Sustained antigen presentation can promote an immunogenic T cell response, like dendritic cell activation

Prolonged Antigen Expression Can Replace DC Activation for Effector Cell Differentiation.

To investigate how DC activation might affect the kinetics of MHC class II-restricted antigen presentation in vivo, we used a double-transgenic mouse line (hereafter dtg) described previously, wherein DC presentation of a peptide/MHC-II complex to CD4⁺ T cells can be manipulated over time (8). In the presence of doxycycline (dox), a reverse tetracycline-dependent transactivator (Ii-rTA) induces the transcription of a second transgene that encodes an invariant chain (Ii) cDNA whose CLIP region was replaced by the H2-E^k-binding epitope of moth cytochrome c, (MCC)_93–103 (TIM) (15, 16). The tetracycline-inducible invariant chain with MCC (TIM) protein also carries a C-terminal amino acid replacement, such that it effectively shuttles the MCC peptide into the MHC class II processing pathway but does not compete with endogenous Ii (17). TIM expression was found almost exclusively in CD11c⁺ DCs (8) and about equally in the CD8⁻, CD8⁺, and CD11c^int120G8⁺ subsets (data not shown). Dtg animals were used as recipients of T cells transferred from the AND TCR transgenic line (18).

In dtg mice, expression of the MCC/E^k epitope is extinguished with a half-life of ≈1 day after dox removal. In our previous experiments with fluorescein (CFSE)-labeled AND T cells, the early termination of antigen exposure in vivo led to incomplete T cell activation (8). We asked whether DC activation, for example by triggering CD40, might be having the same effect as persistent antigen. Thus, we compared CD4⁺ T cells expanding under brief or long antigen presentation, by DCs that were activated with the stimulatory αCD40 mAb FGK45.5 or not. Sixty hours after transfer, labeled AND T cells exposed to waning numbers of MCC/E^k complexes divided 3–5 times and stopped, whereas those transferred into cage-mates that had received an αCD40 injection proliferated more extensively (Fig. 1A, conditions 3 and 4). AND cells transferred into animals wherein DCs expressed TIM throughout the experiment divided 6–8 times, but αCD40 injection had little effect in this instance (Fig. 1A, conditions 5 and 6). IFNγ, an indicator of effector cell differentiation, was found in cells that had divided extensively, resulting from either long exposure to antigen or αCD40 treatment (Fig. 1B).

These data suggested that persistent antigen was having the same effect as voluntary DC activation. One caveat was that DCs were being indirectly activated by the antigen-specific T cells to which they were presenting cognate peptide. Two lines of evidence showed that this was not the case. First, DCs from dtg recipients of AND T cells were compared with those from PBS- and αCD40-treated animals [supporting information (SI) Fig. 6A]. Although the AND T cells were activated and expanded, no significant up-regulation of activation markers CD80, CD86, and MHC class II could be observed on DCs 24 or 60 h after T cell transfer. Second, and more directly, we asked whether pretransfer of AND T cells into the dtg mice would enhance presentation by DCs to a second wave of AND T cells transferred 60 h later in the same hosts (and distinguishable by the CD90.1 marker). Rather than being enhanced, the proliferation of this second wave of AND cells was actually suppressed (SI Fig. 6B). Thus, and in keeping with previous results (19), responding AND T cells do not mimic αCD40 to license them for improved CD4⁺ T cell priming later on. In addition, not all DC activators behaved similarly, and TLR ligands such as CpG oligonucleotides only poorly extended MCC presentation (data not shown).

Because it is possible that the precursor frequency of T cells encountering antigen-presenting cells (APCs) impacts their behavior (20, 21), we performed the same transfers with graded cell numbers. The data in SI Fig. 7 show that cell proliferation was similarly enhanced by αCD40 treatment and persistent antigen, irrespective of the number of transferred cells. This indicates that the TIM turn-off is detected by T cells at high and low precursor frequencies in a similar way and that competition or crowding effects do not operate in our transfer system.

Gene Expression Analysis Demonstrates Antigen Persistence as a Key Component of DC Activation.

There are two possible explanations for the effect of αCD40 in the context of brief antigen expression: that the activated DCs delivered costimulatory signals that allowed the T cells to overcome inadequate antigen exposure, or that activated DCs retained and presented the agonist peptide for longer periods of time. Stated in classical terms, the DCs were providing Signal 2 or were just stabilizing Signal 1. To distinguish between these two possibilities, we assessed the gene-expression profile of AND T cells responding under the four conditions described above (Fig. 2A). Should the first explanation apply, we would expect to find a distinct signature of costimulatory influences; explanation 2, on the other hand, would predict similar profiles elicited by long exposure and by short exposure plus DC activation. Congenically marked AND lymph node cells were transferred into control hosts or into dtg recipients whose DCs expressed the TIM transgene transiently or continuously and were treated with the αCD40 mAb or not (Fig. 2A). An overview of the results is displayed in Fig. 2B, where the gene-expression profile for each condition is compared with that of cells transferred into control recipients (condition 1). Widespread differences were observed under conditions 4–6, less extensive changes were seen with the transient TIM expression of condition 3, and very few changes were detectable in condition 2.

