Memory CD8+ T cells from mice previously primed with alloantigen (alloAg) can respond in vitro to IL-2 and purified class I alloAg presented on microspheres, while no response can be detected using cells from naive mice. Similar results have been obtained using cells from OT-1 mice expressing a transgenic TCR that is specific for OVA257–264 (SIINFEKL) peptide bound to H-2Kb. A population of resting memory cells (defined on the basis of low forward scatter and CD44high, Ly-6C+, CD25−, CD69− surface phenotype) that is present in the OT-1 mice exhibits a substantially higher sensitivity to Ag-stimulation than do naive cells (CD44low, Ly-6C−) expressing the same TCR. CD44high cells respond vigorously to H-2Kb immobilized on microspheres and pulsed with peptide, while CD44low cells respond weakly and only at high class I density and peptide concentration. The Ag-presenting surface only has ligands for TCR and CD8 (class I and peptide), thus ruling out the possibility that differences are due to ligand binding by other adhesion or costimulatory receptors that are expressed at high levels on the memory cells. Experiments using anti-TCR mAb as the stimulus and coimmobilized non-Ag class I as a ligand for CD8 suggest that the difference between naive and memory cells may be at the level of stimulation through the TCR. Thus, in addition to expressing increased levels of adhesion receptors that may enhance responses to Ag on APCs, memory CD8+ T cells appear to be intrinsically more sensitive than naive cells to stimulation through the TCR/CD8 complex.
T cells from animals that have been previously exposed to Ag make a stronger and more rapid memory response upon reexposure to the same Ag (1, 2, 3, 4). This more vigorous memory response is due in part to the increased frequency of Ag-specific cells following an initial response (5) and may also involve some selective expansion of clones with higher affinity TCRs (6). However, memory T cells also differ qualitatively from naive cells. Naive and memory cells have distinct surface phenotypes, with memory cells expressing higher levels of several adhesion molecules (7). Differing expression of surface receptors results in naive and memory cells having distinct recirculation pathways that may contribute to more rapid memory responses in vivo (8, 9).
In addition, there is evidence that memory cells are more readily activated than naive cells, as evidenced by their ability to respond to lower doses of Ag (10, 11), by their lesser dependence on costimulatory receptor-ligand interactions (12, 13, 14), and by their enhanced responses to stimulation with anti-TCR Abs (15). The increased expression of adhesion receptors on memory cells promotes their adherence to APCs, leading to higher levels of engagement of TCR and other receptors that might be involved in activation; this is likely to contribute to their enhanced responses in comparison with those of naive cells.
Whether naive and memory cells might also differ with respect to their sensitivity to signaling through the TCR and CD4 or CD8 coreceptors, independently of contributions from other adhesion molecules, is less clear. One approach to addressing this question is to determine whether activation can occur when purified MHC Ag is presented to T cells on a surface that lacks ligands for adhesion molecules or other receptors and also whether this differs for naive vs memory cells. In previous studies using this approach, it was demonstrated that purified class I alloantigen (alloAg) incorporated onto cell-size microspheres could stimulate in vitro generation of a cytolytic response by CD8+ cells from previously primed mice, provided that exogenous cytokines were added, but could not stimulate a primary in vitro response by cells from naive mice (16). While this observation suggested the possibility that naive and memory cells might differ in their sensitivity to Ag-dependent stimulation, it could not be ruled out that the observed differences might have resulted from differing precursor frequencies and/or selective expansion of clones with higher affinity TCRs during the primary in vivo response.
The use of cells from mice expressing a transgenic TCR eliminates issues of precursor frequency and potential differences in TCR affinities for Ag in the populations being studied. Using this approach and taking advantage of the fact that OT-1 transgenic mice with a TCR that is specific for OVA257–264 (SIINFEKL) complexed to H-2Kb have an endogenous population of TCR transgene-positive memory cells, we have quantitatively compared the requirements for Ag-dependent activation of naive vs memory CD8+ T cells to respond to IL-2 in the absence of contributions from receptors other than TCR and CD8. The results demonstrate that memory cells are much more sensitive to stimulation and can be effectively activated at levels of Ag that do not activate naive cells.
