Naive T Cell Maintenance and Function in Human Aging

Aging, defined as progressive functional decline over time, affects all organ systems and is the major cause of, or at least contributes to, most diseases in the adult. The immune system is a prime example; immune competence declines with age, causing increased morbidity and mortality from infections, as well as being a factor in the increased incidence of malignancies (1–3). Less intuitively, the aging immune system is also more inclined to elicit nonspecific inflammation, which accelerates degenerative diseases, most prominently seen in cardiovascular and neurodegenerative disorders (4–6). Moreover, immune aging can impair tolerance mechanisms and is a risk factor for autoimmunity (7, 8). Generally known as “immunosenescence,” this term is too narrow to reflect the multitude of mechanisms involved and may even be misleading, implying cellular senescence as the main pathological event.

To describe our current understanding of the aging process in its complexity, López-Otín et al. (9) define cellular and molecular hallmarks that describe common pathways that, in turn, signify aging over a range of tissues and species: stem cell exhaustion limiting regenerative capacity; various forms of genomic instability, including telomere attrition, DNA damage, mitochondrial dysfunction, and epigenetic changes; loss of proteostasis; nutritional sensing; cellular senescence; and altered intercellular communication (Table I). In this review, we discuss how these general aging mechanisms help explain age-associated changes in the immune system and, conversely, how studies on T cell aging can expand this conceptual framework. We focus exclusively on human naive T cells and refer to recent broader reviews for comprehensive reading on immune aging (10–14).

As pointed out by López-Otín et al. (9), a decline in regenerative capacity is a well-appreciated hallmark of aging, and attrition of stem cells with age is a universal finding in virtually all tissues (Table I). To prevent stem cell exhaustion, mechanisms are in place to preserve cell quiescence (15). Failure of these mechanisms leads to premature exhaustion and accelerates the aging process. The adaptive immune system is special, in that generation of novel naive T cells is entirely dependent on thymic function. Because thymic output peaks at puberty and progressively declines thereafter, thymic involution may be independent of and precede stem cell aging. The naive T cell emerges as a quasi–stem cell regenerating the T cell system, and principles of stem cell aging apply to naive T cell aging.

The dramatic loss of the thymus prompted a natural supposition that thymic involution is responsible for the age-associated failure of the adaptive immune system (16, 17). Indeed, the naive T cell compartment in the mouse is dependent on thymic emigrants throughout life. Insufficient production of new cells by the thymus during aging is associated with compartment shrinkage and eventually leads to holes in the murine T cell repertoire (18, 19). Several lines of recent evidence have challenged the importance of thymic involution in human immune aging (20). Although vital for building a T cell repertoire during the growing phases of the host, thymic output appears unnecessary for repertoire maintenance during adulthood, and T cell regeneration is nearly entirely derived from homeostatic proliferation of the existing T cell pool, which is sufficient to maintain a large compartment of naive CD4 T cells (Fig. 1) (21).

Surgery removing or reducing the thymus in early childhood changes the composition of the T cell compartment, mimicking immune aging in young adults, but only in individuals chronically infected with CMV (22). Similarly, the relative expansion of memory and effector T cell populations and the relative decline in naive CD4 T cells, attributed to age in earlier studies, is entirely due to chronic CMV stimulation and not found in older individuals who do not have Abs to CMV (23). More importantly, data on T cell turnover and decline in TCR excision circles (TRECs) best fits to the model that T cell generation in the adult relies mostly, if not entirely, on homeostatic proliferation of existing naive T cells. Despite a decrease in daily thymic output by a factor of 10–100 during adult life, peripheral naive T cell turnover does not measurably increase, suggesting that that peripheral proliferation does not need to compensate for thymic failure and that even in early adulthood, thymic output contributes little to the maintenance of the size of the naive T cell compartment (24–26). Only in the eighth and ninth decade of life does the frequency of cycling cells increase, probably as a consequence of increased cell death rather than lack of thymic function (27, 28).

