Abstract
Ag recognition via the TCR is necessary for the expansion of specific T cells that then contribute to adaptive immunity as effector and memory cells. Because CD4+ and CD8+ T cells differ in terms of their priming APCs and MHC ligands we compared their requirements of Ag persistence during their expansion phase side by side. Proliferation and effector differentiation of TCR transgenic and polyclonal mouse T cells were thus analyzed after transient and continuous TCR signals. Following equally strong stimulation, CD4+ T cell proliferation depended on prolonged Ag presence, whereas CD8+ T cells were able to divide and differentiate into effector cells despite discontinued Ag presentation. CD4+ T cell proliferation was neither affected by Th lineage or memory differentiation nor blocked by coinhibitory signals or missing inflammatory stimuli. Continued CD8+ T cell proliferation was truly independent of self-peptide/MHC-derived signals. The subset divergence was also illustrated by surprisingly broad transcriptional differences supporting a stronger propensity of CD8+ T cells to programmed expansion. These T cell data indicate an intrinsic difference between CD4+ and CD8+ T cells regarding the processing of TCR signals for proliferation. We also found that the presentation of a MHC class II–restricted peptide is more efficiently prolonged by dendritic cell activation in vivo than a class I bound one. In summary, our data demonstrate that CD4+ T cells require continuous stimulation for clonal expansion, whereas CD8+ T cells can divide following a much shorter TCR signal.
Introduction
T cell receptor–mediated recognition of antigenic peptides presented by MHC class I and II molecules is necessary for clonal expansion and effector cell differentiation of CD8+ and CD4+ T cells, respectively. Initial TCR signals, integrated by microclusters and synaptic structures, arrest the T cells in contact with the APC allowing for sustained signals initiating blasting and massive de novo gene transcription. One to 2 d later, the cells regain motility and divide every 4–6 h, if not faster (1), for several d to expand the 15–1500 clonal precursors per mouse up to 100,000-fold. This expansion and differentiation of rare pathogen-specific T cell precursors to effector clones is crucial for adaptive immunity and survival (2, 3). On the relevance of Ag recognition following the initial divisions, however, a consensus has not been reached.
For CD8+ T cells, it has been shown that antigenic signals beyond an initial period are not required for proliferation (4–10), thus supporting the notion of early programming of CD8+ T cell expansion (11, 12). However, there is also data that CD8+ T cell expansion is correlated with the continued duration of TCR stimulation in the context of both infections and sterile vaccinations (13–21). The reason for this discrepancy is currently unclear and cannot be explained by differences of protocol, experimental systems or availability of help or inflammation.
The use of an Ag-independent proliferative phase of T cells has been instrumental for the expansion of lines and clones in vitro, especially by including growth factors and cytokines in the cultures. Two studies supplied evidence for the notion that TCR-triggered CD4+ T cells keep proliferating in the absence of Ag and cytokines (22, 23), whereas others did not support such an early-programming scenario but rather supplied evidence for the importance of maintained TCR signals (24–27). Most of the experiments done in vivo supported the Ag dependency of CD4+ T cells throughout the expansion phase (8, 13, 28–33).
Although clonal expansion is at the center of adaptive immunity, very little is known about cell cycle regulation in normal lymphocytes (34). It has been noted in many infections of mice and humans that the magnitude of the CD8+ T cell response is larger than that of CD4+ T cells and an intrinsic difference between the subsets has been invoked (35, 36). However, whether Ag requirements in the expansion phase differ between CD4+ and CD8+ T cells has been asked in few studies that came to different conclusions (8, 13, 32). Thus, we examined the Ag dependency of CD4+ and CD8+ T cell proliferation side by side in a way that excludes cross-talk between the subsets. Because pathogens have developed numerous ways to undermine T cell priming and affect the immune response by inflammation, we used noninfectious methods where priming conditions and Ag removal are clearly defined, are the same for both subsets, and are not affected by APC type and MHC class differences. We found that following strong priming, CD8+ T cells are able to divide in an Ag-independent way whereas CD4+ T cells are not. These findings correlated with the more efficiently extended presentation of a MHC class II–restricted epitope following dendritic cell (DC) activation compared with a class I–restricted peptide. This suggests, we speculate, a coevolution of MHC class and the proliferative responses of the corresponding T cell subset.
