Dendritic Cells Amplify T Cell-Mediated Immune Responses in the Central Nervous System¹ Free

Neuroinflammatory immune responses in the CNS progress in sequential steps, including the infiltration of T cells, recruitment of macrophages, and development of inflammatory lesions (1, 2). T cells that initiate neuroinflammation must be restimulated locally in the CNS to survive (3, 4, 5). Encephalitogenic T cells migrate into the CNS, recognize their Ag locally presented by CNS-resident APCs in the nervous tissue, and cause disease even in the absence of a functional peripheral lymphoid system (4). Dendritic cells (DCs)⁴ are the most critical APCs that mediate the initiation and maintenance of CNS immune responses as they are sufficient to present Ag to T cells to induce CNS inflammation and epitope spreading (4, 6). In this work, we studied whether encounter with cognate Ags on DCs in the CNS would lead to increased activation of T cells with different Ag specificities. This type of T cell cooperation could lead to an amplification of the neuroinflammatory process.

Under steady state conditions, DCs are present in the cerebrospinal fluid, meninges, and choroid plexus (7, 8, 9, 10) and may play a tolerogenic role (7). DCs with a mature, activated phenotype also appear in the brain parenchyma under inflammatory conditions (11, 12, 13, 14, 15). In experimental autoimmune encephalomyelitis (EAE), brain-resident DCs have been shown to induce tolerance against autoantigens after the onset of clinical disease (16). Our own studies and the studies of others have shown that DCs are able to capture Ags in the CNS and migrate to secondary lymphoid organs where they present Ags to naive T cells (17, 18).

Despite these data, the exact role of DCs in the activation of Ag-specific T cells in the CNS and the consequences of brain DC-mediated Ag presentation on T cell functions are not understood. To study these questions, we used various TCR-transgenic recombination-activating gene-1 (RAG-1)-deficient mouse strains, in which all T cells recognize a single specific Ag. Adoptive transfer of defined amounts of naive T cells from one TCR-transgenic mouse strain into another of different specificity enabled us to follow the activation of T cells with defined specificities simultaneously. We demonstrate that DCs presenting CNS-derived Ags mediate either cooperative or competitive interactions between T cell populations with different specificities depending upon MHC-restricting element usage. This suggests that specific interference with T cell responses can significantly influence the overall outcome of immune response by affecting the behavior of other T cell populations, and that there is a significantly earlier DC-mediated enhancement of T cell responses to CNS-derived Ags than previously thought.

To create model networks having two T cell specificities, we used three different TCR-transgenic mouse strains backcrossed to B10.BR RAG-1^0/0 background (Table I). 5C.C7 RAG-2^0/0 TCR-transgenic mice (19) were purchased from Taconic Farms. D10.G4.1 RAG-1^0/0 TCR-transgenic mice (20) were a gift from D. Sant’Angelo (Sloan-Kettering Cancer Center, New York, NY). 3A9 TCR-transgenic mice (21) were obtained from The Jackson Laboratory. Animals were used for experiments between 6 and 12 wk of age and were housed in a specific pathogen-free facility at the University of Wisconsin Medical School Animal Care Unit (Madison, WI), according to the guidelines of the National Institutes of Health and the University of Wisconsin Medical School Animal Care and Use Committee.

We adoptively transferred 10⁶ naive T cells i.v. into TCR-transgenic mice of different specificity. Donor T cell populations were obtained from spleens and lymph nodes of D10.G4.1 RAG-1^0/0- or 3A9 RAG-1^0/0-transgenic mice. The percentage of T cells was measured by flow cytometry after staining with Abs against CD4 and Vβ8 TCR.

Anti-CD4 (clone RM4-5), anti-LFA-1 (clone 2D7), anti-Vβ3 (clone KJ25), anti-Vβ8.1/8.2 (clone MMR5-2), anti-IFN-γ (clone R4-6A2), anti-Vα2 (clone B20.1), anti-CD40L (clone MR-1), anti-CD40 (clone FGK45), along with streptavidin-allophycocyanin and streptavidin-PE conjugates, were purchased from BD Pharmingen. Anti-D10.G4.1 clonotypic Ab (clone 3D3) and anti-3A9 clonotypic Ab (clone 1G12) were gifts from D. Sant’Angelo (Sloan-Kettering Cancer Center) and P. Allen (Washington University, St. Louis, MO), respectively. Anti-GFAP Ab was purchased from Sigma-Aldrich. Anti-I-A^k and anti-CD11c Abs were purified from supernatant of hybridoma clone 10-3.6 and N418, respectively. TO-PRO-3 was purchased from Molecular Probes. Purified pigeon cytochrome c (PCC), chicken conalbumin (CA), and hen egg lysozyme (HEL) were obtained from Sigma-Aldrich. PCC_88–104, CA_134–146, and HEL_52–61 peptides were purchased from BioSource International.

