|Home | About | Journals | Submit | Contact Us | Français|
LPS-activated DCs are thought to follow a set program in which they secrete inflammatory cytokines (such as IL-12) and then become refractory to further stimulation (i.e. “exhausted”). Here we show that mouse DCs do indeed lose their responsiveness to LPS, but nevertheless remain perfectly capable of making inflammatory cytokines in response to signals from activated T cells, and to CD40-ligand and soluble T-cell derived signals. Furthermore, far from being rigidly programmed by the original activating stimulus, the DCs retained sufficient plasticity to respond differentially to interactions with TH0, TH1, TH2 and TH17 T cells. These data suggest that LPS activation does not exhaust DCs but rather primes them for subsequent signals from T cells.
In order to initiate immune responses, dendritic cells (DCs) must first be activated, a process that has been widely studied. Currently, there are two sets of conflicting data regarding the changes that these cells undergo after activation. One set suggests that DCs are programmed to respond to certain activating stimuli, such as LPS, by undergoing a two-stage set of maturation changes. In the first stage, they capture antigens from the peripheral tissue, migrate to secondary lymphoid organs, and up-regulate the production of co-stimulatory molecules and certain pro-inflammatory cytokines (e.g. TNFα, IL-12, IL-6). Twenty four to forty eight hours after LPS stimulation, they are thought to pass to a ‘tolerant’ or ‘exhausted’ state in which they can no longer produce inflammatory cytokines, but instead can only generate TH2-inducing cytokines (1). Similarly, macrophages are thought to be pre-programmed to pass from an early pro-inflammatory state to a second stage, in which they switch to the production of anti-microbial substances (2).
However, there are also data suggesting that macrophages and DC responses can be enhanced, suppressed, or changed by signals from T cells and/or other cells in their immediate environment. For example, macrophages can be changed from “M2” to M1” by the presence of T cells (3). DCs can be ‘licensed’ to activate CTL by signals from TH1 CD4 T cells, or by CD40 cross-linking (4-6). They can be induced to produce copious amounts of IL-12, by interaction with activated CD4 T cells (7, 8) or IFN-γ (9) or CD40 Ligand plus IL-4 (10). They can also be educated to become inducers of TH2 or TH3 responses by interactions with “fed” T cells (11), or with prostaglandin E2 (PGE2) (12), IL-10 (13) or IL-4 (14) (from T cells or basophils).
Thus, on the one hand, there are data showing that DC responses to stimulation are preprogrammed while, on the other, there are data showing that they can be influenced or ‘educated’ by local cues. Because most of the education experiments were done with resting/immature DCs, one way to reconcile these two sets of data would be to suggest that there is a short window, early after LPS stimulation, during which DCs can be re-programmed by T cells, but that 24-48 hours later they are no longer susceptible, and are inexorably destined for “exhaustion”. Alternatively, the early window may be limited to stimulation with LPS such that an LPS-activated DC becomes refractory to further stimulation by LPS, but remains receptive to other cues.
To distinguish between these possibilities, we had a closer look at the phenomenon called “exhaustion”. We prepared bone marrow derived DCs from RAG deficient mice (which lack T cells), stimulated them with LPS and then re-stimulated them in the presence of various kinds of T cells. We found, in fact, that stimulating DCs with LPS does not lead to DC-“exhaustion” but rather to “selective responsiveness”. Although LPS-stimulated DCs do indeed develop into a state in which they no longer respond to LPS or to LPS plus IFNγ, they become highly responsive to a variety of signals from T cells and T-cell products.
All mice were kept in NIH animal facilities (an AAALAC accredited facility) in an SPF barrier colony. Adult 8-14 wk old male and female B10.BR, B10.A RAG2−/−, 5C.C7 RAG2−/− TCR transgenic mice (specific for peptide 88-103 of moth cytochrome c (MCC), 5C.C7 CD40L−/− mice were generated at the NIAID/Taconic Farms, Inc. These mice can be purchased from the NIAID/Taconic Farms, Inc. exchange.