Peripheral expansion of naive T cells is driven by tonic TCR signals and homeostatic cytokines, in particular by IL-7 (29, 30). IL-7 concentrations do not decline with age and therefore do not become limiting, consistent with the observation that the CD4 T cell compartment does not reduce in size. IL-2 stimulation may also play a role because naive CD4 T cells increasingly express the high-affinity IL-2 receptor CD25 with age and become IL-2 responsive (31). IL-7 drives T cell proliferation without inducing a switch to a memory phenotype in vitro (32). CD31, expressed on a subset of naive CD4 T cells, identifies recent thymic emigrants or at least cells that have not peripherally proliferated. TREC concentrations decrease slightly in CD31⁺ CD4 T cells with age, consistent with only limited replication in this population (33). As one would expect, the CD31⁺ population progressively shrinks with age, consistent with the idea that increasingly more T cells participate in proliferation and lose CD31 (33, 34). In parallel, the CD31⁻ naive T cell population increases in size and its TREC content more drastically declines (33). The loss of CD31 expression in naive T cells is generally not associated with the acquisition of phenotypic markers of memory T cells.

In contrast to CD4 T cells, the naive CD8 T cell compartment shrinks with age (Fig. 1) (23, 35). The reason for this subset difference is unclear. Theoretically, homeostatic proliferation conditions are not worse for CD8 T cells. On the contrary, CD8 T cells should receive more signals from abundant MHC class I than CD4 cells receive from restrictedly expressed MHC class II molecules. Indeed, compared with CD4 T cells, naive CD8 T cells develop a higher degree of clonal size inequality with frequent clonal expansions, indicating that naive CD8 T cells do proliferate at a higher rate than CD4 T cells (Fig. 1) (36). Shrinking of the naive CD8 T cell compartment may therefore not be due to defective regeneration. Alternatively, naive CD8 T cells may acquire phenotypic memory markers due to increased turnover. In murine models, such virtual memory T cells preferentially develop in the CD8 compartment (37, 38). However, TCR sequencing studies do not support the notion that development of CD8 virtual memory T cells is a frequent event in older humans. Clonal expansions originating in the naive CD8 T cell compartment largely maintain a naive phenotype and only a small fraction of these clones express memory cell markers. Moreover, TCR diversity in CD8 memory T cells is ∼10-fold lower than that of CD4 memory T cells, which would not be expected if a larger fraction of CD8 than CD4 naive T cells is converted into T cells with a memory phenotype (36).

Alternatively, increased proliferation may be associated with increased cell death, causing depletion of naive CD8 T cells. In modeling naive T cell division and survival, Reynolds et al. (39) proposed that increased IL-7–driven proliferation, possibly due to a lower threshold for IL-7 receptor signaling, leads to abrupt declines in compartment size several years later. Taking this thought one step further, tonic TCR and homeostatic cytokine stimulation need to be optimally tuned to find the balance between providing sufficient signals for survival and replacement without triggering increased turnover (38). Excess cytokines, such as those produced in inflammatory conditions, increased expression of cytokine receptors such as CD25, or lowering of TCR and cytokine receptor thresholds may be detrimental for homeostasis and accelerate immune aging, similar to stem cells where cellular quiescence is important for delaying stem cell exhaustion and aging (9, 40). Accordingly, accelerated aging has been described in several autoimmune diseases (7, 8).

Obviously, these data are based on measurements of PBLs, which represent <3% of the total T cells in the body. A recent study by Thome et al. (41) quantified the T cell subset composition at nine anatomic tissue sites in 56 individual organ donors over a wide age range. Naive T cells were only found in blood and lymphoid tissue. The distribution variance of naive CD4 and CD8 T cells between circulation and lymphoid tissue was low. The impact of age on naive CD8 T cell decline was similar in blood and tissue, whereas the blood frequencies of naive CD4 T cells may underestimate an age-associated decline in the lymphoid tissue. However, these tissue data show relative frequencies and may therefore reflect increased frequencies of T effector memory cells, in particular because CMV⁺ donors were not excluded. In conclusion, homeostatic proliferation appears to be sufficient for maintaining a sizable naive CD4 T cell compartment whereas the naive CD8 compartment declines in size, possibly as a consequence of increased cell loss rather than defective generation (Fig. 1). A lack of regenerative capacity and a decline in immune-competent cells does not appear to cause the immune defects in adaptive immune responses, in particular the CD4 T cell–dependent B cell responses after vaccination.

