Immature dendritic cells (DC), in contrast to their mature counterparts, are incapable of mobilizing a CD8+ CTL response, and, instead, have been reported to induce CTL tolerance. We directly addressed the impact of immature vs mature DC on CTL responses by infusing adenovirus peptide-loaded DC (of the D1 cell line) into mice that had received adenovirus-specific naive TCR-transgenic CD8+ T cells. Whereas i.v. injection of mature DC triggered vigorous CTL expansion, immature DC elicited little proliferation involving only a minority of the TCR-transgenic CTL. Even though the latter CTL developed effector functions, including cytolytic activity and proinflammatory cytokine secretion, these cells differed significantly from CTL primed by mature DC in that they did not exhibit down-regulation of CD62L and CCR7, receptors involved in trapping of T cells in the lymphoid organs. Interestingly, adoptive transfer of CTL effector cells harvested after priming by either mature or immature DC into naive recipient mice, followed by exposure to adenovirus, yielded quantitatively and qualitatively indistinguishable CTL memory responses. Therefore, in vivo priming of naive CD8+ T cells by immature DC, although failing to induce a full-blown, systemic CTL response, resulted in the formation of central memory-like T cells that were able to expand and produce IFN-γ upon secondary antigenic stimulation.
Dendritic cells (DC)4 appear to be the most important family of professional APC in the orchestration of T cell immune responses. A main feature of DC is their phenotypic and functional plasticity. In the absence of any inflammatory or pathogenic element, most DC found in peripheral tissues and lymphoid organs have a resting, immature phenotype characterized by high endocytic capacity and low surface expression of MHC molecules and costimulatory molecules such as CD86 and CD40 (1). However, upon interaction with microbial ligands, including TLR ligands (e.g., LPS, CpG-DNA), or upon ligation of CD40, or under influence of proinflammatory cytokines, DC rapidly acquire an activated phenotype (2, 3, 4, 5, 6, 7). These mature DC have a decreased MHC class II Ag-processing capacity, but a very efficient T cell-priming ability due to up-regulation of costimulatory and MHC molecules on the cell surface (8, 9).
The unique capacity of DC to initiate primary T cell responses in infectious diseases follows from the finding that conditional depletion of CD11c+ cells in mice in vivo results in complete abrogation of CD8+ T cell priming after infection with Listeria monocytogenes or Plasmodium yoelii, hallmarking the essential involvement of DC in the control of immunity to invading pathogens (10). Indeed, many pathogens appear to be presented mainly by DC that cross-present pathogen-derived Ags without being infected themselves (reviewed in Ref. 11). In parallel, tolerance mechanisms driven by immature DC in the steady state are considered to be of crucial importance for self-tolerance. Studies based on transgenic (Tg) expression of the model-Ag OVA in the pancreas under the control of the rat insulin promoter have provided the first evidence that DC can be responsible for peripheral tolerance in steady state conditions (12, 13). In this experimental setting, adoptively transferred naive specific CD8+ T cells are deleted in draining lymph nodes (LN) upon interaction with DC that cross-present cell-associated self-Ag in the absence of an inflammatory environment. Another important situation in which inflammatory stimuli are also lacking is in case of tumors. As tumors often masquerade as healthy tissues, immunological tolerance against tumors can be compared with tolerance induced against autologous tissues. Therefore, cross-presentation of tumor Ags by immature DC probably lies at the base of the inadequate T cell response against tumors.
Based on the paradigm that recognition of Ag in the absence of proper costimulation leads to impaired T cell activation (14), it has been proposed that the balance between tolerance and immunity may rely on the maturation status of DC (15, 16). According to this hypothesis, immature DC would induce T cell tolerance, while their mature counterparts are able to prime potent T cell responses. The issue of tolerance vs priming of CTL after Ag encounter on immature or mature DC has been subject of discussion in the past years (7, 16, 17, 18). To obtain insight into this issue, it is important to use a fully immature DC population, as was done in the DEC-205-targeting experiments performed by Bonifaz et al. (19). Isolation of DC from lymphoid organs or generation of DC from bone marrow generally leads to spontaneous maturation of some or all of them. In our hands, 10–30% of bone marrow-derived DC show a mature phenotype in short-term culture, and FACSort of the immature subpopulation further enhances their maturation (our unpublished observations). The use of Tg mice constitutively expressing a given Ag, as in Redmond et al. (20), offers the possibility to study the effect of Ag presentation under tolerizing and immunizing conditions, although the influences exerted by the immunizing conditions on cell types other than DC are not defined. Given these considerations, the spleen-derived D1 cell line is currently the best suitable candidate as a source of homogeneously pure immature DC. D1 constitutes a well-characterized, growth factor-dependent cell line derived from splenic C57BL/6 mouse DC that behaves like freshly isolated bone marrow DC (21, 22, 23, 24). Stimulation of D1 cells with TLR ligands, such as LPS, results in their full maturation (25). The use of this unique tool allowed us to thoroughly compare the phenotype, function, and survival of CD8+ T cells in vivo following interaction with immature or LPS-matured peptide Ag-loaded DC. Our results clearly indicate that injection of fully immature DC per se does not induce deletional Ag-specific CD8+ T cell tolerance or T cell anergy, nor is associated with T cell ignorance. Rather, naive CD8+ T cells that encounter peptide Ag presented by immature DC proliferate and acquire cytotoxic T cell function, but do not down-regulate CD62L and CCR7 lymphoid-homing receptors, compatible with a central memory T cell phenotype. Upon exposure to adenovirus in secondary hosts, these T cells acquired full-blown effector function and expansion capacity similar to adoptively transferred cells that were originally primed by mature peptide Ag-loaded DC.
