Two Separate Defects Affecting True Naive or Virtual Memory T Cell Precursors Combine To Reduce Naive T Cell Responses with Aging

Infectious diseases remain among the leading causes of morbidity and mortality in older adults. T cells, critical for defense against intracellular pathogens, are profoundly affected by age (reviewed in Refs. 1, 2). Importantly, differences in the composition and maintenance of the T cell pool in mice are observed with aging in the absence of immunization (reviewed in Ref. 3). These changes result from an incompletely understood interplay of 1) reduced naive T cell production caused by thymic involution; 2) lifetime use of the existing naive T cells to respond to infections, including persistent latent infections; and 3) homeostatic mechanisms that normally attempt to balance and maintain T cell pools, but toward the end of life often distort an already reduced and diminished naive T cell pool (4, 5). Functional consequences of these changes for immune defense remain to be fully elucidated.

A diverse TCR repertoire is important for optimal protective responses to a variety of pathogens; holes in the TCR repertoire can result in reduced, absent, or ineffective immune responses (reviewed in Refs. 6, 7). The TCR repertoire becomes constricted with aging, but the extent, mechanisms, target populations, and the consequences for immune defense of this constriction remain unclear. Decreased thymic output requires naive CD8 T cells to rely upon homeostatic mechanisms to maintain the peripheral T cell pool, which may be particularly important in humans (8), and we understand relatively little about how the homeostatic mechanisms may change with aging.

We have reported that aging leads to >70% reduction of Ag-specific T cell precursors in unimmunized old mice, and that many of the remaining Ag-specific cells acquire a central memory–like CD44^hiCD62L^hiCD11a^hiCD127^hiCD122^hi phenotype and the immediate responsiveness to TCR ligation by IFN-γ secretion (9). Moreover, some of these precursors were preferentially maintained and survived, and then dominated the response to infection in old mice (9). Cells of the corresponding phenotype in adult mice were named “virtual memory” (VM) cells and were shown to respond to stimulation by superior proliferation and effector function compared with naive T cells in young animals (10). Because these cells persisted in germ-free adult mice (10) and responded briskly to IL-7 and IL-15, the authors concluded that they likely are generated/maintained by homeostatic cytokines.

In this study, we examined the rules guiding long-term maintenance of naive cells and the emergence of VM cells in unimmunized old mice. Naive Ag-specific precursors are very rare in unimmunized mice and are generally further reduced with aging to as few as a few 10s per animal, severely limiting experimental analysis. We therefore initially used TCR transgenic (TCRTg) mice, which provide abundant copies of a single clone of naive T cells, and validated the results in wild-type (wt) mice. Our results demonstrate that an age-related increase in frequency of VM T cells occurs in TCRTg mice, and that aging directly curtails the proliferation capacity and thus the potential immune defense ability of VM precursors in both TCRTg and wt mice. In contrast, proliferative ability of true naive T cells (TNa) was intact but their numbers were drastically reduced with aging. We discuss the existence of several subsets of naive (deemed naive due to lack of exposure to cognate Ag) CD8 T cells, which are differentially maintained with aging.

Mice were bred and maintained in the animal facility at the University of Arizona, and experiments were conducted under guidelines and approval of the Institutional Animal Care and Use Committee of the University of Arizona. B6.OT-I.Rag-knockout (KO) × B6.Ly5-1 F₁, B6.P14.Rag-KO × B6.Thy1.1 F₁, and B6.gBT-1.Rag-KO × B6.Thy1.1 F₁ mice were bred from the stocks of B6.OT-I.Rag-KO (11), B6.OT-II (12), B6.Ly5-1, and B6.PL mice, purchased from Taconic, the National Cancer Institute, and The Jackson Laboratory, respectively, and from P14 (13) and gBT-I (14) stocks provided to us by Dr. H.P. Pircher via Dr. J.A. Frelinger and by Dr. F.R. Carbone, respectively. Old (18–23 mo) C57BL/6 (B6) mice were purchased from the National Institute on Aging, and adult (12 wk) B6 mice were purchased from The Jackson Laboratory. Mice with large spleens or other obvious abnormalities (e.g., tumors) were excluded from the study. Survival bleeds were performed retro-orbitally.

