No Intrinsic Deficiencies in CD8⁺ T Cell-Mediated Antitumor Immunity with Aging¹

T cells are pivotal mediators of host defense against uncontrolled cancer growth (1, 2). Although most murine models used for cancer research use young mice, cancer is primarily a disease of aging individuals. In the U.S., >50% of cancer diagnoses are made after the age of 65 years (3). Concurrently, aging is associated with numerous alterations in immune function, resulting in a diminished capacity of the aged immune system to respond to infections or vaccinations (4, 5). We therefore wanted to assess age-related changes in APC and CD8⁺ T cell function in vivo and determine how these alterations impact the generation of antitumor immunity, if at all.

Many age-related changes within the T cell compartment occur at the population level. Advanced age is characterized by thymic involution and a progressive accumulation of memory phenotype T cells, both of which decrease the frequency of naive T cells available to respond to novel antigenic challenges (6, 7, 8). Aging is also associated with the appearance of CD8⁺CD28⁻ Ag-experienced T cells in humans (9, 10) and oligoclonal expansions of CD8⁺ T cells in both mice and humans (11, 12, 13); these create a further diminution in the capacity to respond to new Ags.

Deficiencies in T cell function on a cellular level are also known to occur, particularly for CD4⁺ T cells. For example, aged CD4⁺ T cells exhibit decreased proliferation and effector function after TCR stimulation (14, 15) due to impaired production of IL-2 (16). Even in the presence of exogenous IL-2, CD4⁺ memory T cells generated from aged effectors display defects in proliferation and cytokine production (17).

In contrast, one recent report demonstrated that during primary stimulation of CD8⁺ 2C TCR transgenic cells with peptide, there is no loss of in vitro proliferation, IL-2 production, or cytolytic activity with age (18). The major defect seen during secondary stimulation in this system was a slight deficit in IL-2 production (18). However, in vivo generation of primary CD8⁺ T cell-mediated immunity to viral infections in aged mice does appear to be impaired and is marked by decreased numbers of Ag-specific effectors that result from defective expansion of precursor cells (19, 20). Because of this, the numbers of memory cells generated and the magnitude of the recall response are also correspondingly diminished in aged mice (19). Thus, further study is needed to fully characterize the effects of age on CD8⁺ T cell function in vivo.

It is unclear to what extent alterations in APC function contribute to the loss of T cell responsiveness observed during aging. Several studies have described a reduced ability of aged unfractionated APCs or macrophages to stimulate T cell effector function (21, 22, 23). However, human monocyte-derived DCs from aged donors appear phenotypically similar to DCs generated from young donors and are equally able to stimulate Ag-specific T cell responses in vitro (24). Conflicting in vivo APC functions have been reported in aged mice and humans (25, 26, 27); hence, the ability of old APCs to prime naive T cells in vivo remains uncertain.

Previous studies investigating antitumor responses in old mice found protective immunity to be impaired. Multiple reports showed that rejectable syngeneic tumor loads in young mice were able to grow progressively in aged mice (28, 29, 30, 31). One study using tumor cells engineered to express IL-2 found that primary responses to tumor vaccines were similar in young and old mice, but that secondary responses to tumor rechallenge were impaired with age (32). Additionally, old mice may mount a more oligoclonal T cell response to tumors than young mice (33). Therefore, T cell-mediated antitumor immunity in aged mice may be deficient on multiple levels.

Although these previous studies have contributed considerably to our knowledge of the ability of aging T cells to control tumor growth, several key issues remain unresolved. Early reports in this field examined endogenous antitumor responses in non-TCR transgenic mice and were consequently unable to control for the numbers of tumor-specific T cells present in young and aged mice or to characterize their phenotype. Therefore, it is possible that fewer naive, tumor-specific CD8⁺ T cells were present in aged mice in these studies, accounting in large part for the diminished antitumor responses observed. Additionally, the ability of aged APCs to support induction of antitumor T cell responses has not been thoroughly investigated. Thus, multiple defects, affecting both APCs and effector T cells, could factor into the observed impairments in antitumor immunity.

