Abstract
We have previously reported that the CD8+ T cell response elicited by recombinant adenovirus vaccination displayed a delayed contraction in the spleen. In our current study, we demonstrate that this unusual kinetic is a general phenomenon observed in multiple tissues. Phenotypic analysis of transgene-specific CD8+ T cells present 30 days postimmunization with recombinant adenovirus revealed a population with evidence of partial exhaustion, suggesting that the cells had been chronically exposed to Ag. Although Ag expression could no longer be detected 3 wk after immunization, examination of Ag presentation within the draining lymph nodes demonstrated that APCs were loaded with Ag peptide for at least 40 days postimmunization, suggesting that Ag remains available to the system for a prolonged period, although the exact source of this Ag remains to be determined. At 60 days postimmunization, the CD8+ T cell population continued to exhibit a phenotype consistent with partially exhausted effector memory cells. Nonetheless, these CD8+ T cells conferred sterilizing immunity against virus challenge 7–12 wk postimmunization, suggesting that robust protective immunity can be provided by CD8+ T cells with an exhausted phenotype. These data demonstrate that prolonged exposure to Ag may not necessarily impair protective immunity and prompt a re-evaluation of the impact of persistent exposure to Ag on T cell function.
CD8+ T cells play an important role in host defense against tumors, viruses, and intracellular bacteria (1, 2, 3, 4, 5). Successful manipulation of the CD8+ T cell compartment for the purpose of vaccination, either in a prophylactic or therapeutic setting, will rely on an in-depth understanding of CD8+ T cell activation, survival, and function (6). Much of the current knowledge regarding the dynamics of CD8+ T cell response has been derived from murine models using infectious agents, such as lymphocytic choriomeningitis virus (LCMV)3 and Listeria monocytogenes (LM). In these models, the peak CD8+ T cell response is followed by a dramatic contraction phase where ∼90% of effectors at peak response die off within a 3–4 day period, leaving the remaining 10% of CD8+ T cells to form a stable memory T cell population (7, 8, 9). Through a combination of in vitro and in vivo models, it has been estimated that 20 h of antigenic stimulation are sufficient to engage a naive CD8+ T cell to undergo its full proliferative program (including the expansion and contraction phase), acquire effector function, and progress to a memory state (10, 11, 12, 13, 14, 15). The kinetics of the CD8+ T cell response appear to be independent of both the initial magnitude of Ag exposure and the subsequent presence of Ag (13). According to the current model, during a natural course of infection, the kinetics of T cell expansion and contraction are an intrinsic property of naive CD8+ T cells that is programmed within the first 24 h of priming and will proceed without the requirement of additional signals (10, 11, 13).
A stable memory CD8+ T cell population is formed after the contraction phase. Based on phenotype and location, memory T cells can be divided into central memory (Tcm) and effector memory (Tem) T cells (16, 17, 18). Although Tcm express high levels of CD62L and CCR7 and mainly reside in secondary lymphoid tissues, Tem cells are CD62LlowCCR7low and are mostly localized in peripheral organs such as the lung and the liver (16, 17, 18). The current literature supports the notion that Tcm are the truly protective memory population and over time, the CD8+ T cell population becomes enriched for cells with a central memory phenotype (19). Furthermore, as compared with Tem, Tcm exhibit greater basal homeostatic proliferation, mount a larger magnitude of expansion upon Ag exposure, and confer more protection against viral and bacterial rechallenge (19).
Replication-deficient adenovirus (rAd) vectors have been shown to be highly effective in rodent and simian preclinical models (reviewed in Refs.20 and 21). These results have fuelled interest in pursuing rAd as a vaccine platform in the clinic. Although application of the typical Ad5 vectors may be complicated by the high frequency of immunity to Ad5 within the community, the development of new vectors based on low-prevalence serotypes and improved methods for switching the rAd serotype should overcome this issue (22, 23, 24, 25, 26, 27). Our laboratory has been investigating the profile of the immune response elicited by rAd vaccines in murine models to gain insight into the mechanism of action of this promising platform. In a previous study, we reported that the CD8+ T cell response in the spleen following i.m. immunization with a recombinant rAd vaccine peaked between days 10 and 12, followed by a protracted contraction phase that displayed a less precipitous loss of the CD8+ T cell population than shown in previous reports (28). In the current study, we have further investigated the nature of the CD8+ T cell population elicited by rAd and begin to address the mechanism underlying the longevity of the postpeak T cell population.
Materials and Methods
rAds
The Ag used for these studies, SIINFEKL-Luc, is a modified version of luciferase bearing the immunodominant class-I epitope from chicken egg OVA (SIINFEKL) tagged to the N terminus. The control adenovirus (Ad), AdBHG, carries deletions in E1 and E3 but no transgene. Ad vectors expressing SIINFEKL-Luc (AdSIINFEKL-Luc-004) and luciferase (AdLuc) have been described previously (28, 29). All recombinant Ad vectors were propagated using 293 cells and purified using CsCl gradient centrifugation as described previously (30).