We first compared in detail the changes that occurred with transient versus sustained expression of antigen, using the ratio or fold change (FC)/FC plots that visualize the FC relative to a control condition (Figs. 2C and 3). In such plots, data points crowd around the x = y diagonal if there is no or little difference in gene expression under the two induction regimens. However, most of the induced genes were more strongly induced (or repressed) in AND T cells confronted with antigen for the whole of their expansion phase (condition 5 in Fig. 2A) versus with transient exposure to antigen (condition 3). The bulk of the induced genes fell off the ratio plot's diagonal as did most of the repressed genes (Fig. 2C Upper), indicating that interruption of antigen presentation dramatically altered the expression of hundreds of genes.

We found that several groups of genes could be distinguished on the basis of their behavior in response to disappearing antigen (Fig. 2C Lower; a full listing is presented in SI Tables 1–3): the expression of some genes appeared much more dependent on continuous TCR signals than others. The genes represented by green data points (group A) (Fig. 2C Lower) responded essentially the same to the two conditions. Thus, because expression of these genes (including CD122, Stat4, Cxcr3, Ctla2, and Ctla4) did not require sustained TCR engagement, their regulation departs from that of the bulk of genes induced upon activation of CD4⁺ T cells. In contrast, the group B transcripts, labeled blue in Fig. 2C Lower, were expressed ≈2-to 4-fold higher in T cells continuously triggered by peptide/MHC. Genes implicated in effector T cell differentiation like Maf, Tbx21 (encoding T-bet), and Icos lie in this group as well as IFNγ, confirming the notion that persistent antigen is necessary for elicitation of effector function, as shown in Fig. 1. Interestingly, this group was also highly enriched in genes related to DNA metabolism and cell cycle progression (SI Table 2), a finding that was confirmed by statistical analysis of Gene Ontology identifiers (SI Table 4) and that is consistent with the dependence of CD4⁺ T cell proliferation on continued TCR engagement shown before (Fig. 1). Group C genes, depicted as red in Fig. 2C Lower, completely depended on persistent antigen. Here, genes involved in DNA replication and cell cycle control were significantly enriched (SI Table 3). IL-21 and the granzyme A, B, and K genes were in this group. In summary, genes involved in CD4⁺ T cell priming were responsive to waning TCR signals with different kinetics, and it is tempting to speculate that these groups of transcripts are regulated by separate mechanisms.

Next, we compared the gene-induction patterns of the conditions 3–5 depicted in Fig. 2A to the fully stimulatory treatment with αCD40 and dox given throughout, condition 6 (Fig. 3). In the comparison between gene induction to transient or persistent antigen presentation (conditions 3 and 5), the slope of a linear regression trend line (0.34) indicated an ≈3-fold-lower level of gene expression resulting from antigen removal (Fig. 2C Upper). When αCD40 was given with persistent antigen, the data plotted similarly (conditions 3 and 6, Fig. 3 Left). However, DC activation by αCD40 had a strong effect on this “dampening” of the activation signature when TIM was turned on only transiently: in the FC/FC plots comparing conditions 4 and 6 with the control, most of the data points lined up around the x = y diagonal and, e.g., Cxcr3 and IL-21 were now expressed as they were in cells persistently stimulated with antigen. The trend line of the data points now had a slope of 0.88 (Fig. 3 Center). This high degree of similarity in gene-expression profiles suggests that persistent display of peptide/MHC complexes by DCs and their activation by αCD40 (coupled with transient expression of TIM) had the same consequence on a T cell response. This interpretation was confirmed when antigen presentation was maintained by dox treatment and αCD40 was omitted, most of the data points crowded tightly around the trend line with the slope of 0.89, i.e., DC activation did not induce additional genes (Fig. 3 Right).

Long-Term CD4⁺ T Cell Survival Is Influenced by Antigen Persistence, but Not Exclusively.

We asked next whether αCD40 and/or persistent antigen promoted long-term survival and memory formation. Congenically marked AND T cells were transferred and followed over time by serial removal of s.c. lymph nodes between days 3 and 31. After the last time point, the transgene was turned back on to challenge the remaining AND T cells (Fig. 4). Irrespective of DC activation, AND cells transferred into control animals not expressing TIM did not persist long term, in agreement with a recent report suggesting that self-peptide/MHC complexes are limiting for the survival of naïve TCR-transgenic T cells (21). In all dtg mice induced to express cytochrome peptide the AND cells had expanded by day 3. However, in hosts where the cells were exposed to antigen only transiently in the absence of DC activation (condition 3), the cells had essentially disappeared by day 10, and there was no response to the secondary challenge. In contrast, a persistent memory T cell population was detected under all three conditions of longer antigen exposure, whether elicited by αCD40 treatment or by 60 h of TIM induction alone (Fig. 4B). Such cells could expand and produce IFNγ in response to re-expressed antigen (Fig. 4A). In hosts that had been initially exposed to persistent antigen and treated with αCD40, the number of memory cells was higher than in the absence of DC activation (Fig. 4 A and B, compare animals 5 and 6) indicating a synergistic effect of maintained antigen presentation and costimulation for memory-cell formation. Note that the T cells transferred into animals 5 and 6 were indistinguishable at 60 h, including in their mRNA profiles, but generated memory cells with a different efficiency. This quantitative difference might be explained by changes in the T cells themselves, like very subtle changes in gene expression, or changes in lipid, carbohydrate, or protein components. Alternatively, later steps of memory formation may be affected by the αCD40 treatment.