T cell replenishment is unique in that it not only has to maintain compartment size, but also TCR diversity. Unlike in B cells, generation of new T cell specificities entirely relies on thymic production and cannot be accomplished by expansion and modification of naive T cells. For T cell homeostatic mechanisms to maintain a diverse repertoire, at least three conditions must be met: 1) the initial clonal size in the naive T cell compartment must be of sufficient size to confer robustness to loss of T cells; 2) the initial TCR repertoire must be diverse and redundant, such that loss of some clones does not generate holes in the repertoire; and 3) peripheral T cell homeostasis must be nonselective. The first two factors are highly interdependent and are influenced by the size of the species; conditions for mice and humans are therefore very different. Initial clonal size depends on intrathymic proliferation after TCR rearrangement, as well as the peripheral proliferation within the compartment that is at least partially empty, that is, in the fetus as well as during the subsequent growth period up to early adulthood. Data on increased naive T cell proliferation in newborns support the notion that recent thymic emigrants proliferate in the periphery to establish a minimal clonal size. More than 10% of cord blood T cells of infants born prematurely in the third trimester, when the human thymus starts to be active, are in the cell cycle (42). By fetal maturity this frequency has declined, but it is still much higher than in young adults. Data on naive T cell turnover in children are not available, but clonal expansion of thymic emigrants may reach clonal sizes of 100 cells, a number that is consistent with TREC measurements in CD31⁺ T cells that presumably have proliferated little or not at all after the initial seeding from the thymus (33). To determine how age affects TCR diversity in the adult, we performed an in silico simulation based on current estimates of kinetic parameters and clonal sizes (43). Even under the extreme scenario that thymic production stops at age 20 y and compartment size shrinks, we found that diversity is only mildly reduced over lifetime, as long as homeostatic proliferation is not highly selective.

Recent progress in next-generation sequencing has allowed testing of these predictions and estimating the overall richness of the naive TCR repertoire. Initial estimates hovered around a few million different TCRβ-chains, an estimate of the same magnitude as in the mouse, but much lower than those derived from concentrations of TRECs (44–46). The major challenge of these sequencing studies continues to be extrapolating from the small analyzed sample to the entire compartment of close to 10¹² T cells while not excluding infrequent sequences as possible PCR or sequencing errors. Using a nonparametric analysis and multiple sampling, we arrived at a much higher richness estimate of close to 100 million different naive TCRβ-chain sequences in young adults, suggesting that the human TCR repertoire is extremely diverse and it would require a massive contraction with age to be functionally relevant (36). However, we only found a modest 2- to 5-fold contraction in repertoire richness for both the naive CD4 and CD8 compartments when we analyzed replicate samples of highly purified naive T cells from lymphopheresis samples of 70- to 85-y-old individuals. Although these individuals are certainly above average in their health status, these data document that peripheral homeostatic mechanisms are able to sustain a diverse repertoire. Based on these studies we predict that thymic involution does not have a detrimental influence on TCR diversity with age (Fig. 1). Differences in study design may explain why our conclusion contradicts several earlier and also more recent studies that found a larger degree in diversity decline with age. Most of the studies that describe a major contraction in the repertoire based their analyses on total T cell population instead of purified naive and memory T cell subsets. The observed repertoire contraction could therefore only reflect a lower proportion of naive cells in the sample (46). Moreover, richness in older individuals is often underestimated because clonal sizes become increasingly non-Gaussian–distributed with clonally expanded T cells occurring even within the naive repertoire. The increased clonality may result in a low estimate of richness when the repertoire is not analyzed in sufficient depth. Estimates from studies, in particular, from the era prior to next-generation sequencing, therefore reflect the increased clonality while not allowing conclusions on richness (28).

Increased clonality is frequent with higher age, in particular, in the naive CD8 compartment (Fig. 1); some naive T cell clones can reach a size usually only seen for memory T clones (36). Clonal selection may occur due to higher avidity to self-antigens, which in fact has been shown in animal studies (47), or increased responsiveness to cytokines, which we have shown in simulation experiments, but which has not been studied in vivo. The increasing unevenness in clonal sizes in older individuals is likely to bias the repertoire of T cells responding to a new antigenic challenge; more frequent T cell clones will outcompete less frequent specificities. In principle, such a mechanism would be similar to what has been described as “original antigenic sin” where T cell responses are biased toward previous encounters of related Ags, although in this case the bias would derive from uneven homeostatic proliferation, presumably due to self-recognition (48). One possible complication from excessive clonal expansion is a collapse of the repertoire diversity as predicted by in silico simulation (43). Moreover, owing to the increased fitness of peripherally selected clonotypes, ongoing thymic activity or thymic rejuvenation would not be efficacious to prevent or restore such a scenario because new thymic emigrants will not be able to compete (49). We predict that healthy immune aging will require prevention of increasing clonality. Because clonality is variable, valuable information should be learned from understanding why clonality occurs more in some older individuals than in others.

