T Cells Signaled by NF-κB⁻ Dendritic Cells Are Sensitized Not Anergic to Subsequent Activation¹

Ligation of TCR and costimulatory signaling by CD80/86, CD40, and IL-12 are key requirements for immune response control. In particular, the absence of either CD80/CD86, CD40, or IL-12 expression by APCs may lead to the induction of tolerance. Blockade of key T cell signaling components may be exploited directly or through targeted inhibition of upstream intracellular signaling pathways in APC for therapeutic benefit (1, 2). The NF-κB family member, RelB, is an essential upstream factor controlling expression of CD40 by dendritic cells (DC).³ Furthermore, when DC are generated from bone marrow (BM) precursors in the presence of an inhibitor of NF-κB nuclear translocation (Bay 11-7082) they lack CD40 but express CD86 and MHC class II (3, 4). The NF-κB inhibitor, Bay 11-7082, used in the current studies is an irreversible specific inhibitor of IκB phosphorylation. It, and a related compound, have potent systemic anti-inflammatory effects, equivalent to those of glucocorticoids (3, 5). In vivo, myeloid DC in afferent lymphatics and in resting peripheral LN share a similar CD40^lowCD86⁺MHC class II⁺ phenotype (6, 7). In the absence of NF-κB inhibition, CD40^low immature DC up-regulate CD40 expression upon exposure to bacterial signals mediated by TLR (Ref.7). In addition, CD40 signaling by CD154⁺ T cells, following engagement of the TCR, provides critical stimulation for sustained activation of NF-κB, up-regulation of costimulatory molecules, and secretion of IL-12 and proinflammatory cytokines, “conditioning” the DC for efficient induction of CTL and Th1 responses (7, 8, 9).

Recent studies have shown that in the presence of autologous DC and in the absence of exogenous Ag, T cells are signaled for Ca²⁺ response, tyrosine phosphorylation, survival, and weak proliferation. Signaling is mediated by immunological synapses between naive T cells and DC, and can occur in the absence of MHC molecules (10, 11). Moreover, following adherence of resting primary T cells to various substrates or to DC, T cells increase intracellular Ca²⁺ stores and become more sensitive to stimulation through the TCR (12). In vivo, self-peptide recognition in the context of MHC class II molecules can sensitize T cells for exogenous Ag reactivity (13, 14). Therefore, it has been proposed that autologous MHC class II⁺ APC, presenting self-peptides, provide constitutive T cell stimulation in the periphery, thereby maintaining naive T cells in a state of heightened responsiveness. Upon Ag exposure in the presence of activating signals, T cells would be poised to mount a swift and effective response. This paradoxical conclusion contrasts with the suppression of T cell responsiveness by self-peptides proposed to occur as a mechanism to maintain self-tolerance in the periphery.

CD40 is an important costimulatory molecule expressed by APC. Murine BM DC treated with an NF-κB inhibitor, presenting exogenous Ags in the absence of CD40 are able to induce Ag-specific tolerance through induction of IL-10 producing regulatory T cells in vivo. We previously speculated that CD40^low DC would constitutively present apoptotic self Ags in the periphery (15), and that regulatory T cells specific for somatic self Ags would be constitutively cross-tolerized by CD40^low APC, and contribute to active postthymic self tolerance (16).

Given that MHC class II⁺ APC-bearing self Ag, and populating lymphoid organs of naive animals, might have the capacity either to sensitize or to anergize T cells, we examined interactions between CD40⁻ DC in which NF-κB signaling is blocked and T cells. Based on our in vivo studies, we hypothesized that CD40⁻CD86⁺MHC class II⁺ DC would induce T cell anergy and regulatory function in vitro. By generation of CD40⁻CD86⁺HLA-DR⁺ DC in GM-CSF, IL-4, and Bay 11-7082, an inhibitor of NF-κB (Bay-treated DC), we show that T cells are signaled for TCRζ phosphorylation, but low levels of proliferation and IFN-γ production. Surprisingly, T cells are neither anergic nor regulatory, but are sensitized for subsequent IFN-γ production.