DC Activation Extends the Window of Effective in vivo Presentation of Some MHC Class II-Restricted Antigens.

Thus far, our results are in agreement with the idea that DC activation stabilizes peptide/MHC complexes in vivo. To test this more directly, we transferred labeled AND T cells into dtg recipients 3 days after dox removal. The cells hardly proliferated (Fig. 5A, ctrl) unless the recipients were treated with αCD40 at the time of TIM turn-on (Fig. 5A, condition 1), suggesting that DC triggering via αCD40 was inducing the persistence of stimulatory peptide/MHC complexes. The counterinterpretation that this might be due to costimulatory effects was ruled out by the administration of αCD40 mAb just before T cell transfer, which did not enhance the response (Fig. 5A, condition 2). CD40 triggering did not induce aberrant TIM expression, as judged by RT-PCR (data not shown). Thus, DC activation through CD40 engagement extends in this system the duration of antigen presentation in vivo.

To determine how long the stimulatory capability of activated DCs persists, we performed time-course experiments in which labeled AND cells were transferred at different times after TIM turn-off (Fig. 5B). Although a strong drop in stimulatory capacity was observed beyond 3 days, a clear effect of αCD40 was still detectable at 9 days. We also found that the injection of αCD40 coincident with switching on TIM was effective (Fig. 5C, condition 1), but injection 3 days before resulted in little enhancement of T cell activation (condition 2). This finding indicates that the mAb's effects on the population as a whole are quite transient, which is paralleled by waning activation markers on DCs such as CD86 (SI Fig. 6). Together with the results of Fig. 5A, these data suggest that the engagement of CD40 functionally prolongs the window during which the peptide/MHC complexes present on DCs at the time can stimulate a T cell response.

We then asked whether this effect might be observed directly by visualizing peptide/MHC complexes. An independent system of transient antigen exposure was used, where cognate peptides from hen egg lyzozyme (HEL) are targeted to the endocytic pathway of DCs in vivo, fused to the 33D1 antibody as chimeric molecules (22). Here, HEL_48–62/A^k complexes can be detected directly on the surface of splenic CD8⁻ DCs with the Aw3.18 mAb, even after 4 d. However, no effect of αCD40 was seen, as the HEL^48–62/A^k epitope decayed similarly, with or without αCD40 (Fig. 5D). Although the reason for this is not obvious and possible mechanisms will be discussed, this finding poses the caveat that the findings from experiments done in vitro as well as in the dtg mice cannot necessarily be generalized to other ways of antigen delivery.

Mice and Reagents.

The AND TCR-, Ii-rTA- and TIM-transgenic mouse lines were described (8, 16). Experimental animals were injected i.p. with 50 μg of FGK45.5 or control mAb in endotoxin-free PBS or PBS alone. Control animals were single- or nontransgenic littermates.

Cell Transfer and Flow Cytometry.

The equivalent of 2 × 10⁶ AND T cells were transferred except where indicated otherwise. For DC stainings, organs were digested with 100 units/ml collagenase D (Roche, Indianapolis, IN) in DMEM for 30 min at 37°C. Flow cytometry was performed as described (8).

Cell Sorting, RNA microarray analysis.

Lymph node cells from AND TCR-transgenic mice were transferred into recipients and sorted 60 h later on a MoFlo cell sorter (DakoCytomation, Fort Collins, CO). Three independent experiments were performed, except for the conditions 2 and 4 depicted in Fig. 2, for which two independent sorts were done. Sorted cells were lysed in TRIzol, total RNA was prepared, amplified, and biotinylated. Five to 15 μg of aRNA was hybridized to the Affymetrix Mouse Genome 430Av2.0 GeneChip (Affymetrix, Santa Clara, CA). Data preprocessing and analysis were performed within the GenePattern package, v2.0.1 (www.broad.mit.edu/genepattern). Affymetrix .cel files were collectively normalized and unexpressed and Y-chromosomal genes removed from the analysis with the GenePattern Multiplot module (R.M., unpublished work) and S-Plus 6.2 (Insightful) for linear regression. The microarray data have been deposited in GEO data bank with accession no.GSE5245.