From “truly naïve” to “exhausted senescent” T cells: When markers predict functionality



The study of T cell biology has been accelerated by substantial progress at the technological level, particularly through the continuing advancement of flow cytometry. The conventional approach of observing T cells as either T helper or T cytotoxic is overly simplistic and does not allow investigators to clearly identify immune mechanisms or alterations in physiological processes that impact on clinical outcomes. The complexity of T cell sub-populations, as we understand them today, combined with the immunological and functional diversity of these subsets represent significant complications for the study of T cell biology. In this article, we review the use of classical markers in delineating T cell sub-populations, from “truly naïve” T cells (recent thymic emigrants with no proliferative history) to “exhausted senescent” T cells (poorly proliferative cells that display severe functional abnormalities) wherein the different phenotypes of these populations reflect their disparate functionalities. In addition, since persistent infections and chronological aging have been shown to be associated with significant alterations in human T cell distribution and function, we also discuss age-associated and cytomegalovirus-driven alterations in the expression of key subset markers. © 2013 International Society for Advancement of Cytometry


The study of lymphocyte biology has now progressed far beyond the original concept that CD3+ T cells can be simply divided in CD4+ and CD8+ subsets that exhibit helper and cytotoxic functions, respectively [1, 2]. The development of increasingly sophisticated analytical techniques and the discovery of novel markers of distinct lymphocyte populations have literally reshaped the field T cell biology. Investigators must now take into account the extreme heterogeneity of the T cell pool together with the fact that interactions between the component subsets are far more complex than previously thought. Flow cytometry has been a key technology in the identification and characterization of T cell subsets [3]. With this critical advancement in single cell analysis, ever increasing numbers of markers appear able to differentiate and characterize distinct T cell populations, bringing with them increasing complexity and generating new questions regarding the biological relevance of these markers [2]. In this review, we summarize our current knowledge of the various markers used to identify and characterize T cell sub-populations, especially those that indicate distinct stages of differentiation and functionality. To this end, we will focus on the markers that delineate T cell subsets during their progression from antigen-naïve cells into populations that have entered replicative senescence and exhibit critical loss of function. Finally, we relate the presence of these populations to aging and chronic infections that have been linked with the expansion of late-differentiated T cell subsets.

Consensus on T Cell Markers Used for Phenotyping?

Classical Surface Markers of CD4+ and CD8+ T Cells

Multi-color flow cytometry allows single cell analysis of markers that are differentially expressed during T cell maturation, activation and differentiation, thereby shedding light on the distinct functional properties and relative senescence of these diverse populations [4]. The ability of T cells to differentiate into different types of memory cell is a critical determinant of host protection against pathogens and autoimmunity, and is thus a key driver of the extensive research interest in delineating sub-populations of T cells [5]. Infectious disease scientists are still actively engaged in defining the precise specificity of protective T cell responses, in increasing our understanding of immunological memory, and in identifying the mechanisms that maintain effective immune surveillance and control infections [6-9]. Antigen-specific T cells display very different phenotypic and functional profiles from their antigen-naïve counterparts. In immunological studies, several classical markers are commonly used to identify T cell sub-populations [4]. These markers may be considered to reflect T cell differentiation, activation and functional status. A similar set of classical markers can also be used to divide both the CD4+ and CD8+ T cell populations into distinct subsets, although the majority of the data on which this approach is based come from studies of the CD8+ T cell population alone [10]. Recent studies have shown that while CD4+ T cells do indeed share a significant number of markers with CD8+ T cells during their differentiation and activation, several key markers are in fact subset specific [11, 12]. Below, we will describe the classical markers used to describe the various CD8+ T cell sub-populations and highlight differences, when they exist, for the CD4+ T cells.

CD8+ T cells have been successfully sub-divided into four main populations using a limited number of surface markers, namely C-C motif chemokine receptor (CCR)−7 and CD45RA. These populations are designated as naïve cells (N), central memory cells (CM), effector memory cells (EM) and terminally differentiated effector memory cells re-expressing CD45RA (TEMRA), although several intermediate subsets may also exist in vivo [4]. This model also applies to CD4+ T cells, although the equivalent TEMRA sub-population has not yet been clearly identified, at least in healthy adults. In addition to these markers, the CD27 and CD28 co-receptors for T cell activation may be used as further indicators of cellular history. The surface markers typically used to define the four main sub-populations of T cells can be summarized as follows; N (CCR7+, CD27++, CD28++, CD45RA+), CM (CCR7+, CD27+++, CD28+++, CD45RA−), EM (CCR7−, CD27+/−, CD28+/−, CD45RA−) and TEMRA (CCR7−, CD27−, CD28−, CD45RA+) [4, 13-15]. Not only does the presence of co-receptors differ between these sub-populations but also the level of marker expression is variable (Table 1). This is particularly true for CD27, CD28 and CD45RA, and we discuss later how variation in the expression of these markers can in fact be very informative to investigators. The above-mentioned classification has received the unanimous approval of the scientific community based on the widespread application of this approach in publications in the field.