Materials and Methods
Animals and cell lines
C57BL/6 and AKR mice were purchased from Charles River Laboratory (Wilmington, MA). OT-1 mice (17), originally provided by Dr. Francis Carbone (Monash Medical School, Victoria, Australia), were bred under specific pathogen-free conditions at the University of Minnesota. Mice were between 2 and 6 mo of age when used. RDM4 (H-2k), an AKR lymphoma, and EL4 (H-2b), a C57BL/6 thymoma, were maintained in vitro in RPMI 1640 supplemented with 10% FCS, 4 mM l-glutamine, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 U/ml penicillin and streptomycin, 10 mM HEPES, and 50 μM 2-ME (RP-10). E.G7 cells (EL4 cells transfected with OVA) were maintained in vitro in RP-10 supplemented with 250 μg/ml G418.
Preparation of Ag-bearing microspheres
H-2Kb, Db and Kk class I MHC proteins were purified by mAb affinity chromatography from EL4 (H-2b) and RDM4 (H-2k) tumor cells, as previously described (18, 19). Purified Ags were eluted from the affinity columns in 0.5% deoxycholate and 0.15 M NaCl in 10 mM Tris (pH 8.2) and stored at −20°C until used. Anti-Vα2 mAb B20.1 was a kind gift of Dr. Kristin Hogquist, Center for Immunology, University of Minnesota, Minneapolis, MN. Class I Ags and anti-TCR mAb were immobilized on 5-μm diameter sulfate polystyrene latex microspheres (Interfacial Dynamics Corporation, Portland, OR) by adding protein to beads in suspension at 107/ml in PBS and incubating for 1.5 h at 4°C with continuous mixing. Next, 0.5 ml of 1% BSA in PBS per ml of bead suspension was added to block remaining sites on the beads, and mixing was continued for an additional 0.5 h at 4°C. Coimmobilization of both B20.1 mAb and H-2Db was completed by sequential immobilization of first B20.1 and then H-2Db. Following immobilization and blocking, beads were washed and suspended in RP-10 for addition to cultures. Final bead preparations were quantitated by counting using a hemacytometer, and levels of protein incorporated onto the surface were determined by flow cytometry.
Flow cytometry analysis and cell sorting
Beads or cells (∼2 × 105) were washed 1× in HBSS containing 2% newborn calf serum and 0.1% NaN3 (flow buffer). Beads coated with class I protein were incubated for 20 min on ice with mAb specific for H-2Kk (11–4.1 mAb), H-2Kb (Y-3 mAb), or H-2Db (28–14–8 mAb). All anti-class I Abs were purified from hybridoma culture supernatants by passage over protein A-Sepharose columns. Beads were washed once with flow buffer and incubated for an additional 20 min on ice with goat anti-mouse Ig-FITC (Jackson ImmunoResearch Laboratories, West Grove, PA). Cells were stained for 20 min on ice with directly conjugated Abs. Abs used were obtained from PharMingen (San Diego, CA) and included 53–6.7-FITC or -phycoerythrin (PE) (anti-CD8 mAb), H129.19-PE (anti-CD4 mAb), B20.1-PE (anti-Vα2 TCR mAb), 7D4-FITC (anti-IL-2Rα mAb), IM7-FITC or -PE (anti-CD44 mAb), and AL-21-FITC (anti-Ly-6C mAb). The analysis was performed on a FACScan using CellQuest software (Becton Dickinson, San Jose, CA). Mean channel fluorescence values for beads or cells were converted to molecules of equivalent soluble fluorochrome (MESF) by comparison with a calibration curve generated using Quantum 26 calibrated fluorescent standard microbeads (FCSC, San Juan, Puerto Rico).
OT-1 lymph node cells (LNC) were column-enriched for CD8+ cells as described below, stained with anti-CD8-FITC and anti-CD44-PE mAbs, and sorted using a FACStarPlus flow cytometer. Cells were gated for size and viability by forward and side scatter and sorted into CD44lowCD8+ and CD44highCD8+ populations. For some experiments, CD44low and CD44high populations were obtained by sorting column-enriched CD8+ cells using only anti-CD44-FITC mAb.
