INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The CD40 ligand (CD40L) CD154 is a glycoprotein that is transiently expressed at high levels on the surface of CD4⁺ T cells when they are activated (2, 30, 39, 51, 53). This protein is also expressed (although at lower levels) on a subset of CD8⁺ T cells following activation (2, 28, 39, 53), and its expression has been documented on several other cell types, including mast cells, eosinophils, basophils, and B cells (reviewed in reference 66). CD40L is a member of the tumor necrosis factor family (2) and binds to CD40, a member of the tumor necrosis factor receptor family (60). The latter is expressed on a variety of cell types with antigen-presenting cell function, including B cells, dendritic cells, activated macrophages, follicular dendritic cells, and endothelial cells (reviewed in reference 66). The fact that CD40L and CD40 are expressed in a tightly controlled fashion on T cells and on many different cell populations with which they interact suggests that CD40L-CD40 interactions are probably involved in the regulation of a number of aspects of the immune response. This is becoming increasingly apparent as research into the functions of this receptor-ligand pair progresses (17, 22, 23, 38, 50).

CD40L-CD40 interactions were originally shown to play a key role in thymus-dependent humoral immune responses, mediating cognate interactions between CD4⁺ T cells and B cells that are essential for B-cell activation and differentiation, class switching, germinal center formation, and the generation of B-cell memory (reviewed in references 21 and 31). More recently, roles for CD40L-CD40 interactions in the development of other immune effector functions have been described. For example, they have been shown to be of importance in the inflammatory immune response, regulating the induction of secretion of cytokines, such as tumor necrosis factor alpha, interleukin-1, interleukin-12, and gamma interferon (IFN-), and of nitric oxide by monocytes and macrophages and prolonging the survival of these cells at sites of inflammation (reviewed in references 23 and 61). In addition, CD40L-CD40 interactions have been shown to be involved in the initiation of antigen-specific CD4⁺ T-cell responses (24, 25, 65, 71). A current model for the role of this system argues that CD40L is upregulated upon activation of CD4⁺ T cells following recognition of antigen presented by dendritic cells. CD40L then interacts with CD40 on the dendritic cell surface, leading to the induction of costimulatory activity mediated by both cell surface molecules and cytokines such as interleukin-12 by the dendritic cell (11, 35). This costimulatory activity is necessary for the CD4⁺ T cell to become fully activated and produce cytokines and/or perform other effector functions (reviewed in references 22 and 23). Further, CD40 is also expressed on thymic antigen-presenting cells (19), and it has been demonstrated that CD40-CD40L interactions play an essential role in negative selection in the thymus (18). Here too, they likely act by regulating costimulatory activity on antigen-presenting cells.

Despite the advances made recently in understanding the importance of CD40L-CD40 interactions in the activation and effector functions of mature CD4⁺ T cells, the functions of this receptor-ligand pair are by no means fully understood. CD40L is likely involved in other interactions between CD4⁺ T cells and the increasing number of different cell types on which CD40 expression is being documented; in addition, the role(s) that CD40L plays in the activation and/or effector functions of the subpopulation of CD8⁺ T cells that express it is not well defined. Human CD8⁺ T-cell clones expressing high levels of CD40L were shown to be able to stimulate B-cell growth and differentiation in vitro (29), but the level of CD40L expressed by CD8⁺ T cells is generally low, and the contribution made by CD8⁺ T cells to cognate B-cell help during in vivo immune responses is not likely to be very significant (54). The in vivo functions of CD40L expression on CD8⁺ T cells thus remain unclear.

To gain further insight into the potential roles of CD40L in the in vivo immune responses of both CD4⁺ and CD8⁺ T cells, we have been analyzing the immune response mounted by CD40L-deficient mice (70) after infection with different viruses. In a previous study (6), we found that although the effector CD8⁺ cytotoxic T lymphocyte (CTL) response mounted by CD40L-deficient mice on infection with arenaviruses such as lymphocytic choriomeningitis virus (LCMV) was unimpaired and virus clearance occurred with normal kinetics, the level of memory CTL activity detectable in the CD40L-deficient animals 2 months following infection was lower than that in wild-type mice. Here, we have investigated the basis of the defect in the memory CTL response of CD40L-deficient mice. It is shown that CD40L-deficient mice do not have an impaired ability to maintain memory CD8⁺ CTL precursors (CTLp) at stable frequencies over time but that the initial generation of CD8⁺ CTLp is impaired in these animals. Further, it is demonstrated that CD40L expression on CD8⁺ cells is not required for their proliferation and differentiation into memory CTLp. The reduction in memory CTLp generation in CD40L-deficient mice is instead likely a consequence of defects in the CD4⁺ T-cell response. These results suggest that CD40L plays a previously undefined role in the generation of memory CD8⁺ CTLp following virus infections, the nature of which is discussed.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Mice. Homozygous CD40L-deficient mice (70) and wild-type controls were bred in parallel and used for experiments as young adults (8 to 14 weeks old). A randomly selected subset of CD40L-deficient mice was screened by PCR analysis of tail biopsy DNA to confirm the presence of homozygous mutations in their CD40L genes. C57BL/6 mice used as a source of peritoneal macrophages and feeder splenocytes for the in vitro restimulation of CTLp in limiting dilution assays were obtained either from the closed breeding colony of The Scripps Research Institute (La Jolla, Calif.) or from Jackson Laboratories (Bar Harbor, Maine), as were the C57BL/6 Ly5^a mice used for the creation of mixed bone marrow chimeras.

