Chronic infection results in continuous formation and exhaustion of effector CD8 T cells and in failure of memory CD8 T cell development. Expression of CD70 and other molecules that provide costimulation to T cells is maintained during chronic infection. To analyze the impact of constitutive CD70-driven costimulation, we generated transgenic mice expressing CD70 specifically on T cells. We show that CD70 promoted accumulation of CD8 T cells with characteristics strikingly similar to exhausted effector CD8 T cells found during chronic infection. CD70 on T cells provided costimulation that enhanced primary CD8 T cell responses against influenza. In contrast, memory CD8 T cell maintenance and protection against secondary challenge with influenza was impaired. Interestingly, we found no effect on the formation of either effector or memory CD4 T cells. We conclude that constitutive expression of CD70 is sufficient to deregulate the CD8 T cell differentiation pathway of acute infection reminiscent of events in chronic infection.
The CD8 T cell population significantly contributes to immune responses that resolve acute viral infection. Activation of naive CD8 T cells involves proliferation and differentiation into Ag-specific effector CD8 T cells. These effector CD8 T cells acquire effector functions such as the production of immune stimulatory IFN-γ and cytolytic agents that enable them to eliminate virally infected cells. The effector CD8 T cell population contracts upon Ag clearance, and the remaining CD8 T cells that survive in the absence of Ag provide enhanced protection against secondary challenge (1, 2).
CD8 T cell differentiation is perturbed during chronic infection and this becomes apparent in exhaustion of effector function and in lack of memory development (3, 4). Upon restimulation, virus-specific CD8 T cells in chronic lymphocytic choriomeningitis virus (LCMV)3 models display poor proliferation and cytotoxicity and low IL-2 and IFN-γ production (4). CD8 T cell exhaustion has been associated with up-regulated levels of inhibitory molecules such as programmed death protein 1 (PD-1) and IL-10 on CD8 T cells of HIV and hepatitis C virus patients as well as on CD8 T cells in experimental infection models with chronic LCMV (5, 6, 7). Blockade of PD-1 or IL-10R-mediated signals in chronic LCMV infection establishes pathogen clearance, showing that these inhibitory molecules are involved in the development of CD8 T cell exhaustion (8, 9, 10). The underlying mechanism why memory CD8 T cells fail to develop during chronic infection, however, is less well understood. Low expression of IL-7Rα and IL-2/15Rβ on CD8 T cells during chronic infection indicates inefficient maintenance on homeostatic cytokines in the absence of Ag (3, 11). Indeed, transfer of Ag-specific CD8 T cells of chronically infected animals to naive animals results in the disappearance of transferred CD8 T cells. This shows that removal of Ag is insufficient to restore memory formation (3).
The persistence of pathogens during chronic infection results in continual triggering of TLRs that constitutively up-regulate expression of costimulatory molecules and production of proinflammatory cytokines. One of the up-regulated pathways of costimulation is mediated through CD70 and CD27. CD70 is the unique ligand of the TNFR superfamily member CD27 that is expressed on naive CD4 and CD8 T cells (12, 13). CD70-induced triggering of CD27 enhances the proliferative capacity of T cells and the acquisition of effector functions, such as the production of IFN-γ (14, 15, 16). Primary and secondary CD8 T cell responses against influenza infection as well as secondary responses against acute LCMV infection are impaired in the absence of CD27, demonstrating that CD70-driven costimulation is important for in vivo immune responses (17, 18). Expression of CD70 is restricted under homeostatic conditions but upon infection such as with influenza or acute LCMV, it is found on mature dendritic cells (DCs) and activated B and T lymphocytes (17, 19). In contrast to transient expression of CD70 in acute infection, constitutive expression of CD70 occurs in chronic HIV-1 infection and chronic autoimmune disease (20, 21, 22). The constitutive CD70 expression is primarily detected on T cells (20, 21, 22). To dissect the effects of constitutive expression of CD70 specifically on T cells, we generated transgenic (Tg) mice expressing CD70 under a T cell-specific promoter. We found that constitutive expression of CD70 on T cells drives Ag-dependent formation of CD8 T cells that strikingly resembles the phenotype of effector CD8 T cells during chronic infection. In particular, CD70-driven costimulation resulted in enhanced primary CD8 T cell responses, but impaired memory CD8 T cell responses against acute influenza infection. This demonstrates that constitutive expression of CD70 deregulates CD8 T cell differentiation compatible with events in chronic infection and identifies CD70 as a target for intervention in HIV-1 and other chronic diseases.