Animals and immunizations
Female C57BL/6 mice were purchased form Charles River Breeding Laboratories. Thymectomized C57BL/6 mice and matched controls were purchased from Taconic. OT-I mice were bred in the Central Animal Facility at McMaster University (Hamilton, Ontario, Canada). Congenic B6.PL-Thy1a/Cy (Thy1.1+) or B6.SJL-PtprcaPep3b/BoyJ (CD45.1+) congenic mice were purchased from The Jackson Laboratory and Taconic, respectively. For immunization, 108 PFU virus was diluted in sterile PBS and injected i.m. in both rear thighs of each mouse using 50 c.c. insulin syringes (BD Medical).
Flow cytometry reagents
All flow cytometry Abs were purchased from BD Pharmingen except for anti-granzyme B (GrB; clone GB11) and anti-CD127 (clone A7R34), which were purchased from Caltag Laboratories and eBioscience, respectively. The following Abs were purchased from BD Pharmingen: anti-CD3ε (clone 145-2C11); anti-CD8α (clone 53-6.7); anti-CD43 (clone 1B11); anti-CD44 (clone IM7); anti-CD45.2 (clone 104); anti-CD62L (clone MEL-14); anti-CD69 (clone H1.2F3); anti-CD107a (clone 1D4B); anti-IFN-γ (clone XMG1.2); anti-TNF-α (clone MP6-XT22); anti-IL-2 (clone JES6-5H4); anti-Thy1.2 (clone 53-2.1); and anti-Ly/6C (clone AL-21). Streptavidin-linked PE-Cy5 was also obtained from BD Pharmingen. Kb/SIINFEKL tetramers were obtained from the Molecular Biology Core at the Trudeau Institute (Saranac Lake, NY). Results from stained samples were acquired using either an LSRII or a FACSCanto (BD Pharmingen) and analyzed using FlowJo (Tree Star).
Preparation of tissues for flow cytometry
Splenic and lymph node cell suspensions were prepared as described previously (28). Bone marrow cells were obtained by flushing the femur and the tibia with PBS. RBC were removed from the spleen and bone marrow preparations by treatment with 0.15 M NH4Cl lysis buffer. Blood samples were treated with two treatments of 0.15 M NH4Cl lysis buffer for 5 min to remove RBC. Lung cell suspensions were generated by digesting perfused lungs in Hank’s buffered saline containing 150 U/ml collagenase type I (Invitrogen Life Technologies) for 1 h at 37°C. Following the digestion, the lung material was passed through a 70-μM filter. All lymphocyte preparations were washed in FACS buffer (0.5% BSA in PBS). For Ab and tetramer staining, all cell suspensions were aliquoted into 96-well round-bottom plates at 2 × 106 cells/well (BD Pharmingen). Samples were incubated at 4°C for 15 min with Fc block (clone 2.4G2; BD Pharmingen) diluted in FACS buffer before all further staining.
Tetramer staining
This method has been described by our group previously (28).
GrB staining
Cells were stained with tetramer, and subsequently permeabilized with Cytofix/Cytoperm (BD Pharmingen) according to manufacturer’s instructions. Cells were then incubated for 15 min at 4°C with Fc Block (BD Pharmingen) and stained with anti-GrB Ab (Caltag Laboratories) as well as Abs against various surface molecules; all of which were diluted in 1× Perm/Wash buffer (BD Pharmingen).
Intracellular cytokine staining
Intracellular IFN-γ, TNF-α, and IL-2 production in freshly isolated Ag-specific CD8+ T cells was determined following a brief in vitro exposure to specific peptide using a method that we have described previously (28).
Degranulation assay
We used the protocol described by Betts et al. (31). Aliquots of 2 × 106 cells freshly isolated from various tissues were placed in 96-well U-bottom plates and restimulated with peptide (SIINFEKL or KAVYNFATM at 1 μg/ml) for 5 h in the presence of monensin (BD Pharmingen) and anti-CD107a-FITC Ab (BD Pharmingen). Cells were subsequently stained with tetramer followed by staining for intracellular IFN-γ.
In vivo CTL
SIINFEKL-specific CTL were measured in vivo according to our published method (28).
In vivo Ag presentation
To measure the presence of Ag in vivo, we monitored the proliferation of OT-I CD8+ T cells transferred into rAd-immunized mice at various times postvirus injection. Lymph nodes were harvested from OT-I mice (note: mesenteric lymph nodes were excluded), and CD8+ cells were purified using the StemSep CD8+ Negative Selection Kit (StemCell Technologies). Purified OT-I CD8+ T cells were labeled with 5 μM CFSE (28), and injected i.v. into rAd-immunized congenic mice at 0.5 × 106 purified OT-I cells per mouse. Recipients were sacrificed 72 h later, and OT-I cells were identified in the draining lymph nodes (popliteal, inguinal, and iliac) by flow cytometry by the presence of CD8 and their respective congenic marker (Thy1.2 or CD45.2). Proliferation was determined by dilution of CFSE fluorescence as measured using flow cytometry. For these experiments, 1–2 × 106 events were collected per sample.