Bacterial LPS was from Escherichia (E.) coli Serotype 0127:B8 (Sigma). Recombinant mouse GM-CSF and IFN-γ were from Peprotech Inc. Culture medium used throughout was Iscove’s Modified Dulbecco’s Medium (IMDM: GibcoBRL). Complete culture medium consisted of IMDM supplemented with 10% heat-inactivated Fetal Bovine Serum (FBS: tested to be free of endotoxin, mycoplasma, virus and bacteriophage GibcoBRL), plus L-glutamine, 55μM 2ß-Mercaptoethanol (GibcoBRL), penicillin, streptomycin and gentamicin (Biosource). For the BM-DC culture, we added 30 U/ml GM-CSF and 30U/ml IL-4. The TH2 clone, D10.G4.1, was purchased from ATCC and cultured with irradiated splenocytes from B10.BR or B10.A, plus 100μg/ml Conalbumin (Sigma). These cells can also be grown in the presence of 50pg/ml rIL-1 plus 20U/ml rIL-2 (Peprotech Inc.). The AE7 TH1 clone is stimulated occasionally with irradiated splenocytes plus 10ug/ml PCC and then carried in culture with rIL-2.
Generation of bone marrow derived DCs was as described previously, (Abdi et. al. 2006). Briefly, bone marrow cells were flushed out of the femurs and tibias of 8-12 week old mice into complete medium and pipetted vigorously to make a single cell suspension, then passed through a cell strainer (70-μm Nylon mesh; BD Falcon TM). Erythrocytes were lysed using ACK lysis buffer (Biosource) and cells washed 2x with complete medium. At this point, BM cells were either cultured immediately or cryopreserved for later use. BM cells were cultured at 1×106 cells/well in a 24 well plate in a final volume of 2 ml with complete medium supplemented with GM-CSF 6 ng/ml = 30U/ml and IL-4 3ng/ml = 30U/ml (Peprotech Inc.). The characteristics of such bone-marrow-derived DCs have been described in several publications (15-18). Starting at the 3rd day of culture, half of the medium was replaced with fresh warm medium each day. Cells were activated on day 6. On the day 7, the plate was chilled, and clusters of loosely adherent/suspended cells were carefully removed and used as BM-DCs.
Single cell suspensions of splenocytes were incubated with a cocktail of monoclonal antibodies to B220 (RA3-6B2), CD11b (M1/70), CD11c, DX5, GR1 and anti MHC class-II (I-Ek) (14.4.4S) (BD Pharmingen) followed by binding and negative selection with Dynabeads (Dynal Biotech). In some experiments, T cells were further purified by FACS sorting to >98% purity based on the staining for CD45.1, and the absence of staining for B220, pan-Class II, I-Ak, NK1.1, CD11c, CD11b, GR1and CD21.
Spleens from B10.A/SgSnAi TCR–Cyt 5C.C7-1 RAG-2−/− mice were dissociated into homogenous single cell suspensions and passed through a cell strainer (70-μm Nylon mesh; BD Falcon TM). Erythrocytes were lysed by using ACK lysis buffer (Biosource). Splenocytes were washed 3x in cold complete medium and adjusted to 1×106/ml, then cultured in complete medium plus 1μM MCC 88-103 at 37oC with 5% CO2. Fresh medium was added after 3 days of culture. After 5 days, the cultured cells were harvested, washed 2x with medium and re-cultured in the absence of peptide with 10-20 U/ml of rmIL-2. The medium was replenished every 5-7 days with fresh medium plus rmIL-2. For T-DC coculture experiments, these CD4+ T cells were purified with a CD4+ T cell isolation kit containing anti-CD8a, CD11b, CD45, DX5, and Ter-119 mAbs for depletion (Miltenyi Biotech.) with the addition of anti-CD11c and anti-class-II (Miltenyi Biotech.). We used 3-4 LS columns in tandem instead of using LD columns to deplete the cells. These T cells, which were >90% CD4+, were used at various days of culture as a source of in vitro primed/antigen-activated T cells.