Table 1. Markers of human T cell differentiation
  1. The expression of surface markers associated with differentiation is presented for CD4+ and CD8+ T cells. The sub-populations were categorized based on differential expression of the markers (arbitrary definition) but also for how this is related to the current consensus (classical definition). N: naïve; M: memory; -: no/very low expression; +: unit of expression. Expression levels of the markers have been compared between sub-populations (rather than within the same sub-population).

Arbitrary definitionN1N2M1M1M2M2M3M4M4
CCR7 +++++

The classical combination of markers used to define naïve cell populations has proved very useful in tracking the turnover of naïve T cells and in assessing the loss of naïve T cell populations in the context of aging and infectious disease [16, 17]. However, CCR7+CD45RA+ “antigen-inexperienced” naïve T cells may not be entirely naïve in terms of proliferative history. Two distinct types of naïve T cell can be identified; recent thymic emigrants that have not encountered the cognate antigen for their specific T Cell Receptor (TCR), and cells that may perform immunosurveillance roles and undergo proliferation in response to homeostatic survival signals [18, 19]. Various methods have been developed that enable investigators to identify the proliferative experience of T cells, key among these being the enumeration of T cell Receptor Excision Circles (TREC), which are found only in low numbers in memory T cells (CD45RA-CD45R0+ cells). To simplify this type of investigation, a selection of markers has now been identified that can delineate CD4+ recent thymic emigrant T cells by flow cytometry rather than relying on molecular techniques [20]. T cell expression of CD31 (PECAM-1) is strongly correlated with TREC frequency in these cells, and the level of CD31 expressed is consistent with longer telomeres and telomerase activity, clearly suggesting that CD31 is a good marker of CD4+ T cells with low/null proliferative history (Fig. 1). The proliferative history of T cell populations had long been thought to be indicated by expression of the classical subset marker CD28 [21]. By careful analysis of flow cytometry data, it is possible for investigators to identify that CCR7+CD45RA+ CD4+ T cells contain a fraction of CD28dim or CD27dim T cells that may represent antigen-inexperienced naïve T cells with replicative history. Human aging is an excellent model for the study of T cell sub-populations, and increasing evidence from such studies confirms that CD31 is an effective marker of true T cell naivety, albeit with less than 100% sensitivity for the CD4+ T cell pool. The CD31neg naïve T cell population expands with age, while the fraction of CD31+ T cells is progressively depleted. Despite this phenotypic difference, the CD31neg “proliferation-experienced” naïve CD4+ T cells retain functional features of true naïve T cells [22]. While it is clear that markers other than CD31 can also be used to identify distinct subsets of naïve T cells (including PTK7), the signals that actually induce naïve T cells to enter a replicative stage have yet to be defined. Importantly, no equivalent marker to CD31 has so far been identified that can reliably differentiate subsets of naïve CD8+ T cells, but we cannot exclude the possibility that similar phenotypic hallmarks also characterize this population. Our data suggest that the regulation of CD28 expression differs markedly between the naïve T cells that are activated following antigen encounter (subsequently becoming central memory T cells) and the antigen-inexperienced naïve T cells that proliferate in response to homeostatic stimuli (Fig. 2). While homeostatic proliferation signals tend to maintain consistent levels of CD28 on the surface of naïve CD4+ and CD8+ T cells, significantly higher levels of CD28 are expressed by central memory cells (Fig. 2). It is likely that T cell activation mediated via TCR/CD28 and other costimulatory molecules induces signaling crosstalk that drives and maintains increases in CD28 expression during differentiation into central memory cells, whereas homeostatic signaling bypasses the TCR/CD28 pathway and may therefore fail to induce the signaling events that up-regulate CD28 expression. Compelling evidence suggests that cytokines including IL-15, IL-21, and the associated STAT signaling molecules are involved in mediating such events [23, 24].

Figure 1.

