[Frontiers in Bioscience 3, d59-99, January 15, 1998]

	[Frontiers in Bioscience 3, d59-99, January 15, 1998] Reprints PubMed CAVEAT LECTOR
	T CELLS AND AGING Graham Pawelec ¹, Ed Remarque ², Yvonne Barnett ³, Rafael Solana⁴ 1 University of Tübingen, Tübingen, FRG ²,University of Leiden, Leiden, Holland ³, University of Ulster, Coleraine, Northern Ireland ⁴, University of Córdoba, Córdoba, Spain Received 12/29/97 Accepted 1/5/97 8. ALTERATIONS IN T CELL SUBSETS AND MARKERS WITH AGING Is the senescent phenotype due to increases in memory cells and to their "clonal exhaustion"? Age-associated changes in T cell subsets occurring in rodents and humans have been repeatedly documented. Mice seem to show a relative loss of CD4 cells with age compared to CD8 cells. Whereas the TCR repertoire of the surviving CD4 cells is said to remain unchanged compared to young cells, the CD8 repertoire is markedly altered, suggesting expansion of a small number of CD8 (regulatory?) cells during aging (192). On the other hand, Rosenberg et al . failed to find quantitative differences in CD4⁺ cells of old mice, but noted a striking decline in the ability of CD4+ cells to cause rejection of allogeneic skin grafts, whereas CD8+ cells retained fully their function as effectors (193). Moreover, the defect in CD4+ cells could not be overcome by treating the cells with soluble T cell helper factors (containing many kinds of cytokines as well as IL 2). The TCR1 repertoire may also change with age, although the meaning of this finding is unclear. Thus, Giachino et al . (194) demonstrated that the polyclonal Vd 1 and Vd 2 populations present in children changed to oligoclonal in the elderly (57 - 88 yr). This may reflect "memory" for common antigen. Although not observed by all investigators (195), age-associated changes in T cell repertoire may apply to human as well, according to reports by Posnett et al . (196). Expansion of CD8 cells is common, but it may be the rarer CD4 expansions which are observed at increasing frequency in the aged (197). However, in some disease states, eg. RA, CD4 cells may show striking oligoclonal expansions (198). Whilst not seen in all normal donors, it is interesting to note that such CD4 expansions were also seen in unaffected siblings of RA patients, suggesting that they may be a genetic risk factor for rather than a consequence of RA (198). In CD8⁺ but not CD4⁺ T cells, up to >30% of the entire population may consist of oligo- or even monoclonal cells expressing the same TCR-Vß markers (196). Within the CD8⁺ cells, these oligo- or monoclonal populations are prevalent within the CD28-negative subset (196) and the CD57-positive subset, which often overlaps with the CD28-negatives (199). It is interesting to note that it is this CD28-negative, CD8-positive subpopulation which was identified many years ago as containing suppressor cells (200). These data may explain the observation that alterations in proportions of different T cell subsets may also be more marked in CD8⁺ than in CD4⁺ cells of aged humans (201). However, this phenomenon may not be absolutely limited to CD8 cells. Recent data from Thorbeke´s group suggest that although "forbidden" CD4 clones are not present in 24 month-old mice (usually the uppermost limit in mouse aging studies), they do appear in those few mice reaching 30 months of age (202). It was suggested that these potentially self-reactive CD4 cells were derived extra-thymically because thymectomy increased rather than decreased their numbers. There is a considerable literature on extra-thymic T cell development in mice, and evidence for increased development of self-reactive extra-thymic, but not thymic-derived, T cells with age (203). The possibility of enhancing extra-thymic development with factors such as oncostatin-M may offer the opportunity for manipulation of this pathway (204). Moreover, the realization that mature thymus-derived T cells can re-acquire sensitivity to positive and negative selection outside the thymus, in germinal centers (205), indicates in theory the generation and selection of T cells may be effected even the absence of a functional thymus. There are clearly more CD4⁺ CD45RO⁺ (and in mice, CD44^hi) "memory" cells and less CD45RA⁺ (and, in mice, CD44^lo) "naive" cells in PBMC from elderly individuals. Whole blood analyses confirm the relative paucity of CD45RA+ cells in both the CD4 and CD8 populations of the elderly (129). If the CD45RO⁺ cells represent memory cells, and if exportation of naive CD45RA⁺ cells from the thymus decreases with age, then