Materials and Methods
Mice
AND (Tg(TcrAND)53Hed) (37), OT-1 (Tg(TcraTcrb)1100Mjb) (38), invariant chain-moth cytochrome c (Ii-MCC) (Tg(H2-Ea-Cd74/MCC)37GNnak) (39), Ii-reverse transactivator (rTA) (Tg(Cd74-rtTA)#Doi), tet-inducible modified Ii MCC (Tg(tetO-Cd74/MCC)#Doi) (28), CD11c-β2m and K14-β2m transgenics, and B2mtm1Jae (40), and Cd274tmiLpc mice (41) have been described previously. Animals were kept on the C57BL/6 (B6) and the B10.BR/SgSnJ (The Jackson Laboratory) backgrounds. The CD45.1 or CD90.1 congenic markers for AND and OT-1 TCR transgenic T cells were originally derived from B6.SJL-Ptprca Pepcb/BoyJ and B6.PL-Thy1a/CyJ animals (The Jackson Laboratory), respectively. All animals were housed and bred, and experiments were conducted, at the Institute for Immunology in compliance with German federal guidelines and approved by the government of Upper Bavaria (Az. 55.2-1-54-2531-106-08).
Transgene construct
The tetracycline (tet)-inducible signal sequence with OVA (TSO) transgene consists of the Kb signal sequence (from pUC19-H-2Kb, provided by B. Arnold, German Cancer Research Center, Heidelberg, Germany), followed by the sequence encoding OVA257–264, two stop codons, and the human growth hormone splice substrate under the control of the tet responsive elements of pTRE-tight (BD Clontech). Restriction digested transgene DNA was separated from plasmid sequences by agarose/crystal violet electrophoresis and purified for injection into C57BL/6 zygotes by Nucleobond column purification (Macherey-Nagel). Seven founders were identified by PCR with primers RO235 (5′-CTCATCTCAAACAAGAGCCA-3′) and RO236 (5′-CACTGCTTACTTCCTGTACC-3′) and crossed to Ii-rTA transgenics (28). DCs from double transgenic offspring were tested for proliferation of OT-1 T cells in the presence of titrated amounts of doxycycline (dox). One line (number 7) exhibited dox-dependent Ag presentation of the OVA peptide by DCs and all experiments described were done with this line.
Bone marrow chimeras
B6 and B10.BR animals were irradiated (5 Gy) with a [137Cs] source twice (6–48 h apart), transferred with 5 × 106 bone marrow (BM) cells from double-transgenic (dtg)-O or -M donors and supplied with 2 g/l neomycin sulfate and 0.1 g/l polymyxin b sulfate (Applichem) in the drinking water. The animals were used 4–6 wk later for adoptive transfers.
Abs and flow cytometry
Single-cell suspensions were stained in DMEM/5 mM HEPES/1% BSA on ice. Fluorochrome-conjugated Abs were CD4 (clone RM4-5), CD5 (53-7.3), CD8 (53-6.7), CD11b (M1/70), CD11c (N418), CD44 (IM7), CD45R (RA3-6B2), CD45.1 (A20), CD45.2 (104), CD49b (DX5), CD62L (MEL-14), CD69 (H1.2F3), CD71 (RI7217), CD98 (RL3889) IFN-γ (XMG1.2), and TCRβ (H57-597) (all BioLegend); CD25 (PC61), T-bet (eBio4B10), and Eomes (DAN11MAG) (all eBioscience), Ki67-Al488 (B56), Vβ5 (MR9-4) from BD, Vα2 (CL7213F) from Cedarlane Laboratories, and Vβ3 (a gift from N. Asinovski, C. Benoist, and D. Mathis, Boston, MA). Single-cell suspensions from lymphoid organs were prepared by mechanical dispersion. Erythrocytes were removed from splenocyte suspensions by centrifugation through a Ficoll cushion (LSM1077, PAA, or lympholyte M; Cedarlane Laboratories). All cytokine stainings were performed after a 4-h incubation with 20 ng/ml PMA and 1 μg/ml ionomycin, with 10 μg/ml brefeldin A (all Sigma-Aldrich) included for the last 2 h. Cells were fixed and permeabilized with buffers from BioLegend (fix/perm buffer) or eBioscience (FoxP3 staining buffer), according to the manufacturers’ instructions. Flow cytometric analyses were performed on FACSCanto II flow cytometers using the FACSDiva software (BD). Data were analyzed with the FloJo 8.8.7 software (Tree Star), the cells shown are DAPI or Fixable Viable Dye (eBioscience)–negative singlet lymphocytes, according to forward and side scatter properties. The proliferative index N based on CFSE dilution profiles was calculated as the average number of divisions compared with unstimulated control cells.