Expression of classical markers of T cell differentiation. In addition to CD28/CD27/CD45RA/CCR7, recent thymic emigrant T cells (named ‘truly naive’) also express CD31. The expression profile of central memory (CM), effector memory (EM) and T effector memory re-expressing CD45RA (TEMRA) cells is also depicted. The relationship between expression of surface markers and TRECs/Telomeres is depicted. CCR7: chemokine receptor 7; PD1: Programmed death 1; KLRG1: killer lectin inhibitory receptor 1; TREC: T cell Receptor Excision Circles. [Color figure can be viewed in the online issue, which is available at]

Figure 2.

T cells express higher levels of CD28 in the early memory stage compared with the naïve stage. CD4+ and CD8+ T cells were categorized into the subsets; naïve (CD28+CD45RA+), early memory (CD28+CD45RA-), late memory (CD28-CD45RA-) and TEMRA (CD28-CD45RA+), and the expression level of CD28 is shown. Early memory T cells display the highest levels of CD28 expression. A representative experiment is shown. [Color figure can be viewed in the online issue which is available at]

When T cells encounter their respective antigens they start to differentiate into distinct types of memory cells. It is not yet clear which factors determine specific T cell differentiation fates, but the antigen type and antigenic load are major influences on this process [25, 26]. The early signaling cascade in responding T cells exerts a critical influence on the activation and differentiation process [27], while also being strongly influenced by the environmental milieu (including modulatory cytokines) [28]. The memory cells generated by antigen activation are retained in order to protect against secondary infections by the same pathogens [29], and they persist by undergoing homeostatic proliferation in the presence of IL-7 or IL-15 [14]. The broad diversity of the host T cell repertoire must be preserved by the continuous replenishment of naïve T cells and by the homeostatic maintenance of the memory population [26], or else the original repertoire would be subject to progressive depletion by each successive pathogen encounter. If we conduct careful studies of surface marker expression, we can observe that as T cells become increasingly differentiated, they undergo progressive loss of key surface molecules; initially CD45RA is down-regulated, then CCR7 is lost, and finally CD28/CD27 expression are decreased (excepting re-expression of CD45RA by TEMRA). The progressive loss of surface markers is widely considered to correspond to the impairment or loss of T cell functions [30]. Despite this general downward trend in surface marker expression, it is also possible for differentiating T cells to express new markers that are not merely correlates of impaired function [31, 32].

Which Markers Most Accurately Identify Early, Differentiated, and Senescent Memory T Cells?

Having clarified the phenotypic differences between naïve T cell subsets with distinct proliferative histories, it is useful to determine how memory cell populations can be delineated most effectively, and to better understand the functional significance of the surface marker expression by these cells, (especially in the late stages of CD4+ and CD8+ T cell differentiation). In addition to the previously defined markers, we propose that the different memory cell sub-populations can be further described using CD57, killer cell lectin-like receptor-1 (KLRG-1), and programmed cell death protein 1 (PD1). The central memory T cells can be identified as CCR7+CD27+CD28+CD45RA-CD57-KLRG1-PD1-, the effector memory as CCR7-CD27+/- CD28+/-CD45RA-CD57+/-KLRG1+/-PD1+, and finally the TEMRA as CCR7-CD27-CD28-CD45RA+CD57+KLRG1+PD1+/- (Table 1). While a wide range of additional markers can also be used to define these populations, as we will describe later, it is possible to encapsulate the entire concept of T cell differentiation/senescence/exhaustion through the use of just these five key markers; CD28, CD27, CD57, KLRG-1 and PD1 [32, 33]. Some of the alternative markers used to delineate T cell sub-populations, including CD38, CD62L and CD95, have however proved useful in the identification of clinically relevant T cell subsets [34].

The co-receptors CD28 and CD27 have frequently been used to define sub-populations of memory cells, and historically, down-regulation of CD28 was typically associated with loss of functionality, especially with a diminished proliferative capacity and loss of telomeres [35]. However, proliferative arrest has subsequently been identified as a mechanism of host protection against molecular damage that could lead to cellular transformation, hence T cell regulation of coreceptor expression is not as straightforward as previously thought. The reader is invited to consider references pertaining to “Hayflick's limit” for a better understanding of this topic [36, 37]. CD28 is not the only molecule to be lost upon successive rounds of T cell proliferation, since CD27 exhibits a similar expression profile during repeated cell division. Both CD28 and CD27 bind to molecules expressed by antigen presenting cells (APCs); while CD28 binds to B7 family members, CD27 binds specifically to CD70 which has been shown to be an important modulator of T cell functions [38]. While the underlying mechanism has yet to be identified, we observe that memory CD8+ T cells tend to lose CD28 expression before losing CD27. The reverse is true in CD4+ T cells, which may help to explain why CD28- CD4+ T cells are extremely rare in healthy young individuals. One way to study these sub-populations is to assess samples obtained from chronically infected individuals or from elderly individuals that have been subject to extensive antigen exposure during their lifespan. We will discuss these two cases later in this review.