an accumulation of CD45RO⁺ memory cells would be expected in elderly donors. This would be coupled with a predicted reduced ability to respond to new antigens, and a retained ability to respond to recall antigens, as long as the memory cells remained present. Certainly, the proportion of RA⁺ cells decreases and RO⁺ cells increases with increasing age (206). However, in the oldest old, decreases in memory cell phenotype RO⁺ cells have also been recorded (160) and in whole blood analyses, a relative decrease of CD45RO⁺ cells may also be seen in the CD8 but not in the CD4 population (129). However, in exceptional individuals (healthy centenarians) the decrease of RA⁺ cells, especially in the CD8 subset, may be markedly less than in the ordinary old population (207). The meaning of this finding is unclear, because functional tests were not performed, and it is known from other studies that the CD4⁺ cells responsible for the increase of RO⁺ elements express lower levels of CD45RO than do young CD45RO⁺ cells (208). Whether this is related to their impaired function (114,115) is not yet known (208). More subtle analyses may reveal further differences in surface phenotype and function, which remain to be collated and understood. R. Miller´s group has shown in mice, for example, that aging leads to an increase in the proportion of splenic cells expressing high activity P-glycoprotein (Pgp^hi) and therefore able to extrude rhodamine 123. Pgp is known also to participate in the transportation and secretion of cytokines including IL 2, IL 4 and IFN-g (209). Moreover, mAb against Pgp block IL 2 release in PHA-activated T cells, demonstrating a critical role for this molecule in T cell function (210). Despite this, high Pgp expression in old T cells is linked to dysfunctional status. Possibly the Pgp itself is dysfunctional and over-expression represents an attempted compensation. There is an age-associated increase in the expression of MHC class I molecules on these high-Pgp CD4 memory cells. In spite of this, the levels of TAP1 decrease in old mice (TAP1 transporter is usually required for class I peptide loading). The Pgp and TAP1 molecules are related but whether Pgp is taking over the function of TAP1 in old cells and increasing class I remains an open question (211). The failure of Pgp^hi cells to respond to TCR-mediated stimulation cannot be overcome by CD28 signalling, PMA, IL 4 or IL 12 or combinations of these, and may therefore be considered "anergic". Not only do Pgp^hi cells fail to secrete IL 2 but also show impaired IL 5 and IL 10 production and proliferation. Oddly, however, their ability to secrete IFN-g increases with age (212). However, all these data were gathered exclusively in murine systems and it is not known whether human cells behave similarly. The question of whether antigen-independent functional changes in naive T cells can occur has also been addressed. As discussed above, differentiation of T cells to memory cells coupled with age-related changes in memory cell characteristics may be responsible for much of the altered functional phenotype of the aged individual. Linton et al . looked at TCR-transgenic mice with T cell specificity for pigeon cytochrome C antigen, for which they believed they had good evidence for lack of cross-reactivity with environmental antigens. They found that in aged animals, the TCR-transgenic CD4+ cells were decreased in number and in antigen responsiveness but that they maintained a naive cell phenotype. They concluded that the defects observed were therefore due to aging of the naive cells per se and not to environmental stimulatory influences (213). Such findings are clearly consistent with several studies showing different patterns of cytokine production by young and old cells despite posession of the same "naive" phenotype (214). In human, most T cells are CD7+, but the frequency of CD7-negative cells increases with age (215) and although isolated T cell clones retain stable expression of their CD7-positive or negative phenotype (216), repeated stimulation and propagation of uncloned lines results in accumulation of CD7-negative cells in the CD4 but not CD8 subset (217). Increased proportions of CD7-negative cells are found in situations of chronic antigenic stimulation in vivo, eg. in RA (218) and in kidney transplant recipients (219). Such CD7-negative cells show low proliferative responses to CD3-stimulation, low IL 2 secretion but high IL 4 and IL 10 secretion (220). These results suggest that loss of CD7 expression may be age-associated, but