Generation of mixed bone marrow chimeras. Bone marrow chimeras were prepared by a method based on that described by Gao et al. (20). Briefly, bone marrow was removed from CD40L-deficient (Ly5^b) mice and wild-type C57BL/6 Ly5^a mice and was depleted of T cells by treatment with a mixture of antibodies against Thy-1.2 (clone J1j), CD4 (clone RL172), and CD8 (clone 3.168) plus complement. Viable cells were enumerated and mixed 1:1. Recipient mice (wild-type C57BL/6 Ly5^a mice which had been exposed to 1,000 rads of irradiation 4 to 6 h previously) were inoculated intravenously with 2 × 10⁶ to 3 × 10⁶ mixed cells/mouse. The mice were then kept for at least 8 weeks to allow time for reconstitution to occur before they were used in experiments.

Virus growth, titration, and use for infection of mice. All experiments with LCMV involved the Armstrong 53b strain of the virus, a clone triple plaque purified from ARM CA 1371 (16). Stocks of this virus were prepared by growth on baby hamster kidney cells, and their titers were determined by plaque assay on Vero cells as previously described (16). Mice were infected with LCMV by intraperitoneal (i.p.) inoculation of 2 × 10⁵ PFU of virus in a 200-µl volume.

Assay for LCMV-specific effector CTL activity during the primary immune response. LCMV-specific effector CTL activity was quantitated by an in vitro ⁵¹Cr release assay as described by Byrne and Oldstone (9). The effector cells were erythrocyte-depleted splenocyte suspensions from mice infected 7 or 14 days previously with LCMV. Target cells were ⁵¹Cr-labelled fibroblast cell lines MC57 (H-2^b, i.e., syngeneic to the CD40L-deficient mice) and Balb Cl 7 (H-2^d i.e., allogeneic to the CD40L-deficient mice) which were either uninfected or infected 48 h prior to ⁵¹Cr labelling with LCMV at a multiplicity of infection of 3 PFU per cell. Target cells were plated at 10⁴ per well, and effector cells were added to give effector/target (E/T) ratios between 100:1 and 0.5:1. All variables were tested in triplicate. The assay time was 5 h. Results are expressed as the percentage of specific ⁵¹Cr release, calculated as 100 × (experimental release  spontaneous release)/(maximum release  spontaneous release).

Limiting dilution assay for quantitation of LCMV-specific CTLp frequencies. The assay method used was based on that described by Nahill and Welsh (48). Briefly, thioglycolate-elicited peritoneal macrophages from C57BL/6 mice were infected with LCMV at a multiplicity of infection of 0.1 PFU/cell and plated at 7 × 10⁴/well into 96-well flat-bottomed plates (Costar Corporation, Cambridge, Mass.) in a volume of 100 µl of RPMI 1640 medium supplemented with 7% heat-inactivated fetal bovine serum, 1 mM glutamine, 50 U of penicillin/ml, and 50 µg of streptomycin/ml (termed RPMI 7%). After 2 days of culturing, the plates were irradiated (2,000 rads), and half of the medium was removed from each well and replaced with 50 µl of restimulation medium containing 10⁵ irradiated (2,000 rads) syngeneic erythrocyte-depleted splenocytes. The restimulation medium consisted of RPMI 1640 medium supplemented with 5% concanavalin A (ConA)-stimulated rat spleen supernatant, 10% heat-inactivated fetal bovine serum, 2 mM glutamine, 50 µg of gentamicin/ml, and 50 µg of streptomycin/ml. Test cell populations (either erythrocyte-depleted splenocytes or sorted CD8⁺ T cells from mice previously infected for different lengths of time with LCMV) were diluted in restimulation medium so that they could be added to these plates at various numbers of cells per well in 100-µl volumes. Twenty-four replicate wells were set up at each input test cell number. The plates were incubated at 37°C in a humidified 5% CO₂ atmosphere for 7 days. On day 4 of culture, 100 µl of medium was removed from each well and replaced with 100 µl of fresh restimulation medium. On day 7 of culture, cells from individual wells were split twofold and assayed for cytotoxic activity on LCMV-infected and uninfected syngeneic target cells (MC57) by a ⁵¹Cr release assay. ⁵¹Cr-labelled target cells were added to all wells at 10⁴/well to give a final total volume of 200 µl/well, and ⁵¹Cr release into the supernatant was quantitated after a 5-h incubation period. Positive wells were defined as those whose specific ⁵¹Cr release (calculated as described above) was greater than 10%. CTLp frequencies were estimated from the fraction of nonresponding wells at each input cell number per well by single-hit model Poisson distribution analysis (14).