Splenocytes from naïve 5C.C7 TCR transgenic mice were stimulated with 0.4μM MCC peptide for 3 days under various conditions. To make TH1 cells, we added 10μg/ml anti-IL-4 (R&D systems) plus 10ng/ml rIL-12 (Peprotech). To make TH17 cells, we added 10μg/ml anti-IFN-γ (R&D systems) plus 10ng/ml rIL-6 (Peprotech) and 1ng/ml TGF-β (R&D systems). To make TH2 cells, we added 10μg/ml anti-IFN-γ and10μg/ml anti-IL-12 (Pharmingen) plus 10ng/ml rIL-4 (Peprotech). After 3 days, the cells were washed, Ficolled and maintained in 20U/ml rIL-2 (Peprotech). TH2 culture were supplemented with 10ng/ml rIL-4 (Peprotech), and TH17cultures with 10ng/ml rIL-6 (Peprotech) and 1ng/ml TGF-β (R&D systems). To determine the T cell’s polarization status, the cells were sorted and cultured at 5×104 cells/well plus 5×105/well irradiated CD3ε−/− splenocytes and 0.1μM MCC peptide for 48-60 hrs. The supernatants were harvested and tested with specific ELISAs for IL-4, IFN-γ and IL-17. These polarized T cells were sorted and used in co-cultures with BM-DCs to measure their effects on DC cytokine production.
Six day BM-DC cultures were either left untreated in medium alone (Resting DCs), pre-activated with LPS at a final concentration of 100-200ng/ml LPS for 21-27hrs (“Exhausted” DCs) or with LPS plus 100ng/ml IFN-γ for 21-27hrs (“Hammered” DCs). The next day, loosely adherent cells (consisting of 80-90% CD11c positive cells, Fig. S4) were harvested, washed 2x with cold medium and re-stimulated in various conditions.
Various BM-DCs (Resting, “exhausted” or “hammered”) were cocultured with NIH-3T3 cells (at a ratio of five 3T3 to 1 DC), in 48 or 24 well plates at 1×105-2×105 cells/well with 5×105-1×106 cells/well mock transfected or CD40L-expressing NIH-3T3 cells for 48 hrs in the presence or absence of various cytokines. CSN from these coculture assays were tested for the production of cytokines.
1×105 sorted or bead depleted (CD4+ T cell isolation kit Miltenyi Biotech.) naïve or in vitro cultured 5C.C7 CD4+ T cells were incubated with 2×104/well of BM-DCs/well from B10.A RAG−/− mice ±0.1μM MCC peptide 88-103 in triplicate in 96 well u-bottom plates in a final volume of 200 μl/well for 48 hrs. The supernatants were removed and kept frozen at –30oC before measuring the cytokines by ELISA, or SearchLight (Aushon Biosystems, Billerica, MA) or Ray Biotech (Norcross, Georgia) Multiplex cytokine arrays.
D10 T cell clone was incubated at 8×105cells/well in 24 well plates coated with 10μg/ml of anti-CD3 mAb plus a 1:50 dilution of anti-CD28 ascites for 24-48 hrs. CSN obtained was filtered through a 0.22μM filter, incubated overnight at 4oC with ImmunoPure plus immobilized Protein G (Pierce) that was coupled with 1mg/ml of Rat anti-mouse IL-10 or IgG1 isotype control. Post incubation (and column depletion), the CSN was filtered through a 0.22μM filter and was used in the coculture of BM-DCs with antigen-activated 5C.C7 T cells.
The concentrations of cytokines in the cell-free medium were determined using cytokine-specific ELISA. In most experiments, IL-12 heterodimer was specifically detected using the ELISA kit from R&D systems (Quantikine) with a sensitivity of 7.8 pg/ml. In some experiments the CSN obtained were sent out as a service contract to be tested by SearchLight Multiplex cytokine array (Aushon Biosystems, Billerica, MA) or Ray biotech (Norcross, Georgia).
We computed pair-wise Student’s T-tests on the log10-transformed data for figure 4 and P-values were adjusted for multiple comparisons using the Holm-Bonferroni method. Figure 1B, we used unpaired Student’s T-test.