In vitro formation of Kb/OVA257–264 Ag complexes
OVA257–264 peptide (SIINFEKL) was purchased from Chiron Mimotopes (Clayton, Victoria, Australia). Stimulator cells were prepared using irradiated C57BL/6 spleen cells suspended at 5 × 106/ml and incubated for 1 h at 37°C with OVA257–264 at a final concentration of 0.2 μM. Cells were then washed three times with RP-10 before their addition to cultures. Microspheres having H-2Kb or H-2Db immobilized on the surface were resuspended at 5 × 106/ml in RP-10, incubated for 2 h at 37°C with OVA257–264 at the concentrations indicated in figure legends, and then washed three times with RP-10 before they were added to cultures.
Spleen cells were obtained from naive or alloAg-primed C57BL/6 mice. In vivo priming was performed 6 to 12 wk before use by i.p. injection of 20 × 106 live RDM4 cells in a volume of 0.5 ml of PBS. Spleen cells or LNC were obtained from OT-1 TCR transgenic mice. In all cases, cells were enriched for CD8+ cells by passage over Cellect-plus mouse CD8-enrichment columns (Biotex Laboratories, Edmonton, Alberta, Canada) according to the procedure provided by the manufacturer. CD8+-enriched populations were cultured in RP-10 medium in flat-bottom microtiter wells with either Ag-coated microspheres or stimulator cells. Where indicated, cultures were supplemented with supernatant from Con A-stimulated rat spleen cells (ConASN) (T-STIM, Collaborative Biomedical Products, Bedford, MA) at a final concentration of 5% or with human rIL-2 (R&D Systems, Inc., Minneapolis, MN) at a final concentration of 5 U/ml. Proliferation was measured at the times indicated in figure legends (2–5 days) by the addition of 1 μCi of [3H]TdR per well for the last 8 h of culture. Cultures were harvested onto filters using a LKB-Wallac harvester and counted in a Betaplate liquid scintillation counter (Wallac, Turku, Finland).
Lytic activity of CD8+ cells from in vitro cultures was determined in a standard 51Cr release assay. For allogeneic responses, RDM4 targets were used, with EL4 targets as a control for specificity. For OT-1 responses, E.G7 targets were used, with EL4 targets as a control for specificity. Briefly, varying numbers of effector cells were added to 1 × 104 51Cr-labeled target cells in a final volume of 0.2 ml in V-bottom microtiter wells. After a 4-h incubation at 37°C, plates were centrifuged, 0.1 ml of the supernatant was removed, and radioactivity was counted. Spontaneous release was determined by measuring 51Cr release from target cells incubated in medium alone for 4 h; total release was determined by measuring 51Cr release from target cells incubated in 5% Triton X-100. Percent specific lysis was calculated as follows: 100 × (experimental release − spontaneous release)/(total release − spontaneous release). Data are expressed as LU, where one LU is defined as the number of effector cells required to give 50% specific 51Cr release.
Class I Ag and IL-2 are sufficient to stimulate generation of an allogeneic response by memory, but not naive, CD8+ T cells.
Previous studies demonstrated that spleen cells from mice primed with alloAg could be stimulated in vitro to develop cytolytic activity in response to purified class I alloAg on silica microspheres together with supernatant from ConASN (16). In contrast, no cytolytic activity developed in response to this minimal stimulus when cells from naive mice were used. More recent work has shown that purified class I proteins can be immobilized directly onto latex microspheres, and that a very high surface density of Ag can be achieved (20, 21). In vitro stimulation of naive and primed cells was reexamined using CD8+-enriched cells, to determine whether in vitro primary responses could be obtained at high Ag densities and whether the IL-2 cytokine was sufficient to support generation of a secondary response.