Gene expression in tissue samples
Tissues were assayed for luciferase activity as described previously (32).
Adoptive transfer of CD8+ effectors T cells
Female CD45.1+ mice (donors) were inoculated i.m. with AdSIINFEKL-Luc-004 as described above. Ten days after immunization, CD45.1+ donors were sacrificed, and CD8+ splenocytes were purified using the murine CD8+ T cell Isolation Kit from Miltenyi Biotec according to the manufacturer’s instructions. The efficacy of the CD8+ isolation was verified by flow cytometry and achieved at least 85% purity in all experiments. Purified CD8+ T cells were injected via the tail vein into naive CD45.2+ recipients at ∼5 × 106 cells/mouse. Mice were sacrificed 1 and 12 days after adoptive transfer, and the number of transferred CD8+ T cells were enumerated by staining for tetramer and CD45.1.
Virus challenge
Mice were immunized with AdSIINFEKL-Luc-004 and 45 or 90 days later, they were challenged by i.p. injection of 107 PFU Vacc-ESOVA (a gift from J. Yewdell, National Institute of Allergy and Infectious Diseases, Bethesda, MD), which expresses an endoplasmic reticulum-targeted SIINFEKL epitope. Seven days later, virus replication was measured in the ovaries as described previously (33).
Results
rAd vaccination elicits a CD8+ T cell response that displays a protracted contraction phase
We have been studying the CD8+ T cell response elicited by vaccination with rAd using a vector that expresses luciferase tagged with SIINFEKL (AdSIINFEKL-Luc-004), and observed previously that i.m. immunization with AdSIINFEKL-Luc-004 elicited a sustained CD8+ T cell response in the spleen that did not contract abruptly following the peak response but rather waned slowly leaving ∼40% of the peak number of SIINFEKL-specific CD8+ T cells at 3 wk after the peak response (28). In the current study, SIINFEKL-specific CD8+ T cells were enumerated in various tissues to determine whether this protracted contraction phase is a general phenomenon. Consistent with the results in the spleen, the SIINFEKL-specific CD8+ T cell response peaked in the lymph nodes, blood, and lung between days 8 and 12 after rAd immunization and declined afterward (Fig. 1,A). When cell numbers at all time points were normalized to the peak cell number, it appeared that CD8+ T cell numbers in the lung and blood were more stable than the spleen with >60% of the peak cell number remaining at 3 wk after the peak response (Fig. 1,B). Only 8% of the SIINFEKL-specific CD8+ T cells remained in the draining lymph node (popliteal, inguinal, and ileac lymph nodes) at 32 days post-rAd infection, presumably due to the migration of activated T cells out into the periphery (Fig. 1 B).
The CD8+ T cell population elicited by rAd exhibited a partially exhausted effector memory phenotype 30 days after immunization
To gain further insight into the nature of the CD8+ T cell population elicited by rAd, we examined the phenotype of the cells present 5 wk postimmunization. Kb/SIINFEKL-positive CD8+ T cells were examined for the expression of CD11a, CD44, Ly-6C, CD62L, and CD127. Ag-specific cells in the draining lymph nodes exhibited a primarily effector-like phenotype (CD11ahigh, CD44high, Ly-6Chigh, CD62Llow) (Fig. 2,A). However, a fraction of cells retained CD62L expression (∼20%), and a large fraction of cells (∼50%) expressed CD127, a marker that is associated with the subset of long-lived CD8+ T cells (Fig. 2,B). Similarly, tetramer-positive CD8+ T cells isolated from both the spleen and the lung expressed the memory T cell markers CD11a, CD44, and Ly6C (Fig. 2,A). Interestingly, ∼20% of the CD8+ T cells were also positive for CD127 (Fig. 2,B). We observed very few CD62Lhigh cells in the spleen and the lung, if any (Fig. 2, A and B). These results indicate that the CD8+ T cell population present at 1 month after initial rAd vaccination is mainly composed of effector memory cells, although a population of central memory cells appears to be present within the draining lymph nodes.
Analysis of IFN-γ and TNF-α production revealed that only 31.7 ± 2.1, 43.6 ± 2.6, and 51.0 ± 2.3% of SIINFEKL-specific, IFN-γ-producing CD8+ T cells in the spleen, lung, and blood, respectively, also secreted TNF-α (Fig. 2 C and data not shown; n = 8). None of the SIINFEKL-specific CD8+ T cells produced IL-2 (data not shown). This cytokine profile (limited TNF-α production and no IL-2 secretion) was reminiscent of partial exhaustion as described by Wherry et al. (34).