DCs were derived from mouse bone marrow as previously described (18). DCs were loaded with PCC, CA, or HEL individually or a combination of PCC plus CA or CA plus HEL for 6 h as previously described (18).

For protein Ags, a 20-μl total of a sterile solution was injected 1.5 mm from the surface of the skull into the ventral-posterior region of the right frontal lobe of host mice through an insulin syringe via a penetrating depth controller. In the case of DC-mediated Ag delivery, two populations of 250,000 DCs loaded with different Ags were mixed 1:1 and injected intracerebrally in a final volume of 20 μl into host mice. Further details of experimental designs are addressed in Tables II and III.

For s.c. immunization, mice were injected at the base of the tail with 100 μg of PCC or CA in CFA in a total volume of 100 μl.

Mononuclear cells from brain were isolated and stained for flow cytometry as previously described (18). Absolute numbers were calculated based on the percentage of specific T cells from the total cell population acquired and the weight of tissues.

For intracellular cytokine staining, single-cell suspensions from brain were cultured in complete RPMI 1640 supplemented with GolgiStop (BD Pharmingen) in the presence or absence of 1 μg/ml peptide Ag for 5 h. After surface staining, cell suspensions were fixed and permeabilized by Cytofix/Cytoperm solution (BD Biosciences) followed by staining with anti-IFN-γ.

Immunohistochemistry from frozen tissue samples was performed as previously described (18). Paraffin-embedded sections were deparaffinized and antigenic epitopes were retrieved by boiling samples in 1 mM sodium citrate solution. Section then were stained with anti-glial fibrillary acidic protein (GFAP) and anti-I-A^k as described (18).

Recombinant Mycobacterium bovis strain bacille Calmette-Guérin strains expressing GFP-PCC or GFP-HEL-PCC fusion proteins (J. Karman, H. Chu, E. Heninger, P. Hulseberg, D. Co, L. Hogan, Z. Fabry, and M. Sandor, manuscript in preparation) were used to purify peptide-specific fusion proteins. Mycobacterial lysates were purified on CaptoQ anion exchange columns using an AKTA FPLC system (GE Healthcare) according to the manufacturer’s instructions (application note 11-0026-20 AB). Concentrations of GFP fusion proteins were measured by spectrophotometry.

Equimolar amounts of GFP-HEL-PCC or a mixture of GFP-PCC and purified HEL protein were administered intracerebrally into 5C.C7 RAG-1^0/0 mice that received 10⁶ CD4⁺Vβ8⁺ cells from 3A9 RAG-1^0/0 mice as described above. Five days postinjection, brain, cervical lymph node, and spleen were processed as described above.

siRNA-mediated depletion of CD40L and IL-2 was induced using the pSUPER plasmid (Oligoengine) and the Phoenix-E packaging cell line (gift from A. Rudensky, University of Washington, Seattle, WA)-based retroviral infection system (22). siRNA sequences for mouse CD40L (GenBank accession no. X65453) and IL-2 (GenBank accession no. NM_008366) were designed using algorithms developed by Oligoengine. The following oligonucleotides were annealed and ligated with BglII-HindIII digested and linearized pSUPER.retro.neo+gfp plasmid according to manufacturer’s instructions: CD40L sense, 5′-GAT CCC CGT GGG CCA AGA AAG GAT ATT TCA AGA GAA TAT CCT TTC TTG GCC CAC TTT TTA-3′; CD40L antisense, 5′-AGC TTA AAA AGT GGG CCA AGA AAG GAT ATT CTC TTG AAA TAT CCT TTC TTG GCC CAC GGG-3′ (referred to as pSUPER-CD40L); IL-2 sense, 5′-GAT CCC CCT TCA AGC TCT ACA GCG GAT TCA AGA GAT CCG CTG TAG AGC TTG AAG TTT TTA-3′; IL-2 antisense, 5′-AGC TTA AAA ACT TCA AGC TCT ACA GCG GAT CTC TTG AAT CCG CTG TAG AGC TTG AAG GGG-3′ (referred to as pSUPER-IL-2). Viral particles were produced by transfecting exponentially growing Phoenix-E cells either with original pSUPER (referred to as pSUPER-control), pSUPER-CD40L, or pSUPER-IL2 using GeneJammer transfection reagent (Stratagene) according to the manufacturer’s instructions. Supernatants containing viral particles were harvested 2 and 3 days after transfection and used to infect splenocyte cultures.