Human peripheral blood buffy coats were obtained from Red Cross Australia (Brisbane, Queensland, Australia), or peripheral venous blood (PB) drawn from healthy volunteers; all blood was anticoagulated with heparin (David Bull Laboratories, Melbourne, Australia). The study was approved by the Princess Alexandra Hospital Research Ethics Committee and the University of Queensland Medical Research Ethics Committee (Queensland, Australia). PBMC were prepared using a Ficoll-Paque (Amersham Pharmacia Biotech, Uppsala, Sweden) gradient separation. Cells were washed in 0.9% saline and rosetted using neuraminidase (Sigma-Aldrich, St. Louis, MO)-treated sheep RBC (bioMeurieux, Baulkham Hills, New South Wales, Australia; Ref.17). The rosette⁻ non-T and rosette⁺ T cell populations were separated on Ficoll-Paque gradients. Neuraminidase-treated sheep RBC were lysed with 1 M ammonium chloride. For T cell purification for proliferation assays, the rosette⁺ T cells were enriched by negative selection using mouse anti-HLA-DR (DAKO, Carpenteria, CA), anti-CD14, anti-CD19, anti-CD16, and anti-CD56 (BD Biosciences, San Jose, CA) mAb followed by goat anti-mouse Ig-conjugated immunomagnetic bead depletion using a MACS column to manufacturer’s specifications (Miltenyi Biotec, Bergisch Gladbach, Germany) as described (18). T cell purity was >90% on flow cytometry by anti-CD3-FITC labeling (BD Biosciences). For CD4⁺ and CD8⁺ T cell experiments, anti-CD8 and anti-CD4 (DAKO) mAb were added to the depletion mixture respectively. Myeloid-enriched non-T cells were prepared from PBMC by positive selection with CD14⁺ microbeads (Miltenyi Biotec) or by preparation from rosette⁻ non-T cells by negatively selecting with mouse anti-CD16, anti-CD19, anti-CD56, and anti-CD3 (all from BD Biosciences) mAb followed by immunomagnetic bead depletion (Miltenyi Biotec).

All cells were cultured in RPMI 1640 (Invitrogen Life Technologies, Gaithersburg, MD), supplemented with 10% FBS (CSL, Melbourne, Australia) and penicillin G (200 U/ml; CSL), gentamicin (10 mg/ml), and l-glutamine (0.3 mg/ml; Trace Biosciences, New South Wales, Australia. For immature DC, CD14⁺ monocytes or rosette⁻ non-T cells, enriched for monocytes by immunomagnetic bead depletion, were cultured in medium (as described above) supplemented with 800 U/ml GM-CSF (Schering-Plough, Sydney, Australia) and 400 U/ml IL-4 (Sigma-Aldrich) for 48 h (19). Mature DC were cultured for the final 24 h with 500 ng/ml LPS (Sigma-Aldrich). Bay-treated DC were incubated continuously in the presence of 2–20 μM Bay 11-7082 (BIOMOL, Plymouth Meeting, PA) at 1 × 10⁶/ml in 24-well plates (Table I). DC generated from individual batches of Bay 11-7082 were tested for viability and CD40 expression. The chosen concentration of each batch of Bay 11-7082 fully suppressed CD40 expression and maintained DC viability. DC viability was not compromised at concentrations of Bay 11-7082 up to 20 μM. In some experiments, DC were stimulated with 50 ng/ml soluble CD154 trimer (Immunex, Seattle, WA) for the final 24 h of a 48-h culture.

RelA, RelB, c-Rel, and p50 DNA binding was detected by ELISA using a modification of the Mercury Transfactor p50 kit (Clontech Laboratories, Palo Alto, CA). A total of 20 μg of nuclear extract were bound to wells coated with NF-κB consensus oligonucleotides, then incubated with anti-RelA, anti-RelB, anti c-Rel, or anti-p50, followed by anti-rabbit HRP-conjugated Ig, and then detected by measuring color development of tetramethylbenzidine at 650 nm using a Multiskan plate reader (Labsystems, Chicago, IL). Cytoplasmic extracts were prepared and protein estimations conducted using a protein assay kit (Bio-Rad, Richmond, CA). Twenty micrograms of protein extract were separated by 10% SDS-PAGE, and after transfer to nitrocellulose (Amersham Biosciences, Sussex, U.K.), membranes were immunoblotted with either anti-RelB (sc-226), anti-p50 (sc-7178), anti-RelA (sc-109), anti-c-Rel (sc-70), or anti-α-tubulin (Sigma-Aldrich) Abs (sc Abs from Santa Cruz Biotechnology, Santa Cruz, CA). All Abs were detected by incubation with anti-rabbit HRP-conjugated Ig (Sigma-Aldrich) followed by ECL (Amersham Biosciences).