To address these issues, we made use of our DUC18 T cell/CMS5 tumor model system (34, 35). CMS5 is a methylcholanthrene-induced fibrosarcoma line that originated from a BALB/c mouse (36). It possesses a single amino acid mutation at position 136 in the protein ERK2, generating a novel tumor Ag. A tumor-specific peptide encompassing this mutation, denoted tERK,³ is presented on H-2K^d molecules (37). DUC18 is a BALB/c TCR transgenic mouse in which ∼50% of the CD8⁺ T cells express an αβ TCR specific for tERK/H-2K^d complexes (34). After adoptive transfer, 3 × 10⁶ naive CD8⁺ DUC18 T cells from young mice are able to reject day 4 established CMS5 tumors with highly reproducible kinetics (34, 35). The DUC18 T cell/CMS5 system is thus an ideal setting in which to examine the effects of aging on CD8⁺ T cell function and antitumor immunity.

In the current study we examined both the ability of resident APCs in old mice to prime antitumor T cell responses and the ability of aged naive T cells to become activated and to reject tumor challenges. In contrast to earlier reports, we found no loss of CD8⁺ T cell-based tumor immunity with age when equal numbers of tumor-specific T cells were present, a finding that has positive implications for the development of successful immunotherapies in aging cancer patients.

DUC18 TCR transgenic mice are on a BALB/c background and have been described previously (34). Female DUC18 mice were used at 8–12 wk of age (young) or 16–18 mo of age (old). Young female BALB/c (substrain BALB/CAnNCR, referred to as BALB/c) mice were obtained from Charles Rivers Laboratories (Wilmington, MA) through the National Cancer Institute and were used at 8–12 wk of age. Old female BALB/c mice were obtained at 9–10 mo of age from National Cancer Institute and were maintained in-house until 16–18 mo of age, or were purchased at 16–18 mo of age from the National Institute on Aging. At this age, BALB/c mice have reached ∼80–90% of their predicted life span (www.informatics.jax.org) and are of comparable age to the human population in which the majority of cancer diagnoses are made (3). All mice were housed in our specific pathogen-free barrier facility at Washington University (St. Louis, MO). Mice showing obvious signs of pathology (spontaneous tumors, or enlarged spleens or lymph nodes (LNs)) were excluded from study.

The CMS5 fibrosarcoma cell line has been described previously (36). CMS5 cells were cultured in vitro in RPMI 1640 supplemented with 10% FCS (HyClone, Logan, UT), 2 mM Glutamax (Life Technologies, Gaithersburg, MD), and 50 μg/ml gentamicin (Invitrogen, Carlsbad, CA). For in vivo challenge, cells were trypsinized, washed twice in HBSS, and injected s.c. at 3 × 10⁶ cells/mouse in 200 μl of sterile PBS in the right hind flank. For tumor outgrowth and rejection studies, tumor area (product of longest diameter by orthogonal diameter) was measured every 2 days starting on day 4 postchallenge.

The following Abs were used for DUC18 T cell purification: CD4 (RM4-4), B220(RA3-6B2), I-A^d (39-10-8), CD49b(DX5), and CD11b(M1/70). The following Abs were used for flow cytometry: CD4-FITC or -PE(GK1.5), CD8-FITC or -PE (53-6.7), CD11c-bio or -FITC (HL3), CD19-PE (1D3), CD25-bio (7D4), CD40-bio(3/23), CD45RB-bio (16A), CD62L-bio (MEL-14), and CD86-FITC (GL1), Vβ8.3-FITC or -PE (1B3.3), H-2K^d-FITC (SF1–1.1), and I-A^d-FITC (39-10-8). All Abs were obtained from BD Pharmingen (San Diego, CA) or BioLegend (San Diego, CA) and were used interchangeably.

T cells isolated from aged DUC18 mice are referred to as aged or old T cells throughout the manuscript. The term DUC18 T cells refers to cells from DUC18 mice that are CD8⁺/Vβ8.3⁺ and reactive against the CMS5-derived tERK peptide presented by H-2K^d. Endogenous CD8⁺/Vβ8.3⁺ cells in BALB/c mice do not respond to tERK peptide or CMS5 tumors by proliferating or producing cytokines (L. Norian, unpublished observations).