Changing the Simplistic View of Surface Markers by Understanding Their Functions

T cells are thought to require co-receptor signaling in order to achieve full activation, to modulate apoptosis, to ensure sufficient energy for functioning (through modulation of the mammalian target of rapamycin; mTOR) [39], and finally, to modulate the activity of senescence-associated signaling molecules including the p38 member of the MAPK family [40]. Telomere length and telomerase activity/activation are modulated by CD28 signaling, which implies that reduced CD28 expression is associated with shortening of telomeres and decreased telomerase activity [41]. However, it is unknown whether the loss of CD28 induces these changes in telomere homeostasis, or whether decreased telomere homeostasis leads to the loss of CD28. Recent data have shown that CD28- T cells retain some functionality, including the ability to up-regulate telomerase-dependent p-AktSer394 [42, 43], and that the eventual loss of CD27 is a better correlate of reduced telomerase activity. Thus, the loss of the co-receptors CD27 and/or CD28 can be considered to be indicative of impaired telomere function in T cells and denotes progression towards replicative senescence.

The concept that CD28- T cells are senescent is based on the work of Effros and Walford [35], who demonstrated that T cells subjected to several passages in vitro down-regulated CD28 and lost telomeres/telomerase activity, analogous to the state of replicative senescence first described by Hayflick [36]. In the 1960s, Hayflick reported that fibroblasts subject to ∼50 passages displayed “replicative senescence” since they appeared inert and were unable to proliferate further, and yet were also resistant to cell death [36]. The concept of senescence has since also been applied to numerous arms of immunological research and is now often used to describe T cell populations with reduced proliferative capacity and impaired function.

Among the various T cell surface markers described above, CD57, KLRG1, and PD1 have each been associated with replicative senescence [32, 33, 43]. However, it is unclear which of these markers exerts the largest influence on T cell proliferation, since they are often co-expressed (Fig. 3). Data from the literature suggests that Effector Memory cells are the subset most predisposed to express PD1, whereas TEMRA display higher levels of CD57 and KLRG1 [30, 44]. CD57, KLRG1, and PD1 are typically regarded as “inhibitory” molecules because of their association with poor responses to proliferative stimuli. While cells that express these markers may indeed display weak proliferative capacity, it is now clear that they can also exhibit higher cytotoxic activity and more pronounced cytokine production than cells of a less differentiated phenotype [37, 45]. These findings clearly suggest that the presence or absence of surface markers does not confer uni-directional changes in the cellular function of immune cells, but may instead represent a functional switch from a highly proliferative/low cytotoxic profile towards a low proliferative/highly cytotoxic profile. At present, CD57 and KLRG1 are considered to be inhibitors of proliferation, whereas PD1 is primarily associated with immune exhaustion [46]. Blockade of PD1 with a specific antibody has been shown to restore the functionality (cytokine responses) of PD1+ cells, suggesting that immune exhaustion is reversible, whereas replicative senescence appears to be irreversible. Another important marker of cellular differentiation and senescence is the CD45 antigen which exhibits multiple isoforms, among which CD45R0 and CD45RA have proved particularly useful in defining T cell subsets [47]. It is well established that CD45R0 and CD45RA play crucial roles in the early signal transduction process after TCR/CD3 complex stimulation by antigen [48]. CD45R0 and RA modulate Lck activity in T cells by mediating dephosphorylation of the negative tyrosine phosphorylation site (pTyr505) by moving in and out of lipid rafts and the immune synapse. It remains unclear why these CD45 isoforms also serve as effective markers of T cell differentiation [27]. Which specific CD45 isoforms are responsible for T cell activation? What are the functional differences between CD45R0 and CD45RA? These questions will need to be addressed in future in order to advance our current understanding of T cell activation and subset behaviors. In particular, identification of the early signaling events that are initiated by CD45RA/R0 should provide substantial new data on their distinct functional roles, which may prove critical in influencing T cell progression towards exhaustion or senescence. Nonetheless, it is clear that loss of CD45RA expression in CM and EM T cells renders them functionally different from RA-expressing cells (Naïve and TEMRA subsets). This finding gives rise to several intriguing hypotheses; i) TEMRA sub-populations are directly differentiated from naïve cells, ii) TEMRA cells are pre-programed to become TEMRA upon antigen recognition by unknown regulatory mechanisms (probably involving miRNA), iii) differentiation into TEMRA sub-populations is induced by persistent re-stimulation of individual clones. While many investigators routinely use most of the above-mentioned markers, it is notable that the use of CCR7 as a marker of T cell differentiation may prove difficult under certain circumstances (especially during analysis of frozen samples). In order to harmonize analytical approaches and enable robust comparisons across different studies, the use of CD28 or CD27 and CD45RA is recommended.