Materials and Methods
C57BL/6 mice, C57BL/6 Thy-1.1+/+ (Thy-1.1) mice, and TAP1-deficient C57BL/6 (TAP−/−) mice were purchased from The Jackson Laboratory. Mice expressing a TCR specific for the H-2Db-restricted E1A234–243 adenoviral epitope (E1A TCR-Tg) were bred at TNO-PG. Mice were kept at the Leiden University Medical Center animal facility and used at 2–4 mo of age. Experiments were done in accordance with national legislation and under supervision of the animal experimental committee of the University of Leiden.
The D1 cell line, a long-term growth factor-dependent immature myeloid (CD11b+, CD8α−) DC line of splenic origin and derived from a female C57BL/6 mouse, was provided by P. Ricciardi-Castagnoli (University of Milan-Bicocca, Milan, Italy) and was cultured as described previously (24, 25). When necessary, full maturation was achieved by adding Escherichia coli-derived LPS (serotype 026.B6; Sigma-Aldrich) to the culture medium for 48 h (final concentration 10 μg/ml). Maturation status of the cells was checked by flow cytometry after staining with anti-CD86 PE (clone GL1), anti-CD40 PE (clone 3/23), anti-I-A/I-E PE (clone M5/114.15.2), anti-Kb PE (clone AF6-88.5), and anti-Db PE (clone KH95) Abs (all purchased from BD Pharmingen). IL-12(p40) secretion was measured by ELISA, as described previously (25).
DC tracking studies
D1 cells (10 × 106 cells/ml in 0.1% PBS/BSA) were labeled with 2 μM CFSE (Molecular Probes) for 20–30 min at 37°C. Labeling was stopped by addition of FCS (5%), and cells were washed three times in PBS. Ten million CFSE-labeled DC in 200 μl of PBS were injected in the tail vein of C57BL/6 mice. Twenty-four hours later, spleens were excised from DC-infused mice. Organs were placed in Tissue-Tek containing histomolds (Leica Microsystems) and rapidly frozen by dipping the histomolds in liquid N2. Frozen sections (6 μm) were prepared using a Leica CM 3050 S cryostat. Frozen sections were incubated in a FITC-labeled rat anti-mouse CD4 antiserum (BD Pharmingen). CD4-expressing T cells, as well as CFSE-labeled DC, were simultaneously detected by incubation of the sections in a sheep anti-FITC alkaline phosphatase conjugate. This conjugate was visualized with the substrate Fast Red (Sigma-Aldrich).
Preparation of naive E1A TCR-Tg CD8+ T cells
Single-cell suspensions were prepared by mechanical disruption of spleen and LN of E1A TCR-Tg mice. Erythrocytes were lysed by hypotonic shock using ammonium chloride, and cells were then incubated on ice for 45 min with anti-CD4 (GK1.5; 20 μg/ml) mAb at a density of 10 × 106 cells/ml. After extensive washing, labeled CD4+ T cells were depleted using magnetic beads coated with anti-rat IgG Ab (Dynabeads; Dynal Biotech) following manufacturer’s instructions. Achieved CD4 depletion was >95%. Eighty to 95% of CD8+ T cells were E1A TCR-Tg, as determined by tetramer staining. In experiments using TAP−/− recipient mice, a pure population of E1A TCR-Tg CD8+ T cells containing no APC was isolated using a negative selection-based procedure according to the manufacturer’s instructions (Miltenyi CD8+ T cell isolation kit; Sanquin Reagents; obtained purity >95%).