Spleen cells from C57BL/6 mice, either naive or primed 6 to 12 wk previously with live RDM4 (H-2k) allogeneic tumor cells, were enriched for CD8+ T cells as described in Materials and Methods. CD8+ cells increased from 9% to 57% of total cells following enrichment, while CD4+ cells decreased from 8% to <1%. Stimulation of primed cells with H-2Kk on microspheres resulted in the generation of cytolytic activity within 5 days, provided that IL-2 was added to the cultures (Fig. 1 A). The response to H-2Kk-microspheres was completely dependent upon exogenous IL-2, while irradiated spleen stimulator cells from AKR (H-2k) were able to stimulate a response in its absence. The response to purified class I proteins on the microspheres was Ag-specific; H-2Db microspheres stimulated no greater cytolytic activity than the low level generated in response to IL-2 alone. The response to H-2Kk was also dependent upon the density of class I on the surface of the microspheres.
When proliferation of CD8+ T cells from previously primed mice was examined on day 3 in response to these same stimuli, a strong response was obtained to irradiated AKR spleen stimulator cells, but only minimal proliferation occurred in response to Kk microspheres and IL-2 (Fig. 1,B) despite the development of potent cytolytic activity in these cultures (Fig. 1 A). This might be due to the fact that the frequency of Ag-specific responder cells was too low to detect in this assay, although it is possible that proliferation is not required for the generation of lytic activity by the memory cells, and that class I Ag alone is not a sufficient stimulus to promote proliferation.
Even using microspheres with a very high surface density of Ag as the stimulus along with either IL-2 or ConASN, CD8+ cells from naive mice generated no detectable proliferative or cytolytic response (data not shown). Although a relatively large number of precursor CD8+ cells in naive mice are responsive to a single allogeneic class I protein, it is still possible that a response by naive cells is not detected due to the possibility that the number of specific precursors is too low, while the expanded number of precursors in a memory population is sufficient to allow detection of a response. This might particularly be the case if class I Ag is not a sufficient stimulus to promote significant IL-2-dependent proliferation of the Ag-specific cells. Alternatively, naive and memory CD8+ T cells might have differing requirements for activation via the TCR and CD8, the only receptors involved in stimulating a response when class I Ag is the only ligand on the “APC” surface. To examine these possibilities further, experiments were performed using CD8+ cells from TCR transgenic mice in which precursor frequency was not an issue.
CD8+ cells from OT-1 mice proliferate and develop cytolytic activity in response to class I/peptide Ag complexes and IL-2
OT-1 mice are transgenic for the α- and β-chains of a TCR that is specific for OVA257–264 peptide (SIINFEKL) bound to H-2Kb (17); essentially all of the CD8+ cells express the transgenic Vα2+ TCR, and >90% of the Vα2+ cells are CD8+. CD8+ cells enriched from the lymph nodes of these mice, as described in Materials and Methods, are about 90% CD8+ and <0.5% CD4+. To determine whether class I Ag is sufficient to stimulate proliferation, H-2Kb was immobilized on microspheres and pulsed with varying concentrations of OVA257–264 peptide to form Ag complexes. Then the microspheres were washed to remove unbound peptide, placed in culture with OT-1 CD8+ T cells, and IL-2 was added. The cells proliferated in response to the Ag over days 3 through 5, with the level of proliferation being both dependent upon the concentration of peptide used to form complexes with Kb on the microspheres and maximal at 1 to 2 μM peptide (Fig. 2,A). The level of response obtained was also dependent upon the density of Kb on the surface of the beads (Fig. 2 B). Microspheres with a Kb density in the range of that found on normal cells (made using 1 or 2 μg class I per 107 beads) stimulated a strong response, and this was further increased at higher Kb density, concomitant with peak proliferation occurring earlier.