Forty-eight-hour DC were washed and stained with FITC labeled anti-CD40, anti-CD86 (both from BD PharMingen, San Diego, CA), anti-CD80 (Immunotech, Marseille, France), anti-HLA-DR (Sigma-Aldrich), and control FITC negative (DAKO) and analyzed on a BD FACSCalibur, with a single argon laser, gated on the live population, and analyzed on CellQuest software (BD Biosciences). Viability of T cell populations was determined by staining with propidium iodide (Bender MedSystems, Vienna, Austria) and running the samples on the FL-2 channel. For CD154 expression analysis, fresh T cells or T cells incubated with different DC for 1 h were washed and labeled with anti-CD154 PE (BD PharMingen) and control PE negative (DAKO) and analyzed by flow cytometry on the FL-2 channel.

Purified allogeneic PB T cells were rested for 30 min at 37°C, then immature, LPS-treated, or Bay 11-7082-treated DC were added at a 1:2 ratio. Cells were contacted by brief centrifugation, then incubated for 5 min at 37°C. Control T cells were stimulated with anti-CD3 or no mAb (UCHT1; Biolegend, San Diego, CA) for 5 min. Cell lysates were prepared in M-PER buffer (Pierce, Rockford, IL) containing 1 mM Na₃VO₄ and protease inhibitor mixture (Boehringer Ingelheim, Ingelheim, Germany). ZAP70 was immunoprecipitated overnight at 4°C with anti-ZAP70 (Cell Signaling Technology, Beverly, MA) and analyzed by 15% PAGE under nonreducing conditions. Phosphorylated proteins were detected using a phosphotyrosine-specific Ab (Upstate Biotechnology, Lake Placid, NY) followed by anti-rabbit HRP-conjugated Ig and ECL.

Forty-eight-hour mature, immature, and Bay-treated DC were added at varying concentrations, between 1/20 and 1/80, to 10⁵ freshly purified allogeneic or autologous T cells. DC were washed before incubation with T cells. Cells were cultured in round-bottom 96-well microtiter plates (Techno Plastic Products, Trasadingen, Switzerland). Plates were incubated at 37°C for 5 days for primary stimulations or 3 days for secondary stimulations. T cell proliferation was measured by the uptake of [³H]thymidine (1 μCi/well, 6.7 μCi/mM; ICN Pharmaceuticals, Costa Mesa, CA) which was added for the final 18-h of the culture period for primary cultures and 12 h for secondary cultures. Cells were harvested onto glass fiber filter paper with an automated 96-well harvester (Packard Instruments, Meridien, CT), and [³H]thymidine incorporation was determined by liquid scintillation spectroscopy on a TopCount NXT (Packard Instruments). The responses are reported as the mean cpm ± SEM for triplicate wells. T cells alone or T cells previously stimulated with DC were activated using a combination of 20 ng/ml PMA and 1 μM ionomycin (both from Sigma-Aldrich) for 24 or 72 h.

For cell division analysis by CFSE, labeled T cells were stimulated by DC and after 5 days, analyzed by flow cytometry on the FL-2 channel. Modfit software (Verity Software House, Topsham, ME) was used to optimize the division generations and calculate the proportion of cells in each generation. Cell cycle analysis was performed on T cells stimulated by DC. Cells were harvested, washed in PBS, and fixed in 70% ethanol. Cells were washed again and stained with a solution of propidium iodide (4 mg/ml) and RNase A (400 mg/ml) in PBS. Cells were analyzed by flow cytometry as above, and further analyzed using Modfit software (Verity Software House) for cell cycle analysis by exclusion of debris and apoptotic cells, and determination of the proportion of cells in S-phase of the cell cycle at the time of harvest. For IL-2 experiments, recombinant human IL-2 (Sigma-Aldrich) was added to T cell proliferation assays between 1 and 100 μM.