Figure 3.

PD1 and CD57 expression in T cell sub-populations. Data shown are from a typical experiment demonstrating that PD1 and CD57 are not always expressed concomitantly. In particular, the TEMRA populations (which are defined as the latest stage of T cell differentiation) express CD57 but not PD1, whereas the effector memory-like cells (CD28-CD45RA-) displays the highest levels of PD1 often in the absence of CD57. Arrows represent a potential route by which PD1 and CD57 could be lost/gained during differentiation (no consensus at the present time).

The characterization of T cell sub-populations is rapidly becoming essential in the study of human physiology and pathological conditions. T cell distribution between discrete subsets reflects the overall immunological state of the host and can therefore be used to elucidate pathogenesis and provide useful biomarkers of patient prognosis and treatment efficacy. This is particularly true in the case of persistent viral infections that shape the immune system (including T cells) over a prolonged period, and in natural aging which is an excellent model in which to study the impact of time and poly-infection on host immunity.

Effect of Aging on T Cell Phenotypes

Natural aging has profound effects on many aspects of human physiology and is mainly thought to be associated with declining biological functions. However it is important to distinguish between healthy aging and pathological aging. The study of altered immune function in aging individuals can explain many different manifestations of the overall aging process, as well as the development of age-related life-threatening disorders including infections, cancers and atherosclerosis [49, 50]. Generally, the most striking age-related change observed in studies of the T cell compartment in aging individuals is an increase in the number of CD28- cells, as was predicted by the earlier in vitro studies of Effros and Walford [35]. Since CD28- T cells have been previously characterized as “senescent,” the T cells populations associated with aging were also been designated as senescent cells, eventually leading to the wider adoption of “immunosenescence” as a general biological concept. However, it is important to define key terms in order to better understand the sub-populations we assess when investigating senescence. Where “immunosenescence” describes age-related alterations at the immunological level (alternatively “immune erosion”), “immunological aging” refers instead to the apparent aging of individual immune cells and does not necessarily involve aged individuals or aging studies per se. While immunosenescence primarily refers to replicative senescence, the immunological aging process may include elements of both senescence and exhaustion. Against this backdrop, the initial concept that loss of CD28 was sufficient to identify “senescent” T cells was soon found to be unable to account for full extent of the age-associated changes widely observed in the CD4+ and CD8+ T cell compartments. Ongoing studies now indicate that accumulation of differentiated T cells, especially in the TEMRA subset, are hallmarks of aging.

These findings have had far reaching consequences for our understanding—and indeed misconceptions—of what constitutes immunosenescence. We now appreciate that impaired immune function in the elderly is likely a consequence of constant exposure to pathogens, some of which persist throughout life [50, 51]. This means that repeated pathogen encounter during chronological aging eventually exhausts the immune system, leading to the accumulation of highly differentiated T cells that are characterized by replicative senescence [7, 13, 33]. Clearly, this progressive loss of immune function with increasing age can have dramatic consequences for host protection, especially given that the thymus exhibits a concomitant loss of capacity to generate naïve T cells over time [52]. It has been nicely demonstrated that the naïve T cell pool in elderly humans is comprised of only 10% recent thymic emigrants and as much as 90% homeostatic naïve T cells, whereas in aged mice these proportions are exactly reversed [53]. The biological relevance of these findings can be summarized as follows; (i) animal models provide only limited information on the maintenance of human immune homeostasis, (ii) aging is associated with a significant loss of thymic output, (iii) thymic involution may not be as detrimental as previously thought due to the broad antigen experience of elderly subjects, (iv) the imperative to maintain naïve T cell diversity may be a function of the prevailing environmental conditions of the host. The question therefore arises of whether these age-related changes necessarily mean a loss of function, or whether they represent a necessary adaptation to the host's surrounding environment?