CD8⁺ T cells from DUC18 TCR transgenic mice were enriched by negative selection from single-cell suspensions of splenocytes and LN cells. After a brief RBC lysis step, 30–40 × 10⁶ unfractionated cells were incubated with 10 μg of each purified Ab listed above for 30 min at 4°C. Unbound Ab was removed by two washes, and cells were incubated with 250 μl each of anti-rat and anti-mouse IgG MagaBeads (Cortex Biochem, San Leandro, CA) for 45 min at 4°C. Ab-bound cells were removed using a magnetic concentration system (Dynal Biotech, Oslo, Norway). Enriched CD8⁺ T cells from DUC18 mice contained <5% total of cells staining positively for anti-CD4 (GK1.5), CD19, CD11c, and CD11b. Before use, the percentage of live DUC18 T cells in each preparation was determined by flow cytometry as being propidium iodide negative, CD8⁺, and Vβ8.3⁺. For adoptive transfer studies, total cell numbers were adjusted so that 3 × 10⁶ enriched DUC18 T cells/mouse were injected i.v. in 200 μl of sterile PBS.

DCs were harvested from spleen and LNs after mechanical disruption of organs with a syringe plunger and mesh screen. Homogenates were digested for 60 min at 37°C in 20 ml of HBSS containing 0.55 Wuensch units collagenase (Liberase Blendzyme 3; Roche, Indianapolis, IN) and 10 μg/ml DNase I (Sigma-Aldrich, St. Louis, MO), with vortexing every 10 min. Cells were washed twice in RPMI 1640 plus 10% FCS, filtered, and resuspended in PBS plus 10% FCS and 1 mM EDTA. After a 15-min room temperature incubation, cells were centrifuged and resuspended in FACS buffer before analysis.

For in vitro proliferation, 1 × 10⁵ enriched CD8⁺ T cells/well from DUC18 mice were cultured in 96-well, flat-bottom plates. T cells were cultured in RPMI 1640 with 10% FCS, 2 mM Glutamax, 5 × 10⁻⁵ M 2-mercaptoethanol (Sigma-Aldrich), and 50 μg/ml gentamicin. Cells were left unstimulated or were stimulated with plate-bound Abs (8 μg/ml anti-CD3ε 2C11 plus 10 μg/ml anti-CD28). Cultures were incubated for 2 days, then pulsed with [³H]thymidine at 0.4 μCi/well for the final 18 h, harvested, and counted on a beta counter. Each treatment was performed in triplicate. For in vivo proliferation, enriched naive CD8⁺ T cells from DUC18 mice were labeled with 5 μM CFSE (Molecular Probes, Eugene, OR) and injected at 3 × 10⁶ cells/200 μl of sterile PBS i.v. Recipient mice were young or aged BALB/c mice that had been challenged 4 days previously with CMS5 cells or were left unchallenged as controls. Tumors, draining LNs (right inguinal), nondraining LNs (pooled left inguinal, axillary, and brachial), and spleens were harvested 72 or 96 h after cell transfer. Tumors were collagenase-digested as described above for DC analysis. Cell divisions within the live DUC18 T cell population (defined as being CD8⁺/Vβ8.3⁺/CFSE⁺) were analyzed by four-color flow cytometry on a FACSCalibur (BD Biosciences, San Diego, CA). In this way, endogenous CD8⁺/Vβ8.3⁺ cells in recipient mice were excluded from analysis. The expression of activation markers was determined in some samples using CD8-allophycocyanin, Vβ8.3-PE, and CD25-biotin or CD62L-biotin plus SA-CyChrome instead of propidium iodide.

We investigated the effects of aging on the generation of T cell-mediated immunity using our DUC18 T cell/CMS5 tumor model system. To begin these studies, we characterized the phenotype of DUC18 T cells present in our aged TCR transgenic mice. Although a number of phenotypic and functional alterations have been observed in aging T cell populations, many of these can be attributed to an accumulation of memory phenotype cells resulting from exogenous challenges (8, 38). DUC18 T cells possess a TCR specific for a unique CMS5-derived tERK presented by H-2K^d. Because of this, DUC18 T cells from aged mice should remain phenotypically naive.