CFSE labeling of Tg CD8+ T cells
Enriched CD8+ T cells were washed and resuspended at a density of 10 × 106 cells/ml in a 0.5 μM solution of CFSE in PBS. After a 30-min incubation at 37°C, FCS was added to a concentration of 5%, and cells were washed three times in PBS.
Adoptive transfer of E1A TCR-Tg T cells and DC immunization
Unlabeled or CFSE-labeled E1A TCR-Tg CD8+ T cells were infused in sex-matched CD4-depleted Thy-1.1 mice (or CD4-depleted TAP−/− mice when indicated). CD4 depletion was performed to prevent endogenous CD4+ Th cells from activating the D1 cells in vivo (25) and was achieved by i.p. injections of purified CD4 Ab (GK1.5 clone; 50 μg in 200 μl of PBS) three to five times before and one time after DC injection. A total of 3 × 106 E1A-specific CD8+ T cells (as determined by tetramer staining) was resuspended in 200 μl of PBS and injected in the tail vein of recipient Thy-1.1 mice 2 days before DC immunization. One million immature or LPS-matured D1 cells (or less when indicated) in 200 μl of PBS were then i.v. injected after loading at 37°C for 2 h with the E1A CTL epitope-containing peptide (E1A234–243, 5 μg/ml; 4 × 106 cells/ml), followed by extensive washing.
Ex vivo analysis of CD8+ T cell responses by flow cytometry
Spleens and blood of the animals were isolated. Erythrocytes were lysed before staining. In most cases, donor E1A TCR-Tg T cells were detected by virtue of their Thy-1.2 expression. All Abs were purchased from BD Pharmingen. The percentage of undivided cells was calculated according to the formula: percentage of undivided cells = N0/(N0 + 1/2N1 + (1/4N2 + (1/8N3 + (1/16N4, etc.), in which Nx is the number of cells that had divided x times.
Cell surface stainings.
Cell suspensions were incubated for 30–45 min with anti-CD8α PerCP Cy5.5 (clone 53-6.7) and anti-Thy-1.2 allophycocyanin (clone 53-2.1), and, when indicated, with one of the following reagents: anti-CD69 PE (clone H1.2F3), anti-CD44 PE (clone IM7), or anti-CD62L PE (clone MEL-14). CCR7 expression was detected using a three-step staining process: cells were incubated for 1 h with a CCL19-human IgG1 fusion protein (the supernatant containing this protein was kindly provided by U. von Andrian (Department of Pathology, CBR Institute for Biochemical Research, Harvard Medical School, Boston, MA) and H. Pircher (Institute for Medical Microbiology and Hygiene, Department of Immunlogy, University of Freiburg, Freiburg, Germany)), followed by biotinylated anti-human IgG (Jackson ImmunoResearch Laboratories) and streptavidin-PE (BD Pharmingen). PE-labeled E1A234–243-loaded H-2Db tetramers (E1A-TM) were made in house and combined with anti-CD8 allophycocyanin (clone 53-6.7) staining.
Cells were permeabilized and stained using the Cytofix/Cytoperm Plus kit (BD Pharmingen), according to manufacturer’s instructions. CTLA-4 staining was done directly ex vivo (anti-CTLA4 PE, clone UC10-4F10-11), whereas detection of IFN-γ (anti-IFN-γ PE, clone XMG1.2), IL-2 (anti-IL-2 PE, clone JES6-5H4), and TNF-α (anti-TNF-α PE, clone MP6-XT22) was performed after 6 h of in vitro restimulation at 37°C with the E1A234–243 peptide or a control peptide (5 μg/ml) in the presence of GolgiPlug (containing brefeldin A; BD Pharmingen).
Cells were analyzed using a FACSCalibur apparatus using CellQuest software (BD Biosciences).
In vivo cytotoxicity assay
A mixture of spleen and LN cells from naive C57BL/6 mice was used as target cells. After passage over a nylon wool column, cells were split in two equal parts and loaded for 90 min at 37°C with either the E1A234–243 peptide or an H2-Db-restricted irrelevant peptide, derived from the murine gp100 autoantigen, as control (10 × 106 cells/ml; 5 μg peptide/ml). The cells were then washed four times and labeled by incubation at 37°C for 10 min with either 5 μM CFSE (E1A234–243-loaded cells) or 0.5 μM CFSE (control peptide-loaded target cells) in 0.1% BSA/PBS. Labeling was stopped by adding FCS (1/10 of volume), and cells were washed twice with PBS. The two types of target cells were mixed at a 1:1 ratio before being injected i.v. (5–10 × 106 of each type per mouse). Eighteen hours later, spleens were taken out and analyzed by flow cytometry. Specific killing was evaluated by the decline of the CFSEhigh population as compared with the CFSElow internal control.