Ag load after AdSIINFEKL-Luc immunization
We also noted that the CD8a levels on the tetramer-positive population were reduced relative to the tetramer-negative CD8+ cells (data not shown). We suspected that the reduced CD8a levels on the tetramer-positive cells might be reflective of TCR down-regulation due to repeated antigenic stimulation, consistent with the state of partial exhaustion suggested by the cytokine secretion pattern. Repetitive exposure of CD8+ T cells to Ag might represent a pathway for maintaining the peripheral repertoire by continually simulating circulating cells to expand and replace those that have died. However, chronic infection in other models, such as LCMV, has been associated with exhaustion and deletion of Ag-specific T cells (35). As a step toward understanding the partially exhausted phenotype of the CD8+ T cell population, we examined the persistence of Ag within this model. Following i.m. injection of AdSIINFEKL-Luc-004, transgene expression was detected in the draining lymph nodes (popliteal, inguinal, and ileac lymph nodes) and the muscles, but not in the spleen or the distal lymph nodes (28). To determine the degree of Ag persistence in this system, we measured the kinetics of Ag expression within the draining lymph nodes and muscles over a period of 3 wk. Comparable levels of luciferase activity were measured in the draining lymph nodes and the muscle 3 days postvaccination relative to the tissue mass (Fig. 3,A); however, the muscle expressed higher absolute levels of luciferase (Fig. 3,B) in accordance with our previously published report (28). Luciferase expression dropped by >90% between days 3 and 10 (Fig. 3,C), consistent with the onset of CTL function, and luciferase activity was reduced to background levels by 20 days after AdSIINFEKL-Luc-004 inoculation (Fig. 3,B). The decay of luciferase in the muscle was slightly delayed relative to the lymph nodes (Fig. 3 C). Thus, although the loss of luciferase expression correlated with the onset of CD8+ T cell function, Ag remained detectable for at least 14 days, suggesting that this residual Ag may be involved in maintenance of the T cell population following the peak response.
Prolonged Ag presentation following i.m. rAd immunization
Another possible mechanism for sustaining the population of effector CD8+ T cells is through asynchronous engagement of naive T cells. As such, naive CD8+ T cells would be engaged at different times postimmunization and, thus, some cells would be expanding while others were contracting. The net effect would be a population with a broad peak and a contraction that appears to be delayed. Because Ag is measurable for at least 14 days in our model, it is possible that Ag is available for priming naive T cells throughout that period. To address this putative mechanism, we examined the presence of Ag within the lymph nodes, draining the injection site using TCR transgenic OT-I T cells specific for Kb/SIINFEKL as probes. Congenic mice were immunized with AdSIINFEKL-Luc-004 on day 0, and, at various timepoints postinfection, CFSE-labeled OT-I cells were adoptively transferred into the vaccinated congenic hosts. Seventy-two hours after transfer, OT-I T cell proliferation and expansion was determined by flow cytometry. Following immunization with AdLuc, which does not carry the SIINFEKL epitope, only a single CFSEhigh peak was observed, demonstrating that the adoptively transferred T cells could not be activated by virus-induced bystander inflammation (Fig. 4,B). In striking contrast, AdSIINFEKL-Luc-004-immunized congenic hosts exhibited evidence of transgenic T cell proliferation for at least 30 days after initial vector inoculation, which eventually became negligible by day 40 (Fig. 4,A). The extent of CFSE dilution was not equivalent at all time points, suggesting changing levels of Ag presentation during the primary immune response following rAd vaccination (Fig. 4,A). To provide a surrogate marker for the strength of Ag presentation, we examined the expansion of OT-I cells in the draining lymph nodes of mice immunized with AdSIINFEKL-Luc-004 relative to those that received AdLuc. The absolute number of OT-I cells in the lymph node is shown in Fig. 4,C. At early time points following AdSIINFEKL-Luc-004 immunization (days 2–6) similar expansion was observed, suggesting that Ag presentation was stable during this period. Between days 6 and 8 following immunization, there was a sharp reduction (∼90%) in the magnitude of OT-I T cell expansion, which correlated with the loss of Ag expression (Fig. 3,C). Beyond day 8, we observed low-level expansion for a period of ∼4 wk (Fig. 4,C). Although the high phase of this profile correlated with luciferase levels, the low-phase persists at times when luciferase (i.e., the target Ag) is no longer measurable. No expansion of the OT-I population was observed when cells were transferred 40 days after immunization (Fig. 4 C), suggesting that Ag was cleared from the system, although residual Ag may be present at a level at below the sensitivity of the OT-1 Ag presentation assay. Nevertheless, these results demonstrate that an Ag reservoir exists for at least 32 days in the rAd-immunized host; although the nature of the reservoir remains unclear.
CD69 is up-regulated on Ag-specific CD8+ T cells in the lymph node
If, indeed, Ag persisting in the draining lymph nodes of Ad-immunized hosts was responsible for restimulating Ag-specific CD8+ T cells for an extended time period, then we would expect to detect evidence of recent TCR engagement. To this end, we examined the CD69 expression on tetramer-positive CD8+ T cells 30 and 60 days following immunization, because CD69 is up-regulated within 6 h following interaction with TCR with its cognate peptide (36). Between 20 and 25% of SIINFEKL-specific CD8+ T cells in the lymph nodes stained positive for CD69 at 30 and 60 days following immunization with AdSIINFEKL-Luc-004 (Fig. 5,A). The observed CD69 expression on the Ag-specific cells was substantially higher than the basal level of CD69 expression on bulk lymph node CD8+ T cells (5%). In contrast to the lymph nodes, only 3% of Ag-specific T cells in the spleen were CD69high (Fig. 5 A), further supporting the model that SIINFEKL-specific T cells encounter their cognate Ag in the draining lymph nodes at these late time points following AdSIINFEKL-Luc-004 immunization.