Samples were prepared as previously described (15). We used fluorochrome-conjugated Ab clones: CD8α (53-6.7), CD4 (RM4-5), CD44 (IM7), CD62L (MEL-14), Vβ5.1/5.2 (MR9-4), Vα2 (B20.1), IFN-γ (XMG1.2), CD3 (17A2), TCRβ (H57-597), CD5 (53-7.3), PD-1 (29F.1A12), LAG3 (C9B7W), CD122 (TM-β1), IL-4R (mIL4R-M1), phospho-ser¹³⁹ γH2AX (2F3), and CD49d (R1-2), purchased from eBioscience, Invitrogen, BD Biosciences, or BioLegend. Samples were analyzed on a four-laser custom Fortessa cytometer (BD Immunocytometry Systems, Sunnyvale, CA) using the FACSDiva acquisition (BD Immunocytometry Systems) and FlowJo (Tree Star, Ashland, OR) analysis software. For intracellular cytokine staining, samples were stimulated and prepared exactly as previously described (15).

Prior to sorting, CD8 T cells were enriched by magnetic separation (Miltenyi Biotec) using an autoMACS Pro. CD44^hi VM and CD44^lo TNa CD8 T cells were sorted based on CD44 and CD62L expression using a BD FACSAria. The sorted cells were stained with CFSE as described (15). The cells were cultured at 8 × 10⁵ cells/ml in RPMI 1640 complete medium plus 100 U/ml recombinant murine (rm)IL-2 (eBioscience), with either 10⁻⁹ M SIINFEKL peptide or 10 ng/ml rmIL-12 (R&D Systems) and 10 ng/ml rmIL-18 (eBioscience) for 5 d. Alternatively, cells were cultured with 10 ng/ml IL-7 (eBioscience) and 100 ng/ml IL-15 (eBioscience). For polyclonal stimulation, cells were cultured at 8 × 10⁵ cells/ml in RPMI 1640 complete medium 1:1 with anti-CD3/anti-CD28 immobilized on beads (Miltenyi Biotec) and 100 U/ml rmIL-2 (eBioscience). On each day of analysis, cells were Live/Dead stained (Invitrogen) according to the manufacturer’s instructions, stained with surface Abs, fixed/permeabilized with Foxp3 Fix/Perm, and stained with intracellular Ab. For cell cycle analysis, cells were stained with Vybrant DyeCycle Orange (Life Technologies) at 1.25 × 10⁻³ mM for 60 min at room temperature and analyzed immediately. Cells were enumerated using CountBright absolute counting beads (Life Technologies) according to the manufacturer’s instructions.

Five thousand CD44^hi and CD44^lo cells were sorted from OT-I mice as described above. RNA isolation was performed by standard chloroform/isopropanol purification. For cDNA synthesis, 15 μl containing 10 μl RNA (500 ng–5 μg), 1.5 μl oligo(dT)₂₀, 1.5 μl 10 mM dNTP, and 2 μl H₂O was incubated at 65°C for 5 min and then at 4°C for 2 min. Three microliters 10× PCR buffer, 6 μl 25 mM MgCl₂, 3 μl 0.1 M DTT, 1.5 μl RNase Out (Life Technologies), and 1.5 μl SuperScript III reverse transcriptase (Life Technologies) were added and incubated at 50°C for 50 min and then at 85°C for 5 min. RNaseH (1.5 μl; Life Technologies) was added and the mixture was incubated at 37°C for 20 min. TCR α variable (TRAV) primers and PCR reaction were as in Kedl and Mescher (16) using Platinum Taq (Life Technologies). Each primer was used separately at 5 μM final concentration. PCR conditions were 95°C for 5 min followed by 40 cycles at 95°C for 20 s, 56°C for 20 s, and 72°C for 45 s, followed by a final extension at 72°C for 5 min. PCR products were visualized on a 1.8% agarose gel, and band intensities were calculated using Quantity One software (Bio-Rad). Bands were enumerated by blind scoring by three independent examiners, who scored the presence or absence of bands.

