and can substitute for T cell help at least in some systems (9, 14, 15). Another model on Th1 cell help stipulates that primed CD4+ and naive CD8+ T cells interact simultaneously with mature DCs presenting MHC I and MHC II peptides (19, 20). Against this model it has been argued that the probability of the formation of three-cell clusters is scant due to the low frequency of Ag-specific naive CD8+ T cells, estimated to be ~1 out of 106 (21). However, an intravital two-photon imaging study has shown that the inflammatory chemokines CCL3 (MIP-1α) and CCL4 (MIP-1β) chemoattract activated naive CD8+ T cells to DC–CD4+ T cell conjugates, thereby favoring three-cell cluster formation (19). Such clusters have also been observed and enumerated in another intravital imaging study (20). Interestingly, another study demonstrated that freshly primed CD8+ T cells, as CD4+ Th1 cells, can attract naive CD8+ T cells in a CCR5-dependent manner and promote their priming (22).
Although these studies are compelling, the use of adoptive transfer of T cells from TCR transgenic mice is artificial and leaves open critical issues, such as: 1) What is the mechanism by which activated CD4+ T cells are guided to mature cognate DCs and naive CD8+ T cells to DC–Th1 cell conjugates; 2) What are the factors and mechanisms that induce naive CD8+ T cells to surface express CCR5; 3) What precise roles do Th1 cytokines play in CD8+ T cell priming; and 4) It has been shown that primed CD8+ T cells exit lymph nodes upon expression of sphingosine 1-phosphate receptor 1 (S1P1) (23, 24), but it is not known what factors govern S1P1 expression during CD8+ T cell priming.
To conclusively investigate these matters, we used as a preclinical mouse model HLA-DRB1*0101, HLA-A*0201 (with α3 from Db), H-2−/− transgenic mice (A2/DR1) (25). As TAA we used the cancer germ line Ag NY-ESO-1 (ESO), which has been widely used in cancer vaccine trials, is broadly expressed on diverse tumor cells, and is highly immunogenic (3, 26–28). It is a cytosolic protein of 180 aa, and MHC I and MHC II epitopes have been extensively mapped (http://www.cancerimmunity.org/peptidedatabase/Tcellepitopes.htm). For A2 the immunodominant epitope is 157–165 (157p), which is expressed by most tumor cells (2, 3, 26), and for DR1 various epitopes have been reported, for example, 123–137 (123p), 87–101 (87p), and 87–111 (27, 28).
Our investigations in this highly defined system showed that 1) efficient priming of CD8+ T cells in draining lymph nodes requires available Th1 cells and copresentation of strongly immunogenic MHC I and MHC II epitopes by the same mature DC; 2) the Th1 cytokines IL-2 and IFN-γ, namely in combination, induce CCR5 surface expression on naive CD8+ T cells; 3) the chemokines CCL4 and CCL3 play differential roles in guiding Th1 cells to DCs and activated naive CD8+ T cells to DC–CD4 T cell conjugates, respectively; and 4) these chemoattractions and the strength of Th1 cell help determine the upregulation of S1P1 on freshly primed CD8+ T cells and hence their exit from the draining lymph nodes into circulation.
Materials and Methods
Mice and cells
A2/DR1 mice have been described previously (19). AJ-H1 fibrosarcoma cells were obtained from mice 6 mo previously injected s.c with 3-methylcholanthrene (100 μg/mouse). Single-cell suspensions were derived from solid tumors and cultured in RPMI 1640 supplemented with 10% FCS and antibiotics. A 123p-specific Th1 CD4+ T cell line was derived from lymph node CD4+ T cells from mice immunized with 123p by stimulation with 123p (1 μM) at day 0 and rIL-2 each 2 d (10 IU/ml; PeproTech, London, U.K.) and tetramer-guided sorting after 1 mo.
Groups of mice (n = 5) were immunized by s.c. injections of 50 μl emulsion containing the indicated peptides (50 μg peptide emulsified in 50 μl PBS and 50 μl CFA [Difco Laboratories, Detroit, MI] or IFA and CpG at the base of the tail). DCs were obtained by culturing bone marrow cells for 8–10 d with GM-CSF (20 ng/ml; PeproTech) and activated with LPS (0.5 μg/ml; Sigma-Adrich, St. Louis, MO) overnight. After pulsing with 1 μM indicated peptides at 37°C for 1 h, DCs were washed, resuspended, and injected s.c. in the hind foot pads (0.5 × 106 cells/mouse); in some experiments blocking anti-CCR5, CCL3, and CCL4 Abs (50 μg/mouse) were coinjected.