Day 10 effectors contract more rapidly in an Ag-free environment
To more closely address the relationship between Ag presentation and the maintenance of the CD8+ T cell population elicited by rAd vaccination, we isolated CD8+ splenocytes from CD45.1+ congenic donors 10 days after immunization and adoptively transferred the cells into Ag-free naive CD45.2+ hosts. Interestingly, the number of tetramer-positive congenic T cells present in the spleen 12 days after transfer was reduced by 80% relative to the number of tetramer-positive cells present 24 h after transfer (Fig. 5,B), demonstrating that in the absence of Ag, the rAd-induced SIINFEKL-reactive CD8+ population contracts at a more rapid rate. A similar rate of contraction was also observed in the lungs (data not shown). The loss of tetramer-positive CD8+ T cells following adoptive transfer was not due to simple clearance of the transferred CD8+ T cells because the population of congenic CD8+ T cells present at day 12 had only abated by 40% relative to the numbers 24 h after transfer (Fig. 5 B).
CTLs are fully functional effectors
We have previously demonstrated using in vivo CTL assays that SIINFEKL-specific target cell lysis could be detected in the draining lymph nodes as early as 3 days after Ad vaccination (28). Thus, the persistence of Ag presentation for up to 5 wk was perplexing because one would expect that activated effectors would rapidly clear peptide-loaded cells within the lymph nodes. Our observation that the CD8+ T cell population elicited by Ad exhibits a partially exhausted phenotype suggested that that the CD8+ T cells may be dysfunctional and, therefore, unable to effectively clear Ag, allowing Ag to persist for an extended period.
To verify that the CD8+ T cells were functionally lytic, we examined the intracellular storage of GrB in Ag-specific CD8+ T cells. Active CTLs have been shown in other viral infection models to contain GrB in their granules when examined directly ex vivo (19, 37). In addition, CTL-mediated lysis via the perforin/GrB pathway does not require de novo synthesis of proteins, further supporting the notion that GrB-positive effectors are indeed cytolytic (38). Consistent with the notion that the rAd-induced cells are cytolytic, effectors isolated from the lymph node, spleen, and lung all stained positive for GrB from day 6 to 21 post-rAd immunization, although the intracellular GrB levels gradually decreased as the T cells progress away from the initiation of the cellular immunity (Fig. 6, A and B).
Despite the fact that the Ad-induced CTLs contained GrB, it remained possible that they were unable to release their granule contents, so we measured degranulation at a single-cell level using the CD107a staining method (31). In all of our experiments, restimulation of SIINFEKL-reactive CD8+ T cells with their cognate peptide yielded IFN-γ-positive cells that costained for CD107a, indicating that they have recently degranulated (Fig. 6,C and data not shown). In contrast, the same population of SIINFEKL-specific CTLs, identified by tetramer staining, was negative for both IFN-γ production and CD107a following stimulation with nonspecific peptide (Fig. 6 C). Therefore, in addition to possessing intracellular storage of GrB, the CD8+ T cells were able to release their granule content upon recognition of cognate MHC:peptide complex.
Finally, lytic activity in vivo was verified. Results from both the lymph node and spleen demonstrated comparable CTL function at both 8 days and 31 days after immunization with AdSIINFEKL-Luc-004 (Fig. 6 D), demonstrating that SIINFEKL-reactive CD8+ T cells have maintained a functional effector phenotype.
The sustained CD8+ T cell population is not likely due to continual recruitment of recent thymic emigrants into the response
As mentioned earlier, we postulated that the sustained CD8+ T cell response may result from an asynchronous priming event, where naive T cells are continually recruited into the response and replace the cells that die. To address this hypothesis, we terminated the source of naive T cells by thymectomizing wild-type C57BL/6 mice 8 wk before immunization with AdSIINFEKL-Luc-004. If the sustained T cell population was due to the continual recruitment of recent thymic emigrants into the response over a prolonged period (i.e., 30 days), then thymectomized mice should display a more abrupt loss of the CD8+ T cell population because the pool of Ag-specific T cell precursors should be depleted more rapidly in the absence of a thymus to replenish the pool. Although SIINFEKL-specific CD8+ T cell numbers were slightly lower in thymectomized mice at day 12 of the response (Fig. 7), consistent with the reduced availability precursors, we observed no difference in cell number at 60 days postimmunization, indicating that there is no increased loss of SIINFEKL-specific CD8+ T cells in thymectomized mice (Fig. 7). Thus, continual recruitment of recent thymic emigrants following rAd immunization does not seem to explain the protracted contraction phase.