In vivo assessment of established memory
Thy-1.1 mice were infused with CFSE-labeled E1A TCR-Tg T cells and subsequently immunized with peptide-loaded, immature, or LPS-matured DC, as described above. Three days after DC immunization, spleens of the animals were harvested and RBC lysis was performed, followed by CD8+ T cell isolation using a negative selection-based procedure, according to the manufacturer’s instructions (Miltenyi CD8+ T cell isolation kit; Sanquin Reagents; obtained purity >95%). Dividing effector E1A TCR-Tg CD8+ T cells were then sorted by FACS, based on their level of CFSE and Thy-1.2 expression. Sorted cells were ≥95% pure, and no naive undivided CD8+ T cells could be detected in the sorted population. A total of 75,000 of these cells was reinjected in naive Thy-1.1 recipient mice. Survival and development of memory cells were then assessed 3 wk later by s.c. injection of 108 PFU adenovirus (strain AdTs125) in both flanks. Five days later, presence (Thy-1.2 expression) and function (IFN-γ production) of E1A TCR-Tg memory CD8+ T cells were evaluated by FACS in inguinal draining LN, as described above.
In vivo expansion of naive E1A TCR Tg CD8+ T cells following injection of LPS-matured DC, but not immature DC
We previously reported that myeloid DC matured with LPS or an agonistic anti-CD40 Ab and loaded with the E1ACTL epitope (E1A234–243) efficiently induce an E1A-specific CD8 response in naive mice, whereas immature DC fail to do so (25). To dissect the mechanisms leading to this striking difference, we have set up a well-defined in vivo system using E1A234–243-specific TCR Tg (E1A TCR-Tg) CD8+ T cells (Thy-1.2+) infused into recipient mice whose T cells express the congenic marker Thy-1.1. The use of E1A TCR-Tg T cells as a pool of naive CD8+ T cells of a single specificity permits detection of weak responses, as might happen when immature DC are used for vaccination. As a source of DC, we have chosen to use the D1 cell line, which has the advantage of remaining fully immature in culture, as indicated by very low levels of costimulatory molecules and MHC class I and intermediate levels of MHC class II expressed on their surface (Fig. 1). Furthermore, immature D1 cells are characterized by proliferative capacity, high Ag uptake ability, and low T cell stimulatory efficiency, thus behaving as classical immature DC (22, 24, 25). Upon culture in presence of the bacterial component LPS, D1 cells can be activated to full maturation, resulting in arrest of cell growth, low Ag uptake, and high expression of costimulatory and MHC molecules at the cell surface and IL-12 secretion (Fig. 1) (22, 25).
We first performed a time-course experiment in which we analyzed the in vivo response of the E1A TCR-Tg T cell population following injection of peptide-loaded, immature vs LPS-matured DC. As shown in Fig. 2, Tg CD8+ T cells expanded vigorously in both the blood and the spleen of recipient Thy-1.1 mice when LPS-matured DC were injected. In contrast, the percentage of Thy-1.2+CD8+ cells did not significantly increase following injection of immature DC and remained comparable to the percentage of Thy-1.2+CD8+ cells present in animals that did not receive DC, or that were injected with unloaded LPS-matured DC (data not shown). Also, when together with injection of LPS-matured DC, CD28-B7 interactions were blocked, accumulation of the Tg T cells did not take place (data not shown).
Ag presentation by immature DC is not associated with T cell ignorance, but stimulates weak E1A-specific CD8+ T cell responses
To find out whether immature DC were capable at all to present the E1A epitope to T cells, we next examined in detail the fate of naive E1A TCR-Tg CD8+ T cells following injection of Ag-loaded DC. We made use of CFSE-labeled E1A TCR-Tg T cells to directly study the proliferative response of T cells upon encounter with Ag-presenting immature or LPS-matured DC in vivo. As shown in Fig. 3 A, we detected strong Ag-specific T cell proliferation in the spleen 3 days after i.v. injection of LPS-matured DC. This extensive proliferation resulted from the priming of most naive T cells, as shown by the few undivided cells left, and from proper accumulation of the activated cells, as shown by the increase in cell counts correlating with the decrease in CFSE label for each cell division. The accumulation of cells that had divided more than six times is very clear at day 5 after immunization with LPS-matured DC. At day 8, most accumulated cells had already disappeared from the spleen (data not shown). This was probably due to the death of part of them, as a high number of annexin V+ Thy-1.2+ cells could be detected in the liver, and to the migration of others to nonlymphoid organs (data not shown). The profile of T cell divisions that we observed when immature DC were injected was drastically different. We detected proliferation in the spleen on day 3, indicating that E1A TCR-Tg T cells had encountered and recognized specific Ag. However, this proliferation was very weak and undivided cells were still abundant. Moreover, in contrast to what we observed following injection of LPS-matured DC, accumulation of the dividing T cells was not detectable upon proliferation, neither on day 3 nor on day 5.