Cell preparation and stimulation in vitro
Lymph node (LN) and splenocyte suspensions were prepared in DMEM/5 mM HEPES/1% BSA subsets purified magnetically (Miltenyi Biotec) using positive selection with biotinylated CD4 or CD8 mAbs or negative selection with mAbs against CD11c, CD11b, GR1, CD49b, CD45R, CD4, or CD8 and anti-biotin beads to >90% purity. TCR transgenic or polyclonal CD4+ or CD8+ T cells were cultured for 2 d in 96-well round-bottom plates (Sarstedt) precoated with 10 μg/ml anti-CD3 (145-2C11) and anti-CD28 mAbs (37.51; both from BioXCell) or untreated plates at 1–2 × 105 cells/well in RPMI 1640 medium (PAA) supplemented with 10% FCS (Life Technologies), 2 mM glutamine, 50 μM 2-ME, antibiotics (complete medium), and 5 ng/ml IL-7 (Immunotools). For Th1 and Th2 differentiation of AND T cells in vitro, cells were stimulated with anti-CD3 and anti-CD28 in complete medium containing 5 ng/ml IL-12 and 20 μg/ml anti–IL-4 mAb (11B11) for Th1 conditioning or 50 ng/ml IL-4 and 50 μg/ml anti–IFN-γ mAb (XMG1.2; both BioXCell) for Th2 conditioning. Following stimulation, cells were CFDA-SE labeled and transferred as described previously (42).
Generation of rested effector CD4+ T cells
Rested effector cells were generated according to Ref. 43. Briefly, 0.5 × 105 naive LN AND T cells were cultured with 105 irradiated splenocytes from Ii-MCC transgenic mice that constitutively express Ek/MCC complexes on DCs and B cells (39) in 96-well plates in complete medium containing 80 U/ml IL-2 for 4 d. APCs were removed by centrifugation over a Ficoll cushion (LSM1077; PAA), and cells were cultured for an additional 3 d without IL-2.
In vivo killing assay
Congenically marked B6 splenocytes were cultured for 4 h in complete medium with or without 1 μg/ml SIINFEKL peptide (Peptides and Elephants). Cells were then labeled with 5 nM (pulsed) or 5 pM CFDA-SE (unpulsed) and 2.5 × 106 of each transferred into recipients of OT1 cells 3 d before. Splenocytes were analyzed 1 d later.
Animal treatments
For blocking coinhibitory interactions, animals were injected i. p. with 200 μg blocking mAbs to PD-1 (J43), PD-L1 (10F.9G2), CTLA-4 (UC10-4F10-11), LFA-1 (M17.4), or control Ab (polyclonal Armenian hamster IgG; rat IgG2a; all from BioXCell). For DC activation, 20–50 μg anti-CD40 (FGK45.5; Miltenyi Biotec or BioXCell) were given i.p. Where indicated mice were injected i. p. with 2 × 106 PFU murine CMV (MCMV) (provided by T. Brocker, Munich, Germany). For gene induction, mice were fed with 100 μg/ml dox (Sigma-Aldrich) diluted in water low in divalent cations (Volvic, Danone Waters) supplied ad libitum. The use of dox-treated nontransgenic and untreated dtg recipients did not make a difference so that antibiotic side effects have been excluded.
Cell sorting and RNA preparation
For 2-d transcriptome analysis, lymphocytes were sorted and cultured as described and sorted to >99% purity on a MoFlo cell sorter (Cytomation, Fort Collins, CO). Transferred CD4+ or CD8+ T cells were stained and CD4+/CD8+CD45.1+DAPI− cells were sorted twice, the second time with a purity of 99–100% directly into TRIzol (Invitrogen). RNA from sorted cells was isolated as described previously (42). One to 20 ng high-quality total RNA (RNA integrity number> 7; 2100 Bioanalyzer; Agilent Technologies) was amplified using the two-cycle MessageAmp II aRNA Amplification Kit (Ambion). Amplified aRNA (10 μg) was further processed using the Message Amp II-Biotin Enhanced Kit (Ambion) and hybridized on Affymetrix Mouse Genome 430 2.0 arrays in a GeneChip Hybridization Oven 645, washed, and stained with a GeneChip Fluidics Station 450 and scanned with a GeneChip Scanner 3000. Three biological replicates of stimulated and control CD4+ and CD8+ T cell samples at days 2 and 5 were analyzed as indicated in Figs. 2 and 6 (with two exceptions; 28 samples altogether). The data were analyzed with modules of the Genepattern package (Broad Institute, Boston, MA; Ref. 44). Primary .cel files were normalized with the ExpressionFileCreator module implementing RMA and redundant probe sets collapsed with the CollapseDataset module. One hundred and twenty-eight gene probes for B cell–specific (45) and XY chromosome–encoded genes were removed. Data were visualized with the Multiplot (volcano and fold change [FC]/FC plots), HierarchicalClusteringViewer and HeatMapViewer modules, nearest neighbors clustering was done with the Gene-E module of GenePattern. Regression lines in Fig. 6 were calculated in Matlab (Mathworks).