A variety of controls demonstrated that the proliferation depended upon recognition of the Kb/OVA257–264 complexes on the beads. Kb microspheres that were not pulsed with peptide did not stimulate a response (Fig. 2, legend). Microspheres made using Db, which binds OVA257–264 (22), and pulsed with peptide also did not stimulate a response (Figs. 2 and 5, legends, and data not shown), confirming a requirement for the presence of the Kb restriction element on the bead and making it unlikely that OT-1 cells were responding to peptide that had come off the beads and was being presented by other CD8+ cells in the culture. This possibility was further ruled out in experiments using transwells (data not shown). Kb-microspheres that had been pulsed with peptide only stimulated a response when they were added on the same side of the transwell as the cells. No response was obtained when they were placed on the other side of the membrane, even though free peptide added to the other side could readily cross the membrane and stimulate a response.
Response of OT-1 CD8+ T cells to Kb/OVA257–264 is predominantly from cells having a memory cell surface phenotype
The ability of CD8+ cells from OT-1 mice to respond to class I/peptide complexes despite not having been immunized with OVA raised the possibility that the inability to stimulate naive allogeneic responses with class I alloAg might be accounted for by low numbers of precursors in normal mice. However, some TCR transgenic mice have a pool of memory cells that express the transgenic TCR despite never having been exposed to the relevant Ag (23). There is considerable evidence that these cells also express an endogenous TCR α-chain, and it is this additional TCR that stimulates memory cell generation. Therefore, we examined this possibility in the OT-1 mice and found that they do in fact have a population of CD8+, Vα2+ cells with a memory surface phenotype.
The expression level of CD44 provides a good marker for murine CD8+ cells that have previously responded to Ag, with naive cells being CD44low and memory cells being CD44high (24, 25, 26, 27). Lymph node (Fig. 3,A) and spleen CD8+ T cells (data not shown) from OT-1 mice are about 10% CD44high. The CD44high cells also express Ly-6C (Fig. 3,D), another marker of memory CD8+ T cells (28). The CD44high cells appear to be resting memory cells; they have the forward scatter profile of small, resting lymphocytes (data not shown) and express low levels of CD25 (Fig. 3,B) and CD69 (Fig. 3 C). Essentially all of the CD44high cells express the Vα2+ transgenic TCR (data not shown).
To compare the abilities of naive and memory cells to respond to Kb/OVA257–264 complexes, CD8+-enriched cells were sorted by FACS based on CD8 and CD44 expression. The resulting CD44low population was highly purified, having <1% CD44high cells, and the CD44high population was enriched about threefold with 32% of the cells being CD44high (Fig. 3, E and F). Both populations were >99% CD8+. Proliferation (day 3) of the two populations was identical in response to irradiated C57BL/6 spleen cells pulsed with OVA257–264 (Fig. 4, legend). The two populations were also stimulated with IL-2 and Kb-microspheres pulsed with OVA257–264, varying either the concentration of peptide used for the pulsing (Fig. 4,A) or the density of Kb on the beads (Fig. 4,B). Under all conditions, CD44high cells gave a strong response while CD44low cells responded very weakly. The difference is even more dramatic when it is taken into account that only about 32% of the cells in the CD44high population are, in fact, CD44high. Assuming that the CD44low cells present in this population respond in the same way as the highly purified CD44low cells, then it can be estimated that the CD44high cells had a >20-fold greater response in this experiment. Responses of both the CD44high and CD44low populations to Kb/OVA257–264 microspheres were completely dependent upon the addition of exogenous IL-2 (Fig. 4, legend).
Treating CD8+-enriched cells with anti-CD8 and anti-CD44 Abs in the same way as for cell sorting but without separating populations did not decrease the response to IL-2 and Kb/OVA microspheres in comparison with the response made by unsorted cells in the absence of the Abs (data not shown). Thus, the presence of the Abs on the sorted populations does not appear to alter the responses. Consistent with this, essentially the same results have been obtained in additional experiments examining populations sorted only on the basis of CD44 levels and in experiments using cells obtained by panning using immobilized anti-CD44 mAb (data not shown). Responses of sorted CD44high and CD44low cell populations were compared in a total of 20 experiments; 10 in which sorting was by CD44 only and 10 in which sorting was by CD44 and CD8. Similar results were obtained in all cases, and Figure 4 is representative of these results.