Materials and Methods
Generation of Tg mice
To generate CD70 Tg mice that express CD70 specifically on T cells, a construct was created containing the cDNA of murine CD70 under the control of the human CD2 promoter (Fig. 1 A). The CD70 cDNA was isolated by digestion with EcoRI from the pcDNA3 plasmid (23) and then cloned into the EcoRI site of the hCD2 promoter plasmid (provided by Dr. R. Meuwissen, Netherlands Cancer Institute, Amsterdam, The Netherlands). The hCD2 promoter plasmid also contains the locus control region of the human CD2 gene that confers T cell-specific, copy-dependent, and position-independent gene expression in Tg mice. After linearization and removal of plasmid sequences by digestion with KpnI and XbaI, the hCD2-mCD70 construct was injected into the pronuclei of fertilized oocytes of C57BL/6 mice. A founder was identified by Southern blot analysis of tail DNA and mated with C57BL/6 mice to obtain heterozygous CD70 Tg mice. These mice were backcrossed to create homozygous CD70 Tg mice.
C57BL/6/J, OT-I Tg (The Jackson Laboratory), CD27−/− (18), and homozygous CD70 Tg mice and homozygous CD70 Tg mice crossed with OT-I Tg or CD27−/− mice were maintained at specific pathogen-free conditions at the animal department of the Academic Medical Center (Amsterdam, The Netherlands). Screening of the mice for CD70 Tg and OT-I Tg expression was performed by flow cytometry of leukocytes from tail vein blood using anti-CD70 Abs (3B9) and anti-Vβ5 Abs (MR9-4), respectively. The CD27 genotype of the mice was screened using PCR on genomic DNA as described previously (18). Mice were used at 8–12 wk of age unless stated otherwise and within individual experiments mice were strictly age matched. All animal experiments were performed according to institutional and national guidelines.
The following mAbs from eBioscience were used: anti-CD3 (145-2C11), anti-CD4 (RM4-5), anti-CD8 (53-6.7), anti-CD27 (LG.3A10), anti-CD44 (IM7), anti-CD62L (MEL-14), anti-CD69 (H1.2F3), anti-CD70 (FR70), anti-B220 (RA3-6B2), anti-PD-1 (RMP1-30), anti-IL-7Rα (A7R34), anti-KLRG1 (2F1), and anti-CD40L (MR1). Anti-IFN-γ (XMG1.2), anti-IL-2 (JES6-5H4), anti-IL-10 (JES5-16E3), anti-TNF-α (MP6-XT22), and anti-Ki-67 (B56) were purchased from BD Biosciences.
Single-cell suspensions were obtained from spleen, peripheral lymph nodes (pLNs), lungs, liver, and bone marrow (BM) by grinding tissue over nylon filters (BD Biosciences). Contaminating RBC were removed from these preparations using erylysis buffer (155 mM NH4Cl, 10 mM KHCO3, and 1 mM EDTA). Absolute cell counts were determined by an automated cell counter (CasyCounter; Innovatis). Cells were stained with the indicated fluorochrome-conjugated or biotinylated primary Abs in the presence of anti-CD16/CD32 block (2.4G2, gift from L. Boon, Biosource BV, Utrecht, The Netherlands) for 30 min at 4°C in PBS containing 0.5% BSA. In the case of biotinylated primary Abs, cells were incubated with fluorochrome-conjugated streptavidin (eBioscience) for 30 min at 4°C in PBS containing 0.5% BSA. For staining of the nuclear Ag Ki-67, cells were fixed and permeabilized for 30 min at 4°C with fixation and permeabilization buffer (eBioscience). Expression was measured using FACSCalibur or Canto flow cytometers (BD Biosciences).
Intracellular cytokine staining
Splenocytes were stimulated with 10 ng/ml PMA (Sigma-Aldrich) and 1 μM ionomycin (Sigma-Aldrich) for 1 h. Then, 10 μg/ml brefeldin A (Sigma- Aldrich) was added to prevent cytokine release and after 4 h cells were harvested and stained with Abs against CD4 and CD8. Next, cells were fixed and permeabilized using Cytofix/Cytoperm (BD Biosciences) and labeled for intracellular cytokines using specified Abs.
RNA was extracted using the Invisorb Spin Cell RNA Mini Kit (Invitek), cDNA was synthesized using Superscript Reverse Transcriptase II (Invitrogen) and poly(T) oligonucleotides (Invitrogen), and quantitative real-time RT-PCR was performed on a LightCycler (Roche). Transcription levels were obtained using the LightCycler FastStart DNA Master SYBR Green reagent kit (Roche) and the following primer sets for 18S (forward. 5′-TCAAGAACGAAAGTCGGAGG-3′ and reverse, 5′-GGACATCTAAGGGCATCACA-3′); T-bet (forward, 5′-CAACAACCCCTTTGCCAAAG-3′ and reverse, 5′-TCCCCCAAGCAGTTGACAGT-3′); Eomes (forward, 5′-TGGACTACCATGGACATCCAGAA-3′ and reverse, 5′-TTCTCTTGCAAGCGCTGTTGT-3′); and Blimp-1 (forward, 5′-CCTCATCCCATGCTCAATCCA-3′ and reverse, 5′-GGACTACTCTCGTCCTTCATGCT-3′). Values are represented relative to that of 18S, with the lowest experimental value standardized at 1.