Statistical analysis
Cellular data were analyzed using the Prism 5.0 software (GraphPad, San Diego, CA).
Accession numbers
The microarray data have been deposited in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/gds) under the accession number GSE49063.
Results
Inducible Ag presentation of MHC class II and I epitopes
For comparing CD4+ and CD8+ T cell responses, we used dtg mouse models where DCs present antigenic epitopes under tet- or dox-dependent control of the Ii-rTA transgene described previously (28). Its combination with the tet-inducible modified Ii MCC transgene (dtg-M for brevity) allows for the induction of an epitope consisting of H-2Ek and an MCC peptide to which AND CD4+ T cells respond vigorously (Fig. 1A, 1B). To generate mice with controllable presentation of the H-2Kb/OVA257–264 complex, a transcriptional unit consisting of a signal sequence, the epitope, stop codons and a splicing substrate was put under the control of tet-responsive regulatory elements (TSO) and introduced into the germ line of B6 animals (Fig. 1A). This minigene targets the OVA257–264 peptide directly to the ER in a proteasome- and TAP-independent fashion and excludes indirect presentation of transferred OVA protein by other APCs. TSO+ founder animals were bred to the Ii-rTA line. dtg-O offspring displayed dox-inducible OVA257–264 presentation in vivo as evidenced by proliferation of CFSE-labeled OT1 TCR transgenic T cells (Fig. 1B). However, in contrast to dtg-M recipients, we observed proliferation in the absence of dox in the drinking water and 3 d after turn-off, suggesting unregulated TSO expression in at least some MHC class I+ cells in dtg-O animals. When we restricted the transgenes to BM-derived APCs in dtg-O → wild-type (wt) radiation chimeras, background and residual OVA257–264 presentation was extinguished so that OT1 T cells did only proliferate in dox-treated recipients but not in chimeras that had been treated 3 d before or not at all (Fig. 1B).
To directly compare the effective half-lives of the two epitopes in the steady state, we treated dtg-M and -O → WT chimeras with dox for 24 h and transferred the respective T cells 4–0 d later. We used the CFSE dilution of transferred T cells to estimate the disappearance of pMHC complexes as far as they can be detected by naive TCR-transgenic T cells. In this sense, we estimated the epitopes’ relative survival time to about a 1 d for Ek/MCC, as observed previously in nonchimeric dtg-M animals (28) and 2 d for Kb/OVA (Supplemental Fig. 1), likely because of the slow dissociation of this peptide (46).
Comparable activation of AND and OT1 T cell activation in vitro
We asked next whether CD4+ and CD8+ T cells responded differently to transient signals. Because of the “leakiness” of the dtg-O system, we decided to prime the T cells in vitro by surface-bound mAbs. Thereby, MHC class differences of tissue distribution, cell biology, and stability as well as differential imprecision of the tet systems are circumvented: because the anti-CD3 and -CD28 mAbs trigger T cells independently of αβTCR and coreceptor avidities, this system may be better suited for revealing differences between the two T cell types. Following 2 d of stimulation, AND and OT1 T cells exhibited no significant differences as far as TCR downregulation and the activation markers CD25, CD44, CD62L, CD69, CD71, and CD98 were concerned; the only notable difference was the slightly lower TCR expression levels on the OT1 cells before and after stimulation (Fig. 2A), a finding that also applied to polyclonal B6 T cells (data not shown). We also noted a difference of CD5 expression between the subsets (Supplemental Fig. 2), in agreement with previous work on higher tonic TCR signaling (47) and ensuing CD5 expression by CD4+ T cells (48, 49). Following incubation, the cells had become blasts and had initiated proliferation because they had divided once on average (Fig. 2B). To find out how similar the cells were in terms of their gene transcription at this point, we analyzed their transcriptome with microarrays. Both cell types had induced ∼1400 genes at least 2-fold (Fig. 2C, upper panels). The high accuracy of the reciprocal projections of these gene sets between the T cell subsets, evidenced by infinitesimal p values, indicated the high degree of similarity of the two induced transcriptomes at this time point. Overall, these data indicated that the culture conditions activated both T cell subsets comparably as far as protein, proliferative, and transcriptional parameters can tell and are in agreement with recent data on subset signaling and transriptomes (50).