The tetramer-enrichment protocol was performed exactly as in Rudd et al. (9) using the spleen, inguinal, cervical, and axillary lymph nodes harvested from individual mice.

Following the analysis of flow cytometry data using FlowJo software, GraphPad Prism was used for statistical analysis (unpaired Student t test, paired Student t test, linear regression and best fit line, and one-way ANOVA with a Bonferroni posttest). A p value <0.05 was considered statistically significant.

We examined phenotypic changes with aging in a single, large clone of unimmunized CD8 T cells using OT-I Rag⁺ mice, expressing a transgenic TCR specific for OVA_257–264:H-2K^b (11). Blood from 10-, 12-, 18-, and 22-mo-old unimmunized OT-I mice contained progressively increased CD44^hi T cell fraction compared with 2- to 4-mo-old mice (Fig. 1A), and this was paralleled by an increase in IFN-γ–producing cells responding to the cognate peptide (Fig. 1B). Large variability was observed in VM conversion in individual 10- to 12-mo- old mice (blood CD8CD44^hi, 10–100%). However, with advanced age (in the rare OT-I mice surviving to 18-22 mo) we found uniform and high representation of CD44^hi cells (> 75%; Fig. 1A). In the experiments below we often used mice younger than the National Institute on Aging–defined age cutoff of 18 mo, because 1) OT-I mice, in our colony, have a median lifespan of 15–16 mo (not shown; colony B6 mice, 26–30 mo), which renders OT-I mice >18-mo-old extremely scarce; 2) 10- to 14-mo-old OT-I mice showed a range of VM conversion from <20 to > 70%, allowing us to analyze the results of functional and other assays in individual animals and test whether the VM conversion correlated with functional responsiveness (Fig. 1C); finally, 3) interventions to improve immunity would be implemented not only in old age, but likely also, if not preferentially, in middle-aged individuals that still exhibit reasonable immune function. Importantly, all of our key observations were reproduced in bona fide old TCRTg and/or wt mice.

VM cells were reported to be of central memory (CM) CD44^hi62L^hi phenotype (9). Likewise, CD44^hi cells in TCRTg mice were overwhelmingly CD62L^hi (Fig. 1D), and CD44^hi62^lo effector memory (EM) cells did not significantly accumulate with age (Fig. 1E). Indeed, frequency of VM, but not EM, cells tightly correlated with percentage IFN-γ⁺ production by CD8 T cells in response to the OVA peptide (Fig. 1C and data not shown; p < 0.0001 and p = 0.19, respectively). We also determined the frequency of TCRTg cells that express CD49d, also known as α₄ integrin, which associates with β₇ (α₄β₇ or LPAM) and binds to VCAM-1 and fibronectin, allowing T cells to enter inflamed tissue (16). Recently, it was shown that the CD49d^hi phenotype marked “true” memory T cells arising from Ag stimulation, whereas memory cells formed by homeostatic proliferation (VM) were CD49d^lo (10).We found that most aged OT-I Rag⁺ CD44^hi62L^hi VM CD8 T cells are CD49d^lo (Fig. 1E), suggesting that they are not true, foreign Ag-driven memory, arising instead from alternate mechanisms. In contrast, ∼30% of EM CD8 T cells (CD44^hi62L^lo) were CD49d^hi, consistent with the explanation that at least a portion of this population had reacted to an Ag.

We did not observe reproducible and significant lymphopenia of CD8 T cells with aging (for 12- to 14-mo-old mice, two of four experiments showed mild lymphopenia in blood and one of three experiments also in the spleen; for 18- to 22-mo-old mice, zero of one experiment showed lymphopenia in blood or spleen). Collectively, these data show that TCRTg VM cells exhibiting the CM phenotype increase with age in absolute terms.

VM cells in both adult (10) and old (9) mice exhibit strong immediate effector function, unlike their TNa CD44^lo counterparts. We found that the age of OT-I CD8 T cells and the CD44^hi phenotype correlated tightly with the acquisition of immediate cytokine production in these cognate Ag-inexperienced cells upon peptide stimulation (Fig. 1C). These results show that with aging TCRTg T cells exhibit generalized increased phenotypic and functional conversion into VM cells, similar to T cell precursors in aging wt mice (9).