the fact that long-term cultured T cell clones retain high CD7 levels imply that factors other than merely the number of PD undergone are critical for CD7 expression. Thus far, CD28 is perhaps the closest to a biomarker of aging found for human lymphocytes. Both in vivo and in vitro, the proportion of CD28⁺ cells decreases with age. In monoclonal populations, the density of expression of CD28 decreases with age (221). Effros et al . were the first to observe a decreasing percentage of CD8 cells carrying CD28 in the elderly, paralleling their observations in T cell lines aging in culture (168). Others have confirmed that particularly the CD8 subset shows progressively decreasing CD28 expression with age (222). There is a correlation between shortening of telomeric repeats and age of the donor, which is not confounded by differences in white blood cell count (223). Moreover, telomere lengths (TL) in the CD28-negative cells were less than in the CD28⁺ cells from the same donors, implying that the former had undergone more rounds of cell division than the latter (224) (see section 7). This type of proliferative senescence may therefore be responsible for the commonly observed accumulation of CD28-negative oligoclonal populations in elderly people (196). Although originally described only in CD8 cells, the number of individuals with such clonal expansions in both CD4 and CD8 cells was very similar (ca. 70% of individuals over 65); moreover, these expansions were stable over a two-year observation period (225). In diseases with chronic antigenic stimulation, further circumstantial evidence in favour of the hypothesis of proliferative senescence indicated by downregulated CD28 expression can be garnered. To give some examples: the percentage of CD28⁺cells decreases during Chagasic progression (226); both CD4 and CD8 cells show decreased CD28 expression in chronic B lymphocytic leukemia (227); in rheumatoid arthritis, the percentage of CD4 cells carrying CD28 is reduced (228) and in both RA patients and normal controls the CD4⁺ CD28-negative cells show TCRVB oligoclonality (229); in Crohn´s Disease, the ability of CD28 to mediate costimulation of CD4 cells is compromized (230). These examples suffice to illustrate the range of situations in which T cell proliferative senescence may play a role in modulating immune responses independently of the age of the host. The effects of this kind of "clonal exhaustion" in the elderly may simply be more noticeable than in the young because of thymic involution preventing effective generation of naive T cells and because T cells present in the old may already have undergone many rounds of division. 8.1 Longevity of naive and memory cells An important question raised here concerns the longevity of naive and memory cells. There is some evidence that memory cells are not quiescent long-lived cells, but represent T cell clones in a constant dynamic state of activation. Beverly´s group used a method for measuring intermitotic time to investigate longevity of CD45RA and RO cells in cancer patients following radiotherapy. Irradiation induced dicentric chromosomal lesions which can be visualized cytogenetically. They found small numbers of CD45RA⁺ cells with these lesions up to 10 years after irradiation, consistent with the belief that naive T cells can be very long lived. However, CD45RO+ cells with such lesions had all disappeared by one year, suggesting that they had all attempted unsuccessfully to divide, consistent with memory cells being in cycle (158). A later study extended these data to conclude that proliferation rates of naive cells were 8x lower than proliferation rates of memory cells. They estimated that, on average, naive cells divided once every 3.5 years, whereas memory cells divided every 22 weeks (231). However, as these authors themselves pointed out, there are some problems with these data, viz. they were dealing with cancer patients, whose T cells and immune status were not normal; irradiation can have indirect effects like induction of lymphopenia which further disrupts the system; possibly some cells with dicentric lesions can nonetheless divide; and the proportion of cells dying without attempting to divide is unknown. All these factors could lead to underestimates of memory cell longevity (232). A direct approach to whether memory cells continue to cycle in vivo has been taken using transgenic mice (233). Naive cells deliberately activated with specific antigen were transfered into athymic hosts in the absence of antigen and found to continue to cycle