Mice were intranasally infected with 10× 50% tissue culture effective dose (TCID50) of the H1N1 influenza A virus A/PR8/34 for analysis of primary immune responses. Heterotypic infection with 100× TCID50 of the H3N2 influenza A virus HKx31 and 10× TCID50 of A/PR8/34 was performed to examine secondary responses. At fixed time intervals, body weights of the infected mice were obtained as a measure of disease and blood samples were drawn from the tail vein to determine levels of influenza-specific CD8 T cells. At the indicated days after infection, mice were sacrificed and blood, spleen, mediastinal lymph nodes (mLNs), and lungs were collected for analysis. Viral loads within the lungs were quantified using quantitative PCR as previously described (24). Influenza-specific CD8 T cells were enumerated using anti-CD8 Abs and PE- or allophycocyanin-conjugated tetramers of H-2Db containing the influenza-derived peptide NP366–374 ASNENMETM.
Figures represent means and error bars denote SD. Student’s t test was used to analyze for statistical significance. A value of p < 0.05 was considered statistically significant.
CD70 on T cells regulates peripheral T cell homeostasis
To study the impact of persistent CD70 expression on T cells as described in chronic infections (22), we generated Tg mice that express CD70 under control of the human CD2 promoter (Fig. 1,A). This resulted in expression of CD70 protein on T cells, but not on other cell types (Fig. 1,B). Because the transgene expression of CD70 was low on heterozygous Tg T cells, we generated homozygous Tg mice, displaying T cell-specific CD70 expression at higher levels than heterozygous Tg mice (Fig. 1 B). The levels of CD70 on T cells of homozygous Tg mice were comparable to expression of CD70 on T cells during chronic infection (22). This prompted us to use homozygous Tg mice throughout the study.
The CD70 transgene was expressed early during development of T cells within the thymus and was present on all thymocyte subsets (our unpublished data). In wild-type (WT) mice, CD27 is expressed on thymocytes as well as on T cells (12). In contrast, CD70 Tg mice did not have expression of CD27 on thymocytes, indicating that CD70 induced triggering and shedding of CD27 within the thymus (our unpublished data). Interestingly, this did not induce apparent changes in thymocyte development and thymocyte subsets were similar in size in WT and CD70 Tg mice (Fig. 1 C).
The CD70 transgene was detected on T cells in spleen, pLNs and BM of CD70 Tg mice (our unpublished data). The levels of CD27 were reduced on T cells within these tissues compared with WT mice (our unpublished data), indicating that the CD70 transgene has engaged its ligand. We analyzed whether CD70 on T cells impacted the formation of effector memory (EM) T cells similar to CD70 on DCs and B cells (14, 25). Therefore, the profile of CD44 and CD62L expression was determined on CD4 and CD8 T cells. CD44 and CD62L characterize distinct T cell populations in mice, and CD44lowCD62Lhigh T cells are defined as naive T cells, CD44highCD62Lhigh T cells as central memory (CM) T cells, and CD44highCD62Llow T cells as EM T cells (26). CD70 Tg mice had higher percentages as well as absolute numbers of EM T cells within the CD8 compartment, but not within the CD4 compartment of the spleen (Fig. 1, D–H). The absolute number of EM CD8 T cells, in contrast to that of EM CD4 T cells, was also increased within the liver and BM, but not the pLNs (Fig. 1, E and F). We observed that the absolute number of EM CD8 T cells within the spleen of CD70 Tg animals steadily increased with age, as occurred in WT animals (Fig. 1,G). In contrast to EM CD8 T cells, the size of all CD4 T cell populations and of naive and CM CD8 T cell populations were not changed in CD70 Tg mice compared with WT mice (Fig. 1,H). Crossing the CD70 Tg mice onto the CD27−/− background completely reversed the memory T cell phenotype to WT and CD27−/− levels (Fig. 1, D–F), showing that EM CD8 T cell formation is mediated by CD70-CD27 signaling. Thus, CD70 expression is functional on T cells and induces increased EM differentiation of CD8 T cells.