Peptides and tetramer
Peptides included 123p (LKEFTVSGNILTIRL), 87p (LLEFYLAMPFATPME), 157p with mutation C165V (SLLMWITQV), collagen type II261–273 (AGFKGEQGPKGEP), and CMV pp65495–503 (NLVPMVATV). The fluorescent peptides Alexa 555–ESO157–165 (C[Alexa 555]LLMWITQV) and ESO123–137 (C[Alexa 488]-GG-LKEFTVSGNILTIRL]) were prepared by reacting the thiol peptides with the corresponding Alexa maleimides as recommended by the supplier (Molecular Probes/Invitrogen, Paisley, U.K.). HLA-A2α1,α2/Kb α3 tetramers (PE labeled) were prepared by refolding using human β2-microglobulin and 157p.
Flow cytometry and Abs
Popliteal and inguinal lymph nodes and spleens were homogenized, cells were washed, and for flow cytometric detection of IL-2– or IFN-γ–producing cells stimulated with the indicated peptides (1 μM) for 4 h in the presence of brefeldin A (10 μg/ml; BD Biosciences, San Jose, CA), cells were fixed (5% paraformaldehyde), permeabilized with 0.5% saponin (Sigma-Aldrich), and stained with cytokine Abs (BD Biosciences). For tetramer staining, CD8+ T cells (1 × 106) were incubated in 50 μl FACS buffer with 10 nM A2/157p tetramer for 30 min at room temperature followed by 30 min with FITC anti-mouse CD8 (clone 5H10; BioLegend, San Diego, CA) and Cy5-PE–labeled anti-CD4 (GK1.5) Abs. Anti-mouse S1P1 Ab (clone 11424) was from Abcam (Cambridge, U.K.), PE-conjugated anti-mouse CCL3 (clone 39624) was from R&D Systems (Oxon, U.K.), and PE anti-mouse CCR5 Ab (clone HM-CCR5) was from BioLegend. All samples were analyzed on a FACSCalibur (BD Biosciences) and data were analyzed with the FlowJo 8.7.1 software (Tree Star, Ashland, OR).
ELISPOT and multicytokine detection assay
ELISPOT was performed as described previously (29). All Abs were purchased from Mabtech (Nacka, Sweden). The plates were analyzed by ZellNet Consulting (Fort Lee, NJ). For the detection of multicytokines, cytometric bead array kits were used on supernatants, obtained as described for ELISPOT, following the manufacturer’s instructions (BD Biosciences).
For in vitro cytotoxic assays, 51Cr-labeled AJ-H1 cells (5 × 103 cells/well) were incubated with CD8+ splenic T cells from immunized mice (E:T ratio of 1:10 or as indicated) for 4 h at 37°C with the indicated concentrations of 157p. Specific lysis was calculated from released 51Cr as: 100 × [(experimental − spontaneous release)/(total − spontaneous release)]. In vivo cytotoxic assay was performed as described previously (30). In brief, AJ-H1 cells or syngeneic splenocytes were labeled with 0.1 μM CFSE and pulsed with 157p (1 μM) or with 1 μM CFSE, mixed 1/1, and transferred i.v. in immune or control mice (1 × 107 cells). The specific lysis (%) was calculated as: (1 − ratio immunized/ratio nonimmunized) × 100, whereby the ratio indicates the percentage CFSEhigh/percentage CFSElow cells.
In vitro priming of CD8+ T cells and ELISA
Mature DCs (2 × 104) were incubated in 24-well plates with the different peptides (1 μM) at 37°C for 1 h and washed, resulting in the removal of most nonadhered cells. CD4+ T cells (2.5 × 105), obtained by negative selection (>90%) using MACS and Ab-coated magnetic beads (StemCell Technologies, Grenoble, France) from lymph node cells from 123p-immunized mice and suspended in Iscove’s IMDM medium (Life Technologies, Paisley, U.K.), supplemented with 10% FCS, β-mercapthanol, and antibiotics, were added, and after 12 h, conjugates were enumerated. Alternatively, naive CD8+ T cells (5 × 105), obtained by negative selection from splenocytes, were added and after 6 d of incubation, the percentage of A2/ESO157–165 tetramer+ CD8+ T cells was assessed by flow cytometry. For blocking or neutralization experiments, Abs (10 μg/ml) specific for mouse IL-2 (clone JES6-5H4), IFN-γ (H22), TNF-α (TN3-19.12), CD40 (clone HM40), IL-12 (C17.8), CCR5 (HM-CCR5), anti-CD40L (clone MR1) were from BioLegend, CCL3 (ab10382) and CCL4 (ab10386) were from Abcam, and CD70 (clone 118510) and CCL3 (ab-450-NA) were from R&D Systems. Isotype-matched irrelevant Abs, as recommended by the suppliers, were used as negative controls. Recombinant IL-2, IFN-γ, and IL-12 were from PeproTech. For ELISA, supernatants from cultures were assayed using ELISArray kits as recommended by the supplier (SABiosciences, Frederick, MD).