We generated DC populations by culturing bone marrow from RAG−/− mice in standard cultures containing GM-CSF and IL-4. We chose RAG−/− mice because memory T cells accumulate in the bone marrow of conventional mice (19), and these can greatly influence the DCs’ propensity to produce cytokines (9). After six days of culture, we stimulated the DCs with LPS, waited 24-27 hours (the period in which they have been described to become “exhausted” (1), washed and then re-stimulated them in various ways. We began with an analysis of IL-12p75 production by the DCs, as IL-12p75 (also known as IL-12p70, which we will hence call “IL-12”) was one of the main cytokines analyzed in the original study defining exhaustion (1), and it is thought to be the primary cytokine driving TH1 responses.
Figure 1A shows that mouse bone marrow-derived DCs behave similarly to the human blood monocyte-derived DCs used in the original “exhaustion” study. Resting DCs make large amounts of IL-12 after stimulation with LPS plus IFN-γ. However, when re-stimulated 27 hours after a previous LPS stimulation, the DCs make no IL-12, even in the presence of IFN-γ. Thus, like their human counterparts (1), LPS stimulated mouse DCs appear to become “exhausted”.
Under normal circumstances, however, a DC would be very unlikely to remain isolated from T cells for very long after stimulation. A resting DC that encountered LPS in the periphery would migrate within a few hours to a draining secondary lymphoid organ, where it would encounter both local and migratory T cells. We therefore asked if signals from T cells might change the responses of “exhausted” DCs, using CD4 T+ cells from RAG−/− 5C.C7 TCR Tg mice, which are specific for pigeon cytochrome c [PCC]. To separately test the effects of naïve versus previously stimulated T cells, we isolated naïve T cells from unprimed mice and used them immediately, or cultured them with PCC and APCs for one to several weeks to generate effector/memory T cells. We found that the presence of T cells did indeed change the behavior of the DCs. Figure 1B, composite of 44 separate experiments, shows that activated T cells (but not naïve T cells) helped both resting and “exhausted” DCs to produce large amounts of IL-12. Thus, the DCs seemed to be only conditionally “exhausted”. They did not respond to LPS plus IFN-γ, or to naïve T cells, but they did respond to IL-12-inducing signals from activated T cells.
Because DCs from RAG−/− mice make very little or no IL-12 when stimulated with LPS alone, we considered the possibility that they might not be sufficiently “exhausted”. It might be necessary, for example, to generate conditions, in which a large amount of IL-12 is made, in order to generate the negative feedback conditions leading to “exhaustion”. To see if a stronger primary IL-12-inducing stimulus might induce a more profoundly “exhausted” state that could not be overcome by the addition of T cells, we stimulated resting DCs with LPS plus IFN-γ, a stimulus that induces large amounts of IL-12 (Fig. 1A,). Figure 1C shows that these cells, which we call “hammered” DCs, behaved like “exhausted” DCs. They were unable to produce IL-12 when re-stimulated with LPS plus IFN-γ, but made copious amounts in response to activated T cells.
Figure 1D shows that IL-12 production is not an exception, as pre-stimulated DCs can also produce other pro-inflammatory cytokines when given T cell help. The pattern of TNFα expression, for example, mirrored that of IL-12. “Exhausted” and “hammered” DCs were unable to produce TNFα in response to further stimulation with LPS, or with LPS plus IFN-γ, but responded well to help from T cells. IL-1α behaved somewhat differently, in that “exhausted” and “hammered” DCs actually made more cytokine than resting DCs, and much more in the presence of T cells. Finally, IL-12p40, the subunit that contributes to both IL-12 and IL-23, was made in approximately equivalent amounts by all three sets of DCs.
These data suggest that LPS-activated DCs are not inexorably programmed to follow a particular path to “exhaustion”. Nor, as previously suggested, do they necessarily become skewed towards TH2 responses, (1), or anti-microbial responses (2). Instead, it seems that the DCs lose their responsiveness to LPS, but remain perfectly functional and capable of making pro-inflammatory cytokines if given other appropriate stimuli, such as the signals mediated by activated (but not by naïve) T cells.