Constitutive CD70 induces CD8 T cells that resemble exhausted effector CD8 T cells
We analyzed the phenotype of EM CD8 T cells in CD70 Tg mice to examine how CD8 T cell differentiation under constitutive CD70 costimulation related to that of chronic infection. We observed that EM CD8 T cells of CD70 Tg mice had reduced levels of IL-7Rα and enhanced levels of CD69 and PD-1 compared with those of WT mice (Fig. 2, A and B). This was not observed on other CD4 and CD8 T cell populations (our unpublished data). The low expression of IL-7Rα indicates maintenance of EM CD8 T cells independent of the homeostatic cytokine IL-7 (27), whereas high expression of CD69 and PD-1 indicates recent stimulation on Ag (28). Ag-driven proliferation displays a much higher turnover as compared with cytokine-driven homeostatic proliferation of memory CD8 T cells (29). Therefore, to establish whether T cell proliferation was enhanced during constitutive CD70 costimulation, we analyzed the expression of Ki-67, which is expressed by cycling cells. The expression of Ki-67 was up-regulated in EM CD8 T cells of CD70 Tg mice compared with WT mice (Fig. 2,C). EM CD4 T cells and naive and CM CD8 T cells of CD70 Tg mice also had elevated expression levels of Ki-67, but to a lesser extent (Fig. 2 C). This indicates that naive, CM, and EM CD8 T cell populations may all contribute to the generation of EM phenotype CD8 T cells in CD70 Tg mice. The unaltered Ki-67 levels of naive and CM CD4 T cells and the marginally increased Ki-67 levels of EM CD4 T cells in CD70 Tg mice correspond with the absence of CD70-driven EM CD4 T cell formation. The expression profile of EM CD8 T cells of CD70 Tg mice is reminiscent of that of pathogen-specific CD8 T cells in chronic LCMV or HIV-1 infection: i.e., low levels of IL-7Rα (30), high levels of Ki-67 (31, 32), and high levels of activation-induced molecules including CD69 (31) and PD-1 (5, 6). Thus, costimulation through CD70 enhances the formation of CD8 T cells that phenotypically resemble Ag-dependent and rapidly proliferating effector CD8 T cells in chronic infection.
Up-regulation of inhibitory molecules is another distinctive feature of CD8 T cells of chronic infections (5, 6, 7, 8, 9, 10). We observed up-regulation of the inhibitory molecule PD-1 on EM CD8 T cells of CD70 Tg mice (Fig. 2, A and B). Therefore, we also analyzed WT and CD70 Tg CD8 T cells for the intracellular expression of the inhibitory cytokine IL-10 after short-term PMA and ionomycin stimulation (Fig. 2, D and E). CD8 T cells of both CD70 Tg and WT animals produced IFN-γ, but CD8 T cells from CD70 Tg animals uniquely coproduced IL-10 (Fig. 2, D and E). Although EM CD8 T cell numbers were increased, we did not observe enhanced IFN-γ production in CD8 T cells of CD70 Tg mice compared with WT mice (Fig. 2, D and E). This shows that CD8 T cells that are continually stimulated through CD70 have up-regulated levels of inhibitory molecules that in chronic infection have been shown to induce CD8 T cell exhaustion (8, 9, 10).
To examine whether the constitutive CD70-driven activation of CD8 T cells was indeed Ag dependent, we generated OT-I Tg mice coexpressing the CD70 transgene. The CD8 T cells of OT-I Tg mice contain a Tg TCR that specifically recognizes the MHC class I H-2Kb-restricted OVA peptide OVA257–264 SIINFEKL (33), an Ag that they normally do not encounter. We found that the enhanced EM phenotype of the CD8 T cell compartment of CD70 Tg mice was dependent on TCR triggering, as shown by comparable percentages and absolute numbers of EM CD8 T cells of OT-I Tg mice and CD70 × OT-I Tg mice (Fig. 2,F). Higher levels of EM CD8 T cells were present within spleen, BM, and liver but not the pLNs of CD70 Tg mice compared with CD70 × OT-I Tg mice (Fig. 2 F). Thus, recognition of environmental Ags is required for CD70 to drive the formation of CD8 T cells with an EM phenotype.
We next analyzed whether transcription factors involved in CD8 T cell development were differentially expressed in EM CD8 T cells during constitutive CD70-driven costimulation. We observed that, in contrast to T-bet, Eomes and Blimp-1 were strongly up-regulated in EM CD8 T cells of CD70 Tg mice compared with those of WT mice (Fig. 2 G), as has been previously described for exhausted effector CD8 T cells (34). Taken together, our data show that constitutive signaling through CD70 and CD27 accelerates Ag-driven formation of CD8 T cells that acquire a phenotype similar to exhausted CD8 T cells.
Polyfunctional cytokine responses are impaired under CD70-driven costimulation
Exhausted CD8 T cells in chronic infection produce low levels of cytokines such as IFN-γ, IL-2, and TNF-α and display poor cytotoxicity upon restimulation (4). Polyfunctional analysis of CD8 T cells in HIV-1 patients has shown that, in particular, the ability to produce multiple cytokines such as IFN-γ, TNF-α, and IL-2 simultaneously is impaired (35). Reduction of the polyfunctional CD8 T cell response correlates with poorer effector function of CD8 T cells on a per cell basis (35). Since we found up-regulation of the inhibitory molecules PD-1 and IL-10 on CD8 T cells of CD70 Tg animals, we examined the polyfunctional T cell response in WT and CD70 Tg mice. We observed that the IFN-γ-producing CD4 and CD8 T cell populations of CD70 Tg mice were less polyfunctional than WT mice upon stimulation with PMA and ionomycin (Fig. 3, A and B). In particular, the ability of CD4 and CD8 T cells to coproduce TNF-α and IL-2 along with IFN-γ was hampered (Fig. 3, A and B). The effect was more pronounced within the CD4 T cell population and increased with age within both the CD4 and CD8 T cell population (Fig. 3, A and B). This shows that constitutive expression of CD70 on T cells induces exhaustion in CD4 and CD8 T cells.