To show that the observed proliferation was the consequence of direct presentation of the E1A epitope by the injected DC, and not the result of cross-priming by host APC, we performed similar experiments in TAP−/− mice. APC present in TAP−/− mice are not able to present Ags in an MHC class I context due to their inability to transport Ags into the endoplasmic reticulum. Any E1A-specific response detected in these mice would thus be induced by the injected DC. As shown in Fig. 3 B, E1A TCR-Tg cells showed similar proliferation in TAP−/− and in wild-type mice, indicating that the observed T cell responses were resulting from direct priming by the injected peptide-loaded DC.
Both immature and LPS-matured DC are able to prime naive CD8+ T cells into functional effector T cells
Although proliferation is one important characteristic of CD8+ T cell activation, it does not necessarily correlate with the acquisition of functional properties. Naive CD8+ T cells that encounter specific Ag in a noninflammatory context reportedly can divide, but do not develop effector function (26). In these systems, steady-state DC were suspected to be responsible for the defective CD8+ T cell responses observed. Therefore, we decided to study cytokine production and cytotoxic effectiveness of CD8+ T cells in response to interactions with immature vs mature DC. As shown in Fig. 4 A, LPS-matured DC induced vigorous proliferation and accumulation of effector E1A TCR-Tg cells that were able to produce high levels of IFN-γ from their fourth round of division upon short restimulation with peptide in vitro. Interestingly, the small number of CD8+ T cells that were proliferating in response to the Ag presented by immature DC was able to produce the effector cytokine IFN-γ to the same extent as cells stimulated by LPS-matured DC, as shown by similar frequencies of cytokine-producing cells after each round of division. Similar results were obtained when we tested TNF-α production, except that it was produced from the first T cell division. In all cases, we could not detect any IL-4 or IL-10 (data not shown).
In vivo cytotoxicity assays led us to the same conclusion, in that both E1A TCR-Tg cells primed by LPS-matured DC and those primed by immature DC are endowed with effector function. We could observe specific killing of Ag-pulsed target cells when either LPS-matured or immature DC were used for immunization (Fig. 4 B). The cytotoxic activity was lower in the group of animals that had received immature DC and reached only 59% as compared with 85% of target cells killed when LPS-matured DC were injected. This result most likely reflects a quantitative rather than a qualitative difference between the two types of T cell responses induced, as it correlates with the numbers of CD8+ T cells that become activated in each situation.
In contrast to effector CD8+ T cells primed by LPS-matured DC, effector CD8+ T cells primed by immature DC do not down-regulate molecules involved in homing to lymphoid organs
Next to the study of proliferation and function, we performed a phenotypic analysis of E1A TCR-Tg CD8+ T cells in the spleen at day 3 upon interaction with immature or mature DC in vivo (Fig. 5). When infused into Thy-1.1+ hosts, E1A TCR-Tg CD8+ T cells harbor a phenotype typical of naive CD8+ T cells, which is characterized by low surface expression of CD69 and CD44, and high surface expression of CD62L and CCR7, molecules involved in homing to lymphoid organs. They retain that phenotype when no Ag is introduced in recipient mice. As expected, injection of Ag-loaded LPS-matured DC leads to an effector phenotype with a clear increase of the early activation marker CD69 and of CD44 expression on the surface of dividing T cells. Simultaneously, the T cells down-regulate CD62L and CCR7 expression, which allows them to recirculate and patrol the periphery in search of their specific Ag. In contrast, after encounter with Ag-loaded immature DC, E1A TCR-Tg T cells did not down-regulate the expression of CD62L and CCR7, despite up-regulation of the activation markers CD69 and CD44. These results indicate differential regulation of the expression of some molecules involved in T cell migration upon interaction with immature vs mature DC and suggest that the migratory capacity of CD8+ T cells activated by immature DC is hampered.