To begin to dissect the mechanism by which T cells induce “exhausted” DCs to produce cytokines, we added individual soluble and cell-surface T-cell derived molecules, singly and in combination, to DCs stimulated in various ways. We began with the cell surface molecule, CD40L, which is expressed by activated T cells, and rapidly up-regulated by naïve T cells (20). In the original report on DC “exhaustion”, J558 cells transfected with CD40L were unable to elicit IL-12 from “exhausted” human DCs (1). We were surprised, therefore, to find that CD40L-transfected NIH-3T3 cells were occasionally able to elicit IL-12 from both resting and “exhausted” mouse DCs. Because this effect was sporadic, we studied it further and found that DCs rigorously washed from their growth/differentiation medium were unable to respond to the CD40L stimulus, and that even small amounts of the medium were sufficient to restore the response, suggesting that the medium contained soluble products able to synergize with CD40L to elicit IL-12. Because both IL-4 and GM-CSF are soluble T-cell-derived constituents of the DC growth/differentiation medium that may remain in varying residual amounts from experiment to experiment, and because the combination of IL-4 plus CD40L has been shown to synergize with LPS to elicit IL-12 from resting DCs (10), we asked if these cytokines would also synergize with CD40L to stimulate “exhausted” DCs. We cultured resting, “exhausted” or “hammered” DCs alone, or with various cellular stimuli (3T3 cells, transfected or not with CD40L, or effector T cells). For each of these co-culture conditions, we added various combinations of the cytokines, IL-4, IFN-γ and GM-CSF.
Figure 2, shows that mouse DCs, like their human counterparts, did not produce IL-12 in response to any combination of T-cell-derived cytokines in the absence of CD40L (mock, and also see Figure 1, where the DCs were grown in the presence of IL-4 and GM-CSF, but have no source of CD40L except when T cells are added), or to CD40L alone (grey bars). Thus, CD40L was essential but not sufficient. In its presence, different DCs responded to different combinations of soluble T cell-derived cytokines. For example, resting DCs made a small amount of IL-12 in response to CD40L plus IL-4 (as previously shown (7, 10), Figure 2 red bar). “Exhausted” DCs made more, and hammered DCs made even more, suggesting that pre-stimulation with LPS plus IFN-γ strongly sensitizes a DC to further instructions from T cells. The responses of the “exhausted” DCs were further enhanced by the addition of GM-CSF, and even more by the addition of GM-CSF plus IFN-γ. Overall, the data show that none of the DCs were truly “exhausted”. All three populations were capable of making copious amounts of IL-12 in response to intact activated T cells, and different types of activation rendered them differentially sensitive to particular combinations of T cell products. The combination of CD40L plus IL-4 was particularly illuminating, in that “exhausted” DCs were more sensitive to this combination than resting DCs, and “hammered” DCs were even more responsive. These data indicate that an activated DC is prepared to respond to signals that its resting, tissue-resident counterpart cannot.
The synergy of IL-4 with CD40L in activating IL-12 in “exhausted” DCs suggested that TH2 cells might be more potent inducers of IL-12 than other classes of T cells. To test this, we used a classic representative of a TH2 T-cell clone, D10.G4. (21) (specific for Conalbumin, hereafter called “D10”), comparing it with the TH1 clone, AE7 (specific for PCC (22) and to activated TH0 cells. Figure 3A shows that recently activated TH0 T cells and the AE7 TH1 clone and were both very effective at inducing IL-12 from “exhausted” DCs. To our surprise, however, the D10 TH2 clone was not, even though it expresses CD40L (Fig. S1) and is known to produce large amounts of IL-4 (Fig. S2) and (23). Reasoning that long term culture may select for mutant D10 cells that have lost some of their original functions, we generated short term polarized T cells, using standard polarizing culture conditions to differentiate resting T cells into effector TH0, TH1, TH2 or TH17 populations. When added to “exhausted” DCs, the TH0 and TH1 cells induced production of IL-12, but the TH2 and TH17 cells did not (Fig. 3B).