CD8+ but not CD4+ T cells keep dividing after a transient stimulus
To address whether CD4+ and CD8+ T cells exhibit different proliferative capacities following a transient TCR stimulus, MACS-sorted AND and OT1 T cells were cultured in the absence or presence of anti-CD3 and -CD28 mAbs, CFSE-labeled 2 d later, and transferred into congenic Ag-free (condition 2) or -expressing (condition 3) recipients. Cells cultured without the mAbs and transferred into nontransgenics served as controls (condition 1; Fig. 3A). When the LN cells were analyzed 3 d later, AND T cells had divided once or twice upon withdrawal of the stimulus, whereas OT1 T cells had continued to divide further (mean ± SEM: 1.5 ± 0.4 versus 4.0 ± 0.4 divisions). Because lymphocyte egress from LNs is regulated specifically, the recipients’ spleens also were analyzed, and a similar difference was found (1.3 ± 0.2 versus 4.3 ± 0.4). Interestingly, the OT1 T cells had divided almost as much as in the presence of persistent Kb/OVA presentation (Fig. 3B).
To test whether the observations could be reproduced for cells with a diverse TCR repertoire, we used nontransgenic CD4+ and CD8+ T cells from B6 animals under the conditions 1 and 2 described above and found again the CD4/CD8 split in LN (1.8 ± 0.4 versus 3.8 ± 0.4) and spleen (1.6 ± 0.4 versus 3.6 ± 0.5; Fig. 3C). This could also be reproduced with cells from BALB/c animals (Fig. 3D), indicating that our findings were not confined to particular TCR transgenics or genetic backgrounds.
Because the proliferation data were indicative of an increased propensity for programming in CD8+ T cells, we compared the IFN-γ secretion by AND and OT1 cells under conditions of transient and continued TCR triggering. Although both cell types produced the cytokine in response to the persistent signal, only OT1 but not AND cells could be stained for IFN-γ following the transient stimulus (Fig. 3E). Furthermore, OVA257–264-pulsed target cells were efficiently killed by OT1 cells after both transient and continued stimuli (Fig. 3F). These data indicated that CD8+ T effector cell differentiation can occur following a signal of limited duration.
It is conceivable that CD8+ T cells continue their proliferation beyond the initial priming by interactions with self-pMHC complexes, like under the homeostatic conditions of the steady state and in the case of lymphopenia-driven proliferation. Therefore, we used recipients that were deleted of their β2m genes and hence lack MHC class I surface expression but were replete with a transgene expressed in thymic cortical epithelium to allow for the selection of a full T cell compartment, thus reining in lymphopenia-driven proliferation of adoptively transferred CD8+ T cells. Recipients carrying an additional DC-targeted β2m transgene served as positive controls (40). However, when 2-d-stimulated CD8+ T cells were transferred into such Ag-free hosts, they proliferated, regardless of MHC class I expression on peripheral cells (Fig. 3G). These data indicated that CD8+ T cells were able to proliferate beyond the priming phase independently of any TCR signals delivered by self-pMHC complexes.
Likewise, Ab-mediated blocking of LFA-1–dependent T cell clusters (51) did not interfere with the Ag-independent proliferation phase of OT-1 cells, making it unlikely that T-T synapses are involved in late CD8+ T cell proliferation (Supplemental Fig. 3).
Ag-independent CD4+ T cell proliferation is not affected by Th subtype, coinhibiton, inflammation, or memory cell differentiation
We asked next whether the difference in Ag-independent proliferation between CD4+ and CD8+ T cells was caused by specific inhibition of the CD4+ T cells. Th1/Th2-polarizing conditions in the priming cultures did not affect the impeded proliferation of AND cells in Ag-free hosts, excluding subset-specific cytokines as mediators (Fig. 4A).
Coinhibitory molecules like CTLA-4 and PD-1 interfere with T cell priming and there is evidence that they transmit the initial T cell migratory stop signal and thus regulate time and/or strength of the priming interaction between T cells and DCs (52, 53). To investigate whether these molecules block Ag-independent proliferation of CD4+ T cells, we transferred 2-d-primed CFSE-labeled AND and OT1 T cells into Ag-free hosts, which had received blocking mAbs against CTLA-4, PD-1, or PD-L1 (Fig. 4B). None of these treatments enhanced AND T cell division, indicating that CD4+ T cells were not specifically blocked by any of these molecules. To test whether this also applied to the priming phase of the response, we transferred naive AND T cells into transiently dox-treated dtg-M recipients, as described for Fig. 1D, and injected an anti–CTLA-4 mAb at the same time. In addition, we used animals lacking the gene encoding PD-L1 (Cd274) to assess the effect of both pathways being effectively blocked. The application of the mAb did not reveal any difference to controls, even in recipients lacking PD-L1 (Fig. 4C). Thus, CD4+ T cell proliferation is not specifically held in check by cell-extrinsic coinhibition.