Retroviral infection of splenocytes was performed based on Ref. 23 , with modifications. Splenocytes from D10.G4.1 RAG-1^0/0 mice were cultured for 24 h in the presence of 0.5 μg/ml CA_134–146 peptide and 20 U/ml IL-2. Twenty-four and 48 hours after activation, cells were infected using 1.5 ml of transfected Phoenix-E cell supernatant. Cells were harvested 48 h after the second infection cycle and sorted into GFP⁺ and GFP⁻ fractions on a FACSVantage SE cell sorter (BD Biosciences). Purity of sorted cells was determined based on GFP fluorescence and staining with anti-mouse CD4. A total of 10⁵ GFP⁺ cells was i.v. transferred into recipient 5C.C7 RAG-1^0/0 mice. On the same day, recipient mice were injected intracerebrally with a mixture of 60 μg of PCC and CA in 20 μl as described above. Five days postinjection, cells were isolated from brain and processed for cell surface and intracellular staining.

To quantify the expression of CD40L and IL-2 in sorted GFP⁺ cells, both real-time RT-PCR and flow cytometry were used. For real-time PCR, total RNA was extracted from 10⁵ sorted GFP⁺ cells and reverse transcribed using Superscript III reverse transcriptase (Invitrogen Life Technologies) using oligo(dT) as primer according to the manufacturer’s instructions. PCRs were performed using the following sense and antisense CD40L primers: 5′-CAG TGG GCC AAG AAA GGA TA-3′ and 5′-GGT ATT TGC CGC CTT GAG TA-3′ or real-time PCR IL-2 primer mix (BioSource International) and 10,000-fold diluted SybrGreen stock solution (Molecular Probes) in a Smart Cycler version 1.2f (Cepheid). Relative gene expression values were normalized to β-actin expression (β-actin sense primer, 5′-AGA GGG AAA TCG TGC GTG AC-3′; β-actin antisense primer, 5′-CAA TAG TGA TGA CCT GGC CGT-3′) (24). For flow cytometry, cell cultures at the end of the retroviral infection assay were stained with Abs against CD4, Vβ8 TCR, and CD40L or restimulated with 1 μg/ml CA_134–146 peptide in the presence of GolgiStop and stained for intracellular IL-2.

All data are presented as mean ± SEM. The one-tailed Student t test was used for comparisons of two groups. In all cases, p < 0.05 were considered statistically significant.