Quality of the signal given by DC rather than quantity of Ag available is responsible for different CD8+ T cell phenotypes
As described, naive CD8+ T cells that encountered specific Ag in the context of immature DC showed an altered (down-)regulation of two molecules involved in homing capacity, CD62L and CCR7. Another noticeable difference between stimulation mediated by immature DC vs LPS-matured DC was the number of naive CD8+ cells that enter into division. The peak of undivided cells was much higher on day 3 in the spleen of animals that had received immature DC compared with animals that had received LPS-matured DC (see Fig. 2). These observations could be related to either a qualitative or a quantitative difference in the priming signal given by immature vs mature DC to naive T cells. A quantitative difference could be caused by different intrinsic migratory capacities of immature vs LPS-matured DC, leading to an unequal ability to reach the T cell area in the spleen, where they can interact with naive T cells. To study the homing potential of mature vs immature DC, we studied the intrasplenic localization upon i.v. injection of 1 × 107 CFSE-labeled D1 cells, 24 h after injection. Fig. 6 shows a striking difference in the homing capacity of both subsets of DC. LPS-stimulated DC all localize efficiently in the splenic white pulp (Fig. 6,A). These cells are large in size (10–15 μm), showing cytoplasmic extensions, and they all intermingle with CD4+ T cells. In contrast, only very few of the CFSE-labeled immature DC were found in the splenic T cell zones. These cells were much smaller in size (6–9 μm) compared with mature DC (Fig. 6,B). Moreover, immature DC were found in the splenic red pulp (Fig. 6 C) as small round cells. Although prominent differences in the numbers of DC reaching the splenic T cell zones were observed, the number of DC that was found in the total spleen was much more comparable between the different DC types (0.06 vs 0.08% of total splenic DC consisted of D1 cells, as indicated by the CFSE label). These observations indicate that immature DC have difficulties in localization into the T cell zones upon injection. Even though we cannot exclude the possibility that T cell priming could take place in the splenic red pulp, the extent of CTL activation by immature vs mature D1 cells correlates with the relative numbers of D1 that are found in the T cell zones, suggesting that this location is indeed critical for CTL priming. Consequently, the few immature D1 cells that do reach the T cell zones could be responsible for the observed low-level T cell priming.
To investigate whether this difference in homing capacity could cause the differences found in T cell responses, we decided to titrate down the injected number of LPS-matured DC to identify conditions in which the priming of naive T cells would be quantitatively similar to that obtained with 1 × 106 immature DC, as defined by the number of T cells entering into the division process. As shown in Fig. 7, injection of 0.1 × 106 LPS-matured DC and 1 × 106 immature DC initiated division of similar numbers of E1A TCR-Tg T cells, because in both situations 88% of E1A TCR-Tg T cells are left undivided in the spleen at day 3. Importantly, under these conditions, a clear difference in CD62L down-regulation was still observed between animals immunized with immature DC and those immunized with LPS-matured DC, whereas expression of CD44 (Fig. 7) and CD69 (data not shown) was equally elevated in both T cell populations, indicating that the quality of the signal transmitted by DC, rather than the number of DC arriving at the site of CTL priming, accounts for the difference in phenotype of effector CD8+ T cells.
CD8+ T cells primed by immature DC can be activated to full effector function by adenovirus infection
All experiments described above were focused on the early effector phase of the CD8+ T cell response induced by immature vs LPS-matured DC. We subsequently examined whether CTL priming by immature DC, as compared with mature DC, eventually resulted in long-lasting memory or in deletion or functional silencing of the primed CTL. E1A-TCR Tg CTL, primed in vivo by either immature or mature DC, were isolated from splenocytes by FACSorting on basis of the following three criteria: expression of Thy-1.2, CD8, and at least two cell divisions as determined by CFSE intensity. Obtained purity was ≥95%, and no undivided naive cells were demonstrable in the sorted population. Equal numbers of sorted E1A TCR-Tg cells derived from animals primed with immature or LPS-matured DC were subsequently transferred into naive Thy-1.1 mice, where they were allowed to rest for 3 wk. Their capacity to expand and respond to secondary antigenic stimulation was assessed by immunization of the mice with adenovirus. As shown in Fig. 8,A, only very few Thy-1.2+ E1A TCR-Tg T cells were detectable in the draining LN 5 days after virus injection. However, 3 days later, Thy-1.2-expressing cells had greatly expanded and, more importantly, they had expanded to the same extent in the two settings (T cells primed with immature DC vs T cells primed with LPS-matured DC). According to these results, we conclude that memory T cells have been generated in our system and that naive CD8+ T cells did not die, nor became anergic following encounter with immature Ag-presenting DC. Moreover, memory CD8+ T cells shaped after primary immunization with either immature or LPS-matured DC do not only proliferate upon viral restimulation, but also produce IFN-γ (Fig. 8 B), indicating that they are functional.
To examine the role of DC activation in tolerization vs priming of CTL, we conducted a series of experiments using the well-defined DC cell line D1, either immature or maturated with LPS and ex vivo loaded with the adenoviral E1A234–243 CTL epitope as a source of APC, and TCR Tg T cells with specificity for this H-2Db-restricted E1A epitope.