Induction of anti-H-Y immune responses and assay for H-Y-specific memory CTL activity. To induce immune responses to the male antigen H-Y, female mice were inoculated i.p. with 2 × 10⁶ erythrocyte-depleted splenocytes from syngeneic wild-type male mice in a 200-µl volume of phosphate-buffered saline. H-Y-specific memory CTL activity was assayed 4 to 6 weeks later by a method based on that described by Di Rosa and Matzinger (15), except that cytotoxic activity was measured in ⁵¹Cr release assays. Briefly, erythrocyte-depleted splenocyte suspensions from individual test mice were restimulated in vitro by culture in 24-well plates at 5 × 10⁶ cells/well together with 2 × 10⁶ irradiated (2,000 rads) erythrocyte-depleted splenocytes from syngeneic wild-type male mice/well in a final volume of 2 ml/well of restimulation medium. After 3 days of culture, 1 ml of medium was removed from each well and replaced with 1 ml of fresh restimulation medium. Cells were harvested after 6 days of culture, and the cytotoxic activity mediated by bulk effector populations from individual animals was measured in a ⁵¹Cr release assay with syngeneic male and female ConA-activated blasts as target cells. The latter were prepared by culturing erythrocyte-depleted splenocytes from male and female mice for 3 days in 24-well plates at 2 × 10⁶/well in a 2-ml volume/well of RPMI 7% medium containing ConA at a final concentration of 2 µg/ml. After ⁵¹Cr labelling, the target cells were plated into 96-well plates at 10⁴/well, and effector cells were added to give E/T ratios between 60:1 and 7.5:1. All variables were assayed in triplicate. ⁵¹Cr release into the supernatant was measured after 5 h. Results are expressed as the percentage of specific ⁵¹Cr release, calculated as described above.

BrdU incorporation studies for the analysis of lymphocyte turnover. The turnover of different T-cell populations stimulated by inoculation of mice with polyinosinic-polycytidylic acid [poly(IC)] or infection with LCMV was assessed by administering a DNA precursor, bromodeoxyuridine (BrdU) in the drinking water and then examining the surface markers on BrdU-labelled cells, as previously described (64). Mice were inoculated i.p. with 150 µg of poly(IC) (Sigma Chemical Co., St. Louis, Mo.) in 150 µl of phosphate-buffered saline or infected with LCMV and were given sterile drinking water containing BrdU (Sigma Chemical Co.) at 0.8 mg/ml, which was made fresh and changed daily. At different time points thereafter, mice were sacrificed and spleen and lymph node cell suspensions were prepared, stained in multiple colors with antibodies directed against different lymphocyte markers and BrdU, and examined by fluorescence-activated cell sorter (FACS) analysis as described below.

FACS analysis and sorting of lymphocyte subpopulations. Staining of cell surface antigens for FACS analysis was performed by conventional techniques (13). Primary antibodies used in different experiments were fluorescein isothiocyanate-conjugated anti-CD8 (clone 53-6.7; GIBCO BRL, Gaithersburg, Md.), Cy-5-conjugated anti-CD8 (clone 53-6.7), Cy-5-conjugated anti-CD4 (clone GK1.5), biotinylated anti-CD44 (clone IM7.8.1), phycoerythrin (PE)-conjugated anti-CD40L (clone MR1; Pharmingen, San Diego, Calif.), biotinylated anti-Ly5^a (clone A20), and biotinylated anti-Ly5^b (clone 104). Biotinylated primary antibodies were detected with PE-streptavidin or RED670-streptavidin (both from GIBCO BRL). In experiments that involved costaining for BrdU, surface staining was performed first and then the cells were washed, fixed, permeabilized, DNase treated, washed again, and incubated with fluorescein isothiocyanate-conjugated anti-BrdU antibody as previously described (64). Two- and three-color staining were analyzed with a FACScan flow cytometer, four-color staining was analyzed with a FACSCaliber flow cytometer, and cell sorting was performed with a FACSVantage flow cytometer (all from Becton Dickinson & Co., Mountain View, Calif.).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Impairment of the virus-specific memory CD8⁺ CTL response in CD40L-deficient mice. To clarify the roles that CD40L-dependent signalling events play in the induction and regulation of antiviral immune responses in vivo, we have been examining the nature of the immune response mounted by CD40L-deficient mice following infection with different viruses. In a previous study (6), we demonstrated that CD40L-deficient mice infected with arenaviruses such as LCMV mount effector CD8⁺ CTL responses that are indistinguishable from those of wild-type control animals and clear the infection with normal kinetics. However, preliminary analysis of the memory CTL activity exhibited by bulk splenocytes 2 months following LCMV infection indicated that this function was impaired in CD40L-deficient mice.