It has been shown that single OVA Ag-pulsed DCs microinjected into the brain induced the preferential recruitment of OVA Ag-specific T cells into the CNS (17, 18). To establish whether the intracerebral injection of single vs two brain DC populations pulsed with two different Ags affects different Ag-specific T cell recruitment and function in the CNS, we intracerebrally injected a 1:1 mixture of DCs pulsed with either specific or nonspecific Ags alone or in combination into 5C.C7 PCC-specific I-E^k-restricted TCR-transgenic hosts and adoptively transferred cells from D10.G4.1 CA-specific I-A^k-restricted TCR-transgenic donors (Fig. 1,A). This experimental model allowed us to analyze Ag-specific immune responses of two different Ag-specific T cell populations that were restricted by different MHC class II molecules simultaneously (Table II). Five days following adoptive transfer and intracerebral injection, we examined the recruitment of host and donor T cells into the brain using cytofluorometry. Our results showed that cognate Ag-specific T cells entered the brain in large numbers when a mixture of DCs pulsed with either single Ag PCC or CA together with DCs pulsed with irrelevant HEL were injected (Fig. 1,B, two left plots, and C, two leftmost bars). Injecting a mixture of DCs, each pulsed singly with either specific Ag, resulted only in a slight increase in both host and donor T cell numbers (Fig. 1, B and C, compare plots and bars second from right to the two leftmost). The highest number of T cells of both specificities entered the brain when DCs pulsed with both PCC and CA together were injected along with DCs pulsed with irrelevant Ag (Fig. 1, B and C, rightmost data sets). Notably, the intracerebral injection of DCs pulsed simultaneously with PCC and CA led to an ∼10 times higher increase in the absolute cell number of both T cells with different Ag specificities when compared with single Ag-pulsed DC injection. The additive amount of brain accumulating host and donor T cells together was substantially increased as a response to dual PCC and CA Ag-pulsed DC injections (Fig. 1,C). These data indicate that DCs could mediate cooperative interactions between T cells of different specificities when presenting Ags to both T cell populations simultaneously. These findings were repeated using 3A9 cell adoptive transfer (TCR transgenics specific for HEL and restricted to I-A^k) into 5C.C7 RAG-1^0/0 mice with similar results (data not shown). To further determine the functional consequences of increased T cell accumulation in the CNS, we analyzed IFN-γ proinflammatory cytokine production by T cells from the brain. The production of IFN-γ by donor and host T cell populations in an ex vivo recall response showed similar trends (Fig. 1, D and E) and clearly demonstrated the significance of this nonadditive increase of T cell recruitment and proinflammatory cytokine production. These data suggest that cooperative interactions between T cell populations restricted by different MHC class II molecules are enhanced either by the close proximity of T cells to each other and to the APCs, which may be due to the requirement for either high concentration of local soluble mediator(s), or the simultaneous engagement of cell surface molecule(s) on the surface of DCs, or to both.

To determine whether differences in MHC restriction elements could influence DC-mediated T-T cell cooperation, we intracerebrally microinjected DCs, pulsed with single (HEL or CA) or dual (HEL plus CA) Ags, into HEL-specific and I-A^k-restricted (3A9) TCR-transgenic mice and adoptively transferred CA-specific and I-A^k-restricted D10.G4.1 donor cells at the same time of DC microinjection. Five days following DC injection into the brain and adoptive transfers, we analyzed host and donor-transgenic T cells in the brain. Intracerebral injections with mixtures of specific and nonspecific Ag-pulsed DCs were done as before (Fig. 2,A and Table III). Under these conditions, we detected the highest total accumulation of both host and donor T cells in brain when mice were injected with a mixture of either HEL or CA-pulsed DCs (Fig. 2, second from right data sets), although injections containing only one specific Ag-pulsed DC population resulted in the highest fractional accumulation (Fig. 2,B, two left panels). Levels of IFN-γ production showed similar changes (Fig. 2, D and E, second from right data sets). Taken together, these data suggest that T cells restricted by the same MHC class II molecules may compete for a DC population presenting their Ags simultaneously (Fig. 2,B, right), and may cooperate when separate DC populations present their specific Ags (Fig. 2 B, second from right). These observations are consistent with the involvement of both soluble and cell surface mediators in cooperative T cell interactions.

To compare the role of endogenous APCs in mediating T cell interactions to that observed with DC injection, we intracerebrally microinjected protein Ags into 5C.C7 host-transgenic animals and adoptively transferred D10.G4.1 donor cells at the time of Ag injections. As we discussed above, when the D10.G4.1 TCR-transgenic cells were adoptively transferred into 5C.C7 host mice, T cells were restricted by different MHC class II molecules (Fig. 3,A). Five days following intracerebral injections and adoptive transfer, donor and host T cells in the brain were analyzed by cytofluorometry. We found the most substantial donor and host T cell accumulation and IFN-γ production when a combination of PCC and CA was injected compared with injection of PCC or CA Ags alone (Fig. 3, B and C). Similar results were obtained when 3A9 donor cells were adoptively transferred into 5C.C7 host mice (data not shown).

To analyze whether intracerebrally injected Ag can induce T cell-mediated T cell help similarly to exogenously injected DCs, we intracerebrally injected HEL, CA, or a combination of the two Ags into 3A9 host animals and adoptively transferred D10.G4.1-donor cells (different Ag specific T cells were restricted by the same MHC class II molecule) (Fig. 3,A). In this experimental setup, there was no significant increase either in the absolute number of host 3A9 or donor D10.G4.1 T cells in the brain, or in IFN-γ production by these T cells following combined injections of HEL and CA when we compared these values to HEL or CA Ag injection alone (Fig. 3, D and E). These data show that both exogenous and endogenous APCs can mediate different T cell-T cell interactions when presenting CNS-derived Ags, depending on the MHC class II restriction of the specific Ags recognized.