This question is not easy to address because it is necessary to consider the functionality of the cells in question in combination with the immune reserve capacity (the status of the immune system that has been influenced by previous immunological history), the nature of the antigens involved, and the complex interactions between these factors. We must also consider the possible reversibility of the observed “impairment” of cellular functions, (as revealed by studies that have successfully blocked inhibitory receptors), and the fact CD28-T cells are not in fact committed to replicative senescence, because stimuli such as 4-1BBL, OX40L, IL-2, and IL-15 can promote the proliferation of these cells [23, 54] (although the response of CD28- T cells to IL-15 stimulation and their ability to undergo homeostatic proliferation appear somewhat heterogeneous) [23, 24, 55]. CD28-T cells produce variable quantities of pro-inflammatory cytokines that contribute to the state of low-grade inflammation known to be established during aging [56]. The intracellular signaling machinery of more differentiated memory CD4+ and CD8+ T cells is also known to display a persistent low-level of activation that alters the overall activation threshold of these cells [57]. The state of low-grade inflammation observed in elderly subjects is associated with low-level activation of cell signaling and is reflected by the distribution of T cell sub-populations. When maintained in this primed state, T cells can bypass the requirement for CD28 ligation in order to become activated. The long-term maintenance of this “ready-to-go” status in T cells exhausts the host's reserve of naïve cells, reduces clonal diversity, and leads to impaired functionality. However, from an evolutionary point of view, it may be more efficient to maintain a well-trained memory pool than to generate a new and diverse pool of naïve cells [58]. A concept that has now been largely removed from our current model of T cell differentiation, activation, senescence and exhaustion is the notion of “immunological reserves” that are maintained by cytokines such as IL-15 and IL-7 [59] but undergo progressive depletion with aging. In our own laboratory, we have tested the possibility that expression of the IL-7 receptor (CD127) correlates with other markers of T cell history, namely CD28. In Figure 4 we show using a representative experiment that CD127 expression is maximal in CD28+ T cells and that CD28 mean fluorescence intensity (MFI) strongly correlates with CD127 MFI. This corroborates other studies demonstrating that naïve T cells are more responsive to IL-7 stimulation whereas memory cells are more responsive to IL-15. Technical issues complicate the study of IL-15R expression, but future investigation of the IL-7/IL-15 system in T cell sub-populations should clarify some of the key questions that arise from this discussion.

Figure 4.

CD127 expression correlates with CD28 expression. Freshly isolated PBMC from healthy donors were isolated and stained for CD3, CD8, CD45RA, CD28 and CD127 expression by flow cytometry. Single controls were used for compensation matrix and samples acquired using a BD LSRII Fortessa. A representative donor is shown for CD45/CD28 expression in CD8+ T cells (A). We identified that CD127 expression was maximal in the early-memory sub-populations (b and c) and was also present in naïve cells (a). The values in the histogram show representative mean fluorescence intensities (MFI) for CD127 in the respective sub-populations (B). The expression of CD3 and CD8 controls did not show any significant difference between the sub-populations when assessed in parallel. CD127 expression was not linked to CD45RA but was instead correlated with CD28 expression. The MFI for CD127 was plotted against the corresponding CD28 MFI (C). Data from a representative healthy elderly donor is shown. [Color figure can be viewed in the online issue which is available at]

The immunological changes observed in aged individuals reflect the diversity and intensity of the antigenic challenges encountered during the host's lifespan. Among the many hypotheses proposed in order to explain immunosenescence in the elderly is the role of chronic viral infections, especially cytomegalovirus (CMV), for which accumulating evidence indicates a key role in driving exhaustion of T cells and induction of senescence.

Effect of Chronic Viral Infections (Especially Latent CMV) on T Cell Phenotypes

Various chronic viral infections including Epstein-Barr virus (EBV), hepatitis B virus (HBV), human papilloma virus (HPV), and human immunodeficiency virus (HIV) [50] can be reactivated by host immunosuppression, suggesting that maintenance of immune surveillance is normally capable of suppressing persistent infections [60, 61]. These viruses induce a massive proliferation of antigen-specific CD8+ T cells that can be detected long after the virus itself is controlled [62], although the virus may also continue to persist in the host in a latent form. HIV has been shown to be constantly replicating in the host, thereby inducing an antigen-dependent clonal expansion of memory T cells that resembles that found in the aging immune system [5]. This has led to the concept that HIV patients are subject to premature immunosenescence that occurs independently of aging [63].