CD70 enhances effector CD8 T cell responses against influenza
Stimulation through CD70 on APCs quantitatively and qualitatively enhances CD8 T cell responses against acute viral infection (18, 36). We were interested whether constitutive CD70 on T cells was also able to enhance CD8 T cell responses. For this purpose, WT and CD70 Tg mice were intranasally infected with the influenza virus A/PR8/34 and as a measure of disease the body weight of the mice was monitored. The decrease in body weight upon influenza infection was less severe and resolved at earlier time points in CD70 Tg mice than in WT mice (Fig. 4,A). In addition, the viral loads of CD70 Tg mice were reduced compared with those of WT mice at day 10 (Fig. 4,B), but not at day 7 or 14 (our unpublished data). To measure the magnitude of CD8 T cell responses, we used tetramer staining of peripheral blood cells. Within the blood, influenza-specific CD8 T cells peak at higher levels in CD70 Tg mice than in WT mice, but levels of influenza-specific CD8 T cells of CD70 Tg mice return to WT levels when infection is resolved (Fig. 4,C). At the peak of the CD8 T cell response against influenza, CD70 Tg mice also contained higher numbers of tetramer+ CD8 T cells than WT mice within the spleen and mLNs, but not within the lungs (Fig. 4,D). Moreover, higher numbers of CD8 T cells in spleen of CD70 Tg mice than of WT mice produced IFN-γ upon peptide restimulation (Fig. 4 E). Production of granzyme B and IFN-γ upon peptide restimulation was not different between lung-derived CD8 T cells of WT and CD70 Tg animals, corroborating the tetramer analysis of the lungs (our unpublished data). We have no direct evidence for involvement of CD8 T cells in the increased antiviral response of CD70 Tg mice, although the enhanced CD8 T cell response in CD70 Tg mice correlates with improved clinical performance and viral clearance. Since the influx of granulocytes, monocytes, or macrophages within the lungs was not enhanced and Ab responses were not improved in CD70 Tg mice (our unpublished data), this strongly suggests that the increased antiviral response is due to the enhanced CD8 T cell response.
CD70 did not induce further up-regulation of PD-1 expression on tetramer+ CD8 T cells and did not trigger IL-10 production by CD8 T cells upon peptide restimulation (our unpublished data). Analysis of coproduction of IFN-γ, TNF-α, and IL-2 upon peptide restimulation did not reveal differences in the polyfunctional response between WT and CD70 Tg CD8 T cells of spleen and lungs (our unpublished data). Taken together, this shows that constitutive CD70 on T cells provides costimulation and does not induce T cell exhaustion in a setting of acute infection.
Constitutive CD70 induces waning of CD8 T cell memory against influenza
Differentiation pathways of effector and memory CD8 T cells separate early after infection. At the peak of the CD8 T cell response, KLRG-1 identifies short-lived effector CD8 T cells (SLECs) and IL-7Rα identifies memory precursor effector CD8 T cells (MPECs) (27, 37). Therefore, we analyzed these fractions within influenza-specific CD8 T cells at day 10 after primary influenza infection within the blood. We found that constitutive expression of CD70 enhanced the generation of SLECs as well as MPECs (Fig. 5 A). This indicates that CD70 has a positive effect on memory formation through the generation of larger numbers of memory precursors.
The development of memory is a cardinal feature of CD8 T cell responses against acute infection and transient CD70-driven costimulation has been shown to enhance memory CD8 T cell responses against influenza and acute LCMV infection (17, 18). To examine the effect of constitutive CD70 costimulation on memory CD8 T cell responses, we did a long-term follow-up of tetramer+ CD8 T cells after influenza infection. Remarkably, at day 30, CD70 Tg mice contained fewer total and fewer memory phenotype influenza-specific CD8 T cells than WT mice within the blood (Fig. 5,B). The number of SLECs at this time point is very low, which corresponds with viral clearance in both WT and CD70 Tg mice (Fig. 5,B). Follow-up of tetramer+ CD8 T cells within the blood of WT and CD70 Tg mice beyond 30 days revealed a steady decline in the number of influenza-specific memory CD8 T cells that was much more pronounced in CD70 Tg mice (Fig. 5,C). Around day 60, percentages and absolute numbers of influenza-specific memory CD8 T cells within the blood were ∼3- to 5-fold lower in CD70 Tg mice compared with WT mice (Fig. 5, C and D). Within all other tissues examined such as the lungs, mLNs, and spleen, we detected only very low numbers of influenza-specific memory CD8 T cells within CD70 Tg mice (Fig. 5,D). WT animals contained significantly more influenza-specific CD8 T cells within the mLNs and in particular within the spleen than CD70 Tg animals (Fig. 5,D). Also peptide restimulation revealed strongly decreased numbers of IFN-γ-producing influenza-specific CD8 T cells in the spleen of CD70 Tg mice compared with WT mice (Fig. 5 E). Thus, despite higher primary effector CD8 T cell responses and higher levels of MPECs, maintenance of memory CD8 T cells under constitutive CD70 costimulation was severely compromised.