Our experiments show that, in this experimental setup, naive CD8+ T cells encountering Ag on immature DC are neither ignorant nor tolerized, but instead are activated to central memory-like poised T cells. These CD8+ T cells are retained in lymphoid organs such as the spleen, because they have not down-regulated the lymphoid-homing receptors CCR7 and CD62L. This probably prevents them from exerting appreciable effector function in the periphery, despite their ability to specifically lyse targets that are delivered to the lymphoid organs, and to produce cytokines (IFN-γ and TNF-α). In addition, lower numbers of CD8+ T cells proliferated and accumulated over 5 days when stimulated with immature DC, contributing to a decreased magnitude of the CD8+ T cell response compared with stimulation with mature DC. At first sight, these data seem compatible with previous reports concerning the induction of functional tolerance of CD8+ T cells following priming by immature Ag-loaded DC (12, 19, 27, 28) (reviewed in Ref. 7).
We show that the CD8+ T cells primed by immature DC still fully retain the capacity to develop a robust CD8+ CTL effector response upon appropriate secondary stimulation with Ag, in this case adenovirus, to the same extent as those primed by mature DC. Our findings provide an explanation for the results published by Redmond et al. (20), who also observed that T cells stimulated by quiescent APC could be rescued to cells with full-blown effector function after transfer into an Ag-free environment. Although T cells develop effector function after Ag encounter on immature DC, they do not develop into full-blown effector cells. Because these cells do not down-regulate expression of CD62L and CCR7, they are not capable of exiting from the lymphoid organs and migrating to the periphery to exert peripheral effector function. The suboptimally activated T cells, instead of being deleted, develop into central memory-like T cells, capable of responding to secondary encounter with the specific Ag in immunizing conditions. This is in accordance with data published by Sallusto et al. (29), in which it is indicated that naive T cells, when stimulated, can give rise to central memory T cells that can be further differentiated into effector memory cells upon effective stimulation (reviewed in Ref. 30). Furthermore, the studies presented in this work provide formal proof, suspected from our earlier studies with adenovirus-induced tumors, that Ag presentation by immature DC to either TCR-Tg T cells (this study and Ref. 31) or wild-type naive T cells (31) leads to neither tolerance nor anergy, but to an intermediate type of memory CD8+ T cell. These T cells are confined to the lymphoid organ in which they were primed, poised for action upon appropriate secondary stimulation by adenovirus (this study) or by activation of DC that naturally cross-presented adenoviral Ags in tumor-draining LN through molecularly defined triggers of DC activation such as CpG-DNA, LPS, or agonistic anti-CD40 mAb (31).
In a system based on cross-presentation of a model self-Ag, i.e., the influenza virus-derived hemagglutinin, the importance of Ag persistence in the process of tolerance induction via deletion of naive CD8+ T cells was recently emphasized (20, 32). Because APC, i.e., DC, themselves are the targets of cytotoxic CD8+ T cells in the course of the response and that immature DC are less resistant to this killing due to their low expression of antiapoptotic molecules (33), it is likely that the immature D1 cells we injected were rapidly eliminated following activation of E1A TCR-Tg T cells. In our model, in contrast to Hernandez et al. (32) and Redmond et al. (20), no constant (re-)expression of Ag from endogenous sources occurs. It would be interesting to assess whether in our model the measured T cell response would be different in case of Ag persistence, e.g., by repeated injection of Ag-loaded immature DC.
Recent studies have shown that CD4 T cell help is required during the priming phase to generate powerful memory CD8+ T cells (34, 35, 36). Our data, obtained from experiments performed in CD4-depleted animals, suggest that, in certain situations, a good recall CD8 memory response can be obtained without the initial participation of CD4+ T cells. Our experiments do not exclude that CD4+ T cells may have played a role in the survival of activated CD8+ T cells once these have been transferred into naive CD4-proficient recipient mice, as well as during the re-expansion of the generated memory CD8+ T cells upon virus exposure.
T cells that interacted with immature D1 cells in vivo could still be reactivated to proliferate and to produce IFN-γ. As the cells did not down-regulate the expression of CCR7 and CD62L, they can probably best be compared with central memory T cells, as described by Sallusto et al. (37). Indeed, in different studies, the development of central memory cells was described to be the result of T cell differentiation after priming by a signal of suboptimal strength (29, 38, 39), like that provided by immature DC. The central memory-like phenotype of T cells could be accountable for capture of tumor-specific T cells in tumor-draining LN (40). The absence of tumor-specific T cells outside tumor draining LN in tumor-bearing animals can probably be explained by lack of down-regulation of molecules involved in migration into lymphoid organs, such as CCR-7 and CD62L on T cells interacting with their specific Ag on immature DC.