To determine whether differing response levels of naive and memory cells might result from different kinetics of response, proliferation was measured over days 2 through 5. Responses of CD44low and CD44high cells to stimulation with irradiated C57BL/6 spleen cells pulsed with OVA257–264 were virtually identical with respect to time course and magnitude (Fig. 5,A). In contrast, CD44high cells again responded well to IL-2 and Kb microspheres pulsed with OVA257–264, while CD44low cells responded only marginally at all times (Fig. 5 B).
The differential ability of naive and memory cells to respond to a minimal stimulus is quantitative, and CD44low cells do make a weak proliferative response to a high surface density of Ag (Fig. 4). To determine whether lytic effector function develops in the two cell populations, CD44high and CD44low cells were obtained by FACS and stimulated for 3 days with IL-2 and Kb microspheres pulsed with OVA257–264. CD44high cells developed potent Ag-specific cytotoxicity by day 3, and this was dependent upon stimulation with Ag, since no cytotoxicity was obtained in cultures stimulated with IL-2 alone (Fig. 6). In contrast, CD44low cells did not develop significant cytotoxic activity when stimulated with Ag and IL-2.
Both CD44high and CD44low cells are efficiently stimulated by immobilized anti-TCR mAb
Class I/peptide Ag is a ligand for both the TCR and CD8 coreceptor, and both are likely to contribute to signaling for activation. The large quantitative difference in the ability of naive vs memory cells to respond to class I Ag complexes (Figs. 4 and 5) could result from CD44high cells expressing higher levels of these receptors on their surfaces; this was examined by flow cytometry. CD44high cells express slightly higher levels of CD8 than do CD44low cells, while CD44low cells actually have slightly higher levels of the transgenic Vα2 TCR. (Fig. 7). Somewhat lower expression of the transgenic TCR on the CD44high population is consistent with the suggestion that these cells also express a TCR having an endogenous α-chain. Thus, the slightly higher level of CD8 on the memory cells could potentially contribute to a greater response to class I Ag. However, differences in expression levels of the Kb/OVA257–264-specific TCR would, if anything, predict a lower sensitivity of the CD44high population to antigenic stimulation, rather than the substantially higher sensitivity that is observed.
In an attempt to determine whether the difference between CD44low and CD44high cells was at the level of the TCR or the CD8 coreceptor, the ability of the two populations to respond to IL-2 and immobilized anti-TCR mAb was determined. Surprisingly, both responded well to this stimulus. The response of CD44low cells remained lower than that of CD44high cells (Fig. 8), but the difference was not as dramatic as that seen when Ag was used as the stimulus (Figs. 4 and 5). Although the differences are not great, the response of CD44low cells to anti-TCR mAb was consistently found to be lower than that of CD44high cells in multiple experiments. Similar results were obtained when the surface density of the immobilized anti-TCR mAb was varied; the differences between CD44low and CD44high cells were not increased when the level of stimulus was decreased (data not shown). Potential differences in CD8 contributions were examined by including non-Ag class I protein, a ligand for CD8 but not TCR, on the surface of the microspheres along with the anti-TCR mAb. For both CD44low and CD44high cells, non-Ag class I caused about a 50% increase in the level of response over that obtained with anti-TCR mAb alone (Fig. 8). This increase occurred even when anti-TCR mAb was at a level where further increasing its density did not increase responses. It is likely that the class I protein contributes by supporting adhesion and/or signaling mediated by CD8, but the naive and memory cells do not appear to differ significantly in this respect.
The results described in this report demonstrate that CD8+ memory T cells are much more sensitive than naive cells to stimulation by isolated class I Ag to proliferate in response to IL-2. While this difference in allogeneic responses (Fig. 1) might be attributable to differences in precursor numbers or in vivo selection for clones with higher affinity TCRs, these factors are eliminated when cells from TCR transgenic mice are used. Despite having the same TCR, OT-1 cells with a memory phenotype respond much more readily to Ag than do cells with a naive phenotype (Figs. 4 and 5). These differences are seen using purified class I Ag on microspheres as the stimulus. Since these “artificial APCs” lack ligands for other adhesion molecules or receptors, the results strongly suggest that CD44low and CD44high cells differ at the level of signaling via TCR and/or the CD8 coreceptor.