Assessment of CCL3 and CCL4 messages by quantitative PCR
Experiments were performed as described previously (31). CD4+ T cells (1 × 106) from a 123p-specific line and mature 123p-pulsed DC (1 × 106) were coincubated and after 6 h, FACS sorted after pipetting in cold PBS and EDTA (5 mM). Total RNA was extracted from CD4+ T cells (5 × 105 cells) and DCs (5 × 105 cells) using TRIzol (0.5 mM; Invitrogen). The values of CCL3 and CCL4 were normalized to the expression of TATA binding protein.
Mature DCs were labeled with Cy5 succinimidyl ester (GE Healthcare, Otelfingen Switzerland) (5 μM, 10 min), pulsed or not with peptides (1 h at 37°C), and plated at the center of glass-bottom dishes (MatTek, Ashland, MA) (5–10 × 103 cells in 50 μl). After 3 h of incubation at 37°C, the bulk of nonadherent DCs was washed off. Purified 123p-primed CD4+ T cells (5 × 105) were labeled with CFSE (Invitrogen) (5 μM, 8 min) and added in the center of the dish. After 16 h of incubation in the absence or presence of Abs, the bulk of nonadherent cells was washed off. Purified naive CD8+ T cells (5 × 105), preincubated for 6 h with IL-2 and IFN-γ (10 IU/ml and 100 ng/ml, respectively), were labeled with CellTracker Orange CMTMR (Invitrogen) (2 μM, 15 min) and added to the dishes. Images were acquired on a Zeiss Axioplan microscope equipped with an AxioCam MRm camera and analyzed on AxioVision LE 4.5 software. DCs and conjugates were quantified using images (×20) from the four quadrants near the center. Analyses and imaging of draining lymph nodes were performed as described (32).
Migration of lymphocytes was performed as previously described (33). Briefly, supernatants were obtained from mature DCs cultured alone or together with CD4+ T cells from 123p-immunized mice in the absence or presence of 123p. After 6 h, supernatants were placed on the lower chamber of Transwell plates (96 wells, 5-μm pores) (ChemoTx; NeuroProbe, Gaithersburg, MD). Cells (2.5 × 105) with or without Abs specific for CCL3, CCL4, and CCR5 were added on the upper side of the membrane and incubated for 4 h at 37°C. The numbers of migrated cells in the lower wells were analyzed by flow cytometry.
An unpaired Student t test was used to determine the statistical significance of the data. The level of significance was set at p values of <0.05.
CD4+ and CD8+ T cell responses elicited in A2/DR1 mice by immunization with ESO peptides
We first immunized A2/DR1 mice with the A2-restricted 157p peptide in IFA and CpG and 7 d later observed >2% of A2/157p tetramer+ CD8+ T cells in the draining lymph nodes (Fig. 1A), and upon 157p stimulation a comparable fraction of IFN-γ+CD8+ cells was found (Fig. 1B). The frequency of CD8+IFN-γ+ cells was nearly 10-fold lower on splenocytes and almost undetectable on PBL. We next immunized mice in a like manner with the DR1-restricted 123p and 10 d later found 0.16% IL-2+CD4+ cells in the draining lymph nodes (Fig. 1C). When 87p was used, only 0.07% IL-2+CD4+ T cells were observed, indicating that this peptide is less immunogenic (Fig. 1D). When using the more sensitive ELISPOT assay an ~4-fold difference in IL-2 response was observed (Fig. 1E). Most IL-2+CD4+ T cells were in the draining lymph nodes. The IL-2 responses were maximal 7–11 d after immunization and contracted thereafter (Fig. 1F). Lymph node cells 10 d after 123p immunization produced upon 123p stimulation high amounts of IFN-γ, IL-2, and TNF-α, but no detectable IL-4 and IL-5 (i.e., they were typical Th1 cells) (Fig. 1G). Conversely, 87p elicited ~10-fold lower IFN-γ and IL-2 responses and no detectable TNF-α.