IFN-γ concentration was assessed in supernatants obtained from DC stimulated T cell cultures after 48 h for primary stimulations, and 24 h for secondary stimulations, IFN-γ production was assessed using the Quantiferon kit (Cellestis, Melbourne, Australia) according to the manufacturer’s instructions. The Quantiferon plate was read using a Multiskan plate reader (Labsystems) at 492 nm. Cells activated as described above were also stained for intracellular cytokines after the addition of brefeldin A 10 μg/ml for the final 24 h of activation to inhibit cellular secretion. Using a Cytoperm/Cytofix kit (BD PharMingen) cells were fixed and permeabilized according to the manufacturer’s instructions, and stained with anti-IFN-γ FITC, anti-IL-2 FITC (BD PharMingen), or control FITC negative (DAKO), and analyzed by flow cytometry on a BD FACSCalibur.

Nuclear translocation of RelB is required for the up-regulation of Ag presenting ability and associated up-regulation of MHC and costimulatory molecule expression by myeloid DC (8). Immature DC generated from monocytes in the presence of GM-CSF and IL-4 for 48 h, display low levels of RelB DNA binding activity, which is inducible after 18–24 h incubation with LPS, TNF-α, or soluble CD154 (8). DC generated from monocytes in the presence of GM-CSF, IL-4 without (immature DC, Table I) or with Bay 11-7082 for 48 h (Bay-treated DC) demonstrated low levels of RelA, RelB, c-Rel, and p50 DNA binding. NF-κB DNA binding did not increase in response to LPS in Bay-treated DC in keeping with the inhibition of NF-κB activity (Fig. 1,A). In contrast, mature DC, generated from monocytes in the presence of GM-CSF and IL-4 for 48 h and exposed to LPS for the final 24 h, displayed high levels of NF-κB DNA binding relative to immature or Bay-treated DC (Fig. 1 A), consistent with previous observations (8).

To determine the effects of Bay 11-7082 on cytoplasmic NF-κB levels, cytoplasmic extracts from normal and Bay-treated monocyte-derived dendritic cells (MDDC) were immunoblotted for NF-κB (Fig. 1 B). Immature MDDC expressed cytoplasmic RelB, p50, p105, and RelA (lane 1). Stimulation of MDDC for the final 24 h of culture with LPS induced cytoplasmic RelB, p50, RelA, and c-Rel and p105 (lane 2). Culture with Bay decreased basal cytoplasmic levels of RelB, p50, RelA, and p105 (lane 3). Except for RelA, no induction of NF-κB was observed in Bay-treated DC following the addition of LPS during the last 24 h of culture (lane 4). Taken together, the data demonstrate that treatment of MDDC with Bay reduces both the cytoplasmic pool and nuclear DNA binding of NF-κB compared with either immature or LPS-treated DC, and are consistent with previous analysis of Bay-treated murine BM DC (4). The reduced cytoplasmic pools likely reflect transcriptional autoregulation of NF-κB coupled with cytoplasmic degradation, as previously shown for RelB (20).

Mature DC, as expected, expressed relatively higher levels of HLA-DR, CD40, CD80, and CD86 than immature DC. (Fig. 1,C) Bay-treated DC expressed similar levels of HLA-DR, CD80, and CD86, but much lower levels of CD40. The levels of expression of these molecules are generally lower for DC generated over 48 h than over 7 days (8, 21). All DC expressed CD54 (data not shown). Bay-treated DC resemble immature DC, in that NF-κB activity is low, but differ from mature DC in their capacity to respond to a signal that drives NF-κB activation. We further tested this observation by comparing the phenotype of immature and Bay-treated DC in response to a CD40 signal. Bay-treated DC failed to up-regulate CD86 and HLA-DR following treatment with soluble CD154, in keeping with their lack of CD40 expression and NF-κB blockade. In contrast, immature DC up-regulated CD86 and HLA-DR in response to soluble CD154, while neither DC up-regulated CD80 (Fig. 1 D). Therefore, Bay-treated DC, in which NF-κB activation is blocked, lack CD40 expression but express CD80, CD86, and HLA-DR, and neither alter costimulatory molecule expression nor nuclear NF-κB DNA binding in response to CD40 ligation or LPS stimulation. In subsequent experiments, Bay-treated DC were therefore used as a model to examine the T cell response to DC that lack CD40 and NF-κB signaling, but express MHC class II, CD54, and the costimulatory molecules, CD80 and CD86.