We then tested whether the increased production of inflammatory cytokines may trigger T cell expansion in an Ag-independent way by transferring prestimulated CD4+ and CD8+ T cells into animals that were or were not infected by a high dose of MCMV 2 d before. This virus sets off the release of multiple proinflammatory cytokines including IL-12 and IFN-α (54). However, CD4+ T cells were not triggered to undergo additional divisions in MCMV-infected recipients (Fig. 4D), indicating that cytokines produced by the recipients’ cells do not contribute to the expansion of naive T cells following initial Ag encounter.
To assess the effect of precursor frequency in our system, we lowered the number of transferred AND cells stepwise down to 62,500 per recipient and did not detect any effect on proliferation (data not shown). The results excluded mutual T cell inhibition as a mechanism of blunted CD4+ T cell proliferation and confirmed earlier data on the issue obtained by different methods (31).
We next asked whether CD4+ T cells acquire the capacity to proliferate in an Ag-independent way upon their differentiation into memory cells. We therefore generated rested effector cells, which resemble memory cells in their transcriptome, physiology, and function (43, 55), and tested whether their response to transient Ag presentation, as described for Supplemental Fig. 1, differs from naive cells. This was not the case, making it unlikely that CD4+ memory cells proliferate independently of Ag in secondary responses (32). In summary, our data exclude coinhibition and the lack of proinflammatory cytokines as blocks of Ag-independent proliferation of CD4+ T cells and show that memory cells do not acquire this function, suggesting it is a lineage-specific trait.
Direct comparison of cell cycle status
Although CFSE dilution records the division history of a cell population, direct cell cycle analyses visualize proliferation at the time of dissection. Therefore, we quantified the percentages of AND and OT1 T cells in S and G2-M phases by intracellular DAPI stains at several time points (Fig. 5A). Following 2 d of stimulation, similar fractions of the two cell types were in cell cycle (AND: 43.1 ± 4.2; OT1: 41.5 ± 3.3%). However, after 7 d of TCR triggering, the OT1 cells showed a significantly higher percentage of cycling cells (7.6 ± 0.5%) than AND T cells (2.3 ± 0.3%), suggesting that CD8+ T cells are susceptible to TCR signals for longer periods of time (Fig. 5A).
The Ki67 protein is associated with proliferation in many cell types and is a reliable though indirect probe for division. Following 2 d of stimulation, both AND and OT1 cells express it, and both T cell types lose it upon withdrawal of the stimulus with a t1/2 of days (Fig. 5B, upper panels). The AND T cells, however, lose this marker despite persisting Ag presentation, whereas the OT1 cells maintain it at high levels (Fig. 5B, lower panels). These data indicate that CD8+ T cells are able to translate TCR signals into cell division for longer periods of time than CD4+ T cells, a fact that may well contribute to the better overall expansion of CD8+ T cells. Our data also indicate that Ki67 is a protein with a survival time of several days and not necessarily an authentic reporter of momentary cell division. Similar findings have been reported recently for noncycling blood monocytes that maintain Ki67 expression from their precursor stage in the BM (56).
Gene expression following transient and persistent stimulation
To assess how continued TCR signaling affects the overall transcriptome of the two cell types, we transferred 2-d-stimulated AND and OT1 T cells into Ag-free or Ag-expressing dtg-M and -O hosts, respectively, and sorted them 3 d later for microarray analysis. The FC/FC plots in Fig. 6A compare the cells’ transcriptomes in response to transient (abscissa) and persistent (ordinate) TCR triggering to their respective controls. There are evidently more genes expressed in a stimulus-independent way in the CD8+ OT1 cells compared with the CD4+ AND T cells. This is reflected in the trend lines’ slopes differing by a factor of ∼2 (Fig. 6A). The expressed genes can be grouped as Ag-independent and close to the x = y diagonal on the one hand and rather Ag-dependent aligning around the y = 1 horizontal on the other hand. As noted in a previous analysis on day 3 (42), Ctla2, Ctla4, and Cxcr3 belong in the Ag-independent category, whereas the expression of Grzmb, Grzmk, Irf4, and Irf8 is rather Ag-dependent in both CD4+ AND and CD8+ OT1 T cells (Fig. 6A, 6B). Of note, the group of genes expressed Ag independently by OT1 but not AND T cells include cell cycle-associated Cdc25b, Cdc26, Cdc34, and Cdc37. Both Cdc25 and Cdc34 are regulators of proliferation for which specific inhibitors have identified that block cell cycle progression of tumor cells (57, 58). Tbx21 encoding T-bet, the transcription factor regulating CTL differentiation, also falls into this category (Fig. 6B, 6C). Accordingly, T-bet protein is expressed in a more Ag-independent way in OT1 T cells than it is in AND T cells (Fig. 6D). In summary, these data show a differential Ag-dependency of the transcriptional landscape in AND and OT1 T cells and support the notion of the CD8+ T cells’ propensity to divide and differentiate following an Ag pulse of limited duration.