Primed CD4+ T cells boost CD8+ T cells priming by cognate DCs
To establish an efficient prime-boost immunization regimen, mice were first immunized with a DR1-restricted peptide (123p or 87p) and 10 d later with peptide-pulsed (157p plus 123p or 87p) mature DCs. The LPS-activated, mature bone marrow-derived DCs expressed, as expected, high levels of CD40, HLA I and II molecules, CD80 (B7.1), CD86 (B7.2), and CD70 (data not shown). One day after s.c. injection, 123p- and 157p-pulsed DCs were abundantly found in the draining inguinal and popliteal lymph nodes (data not shown). Four days later ~0.3% of the lymph node cells were IL-2+CD4+, or only 0.05% when 87p was used (Fig. 2C). Remarkably, 6 d after immunization with mature DCs pulsed with 123p and 157p, PBL exhibited 5.9% A2/157p tetramer+CD8+ T cells (Fig. 2A, 2B). When the less immunogenic 87p was used, 1.7% of tetramer+ cells were observed, and in the absence of a first immunization only 0.75% were observed, indicating that the efficiency of CD8+ T cell priming was largely determined by the first immunization with a DR1-restricted peptide.
Efficient priming of CD8+ T cells was induced by inoculation of 123p-preimmunized mice with DCs pulsed with 123p and 157p (cognate DCs), but not a mixture of DCs pulsed with 123p or 157p (separate DCs) or 157p only (Fig. 2D). In the former case, the frequency of IFN-γ+ cells in PBL from mice immunized with cognate DCs reached a high maximum after 7 d, contracted during the following 2–3 wk, and remained stable between 30 and 60 d. In the latter case, the frequencies of IFN-γ+ cells were ~4-fold lower, similar to 157p only-pulsed DCs. A2/157p tetramer+CD8+ T cells 7 d after immunization exhibited a typical effector T cell phenotype, characterized by increased expression of CD44 and CD25, and decreased expression of CD62L (Fig. 2E). After 80 d, the tetramer+ cells showed a typical memory phenotype, characterized by biphasic expression of CD62L corresponding to central and effector memory subsets, increased CD44 and CD127 expression, and CD25 background levels (Fig. 2E). For comparison we immunized mice with 123p plus 157p or only 157p in IFA plus CpG. Upon challenge with 157p the frequency of IFN-γ+ cells was >4-fold lower in PBL from mice immunized with 123p plus 157p in IFA plus CpG than in PBL from mice immunized using the prime-boost regimen (Fig. 2F). Note, however, that while peptide-pulsed mature DCs induced powerful CD8+ T cell responses in mice with preexisting Th1 cells responses, they were clearly less efficient at inducing these as compared with immunization with peptide in adjuvant.
Efficiently primed CD8+ T cells exit the draining lymph nodes
The 123p-preimmunized mice 6 d after inoculation with cognate DCs (123p plus 157p) exhibited ~4,4% IFN-γ+CD8+ T cells in the spleen, but only 0.68% in the draining lymph nodes (Fig. 3A). Conversely, after inoculation of noncognate DCs (157p), only 0.4% of IFN-γ+ cells were observed in the spleen, but twice as many were observed in the draining lymph nodes. Similar proportions were observed at different time points (Fig. 3B). Four days after priming with cognate DCs, IFN-γ+CD8+ T cells in draining lymph nodes expressed high levels of S1P1, whereas those in the spleen did not (Fig. 3C, 3D). Conversely, upon injection of DCs pulsed with 157p alone, S1P1 expression was 7-fold reduced on lymph node cells, but not on splenocytes. Essentially the same findings were obtained when S1P1 expression was assessed on A2/ESO157–165 tetramer+ cells (data not shown). These findings indicate that the strength of available Th1 cell help determines the efficiency of CD8+ T cell priming, their S1P1 expression, and their egress from the draining lymph nodes in the spleen and blood (23, 24).
Prime-boost–induced CD8+ T cells are strongly cytolytic
As assessed in 51Cr-release experiments, CD8+ splenocytes from mice immunized first with 123p and then with cognate DCs (123p plus 157p) efficiently killed AJ-H1 fibrosarcoma cells pulsed with 157p (Fig. 4A). At an E:T ratio of 10:1, half-maximal lysis was observed at ~8 × 10−11 M peptide with a maximal lysis of 34% and a nonspecific one of 2%. The lysis of sensitized AJ-H1 cells increased with the E:T ratio, reaching ~60% at a ratio of 1:100 (Fig. 4B). Similar results were obtained when using as targets AJ-H1 fibrosarcoma cells that were stably transfected with ESO (data not shown). Considerably lower lysis was observed for splenocytes from mice 6 d previously immunized with 157p in IFA plus CpG (Fig. 4A, 4B), which correlated with the lower frequency of A2/157p tetramer+CD8+ T cells in the spleens (Fig. 4C). These results demonstrate that our prime-boost vaccine strategy yields considerably higher frequencies of ESO-specific CTL in the periphery than does vaccination with 157p in IFA plus CpG.