Even in unimmunized wt or germ-free adult mice, 10–20% CD8 cells convert into the CD44^hi phenotype, likely due to stimulation by IL-7 during the neonatal period in the periphery (10, 17) and stimulation by NK-T cell–derived IL-4 of single-positive T cells in the thymus (18). To assess whether TCRTg VM cells show signs of differential cytokine maintenance, we examined expression of CD122, CD127, and IL-4R on TCRTg VM cells. We found variable expression of the IL-4R on VM cells in TCRTg mice, and the expression of CD127 (IL-7Rα–chain) did not reproducibly correlate to the VM phenotype (not shown). By contrast, 14-mo-old OT-I mice showed a significantly increased frequency of splenic CD8 CD122⁺ (IL-2/15Rβ–chain) T cells compared with 5–mo-old animals (40–50 versus 10–20%, Supplemental Fig. 1A), and the mean fluorescence intensity (MFI) of CD122 expression was also higher (on average by 2.5-fold; Supplemental Fig. 1B). When compared between the VM and TNa subsets, expression of CD122 on aged CD44^hi VM cells was the highest (Supplemental Fig. 1C, 1D). However, even the CD44^lo cells of a 14-mo-old animal began to express low levels of CD122 (Fig. 2D; 8- to 10-fold lower than in 14-mo-old VM cells, but still higher than in 5-mo-old counterparts, which were CD122⁻ by flow cytometry). This paralleled our findings in wt cells (11), indicating that CD122 MFI correlated tightly to the extent of VM conversion. That finding is consistent with the possibilities that 1) homeostatic cytokines are playing a key role in VM conversion in TCRTg CD8 cells, or 2) that VM conversion was mandated by other signals (TCR), and that activation of CD122 expression was part of a memory cell differentiation program.

Naive precursors in young mice rely upon trophic signals from both TCR (recognition of self pMHC ligands) and homeostatic cytokines for maintenance and survival (19). These parameters may change with time: the self or environmental pMHC universe (e.g., food Ag, normal flora-derived Ag) could evolve with aging owing to differential colonization, aberrant gene expression, and other factors; likewise, homeostatic cytokine availability, T cell repertoire, abundance (particularly loss of naive T cells (4, 5), and cellular responsiveness to key homeostatic cytokines (9) could all be altered by aging.

We evaluated the expression of TCRTg chains relative to the acquisition of the VM phenotype. Transgenic TCRVβ5 was stably expressed between 4 and 18 mo of age (not shown). Young (4 mo old, Supplemental Fig. 2A) OT-I CD8 T cells were almost exclusively (>99.8%) Vα2^hi and exhibited a relatively low (<20%) and stable fraction of CD44^hi cells (Fig. 1). However, older OT-I mice contained a significant fraction of Vα2^int and Vα2^lo populations (Supplemental Fig. 2A). Loss of Vα2 was accompanied by retention of overall TCR expression (pan-TCRβ staining, Supplemental Fig. 2A), suggesting that these CD8 T cells expressed another TCRα-chain and another potential TCR specificity, a scenario that physiologically occurs on up to ∼30% of human and 15% murine T cells (20–22). Importantly, the Vα2^lo/int phenotype correlated significantly with the CD44^hi phenotype in 12-mo-old OT-I mice (Supplemental Fig. 2C).

To examine directly the role of TCRα replacement, we sorted CD44^lo (TNa) and VM cells from mice of different ages in the presence or the absence of the Rag recombinase and performed PCR amplification of different TRAV gene rearrangements. As expected, 14-mo-old OT-I Rag-KO mice exhibited a dominant TRAV14 band (encoding the TCRVα2 protein) and very few other bands (Fig. 2A); importantly, there was no difference in the number of bands in Rag-KO mice regardless of the age or the CD44 phenotype (Fig. 2D), indicating that these mice, as expected, cannot rearrange endogenous TRAV genes (bands on the gels that are appearing in a few TRAV families are artifacts of the PCR). Moreover, we found comparable low levels of endogenous TCRα rearrangements in 3-mo-old VM samples, suggesting that young adult VM cells also do not exhibit pronounced endogenous rearrangements (Fig. 2B), in accordance with the flow cytometry analysis of protein expression (Supplemental Fig. 2). TNa cells from 14-mo-old OT-I mice also did not carry more rearrangements compared with their younger counterparts (Fig. 2C, top). In contrast, VM cells at 14 mo of age contained rearrangements in many different TRAV families (see example in Fig. 4C, bottom). When quantified across groups of animals (by blind examination of three independent examiners) VM cells from old Rag-sufficient OT-I mice exhibited significantly increased expression of alternative TCRα families when compared with all other populations analyzed (Fig. 2D).