slowly for extended periods. Naive cells, on the other hand, did not begin to proliferate when transferred into the these recipients. The mechanism for the maintenance of slow cycling of the memory cells is unknown. It may be that unlike naive cells, memory cells do not require antigen to survive and proliferate, but only the self MHC molecule (ie. have a lower functional activation threshold, ref. 234). However, it is clear from these experiments that proliferative senescence could play a role in the eventual loss of the memory cells. The above data in man and mouse have to be reconciled with ealier results in the mouse showing that after thymectomy it is the naive population which disappears rapidly, whereas the memory cells are long-lived (235). The reason for this apparent discrepancy is unknown. According to our in vitro data on T cell longevity, where TCC survive on average 35 PD and maximally 80 PD, this would put the life expectancy of a memory cell at 35 x 22 = 770 wk = 15 years average; 80 x 22 = 176 wk = 34 years maximum (without taking the primary immune response into account, for which the number of cell divisions required can only be guessed at (see ref. 236)). What might be the source of stimulation of the memory cells, which need to persist in the absence of antigen? Several groups have argued that the solution lies in the hypothesis that antigen does persist somehow. This is certainly true for many viruses, but it is unclear for which other antigens it may apply. For viral infection, there is clear evidence that memory T cells are indeed maintained in a constant activated state even for very long periods. Thus, Rehermann et al . studied patients up to 23 years after clinical and serologic resolution of HBV infection and still found evidence for recently activated HBV-specific CTL (237). This might well eventually allow enough time for "clonal exhaustion" of the originally responding T cell clones, in the absence of the generation of de novo responses (see calculation above). Data from other chronic infections such as HIV may be consistent with the scenario of clonal expansion leading to clonal exhaustion and lack of replacement by new thymic emigrants eventually resulting in diminution of the repertoire and loss of anti-HIV responses (238). CD8 cells of the same CD28-negative phenotype as seen in HIV-uninfected senescent cultures and possessing similarly short telomeres have also been described in young persons with AIDS, leading to the suggestion that replicative senescence of virus-specific T cell clones in vivo might contribute to disease progression (239). A marked decrease in telomere length has also been recorded in CD4, CD8 and B cells from HIV-infected patients with advanced immunodeficiency, supporting the notion of a high turnover of these cells and suggesting that replicative senescence may be involved in the final immunosuppression of these patients (240). There is also an increase in CD4⁺ CD28-negative cells in HIV infection, albeit not so marked as in the CD8 subset (241). Indeed, disease progression in AIDS is reported to be marked by an accumulation of CD28-negative cells unable to secrete IL 2, whereas long-term non-progressor patients maintain CD28 levels and IL 2 secretion capacity (242). The rate of destruction of HIV-infected cells in young and old patients seems similar, leading to the suggestion that the elderly cannot replace CD4 cells as rapidly as the young (243). It has been found that T cell responses in HIV patients are characterized by severe TCRVB biases and clonal expansions in CD4 cells, and that such responses are exaggerated with disease progression (244). Despite this, others have found no evidence for increased CD4 turnover in HIV infection on the basis of lack of truncation of telomere length in CD4 cells during progressive disease, although this was clearly confirmed in CD8 cells (245). However, virally infected cells may conceivably express dysregulated telomerase. Independent evidence for enhanced cycling of CD4 cells in HIV derives from measurements of mutations in the hprt locus, which showed that mutation rates in the CD8 and CD4 cells were similarly high (246), consistent with an increased division rate in both subsets. The cytokine secretion profile of mutant CD4 clones (from healthy or HIV patients) was predominantly Th2-like, whereas the CD8 mutants had the same pattern as wild-type. For non-viral antigens, and thymus-independent regeneration of T cells in general,it is probable that antigen-driven peripheral expansion commonly occurs (247), implying that a finite proliferative lifespan of the T cells would be of critical import for the functional intergrity of the immune system. Experimental data have implicated a sort of TCR cross-reactivity responsible for maintaining numbers of T cells, eg. in the case of CD4 cells via class II molecules regardless of their peptide loading (248). 