We then examined in vivo killing of 157p-pulsed syngeneic splenocytes in mice 7 d (primary response) or 80 d (memory response) previously immunized by our prime-boost regimen. The specific lysis, as calculated from the ratio of CFSElow-labeled 157p-pulsed splenocytes and CSFEhigh-labeled, unpulsed cells was ~91 and 54%, respectively (Fig. 4D, 4E).
Efficient in vitro priming of naive CD8 T cells requires IL-2, IFN-γ, and CCL3
To investigate the mechanisms of CD8+ T cell priming, we incubated naive CD8+ T cells with Th1 cells from 123p-immunized mice and cognate DCs (123p plus 157p). After 6 d, ~7% of A2/157p tetramer+CD8+ T cells were observed (Fig. 5A). When the DCs were pulsed with 157p only, ~2% tetramer+ cells were detected, and in the absence of primed CD4+ T cells merely 0.2%. Note that in these experiments no cytokines were added and therefore CD8+ T cells died unless they received help from activated Th1 cells, resulting in high proportions of tetramer+CD8+ T cells.
Because 123p-primed CD4 T cells upon incubation with cognate DCs secreted substantial amounts of IFN-γ, IL-2, and TNF-α (Fig. 1G), we examined what effect their neutralization has on priming. Whereas neutralization of TNF-α had no significant effect, neutralization of IL-2 reduced the frequency of tetramer+CD8+ T cells by nearly 50%, neutralization of IFN-γ by 90%, and combined by >95% (Fig. 5B). Moreover, blocking of CD40L resulted in >50% reduction of tetramer+ cells, which is consistent with reports showing that CD40 triggering promotes CD8+ T cell priming (1, 6, 15). In contrast, blocking of IL-12 showed no significant inhibition (Fig. 5B). Blocking of CD70 caused a 6- to 7-fold reduction of tetramer+ cells, which is in agreement with studies showing that CD27 provides crucial costimulation for CD8+ T cell differentiation and memory formation (7, 34, 35). Remarkably, CD8 T cell priming was abolished by blockade of CCL3 or its receptor CCR5, whereas neutralization of CCL4 inhibited priming by only ~40%. As negative controls, isotype-matched irrelevant Abs were used, which in all cases exhibited no effect. In the case of CCL3 blockade, the same results were obtained when using two different Abs (data not shown). Furthermore, the blocking Abs specific for CCL3, CCL4, CCR5, and IFN-γ did not affect the proliferation of CD8+ T cell proliferation driven by anti-CD3 and anti-CD28 Abs in IL-2 containing medium, ruling out nonspecific effects.
After 6 d of incubation of naive CD8+ T cells with 157p-pulsed DCs, only 0.1% A2/157p tetramer+CD8+ T cells were observed (Fig. 5C). Although addition of agonist anti-CD40 Ab, rIL-12, or rTNF-α had no significant effects, in the presence of rIL-2 or rIFN-γ, ~1.8 and 2%, respectively, tetramer+ cells were observed, and in the presence of IL-2 and IFN-γ combined, 4.7% tetramer+ cells were observed (Fig. 5C and data not shown). The IL-2 plus IFN-γ–promoted CD8+ T cell priming was ablated by Abs specific for CCR5 and CCL3, but it was only partially inhibited by anti-CCL4 Ab.
Secretion and expression of CCL3, CCL4, and CCR5
We next examined the production of CLL3 and CCL4 by Th1 cells, mature, cognate DCs, and DC–Th1 cell conjugates. Coincubation of 123p-primed Th1 cells and cognate DCs resulted in secretion of high amounts of CCL4 and less CCL3 (Fig. 6A). Importantly, while DCs alone secreted nearly as much CCL4, they produced much less CCL3 and only transiently. Conversely, CD4+ T cells alone secreted no significant amounts of CCL3 and CCL4.
We then coincubated 123p-specific Th1 cells and mature, 123p-pulsed DCs and after 6 h, we separated the cells and assessed their CCL3 and CCL4 messages by real-time PCR. On CD4+ T cells the CCL3 message increased ~130-fold, but the CCL4 message increased by only 20-fold (Fig. 6B). Flow cytometric analysis showed that the CCL3+CD4+ T cells were mostly those producing IFN-γ, that is, they were fully activated Th1 cells (data not shown). Conversely, on DCs, the CCL4 message increased ~43-fold, but the CCL3 message increased only 17-fold.
Because naive T cells are CCR5−, we investigated what factors induce surface expression of CCR5. As assessed by flow cytometry, incubation of naive CD8+ T cells with supernatant from Th1 cells cocultured with cognate DCs resulted in CCR5 surface expression on the vast majority of cells (Fig. 6C, 6D). Nearly the same CCR5 expression was observed upon incubation with IL-2 and IFN-γ. CCR5 surface expression reached a maximum after 3–4 h of incubation and was partially induced by IL-2 or IFN-γ alone (Fig. 5D and data not shown).