To examine whether the presence of secondary TRAV rearrangements impacted the frequency of VM cells across the lifespan, we aged different TCRTg strains on a Rag-KO background. Up to 19 mo of age, the three different TCRTg Rag-KO mouse strains examined—OT-I, P14 (23), and gBT-I (14)—did not downregulate the Vα2 chain and also maintained a small (<5%) VM population that did not increase with age (Fig. 2E, filled bars), consistent with the results of Haluszczak et al. (10). This sharply contrasted with progressive, age-related VM accumulation in Rag⁺ OT-I mice, reaching >50% of all cells (Fig. 2E, open bars). Therefore, we conclude that secondary TCR rearrangements and, therefore, the TCR-mediated signals are essential for the age-related dominance of VM CD8 cells with, but are not necessary for the generation of, neonatal VM cells in TCRTg mice.

We next examined functional responses of VM CD8 T cells with aging. Because VM cells exhibit increased expression of homeostatic cytokine receptors (Supplemental Fig. 1), we tested whether these cells were more sensitive to IL-7/IL-15. Sorted, >98% pure VM (CD44^hi) and CD44^lo CD8 T cells from adult and old OT-I Rag⁺ mice were labeled with CFSE and cultured with IL-7/IL-15 for up to 7 d. At both 5 d (not shown) and 7 d of culture (Fig. 3) VM (CD44^hi) CD8 T cells from both 3- and 14-mo-old mice proliferated better than did their TNa (CD44^lo) counterparts based on percentages (Fig. 3A, 3B) and counts (not shown) of cells reaching fourth division and above. Moreover, compared with their 3-mo-old counterparts, both VM (Fig. 3C) and TNa (Fig. 3D) cells from 14-mo-old mice exhibited significantly higher propensity to divide four or more times in response to IL-7 and IL-15. Therefore, VM OT-I cells exhibit increased sensitivity to homeostatic cytokines, which increased with age and could contribute to their age-related accumulation.

Our results (Fig. 2) indicated that VM cell phenotype and accumulation in old mice must be driven/maintained by TCR signals. During a lifetime, even low-level proliferation/turnover could exert cumulative effects on old T cells. To examine functional impact of this interaction upon TCR-driven functions, sorted >98% pure VM and TNa cells from 3- and 14-mo-old OT-I mice were stimulated in vitro with the cognate SIINFEKL peptide, or with IL-12 plus IL-18, to elicit IFN-γ secretion and proliferation. An increased proportion of VM OT-I T cells rapidly produced IFN-γ in response to both types of stimulation, compared with TNa counterparts (Fig. 4A, 4B), as was seen in WT VM cells (9, 10). Moreover, aging again accentuated this phenotype, because 14-mo-old VM cells produced IFN-γ at a much higher frequency (∼30%) compared with their 3-mo-old counterparts (5–7%; Fig. 4B).

As seen in young adult VM cells from wt mice (10), young adult TCRTg VM CD8 T cells exhibited significantly enhanced proliferation when compared with TNa cells at both day 3 (not shown) and day 5 of culture (Fig. 4C). Importantly, 14-mo-old (Fig. 4D) or older (not shown) VM cells exhibited delayed and reduced entry into advanced cell divisions, and these cells were clearly outperformed by old TNa cells in proliferation on both days 3 (not shown) and 5 after stimulation (Fig. 4D). Collectively, these results suggest that aging leads to selective proliferative impairment; old OT-I VM cells proliferate more than did their TNa counterparts in response to homeostatic cytokines (likely explaining their gradual dominance with time) but are less capable of proliferating in response to cognate peptide (TCR) stimulation. To assess whether the proliferation difference between bulk adult and old naive T cells may be due to differential behavior of TNa and VM cells, we directly compared adult and old VM (Fig. 4E) and TNa cells and found no significant differences between adult and old TNa cells (Fig. 4F). We were surprised that old TNa cells proliferated similarly to their adult counterparts. However, the current literature never distinguished between VM and TNa cells when comparing proliferation between adult and old T cells (24–27).