Due to the advent of successful anti-retroviral therapy, HIV+ patients can now survive for extended periods after initial infection, hence the virus now represents an important pathogen among the elderly population as well as the young. HIV was initially thought to induce premature immunosenescence of CD4+ or CD8+ T cells in infected patients, since these populations exhibited characteristic changes in surface markers and functions that resembled those observed in noninfected elderly subjects [11, 63]. Many patients suffering from AIDS now survive to become elderly and exhibit the persistent systemic inflammation that is characteristic of the clinically latent or asymptomatic stage of the disease [64, 65]. Taken together, it is clear that HIV-infected individuals are subjected to continuous antigen stimulation accompanied by an ongoing inflammatory process and loss of CD4+ T cells. In addition to providing useful data on the nature of immunological aging and host cell senescence, the real-time chronological aging of HIV-infected subjects also provides a useful model for studying the effects of CMV by allowing investigators to differentiate age-dependent events from virus-dependent mechanisms of immunosenescence [63].

CMV is a highly prevalent and chronic infection with an ill-defined clinical impact. The consequences of CMV infection are primarily observed only in immunosuppressed patients (e.g., HIV and transplant recipients) or during infection in utero (CMV disease). Studies of young infants and transplant patients who received an organ from a CMV-seropositive donor have revealed that virus infection induces a huge expansion of CMV-specific T cells that exert potent bystander effects on the rest the T cell pool [6-9, 13]. A new paradigm has now emerged (based primarily on the OCTO and NONA longitudinal studies conducted in Sweden), to suggest that the age-related changes observed in T cell phenotypes and functions are highly influenced by CMV leading to increased frequency in CD8+ TEMRA cells, with specificity for CMV [29]. It is noteworthy here that EBV-specific T cells are mainly CD8+ T cells of the effector memory phenotype i.e. appear less differentiated than CMV-specific T cells. Previous studies have shown that not only is there is decreased T cell diversity in CMV infection but also that the CD8 clones present are strongly associated with CMV seropositivity [66]. These longitudinal studies have led to the definition of an Immune Risk Profile (IRP) that can be used to predict mortality in elderly individuals (albeit with some limitations). The IRP is characterized by an inverted CD4/CD8 ratio, an accumulation of CD28-CD8+ T cells, poor proliferative responses, low B cell frequency, and seropositivity for CMV. It is well established that persistent/latent CMV infection is associated with the accumulation of large populations of “late-differentiated” cells (measured as CD28– and CD57+ phenotypes) that exhibit decreased functionality [8]. More recently a study by Simanek et al. [67] showed for the first time that CMV positivity coincident with an increased level of C-reactive protein (CRP) was associated with decreased patient survival independently of age. This finding suggests that elderly CMV seropositive individuals may have a higher propensity to develop other age-associated disorders such as cardiovascular disease and cancer that contribute to mortality rates [68, 69]. In contrast, it has recently been proposed that CMV exerts only a minor role in driving host inflammation, morbidity and mortality, at least in the elderly population where “inflamm-aging” appears to be independent of CMV status [70]. The levels of IL-6, IL-10, CRP, and TNF detected in this last study were comparable between CMV seropositive and CMV seronegative elderly individuals, and a longitudinal survey revealed a similar profile over a decade [70]. The impact of CMV-driven T cell differentiation on the regulation of host homeostasis thus remains to be fully elucidated.

What is the Future of T Cell Functional Phenotyping?

Surface Markers

The highly inter-related and classically defined T cell surface markers change with aging, antigen challenge, and altered functionality, suggesting that a better understanding of their functional roles will be highly informative. We propose that the most important candidates for study should include the CD45RA/CD45R0 isoforms. These molecules seem to play important roles in the definition of T cell exhaustion and differentiation by undergoing isoform switching. However, the distinct roles played by each isoform of the CD45 protein tyrosine phosphatase during TCR-dependent activation of T cells remains poorly defined. It seems likely that re-expression of CD45RA in TEMRA cells should drive these cells towards a functional state distinct from that of naïve cells (or alternatively that the function of CD45RA depends on other molecules that are differentially expressed in naïve cells and in TEMRA). Crosstalk with other molecules may well underpin both the regulation and action of CD45RA/CD45R0 in T cell biology.