Differences in T cell accumulation could be explained by differential expression of cytokine receptors on T cells activated under different circumstances (38). Higher expression of IL-2Rα chain, and the common cytokine receptor γ-chain for IL-2, IL-7, and IL-15, observed on T cells that received a longer activation stimulus, correlated well with survival of those T cells in an Ag-free environment. It should be remembered in this work that the conditions for costimulation might differ between in vitro and in vivo systems (41). We did not observe a difference in expression of these receptors in our in vivo system (data not shown). Neither did we observe any differences in expression of markers for apoptosis (data not shown) in response to the different DC maturation states. However, in accordance with the in vitro results of Gett et al. (38), we did see a decline of T cell accumulation when CD28-B7 interactions were blocked in vivo (data not shown).
Although Ag-loaded immature DC can prime naive E1A TCR-Tg T cells into functional effectors, as characterized by cytotoxicity and cytokine production, the amplitude of the response they induce is much lower than that induced by LPS-matured DC. The weak migratory capacity of immature DC compared with their mature counterparts seems to be at least partly responsible for this quantitative difference. As reported previously for intradermally injected human DC (42), we show that i.v. administered immature D1 cells do not efficiently reach splenic T cell areas, whereas LPS-matured DC do. This observation correlates with the levels of homing-related molecules measured on the surface of the immature and mature D1 cells (data not shown). Among these molecules, the chemokine receptor CCR7 may be involved in the entrance of DC into the splenic white pulp. Experiments taking advantage of CCR7-knockout LPS-matured DC and/or CCR7-transfected immature DC are necessary to confirm this hypothesis.
Another strategy to study CD8+ T cell responses induced by DC in steady-state conditions consists of specifically targeting the Ag to DC. In an elegant study with inducible expression of lymphocytic choriomeningitis virus-derived epitopes in resting vs CD40-activated CD11c+ cells in vivo, Probst et al. (28) described CD8+ T cell tolerance vs immunity, respectively. However, the exact identity and phenotype of the DC population involved in tolerance induction in that study were not known. Bonifaz et al. (19) chose to target Ag to DC via the DEC-205 receptor using OVA coupled to an anti-DEC-205 mAb and showed proliferation, followed by deletion of naive CD8+ T cells in a steady-state situation. A very interesting, but not exclusive possibility would be that, in the absence of a DC maturation trigger, the DEC-205 receptor itself does not only mediate Ag uptake and presentation, but also provides some specific signals to DC, rendering them tolerogenic. This has been proposed for integrins, another type of endocytic receptor involved in mechanisms of CD8+ T cell cross-tolerization via capture of apoptotic cells by DC (43). Of note, only a subset of DC, namely CD8+ DC, expresses the DEC-205 C-type lectin, and these cells have been shown to harbor regulatory properties via various mechanisms, e.g., Fas ligand-mediated killing of the CD8+ T cells with which they interact (44). Moreover, the CD8+-expressing DC subset has been reported to tolerize CD8+ T cells through cross-presentation of cell-associated Ag, supporting the theory that T cell tolerance induction may be due to a specialized tolerizing population of DC (45, 46, 47). This could explain the lack of induction of actual T cell tolerance in vivo by immature DC, because D1 cells are CD8α−, i.e., of myeloid-like origin.
In conclusion, our results show that presentation of Ags by immature DC in itself is not sufficient for complete tolerization or deletion of Ag-specific CD8+ T cells. A qualitative difference exists in the type of CD8+ T cells that are generated after interaction with Ag-loaded immature vs LPS-matured DC. This difference fits into the signal strength model proposed by Gett et al. (38), in which naive T cells differentiate into either central memory cells or effector cells upon suboptimal or optimal stimulation, respectively. In view of development of therapeutic immune intervention strategies against infectious diseases and cancer, the good news is that such suboptimally activated CD8+ T cells can still be rescued to launch a robust effector CD8+ T cell response. In fact, we hypothesize that the poised central memory-like CD8+ T cells found following triggering by immature Ag-loaded DC might occur frequently under conditions of appropriate (cross-)presentation, but as yet lack of DC activation, such as in incipient infections. These CD8+ CTL might expand to full-blown effector cells following subsequent DC activation, for example, when the incipient infection is not contained. In other words, these reserve soldiers are lying in wait until real danger arises and alarms them.
The authors have no financial conflict of interest.
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.
This work was supported by Dutch Cancer Foundation Grant RUL 99-2025 and The Netherlands Organization for Scientific Research Grant 901-07-097.
Abbreviations used in this paper: DC, dendritic cell; LN, lymph node; Tg, transgenic.