We next sought to address why old VM cells did not proliferate as extensively as did old TNa cells by examining cell cycle progression and survival in the course of proliferation. In response to peptide stimulation, we saw no difference in the ability of VM cells to enter into cell cycle or to cross to different phases of the cycle when compared with TNa cells (Fig. 5A). However, when we measured viability, we found increased cell death in old VM cells when compared with TNa cells (Fig. 5B, 5C), which corresponded to very few live VM cells by day 3 (Fig. 5D). In contrast, we found that VM cells stimulated with IL-7 and IL-15 exhibited superior survival and reduced cell death when compared with TNa cells (Fig. 5E), which resulted in increased cell numbers (Fig. 5F). These results strongly support our findings in Fig. 3, suggesting that survival advantage in response to cytokines and proliferative capacity in response to antigenic peptide may be dissociated in old VM cells.

We further investigated proliferative characteristics of the old VM cells in response to cognate TCR stimulation in relationship to their in vivo accumulation. These cells are more sensitive to homeostatic cytokines IL-7 and IL-15 (Fig. 3) and have increased expression of CD122 (Supplemental Fig. 1). Because alternative TCRs may also play a role (Fig. 2), we were interested in whether old VM cells could proliferate to polyclonal stimulation. We stimulated sorted splenic old OT-I Rag⁺ VM and TNa CD8 T cells with anti-CD3 and anti-CD28 and found old VM cells proliferated similarly to TNa cells (Fig. 6A), and in fact had significantly increased frequency of cells in later cell divisions. This is likely due to increased viability; that is, we found that old VM cells stimulated with anti-CD3 and anti-CD28 exhibit increased viability compared with VM cells stimulated with peptide (Fig. 6B).

One could argue that many of the above observations could be an artifact of the OT-I TCRTg model. To assess whether the above findings from TCRTg mice also extended to wt mice, we isolated different Ag-specific CD8 T cells from unimmunized adult (2–3 mo of age) and old (18–23 mo of age) mice using the tetramer enrichment method (28) as in our prior work (9). As expected (9), the frequency of CD44^hi VM CD8 T cell precursors specific for B8R (poxvirus immunodominant epitope) (29), OVA, and the West Nile virus NS4b (30) increased with age (9) (Fig. 7A). Importantly, with age, all three types of precursors also exhibited significantly higher expression of the IL-2/IL-15Rβ–chain (CD122; Fig. 7B), consistent with findings from aged OT-I mice (Supplemental Fig. 1). We have also found that old wt VM cells proliferate worse than TNa counterparts both in vitro and in vivo (K.R. Renkema, G. Li, M.J. Smithey, and J. Nikolich-Žugich, manuscript in preparation). Collectively, these results demonstrate the essential similarities found between the TCRTg and wt old VM T cell precursors.

Several additional mechanisms could contribute to cellular proliferative senescence, including exhaustion/anergy; an age-dependent increase in inhibitory and exhaustion receptors on CD8 T cells was reported (reviewed in Ref. 31), even on naive precursors in unimmunized mice (32). Although we observed a significant increase in the frequency of both PD-1⁺ (Supplemental Fig. 3A, 3C) and LAG3⁺ (Supplemental Fig. 3B) OT-I T cells at 14 mo of age compared with adults (Supplemental Fig. 3A, 3B), the representation of cells expressing these markers remained modest (>20% in OT-I 14-mo-old mice, with the exception of a single outlier, Supplemental Fig. 3A; and also in Tet⁺ 20-mo-old wt CD8 T cells, Supplemental Fig. 3C) . There were no significant changes in the expression of 2B4 on these cells from 2 to 14 mo of age or on wt T cells up to 24 mo of age (not shown). We conclude that these inhibitory markers are unlikely to be the sole (or perhaps even the main) explanation for the proliferation defects in aging VM precursors, particularly because we have seen them expressed exclusively on the EM (CD44^hi62L^lo) and not the CM/VM (CD44^hi62L^hi) fraction of CD44^hi CD8 T cell precursors (not shown). They could, however, contribute to the overall reduction in proliferative capacity in old VM precursors.