Recall of influenza-specific CD8 T cells is reduced under constitutive CD70
Reduced maintenance of influenza-specific memory under constitutive CD70-driven costimulation may result in compromised secondary responses upon rechallenge with influenza virus. Therefore, WT and CD70 Tg mice were sequentially infected with the influenza virus strains A/PR8/34 and HKx31. The use of serologically distinct virus strains excludes interference by influenza-specific Abs (38). CD70 Tg animals had more pronounced weight loss and higher viral loads at day 8 upon secondary influenza infection than WT animals (Fig. 6, A and B). Although CD70 Tg mice underwent more severe disease, similar to WT mice, they were able to recover and cleared the influenza virus by day 12 (Fig. 6,B). This indicates that in stark contrast to primary CD8 T cell responses, secondary CD8 T cell responses are impaired through constitutive CD70 triggering. Indeed, percentages of influenza-specific CD8 T cells within the blood were reduced in CD70 Tg mice compared with WT mice early but not late in the secondary response (Fig. 6,C). Enumeration of influenza-specific CD8 T cells by tetramer staining at day 8 after rechallenge also showed a severe reduction in absolute numbers within blood and spleen, but not within the lungs of CD70 Tg mice compared with WT mice (Fig. 6,D). This difference was not apparent or strongly reduced in all compartments at day 12 after rechallenge (Fig. 6,D). This shows that recall CD8 T cell responses are delayed in CD70 Tg mice, which reflects the reduced memory maintenance in these mice. The numbers of IFN-γ-producing CD8 T cells upon peptide restimulation were also reduced within the spleen of CD70 Tg mice at day 8 (Fig. 6 E). Within the lungs, we did not detect differences in cytokine- or granzyme B-producing CD8 T cells, reflecting the numbers of tetramer+ CD8 T cells at this site (our unpublished data). Thus, maintenance of memory CD8 T cells was impaired and, therefore, secondary CD8 T cell responses were delayed but not abolished under constitutive CD70-driven costimulation.
Constitutive CD70 reduces CD4 T cell memory but does not impair CD4 T cell help
In the absence of CD4 T cell help, primary CD8 T cell responses were normal, but secondary CD8 T cell responses were severely compromised (39). To examine whether CD70 impaired CD4 T cell help, we analyzed the CD4 T cell response against influenza. Spleens of WT and CD70 Tg animals contained equal numbers of influenza-specific CD4 T cells at the peak of the response (Fig. 7, A and B). At late time points after infection, the percentage of influenza-specific CD4 T cells in spleens of CD70 Tg mice had declined ∼4-fold below those of WT mice (Fig. 7, A and B). This shows that CD4 and CD8 T cells respond similarly in that they are unable to maintain their memory population, but that CD70 costimulation does not induce helpless CD8 T cell responses through elimination of CD4 T cells.
CD4 T cell help includes signaling through CD40L and IL-2 (40, 41, 42, 43). Therefore, we analyzed expression of these molecules on CD4 T cells of WT and CD70 Tg mice. CD4 T cells under constitutive CD70-driven costimulation up-regulated IL-2 and CD40L upon restimulation with PMA and ionomycin, although levels of CD40L were higher and levels of IL-2 were lower compared with WT mice (Fig. 7, C and D). The lower levels of IL-2 may represent functional exhaustion of EM-type CD4 T cells in CD70 Tg mice. However, this was not observed in influenza-specific CD4 T cells of CD70 Tg mice at the peak of the response (Fig. 7, A and E). Thus, we have no evidence that constitutive triggering through CD70 impairs CD4 T cell help.
In the present study, we have investigated the impact of constitutive expression of the costimulatory molecule CD70 on T cells. CD70 on APCs acts as a costimulatory molecule and induces EM CD8 T cell formation with enhanced effector function (14, 15). We have observed that CD70 on T cells is functional as well and induced costimulation that resulted in EM CD8 T cell differentiation and in enhanced primary CD8 T cell responses against influenza. T cells are non-APCs that in mice do not present Ags to other T cells. This indicates that CD8 T cells acquire TCR stimulation separate from CD70-driven costimulation and that this similar to TCR and CD70 triggering by the same APC induces EM CD8 T cell differentiation. Indeed, in trans costimulation of CD8 T cells by CD70 has been observed by others using soluble CD70 Ig constructs (36). On APCs, codelivery of CD70 and MHC class II molecules to the immunological synapse with T cells occurs, underlining the hypothesis that coexpression of MHC class II and CD70 is required for CD4 T cell activation (44). We did not find that CD70 on T cells significantly enhanced EM formation of CD4 T cells. Constitutive expression of T cell-specific CD70 did not enhance the numbers of EM CD4 T cells within spleen, BM, and at peripheral sites such as the liver. Although CD70 on T cells enhanced influenza-specific CD8 T cell responses, it did not enhance virus-specific CD4 T cell responses against primary influenza infection. In contrast, CD70 on B cells and DCs mediates differentiation of naive CD4 T cells into Th1-type T cells that produce IFN-γ (14, 16). Possibly, CD4 T cell activation requires a single APC to provide TCR stimulation and CD70 costimulation.