Stabilization of pMHC complexes upon DC activation in vivo
The main difference between CD4+ and CD8+ T cells is their respective MHC class II and I ligands. In view of our finding that CD4+ and CD8+ T cells differ in their temporal requirements of TCR signals, we asked whether the kinetics of Ag presentation differs for their respective MHC ligands on DCs, the main APCs for priming. Because MHC class II molecules are stabilized on the cell surface upon DC activation by downregulated oligo-ubiquitination of class II β-chains (59, 60) by membrane-associated RING-CH–like E3 ligases (61, 62) in vitro (63, 64) and in vivo (42, 65), we tested whether DC activation affects molecules of both MHC classes similarly. We activated dtg-M and -O DCs in vivo by treatment with a stimulatory CD40 mAb and fed the animals with dox for 24 h subsequently. Three days after turn-off, CFSE-labeled AND and OT1 cells were transferred as probes for stabilized pMHC complexes. Fig. 7 shows that AND but not OT1 T cells detected remaining pMHC complexes. We have shown previously that this proliferation of AND T cells was not caused by increased costimulation (42). These findings indicate that the turnover of Kb/Ova257–264 complexes is not affected by DC maturation in such a way that OT1 cells might detect them 3 d after dox turn-off. Our data are in agreement with results obtained with human DCs in vitro that showed no extension of bulk MHC class I half-lives upon LPS treatment (63, 66) and a 3-fold increase by direct influenza virus infection (66). Recent data, obtained with the minimally invasive technique of heavy water labeling, found a 2-fold increase of MHC class I, but a more than 10-fold increase of MHC class II half lives in activated DCs (R. Busch, personal communication). We conclude that our in vivo findings with two TCR/pMHC combinations are in agreement with previous in vitro data on bulk MHC protein turnover rates. The differential Ag dependency of CD4+ and CD8+ T cells might thus be reflected by their respective pMHC ligands’ distinct susceptibility to regulation of their stability.
Discussion
Our comparison showed cell-intrinsic differences of proliferation between the CD4+ and CD8+ T cell subsets. The experiments were executed in a way to circumvent confounding differences like APC identity, Ag processing, and cell biology of MHC class I and II proteins on the one hand, and the T cells’ differential costimulatory, adhesion, and third-signal cytokine requirements and CD4/8 coreceptor affinities on the other hand. Although CD8+ T cells do not necessarily keep dividing following a short antigenic pulse, they can be pushed to do so, whereas CD4+ T cells cannot. Transiently stimulated CD8+ T cells are fully functional as they differentiate into cytotoxic and IFN-γ–secreting cells, are independent of self-pMHC complexes and are therefore truly “on autopilot” (11). CD4+ T cells could not be brought into an Ag-independent phase of proliferation by coinhibition blockade and inflammatory conditions, are in S phase of the cell cycle and express the proliferation marker Ki67 for shorter periods of time and show a gene expression profile upon Ag withdrawal clearly distinct from their CD8+ counterparts. It is likely that our results underestimate the differences described as the T cells were stimulated and transferred separately. If anything, CD4+ T cells assist APCs in Ag presentation to CD8+ T cells, whereas CD8+ T cells rather limit APC availability for their CD4+ counterparts (30, 32, 33).
We discuss below how these findings have intriguing parallels in the thymic differentiation of the two subsets, migratory behavior of naive and memory cells, differential importance of immunization parameters, and, importantly, correlate with differential stability of MHC class I and II complexes and their respective Ag sources.
The TCR-guided differentiation of CD4+CD8+ precursors to single-positive mature thymocytes is influenced by the kinetics of the TCR signals to achieve the match of MHC class reactivity with coreceptor expression: A transient signal favors the differentiation of CD8+ cells, whereas CD4+ ones require longer signals (67–69). Although the molecular details are not yet known, it is tempting to speculate that thymic assortment and peripheral response mode share a molecular basis. The theme of higher dependency on MHC-derived signals extends to naive CD4+ T cells that traverse LNs in an MHC-dependent manner and adjust CD5 levels to TCR affinity, whereas CD8+ T cells do not (49, 70). In addition, the CD4+ T memory cells recirculate more freely through tissues, while their CD8+ counterparts are confined in an Ag-independent way at the infection site (71, 72). These findings suggest that the CD4+ T cell lineage is marked for TCR-guided plasticity, whereas its CD8+ counterpart responds in a more predetermined way.