In this study, we made four discoveries germane to our understanding of aging and homeostatic regulation of naive T cell pools and their function. First, we found that TCR signals are key to VM conversion of TCRTg cells with aging, because fixing the original TCRTg specificity on the Rag-KO background prevented accumulation of VM cells with aging. This significantly strengthens and molecularly defines the idea (9) that there is TCR-based selection in the naive and VM T cell pool with aging. Second, we found that IL-15R (CD122) was increased with aging in both TCRTg and wt VM cells, and that this resulted in strong proliferation of such cells to IL-15, which overshadowed their TNa counterparts, providing a mechanism for their preferential outgrowth over TNa cells. Third, whereas the response to homeostatic cytokines was robust, old VM T cell precursors exhibited selective replicative impairment in response to TCR signals relative to TNa counterparts. Fourth, we found that this impairment was associated to increased apoptosis specifically in response to peptide stimulation, but decreased apoptosis in response to homeostatic cytokines in old VM compared with old TNa cells. All this provides strong evidence that the naive T cell compartment in aging fails to respond properly to challenge due to two different intrinsic defects: 1) drastic loss of relatively proliferatively intact TNa precursors, and 2) selective functional proliferative impairment of VM T cell precursors responding to TCR stimulation.

Our present results, in the context of data from the literature (9, 10, 17, 18, 32), shed new light on long-term maintenance of T cell precursors with aging. Collectively, there is evidence that there may be at least four distinct pools of naive (indicating lack of prior contact with cognate Ag) CD8 T cell precursors in old mice and that their maintenance with age is differentially regulated. The TNa CD44^loCD8 precursors are severely depleted by the process of aging (9, 32), suggesting that either the key maintenance factors for these cells are insufficient in old animals, or that age-related lymphopenia drives these cells to convert to VM cells, or both. However, these cells retain much of the proliferative function in middle-aged (12 to 14 mo old) TCRTg mice (this study) and in old wt mice (K.R. Renkema, G. Li, M.J. Smithey and J. Nikolich-Žugich, manuscript in preparation). The IL-4–dependent innate memory cells (18, 33) arise in response to IL-4 produced by NK-T cells and are prominent in BALB/c but not B6 thymi. IL-4 deficiency in young B6 mice reduces by ∼30–40% the Ag-specific precursor pool (17). Therefore, the IL-4 innate memory cells contribute to the peripheral pool in B6 mice. Their functional properties in adult and old mice remain to be elucidated, although we did not notice a significant increase in IL-4R⁺ cells with aging (not shown). These cells are absent in TCRTg mice and therefore were not the subject of the present study. TCRTg Rag^−/− mice in our study had a small (<5%) age-insensitive population of VM cells that are most analogous to the cells that expand during the neonatal period in the periphery (17). These VM cells were showed to be independent of IL-4 (17) and are most likely expanded by IL-7 immediately upon egress into an empty periphery of a neonate (10, 17). We provisionally name these cells IL-7VM. Of interest, whereas the overall percentage and number of VM cells in TCRTg Rag⁺ mice increased with aging, this was not the case with the IL-7VM cells in Rag^−/− mice.