Given that CD40L plus IL-4 was a sufficient combination to induce IL-12 from “exhausted” and “hammered” DCs, the fact that TH2 cells were unable to do the same, in spite of their production of IL-4, suggested that they might also be producing an inhibitory factor. We therefore titrated the supernatant from D10 (stimulated with anti-CD3 and anti CD28 so as to preclude anything made by APCs) into a co-culture of “exhausted” DCs with TH0 effector T cells (which would normally induce good levels of IL-12), and found that the supernatant did indeed produce a dose-dependent inhibition of IL-12 production (Fig. 3C). Using Multiplex cytokine arrays, we analyzed the cytokines produced by D10 and found that, in addition to copious amounts of typical TH2-type cytokines (such as IL-4, 5 and 13) and some pro-inflammatory cytokines (such as IL-6), the supernatant also contained large amounts of IL-10 (Fig. S2), which has been established to suppress secretion of IL-12 by newly activated DCs (24, 25). To see if the inhibitory activity in the supernatant was due solely to the IL-10, we passed it over an anti-IL-10 affinity purification column, and tested the depleted supernatant for inhibitory activity on co-cultures of “exhausted” DCs and activated TH0 cells. Figure 3D shows that the anti-IL-10 column removed the suppressive activity and allowed for production of IL-12, (while an isotype control column did not), and figure 3E shows that the suppressive activity of the depleted supernatant could be restored by the addition of recombinant IL-10. Thus, IL-10 can suppress the positive stimulation of IL-12 that is normally produced by TH0 T cells. Altogether, these data suggest that the production of IL-12 is under tight control by T cells, with both positive and negative T-cell-derived regulatory elements (manuscript in preparation).
Thus far, we had focused on the control of IL-12, which is the primary cytokine that led to the idea of “exhaustion”. However, the originators of the “exhaustion” concept did not limit it to IL-12, but also extended it to other cytokines (1). To test this view, we allowed TH0, TH1 and TH2 cells to interact with resting, “exhausted” or “hammered” DCs and measured the production of a wide variety of cytokines using multiplex arrays. Figure 4A shows that, even in the absence of T cells (grey bars), not all DC-derived cytokines show the classic “exhaustion” pattern after LPS stimulation. For example, IL-12p40, IL-1α and IL-6 were produced at low levels by resting DCs, and in higher amounts by both “exhausted” and “hammered” DCs (p values all < 0.03). Thus, inflammatory cytokine production is not extinguished in LPS-stimulated DCs. “Exhaustion” seems to be specific to certain cytokines, such as IL-12 and TNFα.
The presence of T cells made significant changes in these patterns, and the changes varied depending on the effector class of the T cell, and on the activation state of the DC. TH0 cells, for example, had little effect on the production of IL-12p40, but induced “exhausted” and “hammered” DCs to produce IL-12, and stimulated all three DC populations to increase production of IL-1α, IL-6 and TNFα (p values for all <0.05 when comparing DCs cultured with TH0 T cells versus medium controls). TH1 T cells behaved generally like somewhat more potent versions of TH0 cells, except that they also triggered an increase in IL-12p40 production, a strong increase in IL-12 by resting DCs, and a decrease in IL-12 by “hammered” DCs when compared to TH0 cells. TH2 cells, while causing a ten-fold increase in the production of IL-6 (compared to medium alone), generated minor or no increases in IL-12p40 or p75, IL-1α, or TNFα. These data suggest that it may be time to change the view that the presence of T cells should merely amplify the pattern of DC cytokine production that has already been programmed by stimulation through TLRs (26, 27). Instead, it appears that different types of T cells can make significant changes in the pattern of cytokines produced by DCs while showing few differences in their own proliferative responses to these DCs (Fig. S3).