Splenocytes and LN cells were obtained from young and old DUC18 mice and stained for the markers CD8 and Vβ8.3 to identify the specific DUC18 T cell population. In seven separate experiments, we found a decreased percentage of DUC18 T cells in aged mice relative to young mice (Fig. 1,A), which also translated into decreased total numbers of DUC18 T cells per mouse (Fig. 1,B). Expression levels of CD8 and Vβ8.3 on DUC18 T cells were unaffected by age (Fig. 1,C). These observations highlight the need to perform in vivo functional analyses with equal numbers of young and old tumor-specific DUC18 T cells. Gated DUC18 T cells from young mice were found to be CD45RB^highCD62L^high and CD25⁻ (Fig. 1,D), illustrating their naive status. Some aged mice did show slight increases in the numbers of DUC18 T cells with down-regulated CD62L or up-regulated CD25 (Fig 1,D, second row). Overall, however, the vast majority of DUC18 T cells in aged mice are phenotypically naive, as had been reported in a study using comparably aged CD8⁺ 2C TCR transgenic mice (18). Because adoptive transfer studies used enriched populations of CD8⁺ T cells from DUC 18 mice that contained both transgene-positive and transgene-negative cells, we also examined the phenotype of bulk CD8⁺ T cells from young and old DUC18 mice (Fig. 1 D, bottom two rows). As expected, we did find increased numbers of memory phenotype cells in the CD8 T cell compartment as a whole, but even in aged mice these constituted only 10–20% of the cells analyzed.

We next assessed the proliferative capacity of CD8⁺ T cells from aged DUC18 transgenic mice in vitro. Equal numbers of enriched CD8⁺ cells from young or old mice were cultured in the presence or the absence of plate-bound anti-CD3 and anti-CD28. Because we were performing negative selections and our preparations contained varying amounts of transgene-positive and -negative cells, we used Abs to stimulate all CD8 cells. As shown in Fig. 1 E, aged CD8⁺ T cells displayed ∼50% the proliferation seen in their young counterparts at 72 h. Before plating, we calculated the percentages of aged and young DUC18 T cells present in each cell preparation, and so were able to calculate the amount of proliferation contributed by these subsets. In each of four independent experiments, aged DUC18 T cells proliferated less well than young DUC18 T cells in response to anti-CD3 and anti-CD28, with the mean calculated value being 43% (data not shown). These observations differ from a recent report that demonstrated that QL9 peptide-driven in vitro proliferation of CD8⁺ 2C T cells was unaffected by age (18). Our own studies using tERK to assess Ag-driven proliferation of young and aged DUC18 T cells in vitro were inconclusive (data not shown).

We next focused on investigating whether aging altered the in vivo function of APCs and CD8⁺ T cells to such an extent that it affected the establishment of protective antitumor immunity. The age of the host is known to affect the rate of tumor growth. Some tumors demonstrate enhanced outgrowth kinetics in aged mice, whereas others grow more slowly or are not influenced by host age (39, 40). Therefore, before initiating adoptive transfer studies, we needed to determine whether CMS5 outgrowth was altered in aged vs young BALB/c mice.

When equal numbers of CMS5 tumor cells were injected into young and old recipients, no differences in the rate of tumor outgrowth were observed (Fig. 2). This is important, because before assessing the ability of young and aged DUC18 T cells to reject established tumors, we needed to ensure that tumor loads in young and old recipient mice were equivalent at the time of DUC18 T cell transfer. In this model system, transfer of 3 × 10⁶ naive DUC18 T cells into mice on day 4 post-tumor challenge reproducibly results in tumor rejection by day 12 (35). During this period of time, the outgrowth of CMS5 tumors in young and old BALB/c mice is identical, thereby providing equivalent tumor burdens in both groups of mice.

Naive T cell activation in vivo is dependent upon APC function, particularly that of DCs (41, 42). Because contradictory data exist regarding the effects of age on APCs, we examined APC phenotype and T cell stimulatory capacity in our aged BALB/c hosts. We focused the phenotypic analysis on DCs and found no substantial differences in the percentages of CD11c⁺ cells expressing the myeloid DC marker CD11b or the lymphoid DC marker CD8α in young and old BALB/c mice (Fig. 3, dot plots). Slight differences in surface marker expression were detected on young and aged DCs. For example, as shown in Fig. 3, the percentage of CD11c⁺ DCs expressing H-2K^d is comparable in young and old mice (86% young vs 84% old; Fig. 3), but a greater percentage of DCs from aged mice express I-A^d (55% young vs 72% old; Fig. 3). CD86 expression appears comparable between young and old DCs (Fig. 3). Therefore, no consistent pattern of altered DC subset composition or surface marker expression was detected between young and old mice.