By examining the response of 48-h DC to CD40 ligation, we previously demonstrated an association between increased NF-κB activity, DC maturation, and T cell proliferative responses in vitro (8). Therefore, we used Bay-treated DC, in which NF-κB activation is blocked, to test the hypothesis that DC signaling through NF-κB determines the capacity of T cells to proliferate in vitro. Using [³H]thymidine incorporation as a readout of T cell DNA synthesis, allogeneic CD4⁺ or CD8⁺ T cell proliferation in response to DC was reduced by prior treatment with Bay, when compared with untreated immature or mature DC (Fig. 2,A). T cell IFN-γ production was also reduced in response to Bay-treated DC (Fig. 2,B). Furthermore, when DC were generated in the presence of varying concentrations of Bay 11-7082 (2–20 μM), the allogeneic T cell proliferative response covaried with the concentration of the NF-κB inhibitor, and with the mean fluorescence intensity of CD40 expression by DC (Fig. 2, C and D). To exclude a direct effect of Bay 11-7082 on T cells in this experimental system, Bay-treated DC supernatant was added to PMA/ionomycin-stimulated T cells in varying concentrations. No change in T cell proliferation was observed (data not shown). The data indicate that T cell proliferation and associated IFN-γ production (22) in response to allogeneic DC stimulation in vitro is determined by the NF-κB activity of the DC. These data are consistent with previous observations, using various NF-κB inhibitors (8, 23, 24).

The next set of experiments explored the mechanism by which T cell proliferation in response to allogeneic DC was reduced by prior Bay treatment. Bay-treated DC might fail to engage T cells, and therefore fail to signal them for proliferation, possibly leading to T cell apoptosis by neglect (25). Alternatively Bay-treated DC might actively signal T cell death (26). Therefore, we examined T cell viability and death, after stimulation by DC for 5 days in allogeneic MLR. T cell numbers did not increase after stimulation by Bay-treated DC, and T cell viability was not compromised by comparison with unstimulated T cells. Thus, the lack of increase in T cell number was not due to increased cell death but due to impaired proliferation (Fig. 3, A and B). We next examined whether Bay-treated DC were capable of signaling resting T cells. TCR engagement induces phosphorylation of TCRζ and recruitment of the TCR-associated Src kinase, ZAP70. Depending on the activation signal, distinct patterns of TCRζ phosphorylation have been observed. In human T cells, TCRζ may be partially phosphorylated (p32 kDa) or completely phosphorylated (p38 kDa) depending upon the strength of T cell activation (27). To confirm T cell signaling by Bay-treated DC, rested allogeneic T cells were cultured alone, or with anti-CD3, immature, mature, or Bay-treated DC for 5 min. Anti-phosphotyrosine immunoblotting of extracts immunoprecipitated with ZAP70 revealed low levels of phosphorylated 32 kDa TCRζ in untreated rested T cells, as expected in absence of self-MHC signaling (Fig. 3,C, lane 1; Ref.13). Stimulation of T cells with anti-CD3 or immature DC induced phosphorylated 32 kDa TCRζ (lanes 2 and 3). Thirty-eight kilodalton-phosphorylated TCRζ was observed in T cells cultured with anti-CD3 and anti-CD28 (data not shown), or mature or Bay-treated DC (lanes 4 and 5). Thus, early TCR phosphorylation is induced to a similar extent by Bay-treated or mature DC. To provide further evidence of T cell signaling, markers of T cell activation were analyzed after 24 h incubation with either no stimulus (bar 4), mature (bar 1), immature (bar 2), or Bay-treated DC (Fig. 3,D, bar 3). CD62L expression was reduced after stimulation with each DC, and there were small increases in CD69 and CD154 expression relative to unstimulated T cells. HLA-DR expression was increased in T cells stimulated with mature or immature, but not Bay-treated DC relative to unstimulated T cells (Fig. 3 D). Taken together, the data indicate that Bay-treated DC engage and signal T cell viability, TCRζ phosphorylation, and some markers of activation.