To demonstrate that local CNS DCs could mediate dual presentation of Ags, we intracerebrally microinjected PCC and CA protein Ags into 5C.C7 host-transgenic animals and adoptively transferred D10.G4.1 donor cells at the time of Ag injection. Fluorescent immunohistochemistry was completed using CD11c-, Vβ3-, and Vβ8-specific Abs and brain tissue sections were analyzed by confocal microscopy. As it is shown in Fig. 3 F, CD11c-expressing cells are detected in clusters with Vβ3- and Vβ8-positive T cells at the injection sites in brain. These data indicate that CD11c-positive cells can form clusters with T cells and present Ags simultaneously to two different Ag-specific T cell populations in the brain. Similar results were detected when 3A9 donor cells were adoptively transferred into 5C.C7 host animals (data not shown).

Confocal microscopy of the injection site in brain demonstrated a large increase in the infiltration of both Ag-specific T cell populations when PCC and CA Ags were injected together (Fig. 4, A and B). We analyzed the effects of T cell cooperation in response to endogenous APC-mediated Ag presentation on neural inflammation by examining pathological changes (Fig. 4,C) and expression of MHC class II molecules on astrocytes at the injection site. Our results show that there is a significantly higher expression of MHC class II molecules on GFAP-positive cells in mice injected with a mixture of PCC and CA (Fig. 4 D), indicating that T cell cooperation leads to higher local inflammatory reactions.

Having established that simultaneous Ag presentation by brain DCs to T cells of different Ag specificities enhances T cell responses when T cells are restricted by different MHC class II molecules, we studied whether simultaneous draining of PCC and CA from CNS to cervical lymph nodes is also important in this process. To address this question, one Ag was injected intracerebrally for drainage to cervical lymph nodes and the other Ag was injected s.c. at the base of the tail for drainage to inguinal lymph nodes (Fig. 5,A). The absolute number of T cells infiltrating the brain was significantly reduced when either Ag was injected s.c. as compared with injection of both Ags intracerebrally (Fig. 5, B and C). These data indicate that concurrent presentation of the two Ags and activation of T cell populations with different specificities in the same draining lymph node appears to be necessary for cooperation.

From the above studies, we predicted that there was a cooperative interaction between two different Ag-specific T cell populations when they recognized their cognate Ags in the context of different MHC class II molecules on a single DC. We also had to take into consideration that intracerebral injection of purified protein mixtures led to presentation of epitopes from both proteins by the same APC population, however, the relative molar amount of each epitope presented by individual APCs would vary within the population. To facilitate simultaneous Ag presentation of two epitopes by the same APCs, we generated rGFP constructs that expressed PCC or PCC plus HEL antigenic epitopes. These purified fusion proteins were microinjected intracerebrally into 5C.C7 TCR-transgenic host animals simultaneously with adoptive transfer of 3A9 donor cells (Fig. 6 A and data not shown).

We compared the effect of intracerebral injection of GFP-HEL-PCC fusion protein or a mixture of GFP-PCC and purified HEL protein on the activation of 5C.C7-host and 3A9-donor TCR-transgenic T cells 5 days following brain injections (Fig. 6,A). Injection of GFP-HEL-PCC fusion protein resulted in significantly higher numbers of both host and donor T cells in the brain than the injection of GFP-PCC plus HEL mixture (Fig. 6, B and C). There was also a substantial increase in the absolute number of IFN-γ-producing T cells of both donor and host specificity following the same injections (Fig. 6 D).

To examine the mechanisms of T cell-mediated T cell help through DCs in the CNS, we tested the role of CD40L and IL-2 in the 5C.C7-host/D10.G4.1-donor experimental system (Fig. 7 A). CD40L or IL-2 production was knocked down in the donor D10.G4.1 T cell population by retrovirally mediated expression of siRNA (25). Cells infected by retroviruses were identified based on GFP fluorescence. Sorting of GFP⁺ cells yielded ∼90% pure CD4⁺GFP⁺ cell populations for all three retroviral constructs (data not depicted). Depletion of CD40L and IL-2 was verified at both the mRNA and protein levels using quantitative real-time PCR and flow cytometric analysis (data not shown).