To molecularly define the T cell’s role in shaping cytokine production from “exhausted” and “hammered” DCs, we stimulated the three types of DCs with the artificially reconstituted T cell components, CD40L plus/minus IL-4, or with activated T cells. Figure 4B shows that the three types of DCs responded differently. CD40L alone was sufficient to stimulate IL-6 production from all three types of DCs; but IL-12p40, TNFα and IL-23 mainly from “exhausted” and “hammered” DCs, emphasizing the idea that “exhausted” and “hammered” DCs are in fact uniquely primed to receive some T cell signals that do not affect resting DCs, and underscoring the importance of using multiple markers to examine DC activation. The addition of CD40L plus IL-4 increased the production of IL-12. The combination actually decreased the production of IL-12p40 by “exhausted” and “hammered” DCs, when compared to the effect of CD40L alone, as has been previously shown for resting DCs stimulated with LPS (28). Activated T cells were stronger inducers of IL-12, IL-1α and TNFα than their isolated components of CD40L plus/minus IL-4, but weaker inducers of IL-12p40 and IL-23, suggesting that there are T-cell-derived controlling elements yet to be found. Altogether, these data support the concept that DCs are not pre-programmed indelibly by their initial stimulation to produce a particular set of cytokines, only to be enhanced by T cell signals. They change, depending on what signals they encounter.
We also measured T cell cytokine production (Fig. 5), comparing the relative stimulatory capacities of the resting, “exhausted” and “hammered” DCs. Although it has been suggested that “exhausted” DCs tend to promote TH2, or non-inflammatory responses, we found that the pattern was not as clear as previously thought (Fig. 5). All three DC types stimulated TH0 cells with roughly similar potencies. To see if the cytokine patterns were permanently fixed, we also tested the DCs with known TH1 and TH2 cells, to see if any one of the DC types would pair best with particular T effectors. We found that neither resting, nor “exhausted” nor “hammered” DCs paired best with any particular effector T cell type. When tested with TH1 cells, for example, resting DCs were more efficient at eliciting IL-2 and IL-3 production, but the three types of DCs were equivalent at inducing other cytokines. When tested with TH2 cells, “exhausted” DCs induced twice as much IL-13 as did resting or hammered DCs, while resting DCs induced ten fold more IL-3 than the other DCs, and all three types of DCs induced similar amounts of IL-4, IL-5 and IL-10. Thus, when a large array of cytokines was tested, each DC set promoted a different array of cytokines, and neither “exhausted” nor “hammered” DCs seemed to preferentially promote the production of TH2 cytokines.
Collectively, these data confirm that LPS-stimulated DCs are neither “exhausted” (1), nor paralyzed (29) nor tolerized (2) after stimulation with LPS or LPS plus IFN-γ. Nor do they have any one set response pattern. Instead, the DCs seem to move from a state in which they are responsive to some environmental signals (e.g. LPS), to one in which they are highly responsive to a different set of environmental signals (e.g. signals from T cells Figure 6). The resulting DC cytokine production depends on the combined effects of the original activating signal(s) and the T-cell derived signal(s) they receive later.
These data suggest that it is time to re-examine two well-established views about DC differentiation. First, the data demonstrate that DCs do not inexorably follow a certain program from activated to “exhausted” or “tolerized” when stimulated with LPS, or from inducers of TH1 responses to inducers of TH2 responses (1). Second, they do not respond in a pattern that is pre-set by the activating agent, but are sensitive to different types of cell-surface and soluble T cell signals, and coordinate their responses accordingly. In retrospect, this should not be surprising, as DCs must perform different functions at different times and in different places after activation. “Resting” DCs, residing in a tissue, need to be sensitive to activating signals from pathogens or from damaged cells, and perhaps to stimulatory signals from effector T cells entering the damaged site. Upon activation, the DCs need to take up and process molecules from their environment (30, 31), integrate late tissue-derived “education” signals (32), up-regulate the chemokine receptors that guide them to a local lymphoid organ (usually a local lymph node) and release any anchors that hold them in their resident tissue. Later, when they arrive at the draining lymphoid organs, they no longer need to respond to the original activating signals, and in fact, it would be a waste of energy to simply repeat that early program. Their role now is to present antigens and a selection of co-stimulatory and co-inhibitory signals (33) to induce T cells to choose, from the panoply of possible effector responses, those that are the most useful at that particular time and place. Because this is a two-way conversation between the DCs and the naïve and memory/activated T cells they encounter, the DCs need to remain responsive (or become responsive) to a variety of instructions from those T cells. Our data suggest that they do indeed remain responsive, and can change their cytokine expression depending on the type of T cells they encounter. Far from being “exhausted”, they have instead become selective. They no longer respond to the original stimulus that wrested them from their resting state, but instead respond to a number of relevant T-cell derived signals that they might encounter in the draining lymphoid tissues.