Attrition of telomeres is an accepted hallmark of aging and is generally attributed to replicative history (Table I). Similar to stem cells, naive T cells are able to express telomerase, but their telomeres shorten with age (50). The current interpretation is that stimuli triggering homeostatic proliferation are not sufficient to induce telomerase expression, and that telomere shortening is a consequence of homeostatic turnover. It is intuitive to postulate that telomere attrition at least in part accounts for defective T cell responses in older individuals (51). The ability to exhibit a proliferative burst with >10 cell divisions within a short time frame is an absolute requirement for an effective naive T cell response, and telomeric erosion will curtail this burst. Keeping turnover low as long as it is sufficient to replace cell loss is therefore desirable and a prerequisite of healthy aging, with the additional benefit of avoiding clonality as discussed above. However, a replicative history does not appear to be the only or major cause for telomere attrition. In early studies by Weng et al. (52), age-associated telomere attrition in naive T cells paralleled that of memory T cells, although the kinetics of these two populations are vastly different. Moreover, telomere length did not differ between CD31⁺ and proliferated CD31⁻ naive T cells. In addition to being sensitive to replication, telomeres are also subject to DNA damage, even more so than genomic DNA. Because DNA repair mechanisms at telomeres are limited by shelterin (a telomere-specific protein complex) (53), DNA damage tends to persist and could be responsible for age-associated telomere attrition irrespective of proliferation.

Accumulation of genetic damage in general is a characteristic hallmark of aging in various tissues and settings (Table I) (9, 54, 55). Its relevance is highlighted by the finding that many premature aging syndromes are caused by underlying defects in DNA repair pathways (56). DNA damage responses in naive CD4 T cells have not been widely studied in the context of aging, with the exception of the accelerated immune aging in patients with rheumatoid arthritis that has been linked to defects in telomerase induction after activation and defects in the MRN–ATM DNA repair pathway, both of which rendered naive CD4 T cells susceptible to apoptosis (57–60). There is evidence that the defects identified in patients with rheumatoid arthritis are also relevant for physiological aging. DNA damage, as quantified by comet assay, gradually increases with advancing age in memory T cells from healthy controls. In comparison, DNA damage remains low in naive CD4 T cells up to 65 y, and increased DNA damage comparable to memory T cells is only found in some individuals in their seventh decade of life. The consequences of damage to mitochondrial DNA (mtDNA), also a hallmark of aging, have been explored even less for T cell aging (61, 62). mtDNA is particularly sensitive to cumulative DNA damage for a number of reasons. It is exposed to reactive oxygen species, whose major producer is the mitochondrion; it lacks protective histones; and its repair mechanisms are less efficacious.