Because for successful retroviral infection T cells need to be prestimulated, we first tested whether transfer of in vitro-activated donor cells can participate in the same type of cooperative T cell interaction in vivo as naive T cells. Measurement of brain-infiltrating T cell populations after injection of 10⁵ activated CD4⁺ T cells and intracerebral injection of Ag revealed a significant increase in T cell numbers when PCC and CA were injected together as compared with injection of either protein alone, consistent with our earlier observations (data not shown).

After confirming the model using activated donor cells instead of naive cells, we addressed the role of CD40L and IL-2 in T cell-T cell cooperation. Sorted GFP⁺ cells (10⁵) from retroviral infection cultures were adoptively transferred i.v. into host mice, and these recipient mice were injected intracerebrally with a mixture of PCC and CA proteins as we discussed above (Fig. 7,A). Five days following adoptive transfer and intracerebral injection, donor and host-transgenic T cells were analyzed using cytofluorometry. Fewer infiltrating host 5C.C7 cells and less IFN-γ production were detected after injection of either CD40L- or IL-2 depleted cells compared with injection of control virus-infected cells (Fig. 7, B and E). Donor T cells were detected in equal numbers in brain showing that the decrease in host cell numbers is not due to an absence of infiltration of donor T cells into the brain (Fig. 7, C and D). A low number of donor cells were also detectable in the cervical lymph node and spleen (data not shown). These data suggest that both engagement of the CD40-CD40L molecule pair and high local concentration of IL-2 due to simultaneous production by both T cell specificities can contribute to brain DC-mediated T cell-T cell cooperation.

DCs play a central role in regulating T cell immunity at various levels (26) and are an attractive therapeutic target. Competition by T cells for DCs and DC-mediated Th cell cooperation are both well-described phenomena (27). In this study, we show that presentation of Ags by two distinct MHC class II molecules allows DCs to mediate cooperative interactions of T cells with different TCR specificities, leading to enhanced accumulation and IFN-γ production in the target, nonlymphoid organ. We detected a >10-fold amplification of T cell accumulation and IFN-γ proinflammatory mediator production as a response to simultaneous DC Ag presentation to T cells that recognize their Ags in the context of different MHC restriction molecules. This amplification might have a very important role in the outcome of neuroinflammation. We also show that this mechanism likely includes both simultaneous DC:T cell clustering involving CD40-CD40L interactions and high local IL-2 concentrations. These data support a novel role for DCs in locally enhancing T cell functions in the brain.

The current view of how CNS-specific immune responses are amplified involves the recruitment of phagocytic myeloid cells by an initial infiltration of T cells (1, 2). Our results indicate that amplification may happen significantly earlier in response to simultaneous presentation of several Ags to T cells with different TCR specificities and demonstrate a novel role of Ag presentation in the CNS. APC-mediated amplification of T cell responses to CNS-derived Ags also has important functional consequences in promoting local inflammation (Fig. 4 C). This shows that APCs not only play a role in the initiation of CNS inflammation but also play a role in the early enhancement of T cell responses, implicating CNS APCs as potential targets in CNS inflammation.

The experimental models presented here also provide a new perspective on the role of DCs in autoimmune disease development and the contribution of infectious disease to this process. First, our data suggest that simultaneous T cell responses against infectious agents and self-Ags may lead to enhanced responses to self-Ags. There is a wealth of studies on how infectious diseases contribute to autoimmune responses (28, 29). Second, during epitope spreading, T cells respond to different CNS-derived protein epitopes concurrently (30), which may also enhance responses to both Ags. Third, Ag-specific treatment of autoimmune diseases may lead to decreases in autoimmune responses to other determinants. In EAE, tolerance developed to myelin basic protein results in decreased responses to other CNS Ags as well (31). Also in EAE, the induction of the disease with two Ags in wild-type mice results in significantly more severe symptoms than that induced by either Ag alone (32). Our results provide a possible explanation for these observations. This latter study also shows that even in an intact T cell repertoire, simultaneous immune responses can lead to qualitatively different outcomes, and T cell cooperation is not restricted to TCR-transgenic T cell populations.