Differential chemoattraction of CD4+ and CD8+ T cells by CCL3 and CCL4
Using Transwell migration, we first examined the chemoattraction of 123p-primed Th1 cells toward supernatants from DCs or DC–Th1 cell cocultures. Nearly 50% of primed CD4+ T cells migrated toward supernatants from DCs and >60% to supernatants from DC–Th1 cell conjugates (Fig. 7A). This chemotaxis was strongly inhibited by blockade of CCR5 or CCL4, but not CCL3. Preferential attraction of Th1 cells by CCL4, but not CCL3, was also observed when synthetic chemokines were used (Fig. 7B), consistent with a previous study (36). Moreover, and importantly, after 16 h of coincubation of Cy5-labeled cognate DCs and CFSE-labeled Th1 cells, ample DC–Th1 cell conjugates were visible by fluorescence microscopy (Fig. 7C). Their frequency was inhibited stronger by neutralization of CCL4 and CCR5 than of CCL3 (Supplemental Fig. A). This is consistent with results from the Transwell migration experiments and shows that in this in vitro setting DC–Th1 cell conjugate formation depends at least in part on directed cell movement.
We then examined likewise the chemoattraction of activated naive CD8+ T cells and observed ~34% of cells migrating toward supernatant from DC–Th1 cell conjugates, but merely 12% to supernatant from DCs (Fig. 7D). The chemotaxis toward DC–Th1 cell supernatant was strongly inhibited by blockade of CCL3 and CCR5, but not of CCL4. This is consistent with the observations that DC–Th1 cell conjugates, but not DCs, produce large amounts of CCL3, but not CCL4 (Fig. 6A), and that CD8+ T cells are preferentially attracted by CCL3 (Fig. 7E) (36). Preferential chemoattraction of activated naive CD8+ T cells by CCL3 was also observed when enumerating the frequencies of Th1 cell–DC–CD8+ three-cell clusters formed upon incubation of activated naive CD8+ T cells with preformed Th1–DC cell conjugates. Approximately 55% of the DC–Th1 cell conjugates were associated with CD8+ T cells (Fig. 7F, Supplemental Fig. B). In some of the three-cell clusters, the CD8+ T cells and Th1 cells contacted the DCs in distal sites, and in others in proximal sites, possibly allowing Th1 cell–CD8+ T cell interactions. The frequency of three-cell cluster formation was inhibited more by blockade of CCL3 than of CCL4 (Supplemental Fig. B). Collectively, these results indicate that CCL4 preferentially guides primed CD4+ T cells to cognate DCs and CCL3 naive CD8+ T cells to DC–CD4+ T cell conjugates.
Blockade of CCR5, CCL3, and CCL4 strongly inhibits in vivo priming of circulating ESO-specific CD8+ T cells
To examine whether these findings also apply in vivo, mice were immunized with 123p and after 10 d injected s.c. with Abs specific for CCR5, CCL3, CCL4, or irrelevant specificity (control) together with mature DCs pulsed with 123p and 157p. One week later in control mice ~4.3% IFN-γ+CD8+ T cells were observed in the spleen, but only 0.3% in the draining lymph nodes (Fig. 8A). In mice injected with Abs specific for CCR5, CCL3, and CCL4 the frequency of IFN-γ+CD8+ T cells in the spleen was dramatically decreased 0.3–0.75%) and in the draining lymph nodes the frequency of primed cells was low in all groups (~0.3%). Blockade of CCL4 exhibited markedly stronger inhibition in vivo than in vitro (Figs. 5B, 8A). The S1P1 expression on IFN-γ+CD8+ T lymph node cells from mice injected with control Ab was ~10-fold higher than on splenocytes (Fig. 8B). Blockade of CCR5, CCL3, and CCL4 resulted in a 7- to 8-fold lower S1P1 expression on freshly primed lymph node cells (Fig. 8B). The S1P1 expression on splenocytes was similarly low in all cases, which is explained, at least in part, by internalization of S1P1 on cells upon exposure to the high S1P concentrations in blood and spleen (23, 24). Taken collectively, these results indicate that the CCR5–CCL3/CCL4 axes are crucial for in vivo priming of ESO-specific CD8+ T cells and S1P1 upregulation and thus exit from the draining lymph nodes into circulation.