View larger version (15K): [in this window] [in a new window]   FIG. 1.   LCMV-specific CTLp frequencies in the spleens of CD40L-deficient and wild-type control mice 16 weeks following virus infection. LCMV-specific CTLp frequencies in splenocyte suspensions from individual CD40L-deficient (open symbols) and wild-type mice (closed symbols) infected 16 weeks previously with 2 × 105 PFU of LCMV i.p. were quantitated by limiting dilution assay as described in Materials and Methods. The precursor frequencies estimated for each control (control no. 1 to 3) and CD40L-deficient (knockout [K/o] no. 1 to 3) mouse, the mean frequency of each group of animals (control mean and K/o mean), and the fold difference in mean CTLp frequencies between the two groups are shown.

Analysis of the maintenance of CD8⁺ CTL memory in CD40L-deficient mice. In normal mice, virus-specific CTLp levels are maintained at a fairly constant level after an acute virus infection for the life of the animal (32, 40). To determine whether CD40L-deficient mice had an impaired ability to maintain memory CTL, groups of CD40L-deficient and control mice were infected with LCMV and the frequency of virus-specific CTLp in sets of three mice from each group was measured by limiting dilution assay over time after the acute phase of the infection. Figure 2 illustrates the mean number of LCMV-specific CTLp per 10⁶ splenocytes in the animals tested from each group at time points from 1 to 6 months postinfection. As expected, the LCMV-specific CTLp frequency in the control mice remained very stable over the 6-month observation period. Importantly, the virus-specific CTLp frequency in the CD40L-deficient mice did not decline relative to that in the control animals with time postinfection but was consistently found to be approximately fourfold lower than the mean frequency of the wild-type mice at all time points tested. This result shows that CD40L-dependent signalling is not required for the maintenance of memory CTLp at stable frequencies over time.

View larger version (10K): [in this window] [in a new window]   FIG. 2.   Maintenance of LCMV-specific CTLp at stable frequencies over time in CD40L-deficient and wild-type control animals. Groups of CD40L-deficient and wild-type control mice were infected i.p. with 2 × 105 PFU of LCMV on day 0. At the indicated times postinfection, three mice from each group were sacrificed, and the frequency of LCMV-specific CTLp in the spleen of each animal was quantitated by limiting dilution assay as described in Materials and Methods. The results shown are the mean number of CTLp per 106 splenocytes of the control (closed squares) and CD40L-deficient (open circles) animals tested at each time point.

Analysis of the generation of LCMV-specific CTLp in CD40L-deficient mice. To compare the generation of virus-specific CTLp during the acute phase of LCMV infection in CD40L-deficient and wild-type mice, groups of mice of each type were infected with LCMV and limiting dilution analysis of the LCMV-specific CTLp frequencies in two or three individual animals from each group was performed 7 and 14 days later. The direct effector CTL activity mediated by each test population was also measured. Figure 3 shows the results obtained at 7 days postinfection with two CD40L-deficient and two control mice; a second experiment in which three mice per group were tested confirmed these. Although the direct effector CTL activity exhibited by CD40L-deficient and control mice was indistinguishable (Fig. 3a), the LCMV-specific CTLp frequencies in the CD40L-deficient mice were clearly lower than those in the control animals at this time point (Fig. 3b). Very similar results (data not shown) were also obtained at day 14, although by this time postinfection the levels of direct CTL lysis were not as high and the CTLp frequencies were slightly lower. The processes by which effector and memory CTL are generated during an acute virus infection can thus be differentiated by their dependence on CD40L-mediated signalling. CD8⁺ effector CTL can be generated efficiently in the absence of CD40L; by contrast, although there is not an absolute requirement for CD40L in the production of CD8⁺ memory CTLp, CD40L-dependent signalling is clearly involved in a pathway(s) which influences the overall efficiency with which memory CTLp are generated.