We next asked whether endogenous APCs retain the ability to stimulate naive tumor-specific T cells. To accomplish this, we challenged both young and old BALB/c mice with CMS5 tumors on day 0. On day 4, CD8⁺ T cells from young DUC18 mice were enriched, labeled with CFSE to allow for visualization of cell divisions, and adoptively transferred into tumor-bearing recipients. The extent of in vivo proliferation was assessed in DUC18 T cells from tumor-draining LNs (dLNs), control nondraining LNs (cLNs), spleens, and tumors on day 7. In this way, the abilities of young and old APCs to induce proliferation in vivo in the same population of young T cells could be compared.

Young DUC18 T cells harvested from the dLNs of young and old mice displayed equivalent proliferation profiles (Fig. 4,A), indicating that APC function in aged mice is comparable to that seen in young mice. When the data from six independent experiments were combined, we found that the percentages of live young DUC18 T cells present in dLNs from young and old mice were similar, providing further evidence that DUC18 T cell expansion, and therefore APC function, was unaltered in old mice (Fig. 4,B). No proliferation of DUC18 T cells was observed post-transfer into tumor-free control mice (data not shown). A survey of cLNs, spleens, and tumors revealed that a small percentage of cells in these organs were composed of live DUC18 T cells (Fig. 4,A). Although we observed a general trend in which fewer DUC18 T cells trafficked to these sites in aged mice, these differences were not statistically significant (Fig. 4,B). We also examined activation marker expression on proliferating DUC18 T cells in dLNs as another measure of their functional status and found no substantial differences in the levels of CD25 (Fig. 4,C, upper panels) or CD62L expression (Fig. 4 C, lower panels). Assessment of CD11c⁺ DC and CD11b⁺ macrophage infiltration of tumors by immunohistochemistry showed similar infiltration of both cell types throughout tumors in young and old mice (data not shown). These findings imply that during the aging process, APCs retain the ability to enter tumors, migrate to dLNs, and present tumor-derived Ags to initiate T cell activation.

Given that APC function appeared intact in aged mice, we next wanted to determine whether in vivo DUC18 T cell activation and proliferation were affected by the process of aging. Based on our in vitro studies and other previously published reports documenting impaired CD8⁺ T cell expansion in vivo (19, 20), we predicted that aged DUC18 T cells would exhibit fewer rounds of proliferation after their transfer into aged, tumor-bearing mice. To test this, we challenged young or old BALB/c mice with CMS5 tumors on day 0. On day 4, enriched CD8⁺ T cells from DUC18 mice were CFSE-labeled and transferred into recipient mice (young DUC18 T cells into young BALB/c mice and aged DUC18 T cells into aged BALB/c mice). Proliferation in response to tumor-derived Ag was evaluated on day 8. Because APC function in vivo appeared unaltered by aging, we did not test all possible cross-combinations of young and old donor cells and recipients.

Surprisingly, when we examined in vivo proliferation within the live DUC18 T cell population, we found it was not impaired in aging mice (Fig. 5,A, upper panels). We calculated the percentages of live DUC18T cells present in dLNs from young and old mice in four independent experiments and found that whereas slight differences were present, these were not statistically significant (Fig. 5,B). As controls, DUC18 T cells transferred into tumor-free BALB/c mice did not proliferate (data not shown). We again observed that greater numbers of live, activated T cells were present in tumors from young mice than in tumors from old mice, with the difference shown being ∼3-fold (Fig. 5,A, lower panels). Interestingly, data analysis revealed that this trend of impaired trafficking to tumors in aged mice was sufficiently pronounced to reach statistical significance (Fig. 5,B). Aged DUC18 T cells also displayed impaired trafficking to the spleen (Fig. 5,B). Expression patterns for CD25 on proliferating young and old DUC18 T cells were similar (Fig. 5,C). In dLNs from some young mice, an accumulation of CFSE^intermed/lo/CD25^low cells was observed compared with dLNs from aged mice (Fig. 5,C), although this trend was not always present. No age-related deficiencies in the down-regulation of CD62L on proliferating DUC18 T cells were detected (data not shown). Thus, in contrast to our own in vitro data (Fig. 1 E) and earlier in vivo studies measuring the expansion of viral-specific CD8⁺ T cells in aged mice (19, 20), we found no intrinsic defect in the proliferative capacity of aging CD8⁺ T cells in this model system. We find this to be true even when aged T cells were adoptively transferred into aged, tumor-bearing recipients, again highlighting the observation that in vivo APC function appears to remain intact during aging.