A major finding of the present study is that the magnitude, quality, and localization of the prime-boost vaccine-induced ESO-specific CD8+ T cell response in A2/DR1 transgenic mice was largely determined by available Th1 cell help (Fig. 2). A2/DR1 mice immunized with 123p exhibited much higher frequencies of ESO-specific CD4+ T cells than did animals immunized with the ESO peptides 87–111, 84–101, 89–98, or 87–101 (Figs. 1, 2 and unpublished data). This is at variance with studies showing that in patients with ESO+ malignancies the frequencies of CD4+ T cells specific for these epitopes are similar (27, 28, 37). This might be explained by the fact that the DCs that prime CD4+ T cells generate and present biased repertoires of ESO peptides, thus obscuring the actual immunogenicity of the peptides. Because this is ruled out in our system, immunization with synthetic peptides allows conclusive identification of their immunogenicities, which is crucial knowledge for the design of efficient vaccines.
Our results indicate that efficient priming requires that DCs copresent the MHC I- and MHC II-restricted peptides (Fig. 2). This is consistent with a three-cell cluster model for CD8+ T cell priming, in which activated CD4+ T cells and naive CD8+ T cells interact simultaneously with cognate DCs (Figs. 2, 3A, 4B, 5A, 5B). Such three-cell clusters were observed in vitro and in vivo, and their formation relies on chemoattraction of activated naive CD8+ T cells by CCL3 produced by Th1 cells upon conjugation with cognate DCs (Figs. 7, 8, Supplemental Fig.) (19, 20). Three-cell clusters also provide integral stimulation and costimulation for naive CD8+ T cells by DCs and CD4+ T cells. The 6-fold reduction of ESO-specific CD8+ T cells observed upon blocking of CD70 demonstrates that in our system triggering of CD27 (on naive CD8+ T cells) by CD70 (on mature DCs) is important, which is in accordance with previous in vivo studies demonstrating that the CD27–CD70 axis is crucial for the generation and maintenance of CD8+ T cell responses (Fig. 5) (34, 35). Interestingly, primed CD8+ T cells, as Th1 cells, can attract naive CD8+ T cells in a CCR5-dependent manner and promote their priming if the DCs copresent the respective peptides (22). Because primed CD8+ T cells upon interaction with cognate DCs produce CCL3, as Th1 cells, an analogous mechanism is suggested (22).
Although in our in vitro CD8+ T cell-priming experiments neutralization of CCL4 was less inhibitory than was blocking of CCL3, in the in vivo experiments neutralization of CCL4 and CCL3 exhibited similar effects (Fig. 8A) (19). We argue that this divergence is explained by the fact that CCL3 and CCL4 play different roles in CD8+ T cell priming, which is not discernible in in vivo experiments. The observations that mature DCs secreted high amounts of CCL4, but not CCL3, and that primed CD4+ T cells were strongly chemoattracted by CCL4, but not CCL3, convincingly argues that CCL4 preferentially attracted Th1 cells to mature DCs (Figs. 6A, 7A, 7B, Supplemental Fig. A) (36). Because the density of purified, primed, CD4+ T cells and cognate mature DCs in our in vitro experiments was high (Fig. 1C), they may find cognate DCs by random cell movements with significant probability. Conversely, in lymph nodes the probability of activated CD4+ T cells to find rare cognate mature DCs by scanning in complex three-dimensional structures is expected to be more dependent on CCL4-mediated chemoattraction, hence the stronger inhibition observed upon CCL4 blockade (Fig. 8A). In a second stage, primed CD4+ T cells upon conjugation with cognate DCs secrete high amounts of CCL3, but not CCL4, which in turn chemoattracts activated naive CD8+ T cells to DC–CD4 conjugates (Figs. 6A, 6B, 7D–F, Supplemental Fig. B) (36); because of the very low frequency of naive Ag-specific CD8+ T cells (estimated at 1 out of 10−6; see Ref. 21), this is crucial, as seen by the deleterious effect that CCL3 neutralization had on CD8+ T cell priming in vitro and in vivo (Figs. 5B, 5C, 8A).
It is noteworthy that CCL3 can promote CD8+ T cell priming also in other ways than chemoattraction. For example, it can promote Th1 cell formation and IFN-γ production in an IL-12–independent manner, which is consistent with the observation that in vitro CD8+ T cell priming required IFN-γ but not IL-12 (Fig. 5B, 5C) (38–40). Moreover, CCL3 has been shown to have adjuvant effects on CD8+ T cell priming, to increase CD8+ T cell effector functions, and to be involved (together with other chemokines) in recruiting CD8+ T cells to inflamed sites (39, 41–43). Although CCL3 and CCL4 both bind to CCR5, the former, but not the latter, also binds to CCR1, which may account for their diverging biological effects (36, 39).