That led us to conclude that the original specificities of the TCRTg receptors analyzed in this study were not conducive to age-related VM accumulation on their own, and that VM cells in TCRTg mice depend on the secondary rearrangements that produce other TCR specificities. T cells expressing dual TCRs could be reactive to alternate Ags (34, 35). These cells, perhaps due to self-reactivity, gut flora, or other environmental Ag reactivity or cross-reactivity (36), exhibited the VM phenotype, and we call them here aging-related VM cells . Finally, whereas ∼50% of 10- to 14-mo-old OT-I T cells are CD44^hi (Fig. 1A), only ∼7–15% are Vα2^int or Vα2^lo (Supplemental Fig. 1), begging an explanation of why many of the remaining Vα2^hi cells still convert to VM. Because very little VM conversion occurs in TCRTg Rag⁺ mice, we speculate that many of the Vα2^hi cells also may express dual TCRs. Some have used the endogenous Vα2 to replace the transgenic Vα2 chain (the mAb used does not distinguish between these), whereas the others would express the secondary, endogenous Vα at levels below flow cytometry detection, but still sufficient to provide the VM-converting signal over time. Single-cell analysis will be necessary to address these possibilities. Of importance, our preliminary analysis suggests that secondary rearrangements also play a role in VM conversion in precursors isolated from unimmunized wt mice (K.R. Renkema and J. Nikolich-Žugich, unpublished observations).

Recently, it was shown that in young adult B6 mice deficient in secondary rearrangements (the TCRα^+/− genotype) VM cells persist at the same frequency as in wt mice (∼16% of the total naive Ag-specific precursors) (17). We would predict that upon aging such mice may exhibit reduced VM accumulation, but again sequencing of TRAV in a polyclonal TCRα^+/− and TCRα^−/− setting will be needed to establish whether dual TCRα rearrangements are necessary for VM accumulation in the polyclonal TCR repertoire.

In initial publications it was suggested that aging may not adversely affect TCRTg T cells; that is, old TCRTg T cells were found to be functionally comparable to adult T cells in a variety of strains (37–39) using bulk T cell functional assays. This stood in contrast to old T cells from wt mice, which have long been known to exhibit defects in proliferation and differentiation (40–42; reviewed in Ref. 31). These results can be now explained by the likely confounding effect of massive contamination with functionally mature VM T cells, which have not been separated and separately tested in these studies, and which would account for robust immediate effector function, as well as by the presence of sufficient numbers of TNa cells (due to the artifact of extremely high precursor frequencies in TCRTg mice), which would robustly proliferate in bulk assays.

Perhaps most importantly, our results show that VM conversion is not innocuous for an aging precursor. Whereas in youth VM cells exhibit functionally robust proliferation (10), by middle age their proliferative ability declines (this study) and is even worse in old wt mice (K.R. Renkema, G. Li, M.J. Smithey and J. Nikolich-Žugich, manuscript in preparation). Replicative senescence in vitro occurs owing to a finite number of cell divisions (10–60, depending on species and cell type) and eventual cell cycle arrest (43). In vivo existence and relevance of replicative senescence in lymphocytes remain controversial. In this study, we described a “selective” replicative impairment in old VM CD8 T cells, which exhibit robust proliferation in response to homeostatic cytokines (IL-7 plus IL-15) and can rapidly produce cytokines upon cognate peptide or inflammatory cytokine (IL-12 plus IL-18) activation, but proliferate significantly worse compared with their CD44^lo TNa counterparts when activated with cognate peptide. Selective survival differences in response to peptide stimulation and to IL-7/IL-15 homeostatic cytokines appear to underlie these observations, but additional work will be necessary to mechanistically dissect their signaling basis. The other potential culprit for blunted proliferative responses, the accumulation of inhibitory receptors, showed less impressive differences between VM and TNa cells, albeit its functional relevance remains to be tested. Reversing blunted lymphocyte responses with aging remains an implicit goal of this field, and achieving it by restoring costimulatory signaling, by increasing IL-2 production/responsiveness, and/or by targeting exhaustion receptors all remain viable individual or combination-based treatments. Overall, we conclude that not all flavors of precursors in unimmunized mice are the same, and understanding the attrition with age and functional capabilities of each one of them will be important for our understanding of potential immune intervention in older adults.

We thank Paula Campbell of the University of Arizona Flow Cytometry Core Facilty for expert cell sorting, the National Institutes of Health Core Tetramer Production Facility at the Emory University for outstanding reagent preparation, and past and present Nikolich Laboratory members for support and discussions.

The authors have no financial conflicts of interest.