To evaluate the effector function of young and old DUC18 T cells, we monitored the kinetics and extent of tumor rejection seen after their transfer into BALB/c mice. Previous work in our laboratory established that 3 × 10⁶ young naive DUC18 T cells are needed to reject tumors in 100% of young mice challenged with CMS5 cells 4 days previously (35). In our current experiments, BALB/c mice were challenged with CMS5 cells on day 0, and 4 days later enriched CD8⁺ T cells from young DUC18 mice were transferred into young recipients, and enriched CD8⁺ T cells from old DUC18 mice were transferred into old recipients. When 3 × 10⁶ DUC18 T cells were transferred, all young and aged mice rejected tumor challenges (data not shown). Because it was possible that transfer of this many DUC18 T cells was overwhelming the system and masking any subtle alterations in aged T cell effector function, we titrated back the numbers of DUC18 T cells used for adoptive transfers. With the transfer of only 1 × 10⁶ enriched young DUC18 T cells, tumors were rejected in ∼70% of young recipient mice (Table I, experiments 1 and 2). Using this lower dose, we again found no loss in the ability of aged DUC18 T cells to reject day 4 established CMS5 tumors in aged mice (Fig. 6,A and Table I). The kinetics of rejection were similar (Fig. 6,A), and we actually detected a greater overall percentage of old mice demonstrating complete tumor eradication out to 30 days (Table I).

Prior reports had suggested that the generation or function of memory CD8⁺ T cells may be defective in aged mice (18, 19, 32). To test the recall response in our system, we rechallenged young and aged mice that had rejected their tumors ∼12 wk earlier. We found that 100% of both young and old mice (a total of 11 young mice and nine aged mice) were able to reject these secondary challenges, and all mice remained tumor free during the 30-day monitoring period (Fig. 6 B). Therefore, our studies suggest that antitumor CD8⁺ T cell effector function, as well as memory responses are not affected by the process of aging when equal numbers of tumor-specific T cells are present.

Aging is known to be associated with a decline in the magnitude of T cell-mediated immune responses, especially within the context of infections and poor responses to vaccinations in the elderly. Impaired T cell function may also contribute to the increased incidence of cancer seen in mice and humans as they age. In the current study we used a transplantable tumor/TCR transgenic mouse model to examine the effects of age on CD8⁺ T cell activation and effector function in vivo. We evaluated both the ability of endogenous APC populations in aged mice to prime naive T cells and the ability of known numbers of adoptively transferred old T cells to become activated and eliminate tumor challenges. To our knowledge, this is the first study to examine the function of controlled numbers of naive phenotype aged CD8⁺ T cells in vivo, particularly within the context of antitumor immunity. We found no inherent defects in the in vivo function of APCs or CD8⁺ T cells as a consequence of aging.

Specifically, our results suggest that Ag-driven activation of naive CD8⁺ T cells and acquisition of effector function remain largely intact during the aging process. These findings are in agreement with recent work by Li et al. (18) that demonstrated in vitro responses to Ag were equivalent in young and old naive CD8⁺ T cells. Interestingly, our own in vitro experiments showed a substantial decrease in aged CD8⁺ T cell proliferation in response to plate-bound Abs (Fig. 1 E). The specific DUC18 T cell population also demonstrated a loss of Ab-mediated proliferative capacitywith age (data not shown). Ab stimulation does not recapitulate the myriad interactions between T cells and APCs. It is therefore possible that our in vitro data reflect a deficiency in signal transduction downstream of the TCR that is present in aged CD8⁺ T cells, but that is compensated for during T cell/APC interactions by augmented signals derived from costimulatory molecules. In support of this idea, numerous defects in TCR-mediated signal transduction have been described in aged CD4⁺ T cells (43, 44), and APCs from aging individuals have been noted to display enhanced T cell stimulatory capacity in some studies (26).