The population of CD8+ cells in OT-1 mice that have a memory phenotype very likely results from cells that express both the transgenic TCR and a second TCR that derives from endogenously rearranged α-chains and that have also responded to environmental Ag via this second TCR. Lee et al. (23) have shown that DO11.10 mice, transgenic for the α- and β-chains of a TCR specific for OVA323–339 and I-Ad, have a population of CD4+ T cells with a memory phenotype, and that these cells have a second TCR as a result of expressing endogenous α-chains. No memory cells were observed in mice having the DO11.10 TCR transgenes but lacking the recombinase activating gene that allows endogenous α-chains to be expressed. Similarly, the OT-1 CD8+CD44high population studied in this report is not present in mice having the OT-1 TCR on a recombinase activating gene−/− background (data not shown). Also consistent with this memory population arising as a result of expressing dual TCRs is the finding that the Vα2 transgene product is expressed at somewhat lower levels on the CD44high population than on the CD44low population (Fig. 7).
The CD44high population in OT-1 mice appears to consist of resting memory cells rather than activated cells; they are small, do not express CD69 or CD25 activation markers, and do not proliferate in response to IL-2. Furthermore, no cytolytic activity can be detected when specific killing of EL4 target cells expressing OVA is measured in a 4-h 51Cr release assay using freshly isolated spleen or lymph node CD8+ T cells (data not shown), while potent lytic activity is present by day 3 after in vitro stimulation with Ag (Fig. 6). As others have experienced with TCR transgenic mice, we have been unable to substantially increase the pool of memory cells in OT-1 mice by in vivo priming with Ag. Although the CD44high OT-1 cells do not result from intentional priming, the results obtained using cells from in vivo primed alloAg-specific cells confirm that memory cells can respond to the minimal stimulus of class I Ag and IL-2 (Fig. 1).
The difference in the ability of CD44high and CD44low cells to respond to class I is quantitative, not qualitative; the CD44low cells can make a weak proliferative response when the surface density of Ag is very high (Fig. 4, A and B), although the cells that respond do not develop lytic effector function (Fig. 6). Based on the level of responses as a function of peptide dose and class I density (Fig. 4), it appears that CD44high cells are at least eightfold more sensitive to Ag than CD44low cells. This is probably an underestimate of the difference, since the sorted CD44high populations included a significant fraction of CD44low cells. Furthermore, the somewhat lower expression of the transgenic TCR on the CD44high population may decrease their sensitivity to Ag. T cells having dual TCR have been shown to have a decreased sensitivity to physiologic ligands due to reduced surface density of each TCR (29).
In a comparison of the peptide dose responses for naive and memory CD8+ cells of F5 transgenic mice specific for an influenza nucleoprotein peptide, memory cells were found to have a 10- to 50-fold higher sensitivity for stimulation of IL-2-dependent proliferation (11). Ag presentation to the T cells was via spleen cell APCs in that study, and differential expression of adhesion molecules on the memory cells could therefore potentially contribute to an increased avidity of interaction with the APCs. The somewhat smaller difference between naive and memory cell activation requirements that we observe when only TCR and CD8 interactions are involved suggests that greater sensitivity of signaling via TCR and the CD8 coreceptor and increased avidity due to adhesion receptors binding their ligands may both contribute to the greater sensitivity of memory cells for activation.
A recently reported study (30) has shown that naive CD8+CD44low cells from 2C mice with a transgenic TCR specific for alloAg, Ld, and p2Ca peptide (31) can be stimulated to proliferate in response to purified Ld pulsed with peptide. However, responses were very weak unless a variant peptide p2C-QY5 was used. Ld/p2C-Q is bound by the 2C TCR with very high affinity (32), much higher than the affinity of the OT-1 TCR for Kb/OVA257–264 (33), suggesting that naive cells can be stimulated to respond in the absence of other receptor interactions if sufficient TCR engagement can be achieved. This is consistent with our finding that naive OT-1 cells do make a weak response at very high class I and peptide levels (Fig. 4). 2C mice have a population of CD8+ cells that express high CD44 levels, and this population could respond effectively to Ld and the p2Ca peptide (30). Although no quantitative comparisons of CD44low and CD44high responses were reported, this observation suggests that there may be a differential sensitivity of naive and memory cells from 2C mice similar to what we have demonstrated for OT-1 CD8+ cells.