The size and activation state of other leukocyte populations are affected in the CD70 Tg mice as has been described for CD70 Tg mice that express CD70 on B cells (14). Specifically, the levels of B cells but not of other APCs such as monocytes and macrophages are decreased in CD70 Tg mice. Remaining B cells, monocytes, and macrophages in CD70 Tg mice express enhanced levels of MHC class II, indicating an elevated activation state (Ref. 14 and our unpublished data). This is the indirect consequence of enhanced IFN-γ signaling in CD70 Tg mice (Ref. 14 and our unpublished data). Importantly, formation of EM CD8 T cells is not altered in the absence of IFN-γ, indicating that enhanced EM CD8 T cell formation occurs directly through CD70-driven costimulation rather than indirectly through enhanced activation of APCs (Ref. 14 and our unpublished data).
Little is known on how costimulation through CD70 and CD27 affects the differentiation pathway of naive CD8 T cells into effector and memory cells. In this study, we showed that introduction of CD70 enhances the formation of effector memory phenotype CD8 T cells. Based on further analysis using molecules that are expressed on effector cells such as CD69 and PD-1 and molecules that are present on memory cells such as IL-7Rα, it can be argued that these cells are effector rather than EM CD8 T cells. Ag is required for the formation and maintenance of effector CD8 T cells (29). We observed that CD70 required Ag to induce CD8 T cell differentiation. Ag-driven proliferation of CD8 T cells is tightly linked with differentiation into effector phenotype T cells. We do not know whether CD70 on T cells induces naive, CM, or EM CD8 T cells to differentiate into effector CD8 T cells. However, analysis of Ki-67 expression showing elevated proliferation of naive, CM, and EM CD8 T cells may indicate that all of these populations are stimulated to generate effector CD8 T cells under constitutive expression of CD70.
We showed that EM CD8 T cells under constitutive CD70-driven costimulation have a phenotype remarkably similar to Ag-specific CD8 T cells in HIV-1 infection and other chronic infections. EM phenotype CD8 T cells that develop during constitutive CD70 triggering expressed inhibitory molecules that are involved in the induction of T cell exhaustion in chronic infection such as PD-1 and IL-10. These molecules functionally impair effector CD8 T cells in chronic infection, resulting in low production of IL-2 and IFN-γ and poor cytotoxicity, and this prevents pathogen clearance (8, 9, 10). The level of cytokine production on a per cell basis and the number of cytokines, importantly IFN-γ, TNF-α, and IL-2, coproduced by individual cells are major determinants of the strength of CD8 T cell responses (45). Comparison of IFN-γ-producing CD4 and CD8 T cells of CD70 Tg mice with those of WT mice revealed a reduced ability to coproduce TNF-α and IL-2 indicative of functional exhaustion. The induction of functional impairment of T cells likely requires the presence of persistent Ag. In the absence of persistent Ag, such as during influenza responses, we did not observe that constitutive CD70-driven costimulation resulted in CD8 T cell exhaustion. Introduction of CD70 in acute influenza infection did not induce PD-1 and IL-10 up-regulation (our unpublished data). Moreover, polyfunctional analysis of the spleen and lung-resident influenza-specific CD8 T cells did not reveal functional exhaustion during primary and secondary responses (our unpublished data).
Survival of CD8 T cells in chronic infection depends upon continuous stimulation through persistent Ag (29). Ag-dependent survival of CD8 T cells in chronic infection likely requires contribution of signals from inflammatory cytokines and costimulatory molecules. We hypothesize that CD70 provides such signals and enables Ag-dependent maintenance of CD8 T cells and that this developmental pathway may ultimately result in T cell exhaustion. We reported previously that CD70 drives progressive effector T cell formation that eventually results in depletion of T cells (46). Progressive accumulation of EM CD8 T cells also occurred in CD70 Tg mice with CD70 on T cells, although not as dramatic as in CD70 Tg mice with CD70 on B cells. This may reflect expression levels of CD70, which are higher in B cell CD70 Tg mice or cell-specific expression of the CD70 transgene. Taken together with our current results, this provides a strong argument that exhaustion and depletion of T cells that prevent viral clearance in HIV-1 infection and other chronic infections result from CD70-driven immune activation.