To examine T cell division, CFSE-labeled allogeneic T cells were incubated with Bay-treated, immature, or mature DC. Consistent with the reduced uptake of [³H]thymidine, T cells divided relatively less in response to Bay-treated DC, when compared with immature or mature DC (Fig. 4,A). Therefore, we examined the stage of the cell cycle in which T cells arrested in response to Bay-treated DC. T cells alone or T cells stimulated with DC were stained with a solution of propidium iodide and RNase A each day for 5 days and analyzed by flow cytometry. The relative proportion of T cells in each phase of the cell cycle was determined. After stimulation by mature DC, T cells began to enter S-phase from day 2 (data not shown). At day 5, the proportion of cells in S-phase (DNA synthesis) was similar among T cells either unstimulated or stimulated by Bay-DC (Fig. 4 B), and this proportion was reduced compared with T cells stimulated by mature DC. Taken together, the data indicate that Bay-treated DC signal T cells to express CD154, and to maintain viability, but induce limited proliferation and IFN-γ production, with an impairment in progression past the G₀/G₁ resting phase in the T cell cycle.

It has been previously shown that stimulation of T cells by mature DC in which NF-κB was inhibited, or by APC lacking “signal 2” induces a state of T cell anergy (23, 28, 29). Furthermore, we have previously shown that CD40⁻ DC induce regulatory T cells in vivo, which display Ag-specific anergy when restimulated ex vivo (4). Given that Bay-DC signaled T cell activation but did not induce proliferation, we anticipated that these MHC class II⁺CD86⁺CD40⁻ DC would induce a state of T cell anergy, as well as regulatory function. To test for induction of regulatory cells in vitro, allogeneic T cells were stimulated in primary cultures with Bay-treated or mature DC, then transferred into wells in which allogeneic T cells previously stimulated with mature DC were re-stimulated with mature DC. T cells stimulated in primary cultures by Bay-treated DC did not suppress T cell proliferation in secondary cultures when compared with control T cells stimulated in primary culture by mature DC (Fig. 5,A). This remained the case even if T cells were restimulated by Bay-treated DC for several rounds before analysis for suppressor function (30). To test for anergy, T cells were stimulated in primary cultures by Bay-treated or mature autologous or allogeneic DC, washed, then restimulated in secondary cultures with either PMA/ionomycin or mature DC. Surprisingly, T cells stimulated in primary cultures by Bay-treated DC proliferated to the same extent or greater in secondary cultures than T cells previously stimulated by mature autologous or allogeneic DC (Fig. 5, B and C). Furthermore, IL-2 restored the proliferation of T cells stimulated with Bay-treated DC in primary cultures (Fig. 5,D). Unexpectedly, IFN-γ production by T cells in secondary cultures was higher if T cells had been stimulated previously with Bay-treated DC in primary culture, than if T cells were stimulated previously by mature DC or were unstimulated in primary cultures (Fig. 5, E and F). Of importance, Bay-treated DC had a positive impact on T cell function, rather than no effect at all, as autologous Bay-treated DC dose-dependently enhanced the capacity of T cells to proliferate or to produce IFN-γ when restimulated with mitogen, compared with T cells stimulated with mitogen alone. It should be noted that the absolute levels of IFN-γ produced by stimulated T cells varied between donors, but that the relative IFN-γ production by T cells in secondary cultures, after stimulation by Bay-treated or mature DC in primary culture was very consistent. Moreover, those T cells stimulated by Bay-treated DC in primary culture contained the highest proportion of cells expressing either intracellular IL-2 or IFN-γ upon restimulation with PMA/ionomycin (Table II). To test whether T cells left unstimulated in culture for 7 days were capable of producing an optimal proliferative or IFN-γ response, freshly isolated T cells were stimulated immediately or after 7 days in culture and their proliferative response and IFN-γ production were equivalent in response to PMA/ionomycin (data not shown). Thus, T cells stimulated in vitro with Bay-treated DC presenting either self- or allo-Ag do not suppress T cells stimulated by mature DC, and are not anergic. Rather, they are capable of making normal or increased proliferative, IFN-γ, and IL-2 responses when restimulated with either mitogen or mature DC.