View larger version (11K): [in this window] [in a new window]   FIG. 3.   LCMV-specific effector CTL activity and CTLp frequencies in splenocyte populations from CD40L-deficient and wild-type mice infected 7 days previously with LCMV. CD40L-deficient and wild-type mice were infected i.p. with 2 × 105 PFU of LCMV. Seven days later, two mice from each group were sacrificed, and splenocyte suspensions from individual CD40L-deficient animals (open symbols and dashed lines) and wild-type controls (closed symbols and solid lines) were assayed for both ex vivo effector CTL activity (a) and CTLp frequency (b). (a) Percentage of specific 51Cr release (calculated as described in Materials and Methods) mediated by each splenocyte population from syngeneic LCMV-infected target cells over a range of different E/T ratios. Even at the highest E/T ratios, the specific 51Cr release from control target cells (uninfected syngeneic targets and LCMV-infected allogeneic targets) tested in the same assay never exceeded 20% (data not shown), indicating that the ex vivo cytotoxic activity observed was both virus specific and major histocompatibility complex restricted. (b) Results of a limiting dilution assay used to quantitate the LCMV-specific CTLp frequency in each splenocyte population. The precursor frequencies estimated for each control mouse (control no. 1 and 2) and CD40L-deficient mouse (knockout [K/o] no. 1 and 2), the mean frequency of each group of animals (control mean and K/o mean), and the fold difference in mean CTLp frequencies between the two groups (fold difference) are indicated. The results shown are representative of those obtained in two separate experiments in which a total of five CD40L-deficient and five wild-type mice were tested.

Investigation of the role of expression of CD40L on CD8⁺ versus CD4⁺ T cells in the generation of virus-specific CD8⁺ CTLp. A number of reports have shown that CD40L is transiently expressed on the surface not only of activated CD4⁺ T lymphocytes but also of a subpopulation of CD8⁺ T lymphocytes following activation (2, 28, 39). It has been demonstrated that human CD8⁺ T-cell clones which express high levels of CD40L upon activation have the capacity to stimulate B-cell activation, growth, and differentiation (29), but whether CD40L- mediated signalling is involved in other aspects of CD8⁺ T-cell biology has not been fully investigated. The process by which memory CD8⁺ CTLp are generated during an immune response is poorly understood; transient CD8⁺ expression of CD40L could hypothetically be involved. As a first step toward investigating whether CD40L expression on CD8⁺ T cells may be involved in the generation of CD8⁺ CTLp, we assessed the expression of CD40L on CD8⁺ (and also CD4⁺) T cells from the spleens and lymph nodes of mice at different times after infection with LCMV by FACS analysis following two-color staining with antibodies against CD40L and CD4/8. The extent and kinetics of CD40L expression on T cells in the spleen and lymph node differed. CD40L expression was induced on a much higher proportion of T cells in the spleen, where the percentage of CD40L-positive CD4⁺ cells peaked at >10% on day 7 postinfection. On CD8⁺ T cells, CD40L expression was undetectable in the lymph node, but a low percentage of positive cells were observed in the spleen between days 3 and 14 postinfection. It was thus clear that CD40L expression was much more prominent on CD4⁺ cells than on CD8⁺ cells after LCMV infection. However, the sensitivity of the staining method used in this experiment would not have allowed the detection of low levels of CD40L expression; further, as CD40L is expressed only transiently by lymphocytes following activation, this experiment did not reveal what proportion of cells in fact expressed CD40L at some point during the course of the infection. The possibility that low-level CD40L expression on CD8⁺ cells played a role in the generation of CD8⁺ CTLp thus could not be ruled out.

View larger version (18K): [in this window] [in a new window]   FIG. 4.   Changes in the total number of CD8+ T cells in the spleens and lymph nodes of wild-type and CD40L-deficient mice following infection with LCMV. Wild-type and CD40L-deficient mice were infected i.p. with 2 × 105 PFU of LCMV. At the indicated times (days) postinfection, two mice from each group were sacrificed, and the total number of CD8+ T cells in the spleen and lymph nodes of each animal was determined. The results shown are the mean total CD8+ T-cell counts of the wild-type (closed squares and solid lines) and CD40L-deficient (open circles and dashed lines) mice. Vertical bars indicate 1 standard deviation above and below the mean value for each group of animals.

View larger version (17K): [in this window] [in a new window]   FIG. 5.   Division of total, CD44hi and CD44lo CD8+ and CD4+ splenic T cells over time following infection of wild-type and CD40L-deficient mice with LCMV. Groups of wild-type and CD40L-deficient mice were infected i.p. with 2 × 105 PFU of LCMV and immediately given drinking water containing BrdU to label dividing lymphocytes. At the indicated times (days) postinfection, two mice from each group were sacrificed, and the expression of CD4/8, CD44, and BrdU by spleen cells was determined by FACS analysis. Shown (for total, CD44hi, and CD44lo T cells of the CD8+ and CD4+ subsets, as indicated) are the mean percentages of BrdU-positive cells (percentage BrdU+) in the wild-type (closed squares and solid lines) and CD40L-deficient (open circles and dashed lines) mice tested at each time point. The vertical bars indicate 1 standard deviation above and below the mean value for each group of animals. Standard deviations smaller than the size of the symbols are not apparent.