We have shown that constitutive signaling through CD70 and CD27 is detrimental for long-term CD8 T cell-dependent immunity as it prevents formation of memory CD8 T cells. In striking resemblance, the Ag-specific T cell population in chronic infection such as with LCMV does not contain a memory T cell subset that survives independently of Ag (3). The reason for the lack of memory T cells in chronic infection is unknown and this may result from impaired development or from elimination of memory CD8 T cells. We did not observe that CD70 induced preferential effector cell differentiation or impaired memory development during acute infection with influenza. Rather CD70 enhanced formation of memory precursors early in the primary response. We were also unable to find evidence that CD70 impaired CD4 T cell responses that provide essential help for memory CD8 T cell responses against viruses including influenza (39, 47). CD4 T cell removal in chronic LCMV infection results in exacerbation of disease (4), indicating that CD4 T cell help is functional during chronic infection as well. This argues against defective development of memory CD8 T cells through CD70 signaling and in chronic infection. Indeed, some evidence indicates that elimination of memory CD8 T cells underlies the memory defect. Although FasL and Fas normally do not mediate T cell apoptosis upon acute infection, this pathway of T cell apoptosis was observed upon acute influenza infection in CD70 Tg mice (48), and, recently, it has been proposed that FasL and Fas contribute to apoptosis of Ag-specific T cells in chronic infection (49, 50, 51). It will be interesting to analyze whether FasL and Fas also remove memory T cells upon triggering through CD70 and in chronic infection. However, it needs to be mentioned that blockade of FasL- and Fas-dependent apoptosis has also been shown as a mechanism of CD70 to enhance memory CD8 T cell responses (52).
Impairment of CD8 T cell memory through CD70 signaling is in striking contrast to earlier findings that demonstrated that CD70 enhanced memory CD8 T cell responses. Ablation of CD70-driven costimulation using CD27−/− mice resulted in reduced primary and secondary CD8 T cell responses against influenza and reduced secondary responses against acute LCMV (17, 18). A major difference with these studies is the expression of CD70 that in acute LCMV and influenza infection is found transiently on low percentages of APCs and T cells (17, 19). Signaling through CD70 and CD27 is regulated through expression of CD70 and, thus, differences in the expression level of CD70 and in the window of CD70 expression may influence the outcome of immune responses. Constitutive high levels of CD70 expression such as in CD70 Tg animals and in chronic infection may result in overstimulation of CD4 and CD8 T cells and consequently immunopathology. Indeed, in striking contrast to acute LCMV, CD27−/− mice are protected against chronic LCMV infection (53). The secretion of copious amounts of IFN-γ and TNF-α by CD4 T cells that was attributed to CD70-driven costimulation resulted in disruption of the production of neutralizing Abs (53). Also, memory CD8 T cell development during chronic LCMV was restored in the absence of CD27 signaling, but this was considered secondary to the development of a neutralizing Ab response (53). Our experimental setup prevents neutralizing Abs from contributing to the secondary response against influenza, demonstrating that CD70-induced immunopathology includes impairment of memory CD8 T cell formation. Thus, under high and constitutive expression levels such as occur in chronic infection, CD70 impairs rather than enhances memory CD8 T cell responses.
In conclusion, CD70 is functional as a costimulatory molecule on T cells and enhances effector CD8 T cell mediated-immune responses, but abrogates long-term protection through impairment of CD8 T cell memory. Peptide immunization in the presence of soluble CD70 results in strong primary and secondary responses (36). These adjuvant properties of CD70 have fueled the idea that costimulation through CD70 and CD27 can be harnessed as a strategy to break tolerance in the treatment of tumors (54). However, defective maintenance of long-term memory after constitutive CD70-driven costimulation warrants caution regarding the strength and duration of the therapeutic use of CD70 in vaccination strategies. Moreover, our results indicate that treatment of chronic infection or chronic autoimmune disease may benefit from blockade rather than activation of the costimulatory CD70-CD27 axis.
We thank Cathrien Beishuizen, Natasja Kragten, Felix Wensveen, Alex de Bruin, Sten Libregts and Michiel van Oosterwijk for technical assistance, Gijs van Schijndel for making the influenza-specific tetramers and the staff of the animal facility of the AMC for excellent animal care. We thank Louis Boon, Mireille Toebes, and Ton Schumacher for providing essential reagents. Finally, we thank Drs. Monika Wolkers and Hanneke Schuitemaker for critical reading of this manuscript and helpful discussions.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Vidi and Vici grants of the Netherlands Organization for Scientific Research.
Abbreviations used in this paper: LCMV, lymphocytic choriomeningitis virus; CM, central memory; EM, effector memory; pLN, peripheral lymph node; mLN, mediastinal lymph node; MPEC, memory precursor effector cell; SLEC, short-lived effector cell; Tg, transgenic; PD-1, programmed death protein 1; DC, dendritic cell; TCID50, 50% tissue culture infective dose; BM, bone marrow; WT, wild type.