The NF-κB family of transcription factors consists of five members, which exist in either homo- or heterodimeric forms in the cytoplasm of cells. In this state, they are bound to inhibitory factors known as IκB (31, 32). Following an activation signal delivered by various environmental factors, such as LPS or CD154, IκB is phosphorylated and ubiquitinated, allowing translocation of the NF-κB subunits to the nucleus, where they are involved in the control of gene transcription via binding to DNA κB consensus sites (33). In DC, this results in increased Ag processing and presenting ability. Moreover, the presence of RelB in the nucleus of HLA-DR⁺ cells can be used histologically to define mature DC following exclusion of activated macrophages, follicular DC, and B cells (33, 34, 35, 36). Because of the importance of CD40-CD154 interaction in the induction of the immune response, the absence of CD40 has major effects on APC function. Previous studies using DC in which NF-κB is inhibited, demonstrated that myeloid DC treated with the Bay 11-7082 NF-κB inhibitor, RelB^−/− DC, and CD40^−/− DC all induced Ag-specific tolerance when exposed to Ag and adoptively transferred to wild-type naive or previously primed recipients in vivo (4). The role of CD40 in APC has also been shown to be required for Th2 response induction, and CD8⁺ CTL priming with T help or agonist CD40 Ab (37, 38, 39). Thus, previous work indicates that CD40 expression is essential for the licensing of APC (16). However, the type of APC may influence immune response outcome, as CD40-deficient splenic APC can induce normal T cell proliferation in vitro, in contrast to CD40-deficient B cells, which stimulate reduced T cell proliferation (40).

In contrast to these previously demonstrated effects of CD40 in vitro and in vivo, the current experiments demonstrate the capacity of Bay-treated monocyte-derived DC lacking NF-κB activity, cell surface CD40 or NF-κB responsiveness, to signal resting T cells for viability, TCRζ phosphorylation, alteration of cell surface markers of activation, and indeed to sensitize those T cells for subsequent high levels of IFN-γ production. These data accord with previous observations that CD80 and CD54 accessory signals provided by APC are sufficient for induction of CD154 expression by resting T cells, and with evidence of T cell viability and TCRζ phosphorylation in vivo as a result of interactions between naive T cells and MHC class II⁺ APC presenting self peptides (13, 41, 42, 43, 44). Moreover, TCRζ phosphorylation is associated with T cell sensitization to subsequent encounters with APC (13, 45).

However, allo- or self Ag-specific T cell proliferation, progression past the G₀/G₁ resting phase, and IFN-γ production were inhibited in response to Bay-treated DC. This inhibition covaried with suppression of DC NF-κB and CD40 expression. Unexpectedly, Bay-treated DC did not render the T cells anergic or confer regulatory function, but rather sensitized the T cells for enhanced cytokine production in response to subsequent effective TCR signaling by either mature DC or PMA/ionomycin. The data presented in Fig. 5 clearly demonstrate that Bay-treated DC have a positive impact on T cell function, rather than no effect at all, as Bay-treated DC dose-dependently enhanced the capacity of T cells to produce IFN-γ when restimulated with mitogen, compared with those T cells stimulated with mitogen alone, or those previously signaled by mature DC. Although it is possible that T cells initially signaled by mature allogeneic DC might be exhausted in secondary cultures due to the large amounts of IFN-γ produced, this appears to be an unlikely explanation for the T cell response to autologous mature DC in which cytokine levels are lower.

Although the data contrast with those obtained using Bay-treated DC in vivo, they are consistent with previous publications showing that murine peripheral T cells are sensitized by interaction with class II⁺ APC-presenting self-ligands, that autologous DC signal-resting T cells in the absence of exogenous Ag or MHC class II, and that mature rather than immature DC cross-tolerize CD8⁺ T cells in vitro (10, 13, 14, 25). There are several explanations for the differences in the T cell response to Bay-treated DC in vitro and in vivo. First, it is likely that cells in intact LN other than DC and T cells contribute to the generation of T regulatory cells in vivo, and that this interaction is not reproduced in vitro. Second, architectural differences between LN and a tissue culture well may contribute to differences in T cell outcome in vitro and in vivo. Third, T cells may respond differently to Bay-treated murine BM-derived DC than to Bay-treated human monocyte-derived DC. Preliminary data obtained using murine CD40⁻ DC in vitro do not support the third possibility. In contrast, several recent papers implicate B cells, signaled through the CD40 receptor, in the IL-10-dependent suppression or resolution of autoimmune responses (46, 47). However, it has not yet been demonstrated whether this resolution occurs through induction of regulatory T cells. Expression of CD154 by T cells in response to CD86⁺CD40⁻ DC may be critical for their potential to interact with other APC, with downstream consequences for tolerance. CD40 ligation of either B cells or plasmacytoid DC by CD154 may confer upon these APC the capacity to induce tolerance (48, 49). Thus, the current data strongly suggest that signals derived from cells other than CD40⁻ DC and T cells drive the ultimate generation of Ag-specific regulatory T cells in vivo. Furthermore, the capacity of either immature DC or mature DC in the presence or absence of NF-κB inhibitors to induce T cell anergy in vitro in some previous publications (23, 30), but not others (25), and the equivalent levels of CD80/CD86 expressed by immature and Bay-treated DC, suggests one or more regulatory factors additional to CD40, signal 1 and 2 are likely to be required for T cell tolerance (25). Moreover, their relative expression must be exquisitely regulated among APC populations.