8.2 Activation-induced cell death and aging Further light has been cast on the aging of human CD4⁺ T cells using the mAb PD7/26 specific for the CD45RB isoform. Salmon et al (249) reported that whereas CD45RA expression is lost rapidly after activation of naive cells in vitro, loss of RB expression is gradual, occurring over many cell cycles, and reciprocating the increase in RO expression. The progressive shift from RB^hi+ to RB^lo+ is paralleled by gradual loss of bcl-2 (which protects against apoptosis) and acquisition of fas (which can mediate apoptosis), as well as gradual loss of ability to secrete enough IL 2 to maintain autocrine proliferation (whereas IL 4 secretion remains intact). This eventually results in the cells becoming dependent on exogenous IL 2 not only for growth but for their survival, because without enough IL 2 they undergo apoptosis. Longevity extension for T cell clones might therefore be achievable by upregulation of bcl-2. However, suppression of apoptosis by bcl-2 or bcl-2 family-member bcl-x(L) results in enhanced radiation-induced mutagenesis, consistent with the original isolation of bcl-2 as an oncogene (250). The first report describing the behavior of CD45RB in vivo in elderly humans confirmed that CD4⁺cells in old donors showed significantly decreased 45RB expression (251). In agreement with the in vitro data, the percentage of human CD4 and CD8 fas+ cells also increases with aging in vivo (252,253). The majority of these express CD45RO, and CD25 and CD69 activation markers (254). Parallel to this, the amount of soluble fas in the blood of elderly donors is significantly increased compared to young donors (255). Moreover, Phelouzat et al . (256) have shown increased CD3-mediated AICD and hence decreased proliferation in the elderly. This was found not to be due to IL 2 deprivation, nor was it associated with decreased bcl-2 expression (256). There is a possibility that increased AICD might be associated with decreased levels of IL 6 because IL 6 has been reported to protect neonatal T cells from AICD (257). However, IL 6 production seems not to be decreased in older individuals, as discussed above. Consistent with the findings of increased susceptibility to AICD ex vivo, T cell clones aged in culture also become increasingly susceptible to AICD; moreover, T cell lines derived from old donors become more susceptible more rapidly than those derived from young donors (258). In mice in vivo, evidence for increased apoptotic death of superantigen-stimulated T cells has been forthcoming (259). In humans, similar phenomena may be reflected in pathological states such as chronic phase HIV infection, where the fraction of apoptotic cells is greatly increased, especially amongst those with an activated (CD45RO+, DR+, Fas+, CD38+) phenotype, suggesting that chronic stimulation leads to clonal exhaustion by increased susceptiblity to AICD (260). Consistent with this is the finding that in vitro anti-oxidant treatment, which can inhibit AICD, can to some extent restore the proliferative defect of HIV-infected CD4 cells (261). Also consistent with the idea of clonal exhaustion, monitoring HIV-infected individuals for strength and breadth of proliferative responses to HIV peptides revealed that patients with weaker responses progressed more slowly than those with higher responses (262). This could be interpreted to mean that stronger proliferative responses, while neuroprotective (262), result in more rapid clonal exhaustion and therefore disease progression. However, the idea that increased levels of AICD are detrimental to functioning of the immune system must be reconciled with data from several sources suggesting an age-associated increase in resistance to apoptosis on the part of cells from various tissues including lymphocytes. Thus, Zhou et al . (263) generated fas transgenic mice and compared immunological status of young and old transgenics with wild-type littermates. They found that fas expression (like the expression of some other receptors in the same family, eg. TNF-R) and fas-induced apoptosis was decreased in old wt mice, but not old transgenics. In old wt mice, there was an increase in CD44+ fas-negative