Previous studies have indicated that the concurrent activation of T cell clones with different specificities in the same DC:T cell cluster favors interclonal T cell-T cell cooperation over competition (33, 34, 35). Here, we demonstrate key spatial and molecular requirements of cooperative T cell behavior. Our experiments using GFP fusion proteins containing several T cell epitopes supports a T cell-DC-T cell cluster model of cooperation, in which the most effective cooperative T cell behavior is detected when the same DC presents Ags for both T cells. Local IL-2 concentration and engagement of cell surface molecules on APCs are likely to be much higher when two T cell specificities get activated concurrently by the same APC.

The possibility that astrocytes present Ag to T cells in the CNS cannot be excluded. Astrocytes have been suggested to modulate the activity of CNS pathogenic T cells by presenting myelin Ag peptides in combination with cytokines and additional proinflammatory mediators in the CNS (36). Astrocytes have also been demonstrated to respond to cytokines of the adaptive immune system, such as IFN-γ, with induction of Ag presentation (37). We have demonstrated a significantly increased IFN-γ production following intracerebral injection of DCs pulsed simultaneously with PCC and CA when T cells with different Ag specificities and MHC restriction elements were used. The production of IFN-γ by donor and host T cell populations in an ex vivo recall response showed similar trends (Fig. 1, D and E). We also demonstrate astrocyte activation (increased GFAP) and class II up-regulation on these cells (Fig. 4). This might suggest that Ag is being presented by astrocytes in the CNS and the local expansion of T cells in the brain is a result of astrocyte activation and Ag presentation. This might be a possibility when protein Ags or fusion proteins are injected intracerebrally. However, it was recently shown that astrocytes are poor at processing and presenting whole proteins (38). The significance of astrocytes in DC-mediated T cell function enhancement would need to be further studied.

Data from the injection of exogenous DCs indicate that both cell surface and soluble molecules may play a role in cooperative T cell interactions. CD40-CD40L interactions are crucial in a variety of helper T cell-APC interactions (35, 39, 40, 41) and in development of T cell-mediated pathologies in the brain, such as EAE (42). Our results indicate that this molecule pair plays an important role in brain DC-mediated T cell cooperation (Fig. 7). DCs receiving simultaneous CD40-transduced signals from two activated T cell clones may achieve a higher activation status. We hypothesize that CD40 molecules are only partly occupied on the surface of DCs during the activation of a single T cell clone, with some molecules excluded from T cell-DC interaction zones. However, when two T cell clones are activated simultaneously, recruitment of more CD40 molecules into the T cell-DC contact zone occurs. Confocal microscopy showed no significant changes in CD40 expression on APCs in the brain when both T cell clones get activated as compared with activation of a single T cell clone (data not shown), suggesting that higher CD40 engagement rather than CD40 expression acts during simultaneous T cell activation. Our results indicate that another important factor in T cell cooperation is high local IL-2 concentration, which may induce interclonal T cell cooperation through the enhancement of T cell survival and/or proliferation. IL-2 produced by the host cells alone is insufficient to mediate cooperative T cell interactions, suggesting a threshold model for the effect of IL-2 upon T cell cooperative interactions.

Our results delineate some of the essential mechanisms by which T cells with different specificities influence each other’s behavior during DC:T cell cooperative interactions in the CNS. This multi-Ag presentation step by DCs might represent a natural enhancement of neuroinflammatory process in the CNS that justifies the applicability of Ag-specific targeting therapies in CNS inflammatory diseases, such as multiple sclerosis. Interference with T cell responses may broadly influence T cell responses if these T cell clones cooperate or compete with each other. Therapeutic targeting of DC-mediated cooperation and the molecules that play a role in the process will require careful Ag selection and knowledge of restricting elements linked to autoimmune pathology. Nevertheless, manipulation of immune responses via immunization with several carefully selected Ags should ultimately be useful in both infectious vaccination strategies and down-modulation of self-reactive T cell populations.

We thank Sinarack Macvilay, Khen Macvilay, and Toshi Kinoshita for excellent technical assistance; Kathleen Schell and the University of Wisconsin Comprehensive Cancer Center Flow Cytometry Facility (Madison, WI) for assistance in cell sorting; and Lance Rodenkirch and the W. M. Keck Laboratory for Biological Imaging for assistance with confocal microscopy. We also acknowledge Dr. Laura Hogan and Emily Reinke for critical review of the manuscript.

The authors have no financial conflict of interest.