The SIINFEKL-specific memory CD8+ T cell population induced by Ad maintains an effector phenotype in the periphery but acquires a central memory phenotype in the draining lymph nodes
In many murine models of infection including vesicular stomatitis virus, LCMV, and LM, Tcm can be identified in the spleen as early as 3 wk following initial pathogen inoculation based on increased surface expression of CD62L (16, 18, 19). Because we estimated that Ag presentation within the draining lymph nodes ceased around day 40, we interpreted this outcome as evidence that the Ag reservoir, wherever it exists, had been largely, if not completely, depleted. We allowed the CD8+ T cell population a further 20 days of rest and examined the phenotype at day 60 postimmunization to determine whether the proportion of central memory cells had increased. Surprisingly, most, if not all, tetramer-positive CD8+ T cells remained CD62Llow in the spleen, lung, and bone marrow (Fig. 8). There appeared to be a clear population of CD62Lhigh cells in the peripheral blood, representing ∼10% of total tetramer-positive cells (Fig. 8 B). Cytokine secretion patterns for IL-2, TNF-α, and IFN-γ had not changed from day 30 (data not shown), and 30–50% of the cells continued to express the 1B11-reactive glycosylated form of CD43, which has been associated with effector function (39). Interestingly, 30–40% of the tetramer-positive cells were positive for the IL-7 receptor α-subunit, CD127, which may be important for maintaining the memory population (40). However, we observed no correlation between CD127 and CD43 expression (data not shown).
Interestingly, a higher fraction of tetramer-positive cells in the draining lymph nodes had adopted a central memory phenotype (CD62Lhigh, CD127high). In fact, most (∼80%) of the Ag-specific CD8+ T cells in the lymph node expressed CD127. Thus, it appears that the Tcm population in this model is sequestered primarily in the lymph nodes and does not progress to the spleen or bone marrow. Overall, this phenotypic analysis suggests a mixed population of partially exhausted CD8+ T cells with some capacity for recirculation through the lymphatics and responsiveness to homeostatic factors, such as IL-7.
Ad-induced effector memory cells are protective against challenge
CD8+ T cell exhaustion is one mechanism by which chronic viral infections may escape CD8+ T cell immunity. Therefore, even though the Ag-specific CD8+ T cell population is sustained over a longer period following rAd immunization, the observation that the memory cells exhibit partial exhaustion suggests that this population may not provide effective long-term immunity. To confirm functionality of the T cell population generated by rAd immunization in vivo, mice were immunized with AdSIINFEKL-Luc-004 and challenged with Vacc-ESOVA 45 or 90 days later. The only commonality between the vaccine and the challenge is the SIINFEKL epitope. Thus, this model is reflective of protection that is uniquely mediated through CD8+ T cells. Virus titers in the ovaries were measured 7 days postchallenge (Fig. 9). Although the titers of virus in naive mice were >107 PFU, the mice challenged with Vacc-ESOVA 45 days postimmunization were devoid of virus (Fig. 9,A). When AdSIINFEKL-Luc-004-immunized mice were challenged with Vacc-ESOVA 90 days postimmunization, all mice were protected from challenge, and 3 of 5 mice maintained sterilizing immunity (Fig. 9 A). Thus, the CD8+ T cells in this model appeared to be functional in an acute challenge situation over at least a 90-day period.
To assess the relationship between circulating SIINFEKL-specific CD8+ T cells and protective immunity, tetramer-positive cells were enumerated in the peripheral blood (Fig. 9,B). Although it is clear that fewer SIINFEKL-specific CD8+ T cells were circulating in the peripheral blood before the day 90 challenge relative to the frequency of Ag-specific cells before the day 45 challenge (Fig. 9,B, □), there was no correlation between the frequency of tetramer-positive CD8+ T cells and sterilizing immunity. Interestingly, the expansion of the SIINFEKL-specific CD8+ T cells following challenge with Vacc-ESOVA did not appear to be related to the frequency of CD8+ T cells before challenge (Fig. 9 B, ▪), suggesting that a separate mechanism may function to determine the ultimate size of the secondary expansion. We have reported similar results previously in a flu model (41).
Discussion
In this study, we report a number of interesting observations regarding the CD8+ T cell response elicited by i.m. immunization with rAd. First, the CD8+ T cell response is sustained for a period longer than would be expected based on results in other models. Second, the CD8+ T cell population displays a novel phenotype for a period of at least 60 days, with evidence of sustained effector functions and partial exhaustion, yet this population still provides long-term immunity against virus challenge. Finally, Ag remains available in this system for at least 32 days following immunization (as measured in the OT-I transfer studies), which may explain the prolonged effector phenotype of the responding T cells.