The critical role played by APCs in initiating naive T cell activation as well as discrepancies in the literature regarding effects of age on APC function prompted us to investigate possible deficiencies in endogenous APCs in our system. Although we did detect some slight differences in the percentages of myeloid vs lymphoid DCs and surface expression of MHC molecules in DCs from young and old mice (Fig. 3), these appear to have no physiologic consequence. Our experimental strategy of transferring equal numbers of young DUC18 T cells to young or old tumor-bearing mice allowed us to directly investigate the functional capacity of aged vs young APCs to prime CD8⁺ DUC18 T cells against tumor Ags. We found that APCs in aged mice initiate antitumor T cell responses to the same degree as APCs in young mice (Fig. 4). Even when old T cells were transferred into old mice, they expanded and rejected tumors equivalently to young T cells that were stimulated in vivo by young APCs (Figs. 5 and 6). Additionally, a preliminary evaluation of DCs from young and old tumor-bearing mice revealed no striking alterations in DC subset composition or activation state between age groups (data not shown). Therefore, we conclude that even if differences in the number, phenotype, or Ag presentation capabilities of endogenous APCs in old mice are present, they are not substantial enough to alter the course of naive T cell activation in vivo. This idea is supported by several previous studies that looked at aged APC function in vitro and found no defects in their ability to stimulate T cell proliferation (21, 24). It has been proposed that APCs from elderly individuals who are frail, rather than healthy, may indeed possess a loss of T cell stimulatory capacity (45), but we did not address this hypothesis in our study. It is possible, however, that age alone may not be the critical determinant influencing APC phenotype and function in the elderly.

The one discrepancy we found repeatedly was a decreased number of live DUC18 T cells present in tumors and spleens from aged mice. We observed this trend when either young (Fig. 4) or old (Fig. 5) DUC18 T cells were transferred into old tumor-bearing BALB/c mice, although the differences were only statistically significant with the transfer of old DUC18 T cells into old recipients. The reasons for this alteration in trafficking are not currently known. Data presented in Fig. 4 suggest that aged APCs may be unable to induce the full repertoire of T cell effector functions; T cell proliferation is intact, but perhaps acquisition of chemokine receptors on effector T cells is deficient. Alternately, APC function may be completely unaffected by age, and some extrinsic factor(s) is altered in aged mice that can influence migration of activated T cells. Obvious candidates would be chemokines such as inducing protein-10, monokine induced by IFN-γ, or I-TAC, and we have not yet addressed these possibilities. The data presented in Fig. 5 suggest that the aged T cells themselves may also be deficient in their ability to traffic normally, and this in conjunction with extrinsic alterations may lead to the presence of fewer live DUC18 T cells in spleens and tumors from aged mice.

Although we did detect fewer activated DUC18 T cells within tumors in aged mice, this did not translate into an observed loss of tumor rejection in our system (Fig. 6 and Table I). It is possible that either aged T cells acquire enhanced cytolytic capabilities on a per cell basis, or they become less susceptible to activation-induced cell death, and so possess a net gain of effector function due to enhanced longevity within tumors. Previous in vitro studies by others have shown that aged CD8⁺ T cells are deficient in activation-induced cell death (46). However, we analyzed the percentages of tumor-infiltrating DUC18 T cells that were dead and dying, as measured by propidium iodide positivity, and found no differences between young and old T cells (data not shown). For both T cell types, the majority of DUC18 T cells isolated from tumors were found to be dead or dying (data not shown). We do not yet know whether differences exist in the cytolytic abilities of aged and young tumor-infiltrating DUC18 T cells on a per cell basis.