Taking a wider view, telomeric erosion, mitochondrial dysfunction, in part caused by mtDNA damage, and DNA damage responses are closely intertwined with cell metabolism (63, 64). Deregulated nutrient sensing is associated with organismal aging in many model systems (65). The insulin/insulin-like growth factor (IGF)-1 signaling pathway has been noted as a major aging-controlling pathway that is active in various species throughout evolution (9). In contrast, regulation of metabolic pathways in T cells is less dependent on availability of insulin or IGF-1, but integral to T cell function, complicating the application of metabolic concepts in general aging to immune aging (Table I). The decisions of whether aerobic glycolysis or oxidative phosphorylation predominates and which fuel is being used are strictly linked to T cell activation and differentiation (66, 67). Similar to tumor cells, aerobic glycolysis provides the building blocks enabling proliferation and T cell expansion, an essential element in particular for the naive T cell response. Metabolic pathways are closely intertwined with signaling events after TCR stimulation; for example, reactive oxygen species produced as a result of mitochondrial activity play a key role in TCR-induced signaling cascades. Moreover, many signaling molecules involved in T cell activation are also involved in the regulation of metabolic activity. Equally important, autophagy is a key mechanism controlling T cell differentiation and survival (68, 69). Because the metabolic pathways and their relationship to signaling pathways in basic T cell biology are increasingly understood, it can be expected that these insights will be fertile ground for immune aging studies. However, the scope of such studies will likely be different from the nutrition sensing studies in general aging research that try to curtail metabolic activity to prolong longevity, as exemplified with calorie restriction. Although data on naive T cells are lacking, initial studies have proven the value of understanding metabolic pathways to explain defects in T memory cell function (70). In elderly CD4 memory T cells, the increased expression of dual specific phosphatase 4 that terminates nuclear ERK signaling and impairs proliferation and effector functions is the result of increased activation of the metabolic master regulator AMP-activated protein kinase (71). In the case of CD8 effector T cells that lack the expression of CD28, but regain CD45RA and that have features of cellular senescence, inhibition of p38 signaling induced autophagy and restored energy-dependent functions (72). Finally, at least shown for murine T cells, T cell aging is associated with a defect in chaperone-mediated autophagy (69).

Cellular senescence is a classical aging hallmark (Table I). In addition to stable cell cycle arrest, senescent cells show characteristic phenotypic and functional changes, including the production of inflammatory cytokines (73). Although senescence was originally attributed to telomeric erosion, non–telomeric DNA damage can also trigger it by derepressing the INK4/ARF locus. The INK4A/ARF locus comprises overlapping reading frames of two genes, p16^INK4a and p19^ARF, that play important roles as cell cycle inhibitors in regulating cell growth and senescence and are mutated in various forms of cancer (74, 75). In contrast to other forms of cellular senescence, p16-induced senescence is not associated with an inflammatory phenotype (76). Derepression of INK4/ARF and transcription of p16 are seen with age in a variety of tissues (77). T cells are no exception to this rule and even show higher expression of p16 than do most other cells (78). However, p16 expression is also induced after T cell activation and appears to be a physiological mechanism to curtail clonal expansion, particularly in an immune response of naive T cells (79). Whether naive T cells develop features of senescence in the classical sense is undetermined. As discussed above, there is no evidence that T cell turnover decreases with age; if anything, it increases in the very old. Important for the topic of this review, the evidence that increased constitutive expression of p16 in T cells with age includes naive T cells is only indirect and inconclusive (78). Older individuals may accumulate effector T cells, in particular in the setting of chronic latent infection such as CMV. These cells are, by nature of their differentiation state, less prone to proliferation but are otherwise fully functional (80, 81). T cell exhaustion, as found with chronic, actively replicating viral infections, is characterized by the expression of cell surface molecules such as PD1 and TIM3 on memory and effector cell populations and very different from cellular senescence (82).

Naive T cells have a unique feature that is not appreciated as a general aging hallmark for all tissues, but which they share with stem cells (Table I). To stay truly naive (or to remain a fully pluripotent stem cell), naive T cells as well as stem cells need to avoid differentiation. Stem cells develop an altered lineage potential with age: children and young adults have a dominant lymphoid lineage commitment in contrast to older individuals who are biased toward myeloid differentiation. Changes in lineage commitment and partial differentiation in stem cells are related to DNA damage responses and the expression of the transcription factor BATF (83). Similarly, naive T cells can differentiate without having encountered exogenous Ag (37). Su et al. (84) described that healthy individuals carry memory T cells to Ags they have never encountered in life. These cells may be an example of cross-reactivity rather than the result of nonspecific differentiation of naive T cells. However, virtual memory CD8 T cells have been repeatedly identified in mice (85, 86). What drives the clonal expansion of these cells in addition to recognition of self-antigen is not known, nor is it clear if they are also present in CD4 T cell populations or even at all in humans. TCR sequencing studies have shown that clonal expansion occurs in the human naive T cell compartment, and more so for CD8 than for CD4 T cells (36). Most such clonally expanded cells keep a naive phenotype, with expression of CD45RA, CD28, CD27, and CCR7; however, a small fraction of some clones also expresses typical memory markers. On a broader scale, expression of CD25 on naive CD4 T cells and CD95 on a subset of naive CD8 T cells with age indicates partial activation and/or differentiation (26, 31). Transcriptional and, even more so, epigenetic profiling could address whether naive T cells have begun to differentiate with age and have transcriptional signatures reminiscent of memory genes or an epigenetic landscape in which T effector genes are poised. Surprisingly, conclusive gene expression studies examining the impact of age on naive T cells in humans are missing; most of the array studies have either analyzed total CD4 and CD8 T cells or focused on memory T cells, where phenotypic changes of CD28 loss and expression of NK cell surface markers with age have sparked particular interest (87). The same is true for epigenetic studies. Alterations in DNA methylation patterns have been described with age, but only for unfractionated CD4 T cells (88). It remains to be determined whether they reflect T cell differentiation or age-associated changes in naive and/or memory T cells.