View larger version (14K): [in this window] [in a new window]   FIG. 6.   Experiment comparing the generation of LCMV-specific memory CTLp from CD8+ T cells derived from CD40L-deficient and wild-type mice in the same CD4+ T-cell environment. Mixed bone marrow chimeras were created by reconstituting irradiated wild-type Ly5a H-2b mice with a mixture of bone marrow derived from CD40L-deficient Ly5b H-2b and wild-type Ly5a H-2b mice. These chimeras were infected i.p. with 2 × 105 PFU of LCMV, and at 7 days and 2 months postinfection, the frequency of LCMV-specific CTLp in sorted populations of Ly5a+ (circles) and Ly5b+ (squares) CD8+ T cells from individual chimeric mice was determined by limiting dilution assay as described in Materials and Methods. The precursor frequencies estimated for the Ly5a+ and Ly5b+ CD8+ T cells from each mouse tested are indicated.

Investigation of the memory CD8⁺ CTL response to H-Y in CD40L-deficient mice. Whereas the primary LCMV-specific CD8⁺ CTL response is known to be CD4⁺ T-cell independent (41, 45, 52), in other systems CD8⁺ CTL are much more dependent on help from CD4⁺ T cells. A classic example of this is the CD8⁺ CTL response to the male antigen H-Y (26, 58, 67). We reasoned that if, as suggested by the above-mentioned experiments in the LCMV system, the defect in generation of CD8⁺ CTLp in CD40L-deficient mice was due to the fact that CD4⁺ T cells influence the generation of CD8⁺ CTLp via pathway(s) that involve CD40L-mediated signalling, then the defect in generation of memory CD8⁺ CTLp in CD40L-deficient mice may be even more apparent in the H-Y system. We thus compared the H-Y-specific memory CTL response in CD40L-deficient and wild-type female mice immunized 4 to 6 weeks previously with splenocytes from syngeneic wild-type male mice. As the in vitro expansion of H-Y-specific CD8⁺ CTLp is known to be dependent on cytokines produced by CD4⁺ T cells (15), test splenocyte populations were restimulated in vitro in the presence of cytokine-containing supernatants so that differences observed in their lytic activities would reflect true differences in the input CD8⁺ CTLp content rather than defective expansion of these cells in vitro. Figure 7 shows that although H-Y-specific memory CTL responses were readily detected in bulk CTL assays with effector cells from wild-type mice, they could not be detected in CD40L-deficient animals. Similarly, when limiting dilution assays were carried out, the frequency of H-Y-specific CTLp in CD40L-deficient mice was found to be below the limit of detection of the assay (<1 per 500,000 splenocytes). These results lend further support to the hypothesis that CD4⁺ T cells play a role in the generation of CD8⁺ memory CTLp.

View larger version (11K): [in this window] [in a new window]   FIG. 7.   H-Y-specific memory CTL activity in CD40L-deficient and wild-type mice. Female CD40L-deficient and wild-type mice were immunized i.p. with splenocytes from syngeneic wild-type male mice. Four weeks later, two to three mice from each group were sacrificed, and the H-Y-specific memory CTL activity mediated by in vitro-restimulated splenocyte populations from individual animals was determined as described in Materials and Methods. The results shown are the mean percentage of specific 51Cr release (calculated as described in Materials and Methods) mediated by the wild-type (closed symbols and solid lines) and CD40L-deficient (open symbols and dashed lines) mice from syngeneic male target cells (squares) and, as a control, syngeneic female target cells (circles) over a range of different E/T ratios as indicated. The results are representative of data obtained in two independent experiments.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

CD8⁺ CTL play a key role in control of many virus infections, and the need for vaccines to elicit strong CD8⁺ T-cell responses in order to provide optimal protection in such infections is increasingly apparent. However, the mechanisms involved in the induction and maintenance of CD8⁺ CTL memory are currently poorly understood. In this study, we have investigated the involvement of CD40L-mediated interactions in these processes by analyzing the basis of a defect we had previously observed (6) in the memory CTL response of CD40L-deficient mice. We show firstly that the maintenance of memory CD8⁺ CTLp at stable frequencies over time is not dependent on CD40L-mediated interactions, as this is not impaired in CD40L-deficient mice. By contrast, we demonstrate that CD40L-mediated interactions do impact upon the generation of memory CD8⁺ CTLp, as these cells are produced at lower frequencies during the primary immune response in CD40L-deficient mice than in wild-type animals. Further, we show that expression of CD40L by CTLp themselves is not an essential step during their expansion and differentiation from naïve CD8⁺ cells into memory CTLp. The reduction in memory CTLp generation in CD40L-deficient mice is more likely a consequence of defects in the CD4⁺ T-cell response mounted by these animals.