Which molecules then are the prime candidates for more effective delineation of T cell subsets and functions in future studies? We have already mentioned that the IL-7 receptor (CD127) is a likely candidate for participation in the regulation of host cell homeostasis, proliferation of differentiated T cells, and probably also in cell survival. The close correlation of CD127 levels with CD28 expression also suggests that common pathways and driving forces may regulate the expression of both markers. As T cells differentiate they become increasingly likely to exhibit senescence and may display a Senescence-Associated Secretory Phenotype (SASP). Since these “SASP” cells exhibit a pro-inflammatory profile, it could be suggested that pattern recognition receptors including Toll-like receptors, NOD-like receptors and leucine-rich repeats may be differentially expressed and could indicate differential functionality between the T cell sub-populations defined by the classical markers. At present, it is entirely unknown whether T cells can acquire or lose expression of these innate receptors during progressive differentiation [71].

The better-known inhibitory receptors PD1 and CTLA-4 are still not widely used in the characterization of late-differentiated T cells [72]. The individual roles played by these inhibitory receptors and how their functions are modified by interactions with other molecules are still not fully understood. Other inhibitory receptors that contribute to the pathology of chronic age-associated diseases such as rheumatoid arthritis and cancer (and HIV) are also known to induce premature immune aging [73]. NKG2D for example was shown to be expressed in CD4+ T cells at a very late stage of differentiation. These cells were expended in CMV infection and showed an inflammatory (SASP) activity characterized by marked production of IFNγ [73]. The marker CD95 (Fas) has also been used in combination with CD28 to identify senescent T cells under certain circumstances [2, 74]. The use of intracellular markers such as T-bet could also prove informative in this context, but cytosolic and nuclear molecules are more difficult to include in routine flow cytometric analyses [75].

Intracellular Markers

A combined analysis of surface marker expression and intracellular cytokine production could be foreseen as an effective method of assessing T cell differentiation. As mentioned above, it is thought that senescent “SASP” cells are potent producers of pro-inflammatory cytokines. However, it is unknown whether the quantitative and qualitative production of soluble factors by senescent cells is subset-specific. Which cytokines then are the most likely to enable the differentiation of distinct sub-populations? The best-studied factors to date are IFNγ and TNFα, although it is likely that GM-CSF and IL-1β will also prove very useful in delineating T cell subsets in future. Indeed, it has recently been shown that GM-CSF could represent a good candidate for further characterizing the late-differentiated T cells that arise with aging and in response to CMV infection [76].

More recently, it was suggested that the loss of either CD28 or CD27 alone was a poor predictor of T cell senescence. The clustering together of several related parameters may therefore allow for better discrimination between lately differentiated/senescent T cells with distinct functionalities. In a complex multivariate and system biology analysis, Rivet et al. [27] determined that integrated data on the state of cell signaling (Lck, ERK, and LAT activation) can predict T cell aging and functional capacity. Accordingly, the authors advocated the use of high-throughput analysis of numerous different markers and the subsequent integration of marker expression data in the evaluation of cellular aging and function. On the basis of their model, measuring the expression of CD27 and CD28 alone is not sufficient to accurately predict any of the signaling variables, although conversely, the signaling time course responses were indeed able to predict both CD27 and CD28 surface expression [27]. Cell signaling may therefore also become an important consideration in the study of T cell differentiation in aging, chronic viral stimulation, and disease.


Taken together, the experimental evidence presented here confirm that there are important correlations between the classical definitions of T cell sub-populations and their associated functionalities. However, our current picture of the relationship between T cell phenotype and function is far from complete. Certainly, the loss of specific surface markers and probably also the re-expression of other molecules will contribute to the altered functional status of these cells. While the progressive phenotypic and functional changes driven by antigen exposure over time are typically thought to be detrimental to host protection, they could perhaps also be considered to represent environmental adaptations. This concept may help us to progress beyond the default negative appraisal of the aging T cell compartment (in which all changes are deleterious), to allow for better integration of the diverse changes that occur in host immunosurveillance with aging (and to better apply our knowledge of these processes to clinical settings). Comprehensive integration of predictive markers, especially those for which there is a consensus on biological relevance [77], should allow investigators to design more effective interventions for the therapeutic modulation of discrete T cell compartments. The recent development of novel approaches to the study of T cell sub-populations (such as the use of deep phenotyping by mass-spectrometry based cytometry) [78], will undoubtedly help investigators to redefine the links between cellular markers and functionality. The combined, multi-dimensional approaches (genomics, proteomics, lipidomics, glycomics) that will be applied to these problems over the next few years will be critical in the validation of many current concepts in biology and immunology.


Neil McCarthy of Insight Editing London provided writing assistance. We would like to thank the Flow Cytometry Facility at SIgN for their support. Anis Larbi is part of the International Society for Advancement of Cytometry (ISAC) Scholar program.