The peak CD8+ T cell response often coincides with the eradication of acute pathogens such as LCMV and LM (13, 34) The same is true in our model because Ag expression in the draining lymph node is rapidly cleared between days 3 and 10. Ag in the muscle clears at a slower rate than the lymph node, but this is not altogether surprising because the CD8+ T cells must migrate from the draining lymph node to the muscle to mediate Ag clearance. Nonetheless, luciferase activity, which is a direct function of Ag expression, remains detectable in the muscle for at least 14 days, albeit at very low levels. By day 20, Ag levels have been reduced to background levels in both the muscle and the draining lymph nodes. Thus, the Ag in this model would appear to persist for a longer time than LCMV or LM Ag in acute infection models (13, 42), where the pathogen is cleared within 7–8 days, but shorter than models of chronic LCMV infection, where the virus can persist for >30 days (42) or indefinitely (43). However, in both LCMV and LM models, the presence of Ag is not linked to alterations in the contraction phase (13), although it is clear that chronic infection with LCMV profoundly affects the phenotype and functionality of the responding CD8+ T cell population (34, 42, 43). Indeed, loss of CD8+ T cell function has been shown in other models of chronic virus infection (44) and is believed to be a factor in the persistence of viruses such as HIV and hepatitis C virus (45, 46, 47). The intermediate persistence of Ag that occurs following AdSIINFEKL-Luc-004 immunization appears to cause some degree of dysfunction in the SIINFEKL-specific CD8+ T cell population (lack of IL-2 production and decreased TNF-α production), but they remain functionally lytic and capable of rejecting virus challenge at late time points, suggesting that their key effector functions are not impaired. These observations prompt a reevaluation of the criteria for a functional, long-term memory CD8+ T cell. Current studies in our laboratory are focused on comparing the protective function of CD8+ T cells elicited by rAd to CD8+ T cells with the conventional central memory phenotype (i.e., CD62Lhigh, CD127high, IFN-γ-positive, TNF-α-positive).
Interestingly, a recent report demonstrated that immunization with rAd via the i.v. route resulted in functional exhaustion of CD8+ T cells and abrogated protective immunity in a tumor model (48). We have previously demonstrated that i.v. administration is the least efficient method for eliciting T cell immunity with rAd, whereas i.m. and intradermal were the most efficient (30). Although these results provoke careful consideration of routes of administration when using rAd, this issue may not represent a major concern because it is unlikely that the i.v. route would be used in a clinical setting, and rAd vaccines have successfully generated long-term protective immunity following administration via a number of routes, including intradermal, intranasal, and i.m. (49, 50, 51).
The requisite duration of Ag exposure for optimal CD8+ T cell function following acute infection is unclear because Ag persistence varies with each model. Using in vitro stimulation, van Stipdonk et al. (14) demonstrated that only 20 h of stimulation is required to induce CD8+ T cells to acquire full effector function and undergo normal expansion, but it remains uncertain how long Ag must be maintained in vivo because it is most likely that CD8+ T cell activation is an asynchronous event. In the case of LM, Plasmodium falciparum, and vaccinia virus infection, Ag is only detected within the lymphoid tissue for 2–3 days (52, 53, 54), whereas intradermal infection with HSV or intranasal infection with Sendai virus results in Ag presentation that can be measured for 7–9 days (55, 56). Only immunization with bacillus Calmette-Guérin (BCG) produced a prolonged Ag presentation similar to our observations with AdSIINFEKL-Luc-004 (57). However, a key difference between the results with BCG and our results is the kinetics of Ag presentation. In the case of BCG, Ag presentation is low initially, peaks around day 6, and persists at a low level for almost 60 days. These observations are consistent with the fact that BCG is a slow growing mycobacterium that establishes a chronic infection that is sustained for at least 75 days, providing a persistent source of Ag production (57). By contrast, high-level Ag presentation following AdSIINFEKL-Luc is measurable as early as 2 days following immunization and is sustained for 6 days, at which point Ag presentation is reduced to a low-level that persists for an additional 4–5 wk in the absence of measurable Ag production. Interestingly, the CD8+ T cell response elicited by BCG also exhibited a reduced rate of contraction (57), similar to our results with rAd, suggesting that prolonged availability of Ag may explain the rate of contraction.
Stock et al. (56) demonstrated that excision of the injection site within 24 h following HSV infection reduced the magnitude of the CD8+ T cell response at 7 days, indicating that Ag is required for >24 h to achieve the maximum response, but the exact amount of time that Ag is required remains undetermined. The observation that CD8+ T cells taken from rAd-immunized mice at the peak of the response contract more rapidly following adoptive transfer to Ag-free hosts, supports the concept that the presence of Ag past day 10 is important for maintaining the Ag-specific CD8+ T cell population. Given the fact that only CD8+ T cells in the blood and draining lymph nodes exhibit measurable CD62L expression following rAd immunization, we propose that the sustained effector phenotype is the result of recirculation through the draining lymph nodes and repeated interaction with the same peptide-loaded APCs that stimulate the naive OT-I T cells in our adoptive transfer model. This hypothesis is supported by the observation that the 20–25% of the Ag-specific CD8+ T cells in the lymph node are CD69-positive, suggesting recent antigenic stimulation.