How CD8⁺ CTL memory is maintained, and in particular the question of whether persistent contact with antigen is required, is a subject of intensive debate (1, 37, 59). Persistent contact with specific antigen appears to be required to maintain memory T cells in a state of chronic activation which, it has been argued, may be necessary if they are to provide protective immunity against certain forms of viral challenge (3, 36, 37). However, several studies have convincingly demonstrated that memory CD8⁺ T cells are able to survive at constant levels over time in the complete absence of the antigen to which they were elicited (32, 40, 47). Two theories as to how their survival may be maintained propose the involvement of stimulation with cross-reactive environmental antigens (5, 56) or stimulation by cytokines (such as type 1 IFNs) elicited during immune responses to unrelated agents (63). Our data demonstrate that whatever the mechanism involved, CD40L-mediated interactions do not play an essential role. This observation could be accommodated by either of the two theories above. In the latter case, it would imply that production of the cytokines which stimulate bystander turnover of the memory CD8⁺ T cells is not significantly impaired in the absence of CD40L-mediated interactions; this would not be surprising in the case of innate cytokines such as type 1 IFNs. Alternatively, if the former mechanism is involved, CD40L-mediated signalling must not be required for the memory CTLp to be efficiently stimulated by the cross-reactive antigens. This is in contrast to the stimulation of the generation of memory CTLp during the initial immune response to specific antigen, which we have shown does require CD40L-mediated signalling.

The process by which memory CD8⁺ CTLp are initially generated is just as unclear as the mechanism by which CTL memory is subsequently maintained. Demonstrating a role for CD40L-mediated interactions in this process, we went on to investigate whether expression of CD40L on CD8⁺ CTLp themselves was involved. CD40L is transiently expressed not only on CD4⁺ T cells but also on a subpopulation of CD8⁺ T cells following activation (2, 28, 39); it was possible that signalling mediated via this molecule may play a role in the differentiation of this subset of CD8⁺ cells into memory CTLp. However, this did not prove to be the case; instead, the reduction in memory CTLp generation in CD40L-deficient mice appeared to be a consequence of defects in the CD4⁺ T-cell response mounted in the absence of CD40L-mediated interactions.

The requirement for (and nature of) CD4⁺ T-cell help for CD8⁺ CTL responses is also not well understood. Whether CD4⁺ T cells are required for primary CTL responses to be elicited is dependent on the system studied. Thus, virus-specific CD8⁺ CTL responses are efficiently induced in the absence of CD4⁺ T cells following infection of mice with many (8, 41, 42, 45, 49) but not all viruses (69), and the ability to prime mice with CD8 epitopic sequences in the absence of CD4 epitopes is dependent on the immunization protocol followed (55). Further, although effector CTL responses clearly can be elicited to many viruses without the help of CD4⁺ T cells, CD8⁺ T-cell responses do not seem to be sustained during persistent virus infections in the absence of CD4⁺ T cells (10, 62). In both viral and nonviral systems, CD8⁺ T cells are more readily deleted after a transient response to antigen if CD4⁺ help is not available (4, 34, 44). The role of CD4⁺ T-cell help in the generation and/or maintenance of long-term immunological memory in the CD8⁺ compartment is a controversial area. Two previous studies of the LCMV system found that memory CTL responses were impaired in major histocompatibility complex class II-deficient and CD4-deficient mice (12, 68). The first did not determine whether the generation and/or the maintenance of immune memory was affected, while the second suggested that memory CD8⁺ CTLp were generated at normal levels in the absence of CD4⁺ T cells but subsequently declined in frequency over time. However, the latter result is at odds with the conclusion drawn from a less quantitative study which suggested that LCMV-specific memory CTL activity persisted over time in CD4-deficient mice (4) and with findings in the H-Y system that once memory CD8⁺ CTL are formed, their long-term survival does not require help from CD4⁺ T cells (7, 15). Our data showing that the generation but not the maintenance of LCMV-specific CD8⁺ CTLp is impaired in CD40L-deficient mice in which the CD4⁺ T-cell response is defective support the view that the maintenance of CD8⁺ memory CTLp at stable frequencies over time does not require help from CD4⁺ T cells mediated via CD40L. Further, our results suggest a previously unappreciated role for CD4⁺ T cells in the initial generation of CD8⁺ memory CTLp.

There are several mechanisms by which CD4⁺ T cells could potentially influence the generation of memory CD8⁺ CTLp. Firstly, they may mediate antiviral effects (e.g., via production of cytokines such as IFN-) which affect the antigen load to which CD8⁺ T cells are exposed during the early stages of the infection. Exposure of CD8⁺ T cells to very high levels of viral antigen in the early stages of LCMV infection has been shown to result in complete exhaustion of the CD8⁺ CTL response (46); it could thus be envisaged that if in the absence of optimal CD4⁺ T-cell functioning there was a moderate increase in the viral load, the level of generation of CD8⁺ CTLp may be reduced somewhat. However, two lines of evidence argue against this being the mechanism underlying the CD4⁺ T-cell dictated reduction in CD8⁺ CTLp generation we observed in CD40L-deficient mice. Firstly, the titers of infectious virus and kinetics of virus clearance in CD40L-deficient mice did not differ appreciably from those in wild-type animals infected with the same dose of LCMV (6). Secondly, the phenomenon of CTL exhaustion is manifest at the level of effector CTL activity in addition to CTLp generation, but as shown in Fig. 3, the deficit in virus-specific CTLp generation in CD40L-deficient mice occurred independently of any deficit in primary effector CTL activity.