Earlier reports on antitumor immunity in aging mice described profound impairments with age (28, 29, 30, 31). These studies examined endogenous polyclonal T cell responses, probably mediated by both CD4⁺ and CD8⁺ T cells. In contrast, our study was designed to evaluate antitumor responses mediated by controlled numbers of CD8⁺ T cells specific for a single tumor-associated Ag. When CD8⁺ DUC18 T cells are present in excess, they can eliminate CMS5 tumor challenges independently of CD4⁺ T cells (34). However, when limited numbers of DUC18 T cells (as few as 100-1000/mouse) are present, CD4⁺ T cell help is required for optimal CMS5 rejection (47). In our current study we used transfers of at least 1 × 10⁶ DUC18 T cells to study the effects of aging on CD8⁺ T cell function. Therefore, another factor that may account for the discrepancy between our current findings and previous reports is the relative dependency of CD8⁺ T cells on CD4⁺ T cell help. Additionally, our system uses mice on a BALB/c background, whereas most previous reports used C3H/HeN mice (28, 30, 31). Because of this, strain-specific variations in mice could also contribute to the deviation of our present findings from prior reports. Further differences could be caused by the tumors themselves. Ultimately, we feel that the combined effects of impaired CD4⁺ Th cell function and decreased numbers of tumor-specific CD8⁺ T cells are the most likely factors that led to the drastic reductions in tumor immunity previously described in aging mice.

Our studies were performed in mice bearing small transplanted fibrosarcomas. As such, there are many obvious and important differences between our system and the environment in which T cells must combat tumors in aging cancer patients. Even so, we believe that our findings have positive implications for the use of T cell-based immunotherapeutic strategies in these patients. It is known that substantial population shifts occur in the human CD8⁺ T cell compartment with aging. Oligoclonal expansions and accumulations of memory phenotype, CD28⁻CD8⁺ T cells decrease the numbers of naive T cells available to counteract novel antigenic challenges (6, 7, 8). Our model is not influenced by these factors, so we were able to examine aged CD8⁺ T cell function on a cellular basis. We found that when equal numbers of naive phenotype aged and young tumor-specific T cells are present, the aged T cells expand and reject previously administered tumors to the same degree as do young T cells (Figs. 5 and 6). The finding that Ag-driven proliferation of aged CD8⁺ T cells remains intact (18), even in vivo (Fig. 5), implies that tumor Ag-driven in vitro expansion of tumor-infiltrating lymphocytes from aging cancer patients could be successful. Therefore, even when fewer tumor-specific naive T cells are present initially, in vitro manipulation might be able to compensate for this and allow equal numbers of activated cells to be transferred back into patients. Known deficiencies in CD4⁺ T cell function could be targeted by specific strategies such as 41BB cross-linking (15) or exogenous addition of IL-2 (16).

Protective T cell-based immunotherapies also have the potential to generate memory responses against subsequently arising metastases in patients. Although the numbers of mice used were small, our results imply that CD8⁺ T cell memory against tumor Ags remains intact during aging (Fig. 6 B), again provided that sufficient numbers of naive T cells are present to undergo expansion and differentiation to the effector state. This is critical, because primary expansion is known to directly influence the magnitude of the memory response generated (48). A previous report examining viral immunity in aged mice had demonstrated that the initial expansion of naive LCMV-specific CD8⁺ T cells was defective, so fewer memory T cells were generated, but these maintained recall responsiveness out to 5 mo at a level similar to that seen in young mice (19). Thus, our work supports their finding: that memory CD8⁺ T cells derived from aged effectors function as well as their counterparts derived from young effector T cells.

Overall, results from our model system suggest that age alone does not adversely affect the in vivo antitumor function of endogenous APCs or transferred tumor-specific CD8⁺ T cells. This implies that, in contrast to CD4⁺ T cells, the main age-related defects in the CD8⁺ T cell compartment exist at the population level and not at the level of individual naive phenotype CD8⁺ T cells. Therefore, although fewer naive CD8⁺ T cells remain in the elderly, targeted Ag-specific expansion of these cells could allow for the development of protective immunity not only to tumors, but also to viral agents that commonly cause health problems in this population.

We thank Darren Kreamalmeyer for his expert management of the mouse colony, and Jerri Smith for manuscript preparation. We thank Leigh O’Mara, Silvia Kang, and Drs. Laura Mandik-Nayak, Fei Shi, and Ken Matsui for their critical review of the manuscript.