The current studies clearly demonstrate that contact with MHC class II⁺CD40⁻CD86⁺ DC in the absence of exogenous Ag is sufficient to signal T cells for survival and CD154 expression in an Ag-nonspecific manner. However, capacity of DC to activate NF-κB (thus impacting on the expression of cell surface CD40) is required for the progression of T cells through the cell cycle in response to self- or allo-Ag. The low IFN-γ production by T cells in response to Bay-treated DC is consistent with their failure to progress through the cell cycle, as previous studies have demonstrated this requirement for IFN-γ production (22, 50). The data are also consistent with a lack of production of cytokine T cell growth factors in primary culture, and restoration of proliferation by IL-2, rather than anergy. In contrast to the current data, a large body of evidence suggests that reduced CD86 expression is associated with anergy induction in T cells and in some cases, generation of regulatory T cells (28, 29, 51). Furthermore, APC other than DC, which lack CD86 expression, fail to stimulate T cell CD154 expression, and these T cells are anergic (52, 53). Inhibition of NF-κB activation in mature DC through a proteasome inhibitor results in reduced CD86, MHC class II, and CD40 expression by the DC relative to noninhibited mature DC. These DC induce regulatory T cells in vitro (23). In vivo, there is evidence that variation in the expression of CD40 and CD80/86 by APC modulates the response of T cells. Of importance, DC, which constitutively populate resting tissues and lymphoid organs in vivo, express low levels of CD80/86, similar to Bay-treated DC. Thus, low levels of CD80/CD86 expression by DC are NF-κB independent, and therefore under different control mechanisms to CD40 expression (4, 54). In contrast, CD80/86 or CD40 up-regulation in response to TLR signaling of DC is NF-κB dependent (8, 16).

There are a number of examples of DC found in vivo, with similar phenotype and function to Bay-treated DC. Rat afferent lymphatic DC constitutively reaching the gut-draining LN are loaded with apoptotic bodies derived from somatic cells and express MHC class II, CD86, but lack CD40 (6, 55). However, unlike Bay-treated DC, afferent lymphatic DC have the capacity to respond to LPS in vitro, suggesting that the peripheral tissue and afferent lymphatic microenvironments are constitutively suppressive of NF-κB, perhaps in part due to the suppressive effects of apoptotic bodies (56, 57). The current data are consistent with the hypothesis that CD40⁻NF-κB⁻MHC class II⁺ APC present somatic or endogenous self-Ags in lymphoid organs, where they signal TCRζ phosphorylation and viability (45). Upon receipt of additional signals, they may become regulatory T cells, but the cells are sensitized for future productive antigenic encounters, thereby allowing a rapid and strong response to Ags associated with activation of APC by TLR or other NF-κB activating pathways. We suggest that regulatory T cells do not constitute a default pathway driven by CD40⁻ DC, but that signals derived from cells other than DC likely further moderate Ag presentation by CD40⁻ DC to drive the generation of Ag-specific regulatory T cells in vivo.

These data challenge the current paradigm of DC as the paramount professional APC with capacity to shape the immune response. Rather, their major function may be to prime or ready T cells for subsequent modulation by other (accessory) cells and signals. Understanding and engagement of these collaborative events will be essential for successful application of immunotherapeutic strategies to enhance immunity and tolerance.

We thank Alain Trautmann and Ian Frazer for critical reading of the manuscript.