An alternative, more direct mechanism by which CD4⁺ T cells may influence the generation of virus-specific CD8⁺ memory CTLp is via production of cytokines which stimulate T-cell proliferation and/or affect T-cell survival. The importance of cytokines in determining the fate of CD8⁺ T cells after antigenic stimulation has been illustrated in the H-Y system, in which IL-2 was shown to considerably delay the deletion of specific CD8⁺ T cells activated by antigenic stimulation in the absence of CD4⁺ T-cell help (34). CD4⁺ T cells are the major producers of IL-2 during LCMV infection of mice (33). In CD40L-deficient mice, in which CD4⁺ T-cell activation is impaired (24, 25), the amount of this and/or of other cytokines available to CD8⁺ T cells may thus be limiting. It is notable that although memory CTLp were generated at reduced frequencies in CD40L-deficient mice, the proliferation of CD8⁺ T cells was not affected (Fig. 5). Thus, if a deficit in cytokine production by CD4⁺ T cells does underlie the defect in memory CD8⁺ CTLp generation in CD40L-deficient mice, the cytokine(s) involved must impact on differentiation decisions made by CD8⁺ T cells rather than just controlling their overall extent of activation and proliferation.

A third potential mechanism by which CD4⁺ T cells may affect the generation of memory CD8⁺ CTL is by activating and inducing the expression of costimulatory molecules upon antigen-presenting cells. It has recently become apparent that CD40L-CD40 interactions play a central role in the cognate interactions between CD4⁺ T cells and antigen-presenting cells, which are necessary for both cell types to become fully activated (22, 24, 71). CD40L-deficient mice have been shown to express reduced levels of B7 in their lymphoid tissues during the immune response to an adenoviral vector (71); other costimulatory molecules known to be induced by CD40L-mediated interactions are CD44H and ICAM-1 (27, 57). That CD4⁺ T cells provide help not only for the generation of memory CTLp but also for other CD8⁺ T-cell responses by inducing costimulatory activity on antigen-presenting cells is an attractive hypothesis, as it can account for the differences observed in the requirements for CD4⁺ T-cell help in different systems (examples of which were discussed above). In situations in which the costimulatory activity necessary for a particular CD8⁺ T-cell response is provided by an alternative mechanism, the response will appear to be CD4 independent. For example, certain subtypes of influenza virus are able to induce expression of CD86 on antigen-presenting cells, and the CD8⁺ T-cell response to these influenza subtypes is CD4⁺ T-cell independent, whereas CD4⁺ T-cell help is required to elicit CD8⁺ T-cell responses to other influenza subtypes under the same conditions (69). We favor this mechanism as being most likely to account for the CD4⁺ T-cell dictated deficit in memory CD8⁺ CTLp generation in CD40L-deficient mice because it could readily account for our observation that memory CTLp generation but not effector CTL activity is affected in the early stages of the immune response. The lineage relationship between effector and memory CTL has long been a subject of debate; however, it was recently shown that induction of the two can be differentiated on the basis of their requirements for costimulation (43). We hypothesize that the costimulatory requirements for the effector CTL response are adequately met in LCMV-infected CD40L-deficient mice (likely because they can be induced by CD4⁺ T-cell independent mechanisms), whereas those for memory CTLp generation are not.

In summary, our data illustrate that the generation of CD8⁺ CTLp is a process that can be distinguished from the production of effector CTL by its greater dependence on CD40L. We hypothesize that the role of CD40L (expressed on CD4⁺ T cells) is to activate antigen-presenting cells to express costimulatory activity necessary to drive the differentiation of CD8⁺ T cells into memory cells. How dependent the generation of memory CTLp (and also the effector CTL response) is on CD40L will depend on the nature and level of costimulatory activity induced on antigen-presenting cells by other mechanisms. These findings have important implications for the design of antiviral vaccines, suggesting that in order to induce an optimal memory CTL response, the vaccine should be formulated to stimulate a good primary CD4⁺ T-cell response and/or should include adjuvants which activate the expression of high levels of costimulatory activity on antigen-presenting cells. Further, these findings suggest that it should be possible to induce long-lived CTL responses in situations of CD4⁺ T-cell deficiency (e.g., in human immunodeficiency virus type 1-infected patients) provided that the antigenic challenge is designed to induce suitable costimulatory activity.