The signals to which they respond differ depending on the original stimulus. For example, DCs that were previously stimulated with LPS (“exhausted”) require the combination of CD40L (perhaps for longer survival?), IL-4, GM-CSF and IFN-γ to produce optimal quantities of IL-12 (see Figure 2), whereas “hammered” DCs (resting DCs stimulated with LPS plus IFN-γ, as might occur in areas containing activated NK cells, or TH1 effector T cells during an ongoing immune response) produce IL-12 in response to a minimal combination of CD40L plus IL-4. “Resting” DCs (that have recently been activated by the mechanical manipulations involved in cell harvesting) produce IL-12 in response to CD40L plus IL-4 and GM-CSF, and all of them respond to the addition of effector T cells (7). There are both positive and negative T-cell derived signals regulating the activity of previously activated DCs. For example, although CD40L plus IL-4 induces inflammatory cytokine production, TH2 T cells do not (even though they express CD40L and secrete IL-4), and this seems to be due to their production of IL-10. Perhaps this is also the function of the IL-10 produced by a subset of TH1 cells (34): a means of preventing or arresting the positive feedback loop between IFN-γ and IL-12 from getting out of hand (7).
The malleability of previously activated DCs is reminiscent of B cells, whose default response to LPS stimulation (in the absence of T cells) is to produce IgM and IgG3. With the addition of IL-4, B cells produce IgG1 and IgE. The addition of TGF-β induces B cells to produce IgA, and so on. We suggest that DCs follow a similar pattern. In the absence of T cells, they produce one kind of response (perhaps a response guided mostly by the tissue, in which they reside (32), but in the presence of T cells (or IFN-γ from NK cells, or perhaps cytokines from other resident innate lymphoid cells (35, 36) they are open to the production of a wide array of different responses.
Another aspect of our study that bears some mention is that resting RAG−/− DCs do not produce the heterodimeric cytokine, IL-12, when stimulated with LPS (see Fig. 1). Although this finding does not fit with the commonly accepted view that LPS-activated DCs produce IL-12 as a means to induce naïve T cells to become TH1 effector cells, we have seen it repeatedly, and have published previously on this topic (9, 37). To recapitulate here, we believe that the consensus view is perhaps a misconception based on the fact that 1) many early studies used antibodies that do not discriminate between IL-12 and its subunit IL-12p40 (which is far more easily induced by various stimuli), and 2) a number of studies have been done with cultures of DCs from which the T cells, or T-cell products have not been sufficiently removed. We have found that DCs (whether WT or RAG−/−) do not produce IL-12, regardless of the stimulus, in the absence of either IFN-γ or of T cells or their products (such as CD40L and IL-4). In addition, naïve T cells do not express functional IL-12 receptor (38, 39). Thus the idea that LPS-stimulated DCs somehow “know” that they should stimulate a TH1 response, and that they do this by offering IL-12 to naïve T cells, should perhaps be revisited in favor of the view that DCs respond to a plethora of signals from the invaders, the tissues and the immune cells in their environment; and that the ultimate ensuing effector class of immunity is tailored by all of these influences (32).
Finally, there is a prevailing view that the DCs are the controllers of immunity, and that the early context of DC activation determines the subsequent immune response. Our data, that DCs continue to retain behavioral plasticity long after the initial activating stimulus, paint a different picture. We would suggest that DCs are messengers, rather than controllers, continually integrating signals from the environment and modulating the messages they deliver.
We would like to thank Dr. Ronald Schwartz for critical review of the manuscript. We also would like to thank Mr. Jeff Skinner for helping with statistical analysis and Dr. David Dorward for helping with SEM, NIAID Rocky Mountain lab, Hamilton MT. 1
1 This work was entirely supported by the intramural program of the NIAID, NIH.