Supporting evidence for the partial differentiation hypothesis comes from identification of signaling alterations in older T cells (89). Reduced miR-181a expression is at least in part the cause for subdued TCR signaling in naive cells with age. miR-181a controls the expression of DUSP6 and possibly other phosphatases. Loss of miR-181a with age therefore increases DUSP6 activity, which in turn curtails ERK phosphorylation after TCR stimulation (90). Interestingly, miR-181a expression is regulated along a developmental pathway, being highest in thymocytes and lowest in memory and effector T cells (91). Because TCR signaling in memory T cells relies less on the LAT–ERK pathway and more on the hDlg–p38 pathway (92), the lower concentration of miR-181a is not so important for memory cell activation, but it is detrimental for aged naive T cells. To determine whether age-related changes in differentiation-dependent miRNA species are a general theme for naive T cell aging, we profiled two other miRNAs that are known to change in expression with T cell activation and differentiation. Both miR-146a and miR-21 are expressed at higher concentrations in memory cells than in naive T cells (93). Our preliminary data suggest that expression levels in naive CD8 T cells are age-dependent, with older naive T cells more closely resembling memory T cells (M.M. Cavanagh, C.M. Weyand, and J.J. Goronzy, unpublished observations). These data suggest that aging affects miRNA species that regulate T cell differentiation pathways, and that old naive T cells acquire features of a memory cell without maturing completely. Partial differentiation is an aging hallmark that is shared between stem cells and naive T cells (Table I). Obviously, such partial differentiation may have detrimental consequences because the naive T cells may have lost what is essential for their function, such as the ability to rapidly proliferate and clonally expand, while not fully having gained memory functions such as the rapid expression of poised effector genes. Epigenetic studies are needed to determine whether loss of quiescence and entering the differentiation pathway to memory or effector T cells is an important component of naive T cell aging.

Because thymic T cell production only minimally contributes to T cell generation throughout adult life, naive T cells have to be self-sufficient in regenerating themselves to maintain compartment size. Additional complexity comes from the need to also preserve a diverse repertoire. On a positive note, the system is quite effective in maintaining a diverse repertoire with a huge number of different TCR sequences that should be sufficient to recognize all possible antigenic peptides. However, peripheral fitness selection causes expansions of selected clones that can bias T cell responses. On the cellular level, rules governing the aging of naive T cells are also in part similar to those of stem cells. Telomere attrition occurs in both cell types despite their ability to express telomerase, likely due a combination of cumulative DNA damage and replicative history. Non–telomeric DNA instability also contributes to naive T cell aging; at least in patients with rheumatoid arthritis, defective DNA repair mechanisms in naive CD4 T cells contribute to the accelerated immune aging seen in this disease. Equally or maybe even more important than cellular senescence or exhaustion, incomplete differentiation toward memory and effector T cells in part explains the defects in functions characteristic for naive T cells. In fact, senescence and differentiation pathways are not independent. Regulation of clonal expansion in naive T cells and differentiation into memory T cells involves pathways also concerned with cellular senescence and DNA damage responses. Preliminary evidence suggests that symmetries between T cell aging and differentiation also exist for regulatory programs controlled by miRNA expression. How far this concept can be expanded to include other aging hallmarks, in particular in the recently emerged fields of T cell metabolism and autophagy, remains to be seen. Interventions to keep naive T cells quiescent would therefore be equally beneficial to prevent functional decline as well as repertoire distortion due to clonal expansions.

The authors have no financial conflicts of interest.