Differential sensitivity of naive and resting memory cells at the level of TCR and/or CD8 or CD4 coreceptor signaling has not previously been demonstrated under conditions in which contributions of adhesion receptors to avidity differences can be ruled out. However, differences in the activation requirements for naive vs activated effector cells have been observed for CD4+ T cells using class II/peptide complexes immobilized in microtiter wells (34). Naive cells exhibited more stringent requirements for costimulation than did effector cells but also required higher peptide concentrations for activation when costimulation was provided. Thus, it may be the case that increased sensitivity to stimulation through the TCR and CD8/CD4 coreceptors is acquired upon differentiation to effector cells and is retained following reversion to a resting memory state.
Differences between naive and effector cells have also been observed with respect to signal generation via the TCR. CD4+ T cells that were primed in vivo and then rested exhibited a higher sensitivity than did naive cells with regard to peptide-dependent activation of phospholipase C-γ1 and mitogen-activated protein kinase as well as intracellular Ca2+ increase (35). Class II-transfected fibroblasts were used as APCs in this study, and the observed differences might therefore reflect either differences in contributions by adhesion receptors or intrinsic differences in signaling via the TCR. That there may be intrinsic differences in TCR-dependent signal generation in CD8+ naive vs effector T cells is suggested by a recent study demonstrating differences in tyrosine phosphorylation of CD3 components and differences in the activation of CD8-bound p56lck in response to stimulation with anti-CD3 mAb (36).
CD44low and CD44high OT-1 cells exhibit a large difference in sensitivity to Ag (Figs. 4 and 5). When anti-TCR mAb is used as the stimulus, some difference in the levels of response is still seen, but this difference is less dramatic than that observed using Ag as the stimulus (Fig. 8). The high affinity interaction of the Ab with the TCR may not accurately mimic the interaction with Ag and may obscure differences in signal generation in naive vs memory cells. Coimmobilizing anti-TCR mAb along with non-Ag class I as a ligand for CD8 enhances the levels of response of both naive and memory cells, and the differential in response between the two cell types remains comparable with that seen in response to anti-TCR mAb alone (Fig. 8). These results raise the possibility that the differential sensitivity of naive and memory cells may be predominantly at the level of TCR signaling, and work is in progress to further examine this.
The greater sensitivity of CD8+ memory T cells to stimulation by Ag through the TCR and/or CD8, as demonstrated here, together with increased expression of adhesion receptors (7), will act to insure that effector CTL can be generated rapidly in response to even very low levels of Ag, levels that may be too low to initiate a primary response by naive cells. In the experiments described here, costimulatory ligands are not present on the Ag-bearing surface, and responses therefore depend upon the addition of IL-2. This suggests that an effective in vivo memory CTL response could be generated even though the specific class I/peptide Ag complex is not being presented on professional APCs, as long as a CD4+ T helper response is occurring to provide a source of IL-2. Memory CD4+ T cells have less stringent requirements for costimulation than do naive cells (12, 13, 14), and this may also be the case for memory CD8+ T cells. The ability to construct artificial APCs by incorporating purified B7 ligands onto microspheres (37) will make it possible to examine this in more detail and to determine how costimulation requirements relate to the greater Ag sensitivity described here.
We thank Paul Champoux for excellent technical assistance.
This work was supported by Grant AI34824 from the National Institutes of Health.
Abbreviations used in this paper: alloAg, alloantigen; PE, phycoerythrin; MESF, molecules of equivalent soluble fluorochrome; ConASN, supernatant from Con A-stimulated rat spleen cells; LNC, lymph node cells.