Measurement of Ag presentation following infection has relied primarily on two techniques: 1) identification of MHC/peptide complexes ex vivo using T cell lines (53, 55, 56); or 2) adoptive transfer of CFSE-labeled TCR-transgenic T cells (52, 54, 56, 57). Although these techniques can successfully identify low levels of MHC/peptide complex in the lymphoid tissues, they do not prove that the Ag in lymph node is actually involved in the T cell response. With that in mind, an alternate explanation for the sustained CD8+ T cell response in our model is that effector T cells in the periphery are actually stimulated by Ag within the injection site. This interaction leads to release of Ag that can be taken up by local APCs, which subsequently return to the draining lymph node where they can be identified by adoptively transferred OT-I cells. We also cannot preclude the possibility that Ag is being presented at distal sites, such as the spleen or lung, because we see evidence of CFSElow cells in all tissues following OT-I transfer (T. Yang and J. Bramson, unpublished data), and we cannot determine whether those cells have migrated from the lymph nodes or were stimulated locally. We infer that the draining lymph nodes are the primary site of Ag presentation because we observe more CFSEint cells in the lymph nodes relative to other tissues. However, because we cannot state with certainty the exact location of Ag presentation at this time, the data that we have generated can only be interpreted as proof that Ag remains available to the system for up to 32 days, and that the presence of the Ag is required to sustain the CD8+ T cell population after the peak of the response.
Long-term, low-level persistence of Ag has been suggested to be a key factor for maintaining optimal memory responses (58); however, this concept remains controversial in light of data demonstrating that long-term survival of memory CD8+ T cells is not dependent upon Ag (59). Ag has been shown to be critical for maintenance of LCMV-specific CD8+ T cells in a model of chronic infection (42); thus, it appears that T cells may become dependent upon Ag exposure when Ag persists for longer than necessary. In the case of LCMV infection, CD8+ T cells that develop during a chronic infection are poorly responsive to homeostatic factors like IL-7 and IL-15. The CD8+ T cells in our model regain expression of the CD127 (IL-7Rα) around day 60 and progress to a CD127int/high phenotype past day 100; however, they remain CD62Llow in the peripheral tissues (T. Yang and J. Bramson, unpublished data). Thus, it appears that the CD8+ T cells in our model can respond to homeostatic factors, and this may explain the survival of these cells for >350 days (T. Yang and J. Bramson, unpublished data); however, further studies are required to determine whether the CD8+ T cells elicited by rAd are truly Ag-independent over the long term.
Finally, the lineage relationship between effectors, Tem and Tcm cells, remains controversial. Wherry et al. (19) proposed a linear differentiation model where effectors first differentiate into Tem, which then up-regulate CD62L expression to become Tcm. Baron et al. (60), however, have reported that human circulating effector and central memory CD8+ T cells largely form two distinct clonotypic populations that do not interconvert, favoring the concept that naive T cells are preprogrammed to become either Tcm or Tem. As stated previously, we observed cells with a Tcm phenotype in the blood and the draining lymph node, and we hypothesize that these cells circulate between the blood and the lymphatics, where antigenic stimulation of the Tcm within the draining lymph node promotes proliferation and differentiation to Tem, which then migrate to the periphery. Thus, the absence of Tcm in the peripheral tissues (spleen, bone marrow, lungs) may be due to the continued stimulation of Tcm in the lymph nodes. Although this hypothesis is consistent with the linear differentiation model, the model fails to explain the absence of Tcm in the spleen at day 60–100, because one would predict that more Tcm would be present in the peripheral tissues in the absence of stimulation in lymph nodes. The true relationship between the window of Ag exposure and the subsequent development of memory subtypes in our system can only be addressed, however, once we have developed a model where we can control the length of time that T cells are exposed to Ag in vivo.
The results of this manuscript offer new insights into the biology of the CD8+ T cell response elicited by rAd vaccines. Furthermore, these data also prompt a reevaluation of the criteria for a functional memory CD8+ T cell and optimal vaccine design. Further studies are required, however, to fully understand the relationship between duration of Ag exposure and CD8+ T cell function.
Acknowledgments
We thank Carole Evelegh and Kelley Putzu for preparing the viruses used in these experiments and assisting with some of the animal studies.
Disclosures
The authors have no financial conflict of interest.
Footnotes
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 the Canadian Institutes of Health Research (CIHR) (MOP-42433 and MGC-57082; to J.L.B.). J.L.B. is the recipient of an Rx & D Health Research Foundation/CIHR Career Award in Health Research. Y.W. is a CIHR New Investigator. T.G., T.-C.Y., and J.M. were supported through a PREA award given to J.L.B. from the Ontario Ministry of Enterprise, Opportunity, and Innovation.
Abbreviations used in this paper: LCMV, lymphocytic choriomeningitis virus; LM, Listeria monocytogenes; Tcm, central memory T cell; Tem, effector memory T cell; rAd, replication-deficient adenovirus; Ad, adenovirus; AdLuc, Ad luciferase; GrB, granzyme B; BCG, bacillus Calmette-Guérin; int, intermediate.