The key difference between CD4+ and CD8+ T cells is their MHC ligands and their tissue distribution (73). It is understood that both T cell subsets are primed best by activated DCs, although not necessarily by the same subsets (74, 75). In DCs, the classical class discrimination of the presentation pathways is breached by cross-presentation of exogenous Ags by class I and autophagy of cytosolic Ags for presentation by class II molecules. The half-lives of the MHC molecules, however, are controlled by distinct ubiquitination pathways. The membrane-associated RING-CH family E3 ubiquitin ligases affect not only several aspects of the MHC class II pathway, such as its master transcription factor CIITA, the peptide exchange by DM, class II trafficking and t1/2, but also costimulatory molecules such as CD80 and CD86. In activated DCs, class II and CD80/86 ubiquitination is reduced, allowing for high surface expression of the molecules necessary for efficient CD4+ T cell priming (59, 60). In contrast, the t1/2 of bulk MHC class I molecules is not extended >3-fold on LPS-activated DCs (63, 66; R. Busch, personal communication), with perhaps individual pMHC complexes varying over a wide range (76), and ubiquitination by viral E3 ligases that targets them for destruction has been merely described as a mechanism of immune evasion.
A second difference between class I and class II molecules affecting their presentation kinetics is the source of Ag: While for class II whole proteins are internalized and degraded in the endosomal compartment, most of the substrates for the class I pathway are newly synthesized polypeptides and defective ribosomal products (77). Accordingly, in a side-by-side comparison of presentation kinetics with precisely regulated Ag expression, the presentation via MHC class I correlated closely with active gene transcription and that of class II with the stability of mature viral proteins (78). It is thus tempting to speculate that the capability of CD8+ T cells to respond to a short Ag pulse might reflect transient viral gene transcription and its preferred feeding of the class I pathway. The fact that CD4+ T cells respond much longer to residual Ag following viral clearance than CD8+ T cells do also indicates that antiviral CD8+ T cell responses depend on viral transcription (79, 80).
Further evidence for the importance of extended Ag presentation for CD4+ immunity has been generated by targeting Ags to DCs by mAbs of different half-lives: CD4+ T cell immunity and Abs can be raised in the absence of adjuvants given an extended t1/2 of the immunizing mAb (81). This finding suggests necessary refinements of our understanding of DCs in immunity and tolerance (82) with important implications for peptide vaccination (83).
In summary we have shown that CD4+ and CD8+ T cells differ in their cell-intrinsic capacities to proliferate beyond an initial antigenic stimulus. As CD4+ T cells direct (“help”) other immune cells toward effector functions they appear to be under tighter control of Ag presentation than CD8+ T cells whose main functions IFN-γ production and cytotoxicity unravel even under Ag-free conditions. This difference might be reflected in differences of Ag source for and stability of presentation by their respective MHC protein classes.
Acknowledgements
We thank A. Kollar and S. Pentz for technical assistance, A. Bol and W. Mertl for animal husbandry, B. Arnold, H.-G. Rammensee, C. Benoist, D. Mathis, and T. Brocker for reagents or mice, C. Guo for Matlab-based calculations, R. Busch for sharing unpublished data, J. Johnson and T. Brocker for discussions, and A. Erlebacher and E. Huseby for comments on the manuscript.
Footnotes
This work was supported by German Research Council Grants SFB571-B8 and SFB1054-B7 (to R.O.), Bundesministerium für Bildung und Forschung Grant NGFNplus 01GS0850 (to J.B.), and the Münchner Medizinische Wochenschrift (2010) (to H.R.).
The microarray data presented in this article have been deposited in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/gds) under accession number GSE49063.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- B6
C57BL/6
- BM
bone marrow
- DC
dendritic cell
- dox
doxycycline
- dtg
double-transgenic
- FC
fold change
- Ii
invariant chain
- LN
lymph node
- MCC
moth cytochrome c
- MCMV
murine CMV
- pMHC
peptide-MHC
- rTA
reverse transactivator
- tet
tetracycline
- TIM
tetracycline-inducible invariant chain with MCC
- TSO
tetracycline-inducible signal sequence with OVA
- wt
wild-type.
References
Disclosures
The authors have no financial conflicts of interest.