Considering that CCR5 expression of CD8+ T cells plays key roles in their priming (Figs. 5, 8) (19, 22), but also in directing primed cells to tumors (44) or to sites of inflammation (45), surprisingly little is known about the factors and mechanisms that govern CCR5 expression. Our results demonstrate for the first time that the Th1 cytokines IFN-γ and IL-2, namely when combined, induce within hours CCR5 surface expression on naive CD8+ T cells (Fig. 6C, 6D). Importantly, this occurs in the absence of Ag or TCR triggering as has been described previously (19, 22, 46), and hence is a plausible mechanism by which naive CD8+ T cells become CCR5+ on entering sites where Th1 cytokines are produced, becoming thus susceptible to chemotaxis (Fig. 5B, 5C). The importance of this is demonstrated by the dramatic inhibition of CD8+ T cell priming by neutralization of IFN-γ plus IL-2 or of CCL3 or CCR5 and their ability to partially substitute primed CD4+ T cells in in vitro priming of CD8+ T cells, which again relies on CCR5 expression (Fig. 5C).
The strength of available Th1 cell help also determined the S1P1 expression on primed CD8+ T cells and thus their egress from the draining lymph nodes in the circulation (Fig. 3) (23, 24, 47). This is of importance, because an efficient cancer vaccine must induce CD8+ effector T cells that leave the sites of priming and can access tumors in the periphery (44). Our observation that in mice immunized with 157p in adjuvants, primed CD8+ T cells remained mostly in the draining lymph nodes, cautions that such peptide vaccines may fall short in inducing circulating CTL. Interestingly, even when strong Th1 cell help was available, blockade of CCR5-CCL3/4 severely inhibited upregulation of S1P1 (Fig. 8B). Because this blockade equally strongly inhibited the priming of ESO-specific CD8+ T cells in lymph nodes and in vitro, it appears that CCR5-CCL3/4 orchestrate interactions between Th1 cells, DCs, and naive activated CD8+ T cells and that this is essential for efficient CD8+ T cell priming, S1P1 expression, and CD8+ T cell memory formation (Figs. 5B, 8A) (19, 47, 48). S1P1 is downregulated upon initial T cell activation, but it is strongly expressed after extensive T cell differentiation and proliferation, and subsequent S1P1 expression seems to be programmed during the initial priming events (47). Although this needs to be investigated in further detail, it is interesting that CCR5 can be recruited in the immunological synapse and contribute to the activation of distinct T cell functions (49).
In conclusion, the present study describes a versatile prime-boost vaccine strategy that allows the generation of powerful, high-avidity TAA-specific CD8+ CTL responses. By first inducing Th1 cells by immunization with a MHC II-restricted peptide and then priming CD8+ T cells with peptide-pulsed mature DCs, this allowed us to define key variables for efficient cancer vaccines, such as: 1) the strength of available Th1 cell help determines the frequency of primed CD8+ T cells and, via their S1P1 expression, their egress from the lymph nodes in the circulation, which is required for their efficient interaction with tumors; and 2) efficient priming of CD8+ CTL requires coordinated interactions of activated naive CD8+ T cells, cognate mature DCs, and primed Th1 cells, which are orchestrated by CCL4 and CCL3, guiding Th1 cells to mature DCs and activated naive CD8+ T cells to DC–Th1 cell conjugates, respectively. Our study advocates the use of defined, strongly immunogenic MHC II- and MHC I-restricted peptides for cancer vaccines, because this circumvents the caveat that TAA epitopes cross-presented by DCs may not be the most immunogenic ones and/or may not be presented by tumor cells. Note, however, that ESO is not expressed in mice, but it is expressed in humans on cancer cells, but also on testis and at low levels in a limited number of somatic tissues (50); therefore, conclusions obtained in mouse models should be tempered with regard to applicability in human vaccination of cancer patients, due to potential tolerance in humans. Advances in TAA discovery and peptide predictions make available an ever-growing range of TAA epitopes restricted by diverse HLA molecules, which can be combined to broaden the range of T cell recognition.
We thank Drs. Yu-Chun Lone (INSERM U542, Hôpital Paul Brousse, Villejuif, France) for providing the A2/DR1 mice and Philippe Guillaume (Ludwig Institute for Cancer Research facility, Lausannne, Switzerland) for A2/157p tetramers.
Disclosures The authors have no financial conflicts of interest.
This work was supported by Grants 310030-125330 (to I.F.L.) and 31003A-127474 /1 (to D.F.L.) from the Swiss National Foundation and Grant OCS 01421-08-2003 from the Swiss Cancer League.
The online version of this article contains supplemental material.
Abbreviations used in this paper:
mean fluorescence intensity
- MHC I
MHC class I
- MHC II
MHC class II
NY-ESO-1157–165 with mutation C165V
- + pep
- − pep
sphingosine 1-phosphate receptor 1