cells, decrease in autocrine proliferation, decreased IL 2 production and increased IL 4 and IL 10 production. In the transgenics these changes were not found. Even age-related thymic involution was prevented in the fas-transgenics. It was therefore suggested that some of the manifestations of aging on the immune system were related to downregulated apoptosis (263). However, the lifespans of the transgenics were not increased and this seemed to be associated with enhanced production of IL 6 and other factors in these mice (264). How might the transgenic fas expression exert these effects? Perhaps by removing defective cells by apoptosis and making room for fresh cells? More likely may be the alternative function of fas, ie. that of lymphocyte stimulation rather than killing. Thus, low to intermediate fas expression (and ligation) results in apoptosis, whereas high level expression can protect against cell death (265), and thereby result in enhanced responsiveness. Spaulding et al . (266) provided evidence partly consistent with that of Zhou et al . in normal mice, where they demonstrated that T cell apoptosis induced by irradiation, heat shock or CD3-stimulation was reduced in old compared to young mice, unless the former had been maintained on a calorically restricted diet. Polyak et al. 267) also reported higher levels of in vivo and in vitro lymphocyte apoptosis after irradiation in young compared to old mice. Some further supporting data for the concept of decreased apoptosis in aged cells may be found in the report of Lechner et al . (258) who found decreased inducibility of CD95 after CD3-stimulation of old persons´ T cells compared to young. However, as already discussed above, susceptibility to AICD of T cell lines established from old donors was greater than those from young donors (258). Other tissues may also show increasing susceptibility to apoptosis with age, as is the case with hepatocytes (268). Here, caloric restriction also reverses the age-associated effects and reduced apoptosis, the opposite of its effect on lymphocytes.

[Frontiers in Bioscience 3, d59-99, January 15, 1998]
Reprints
PubMed
CAVEAT LECTOR

T CELLS AND AGING

Graham Pawelec ¹, Ed Remarque ², Yvonne Barnett ³, Rafael Solana⁴

1 University of Tübingen, Tübingen, FRG ²,University of Leiden, Leiden, Holland ³, University of Ulster, Coleraine, Northern Ireland ⁴, University of Córdoba, Córdoba, Spain

Received 12/29/97 Accepted 1/5/97

8. ALTERATIONS IN T CELL SUBSETS AND MARKERS WITH AGING

Is the senescent phenotype due to increases in memory cells and to their "clonal exhaustion"?

Age-associated changes in T cell subsets occurring in rodents and humans have been repeatedly documented. Mice seem to show a relative loss of CD4 cells with age compared to CD8 cells. Whereas the TCR repertoire of the surviving CD4 cells is said to remain unchanged compared to young cells, the CD8 repertoire is markedly altered, suggesting expansion of a small number of CD8 (regulatory?) cells during aging (192). On the other hand, Rosenberg et al . failed to find quantitative differences in CD4⁺ cells of old mice, but noted a striking decline in the ability of CD4+ cells to cause rejection of allogeneic skin grafts, whereas CD8+ cells retained fully their function as effectors (193). Moreover, the defect in CD4+ cells could not be overcome by treating the cells with soluble T cell helper factors (containing many kinds of cytokines as well as IL 2). The TCR1 repertoire may also change with age, although the meaning of this finding is unclear. Thus, Giachino et al . (194) demonstrated that the polyclonal Vd 1 and Vd 2 populations present in children changed to oligoclonal in the elderly (57 - 88 yr). This may reflect "memory" for common antigen. Although not observed by all investigators (195), age-associated changes in T cell repertoire may apply to human as well, according to reports by Posnett et al . (196). Expansion of CD8 cells is common, but it may be the rarer CD4 expansions which are observed at increasing frequency in the aged (197). However, in some disease states, eg. RA, CD4 cells may show striking oligoclonal expansions (198). Whilst not seen in all normal donors, it is interesting to note that such CD4 expansions were also seen in unaffected siblings of RA patients, suggesting that they may be a genetic risk factor for rather than a consequence of RA (198). In CD8⁺ but not CD4⁺ T cells, up to >30% of the entire population may consist of oligo- or even monoclonal cells expressing the same TCR-Vß markers (196). Within the CD8⁺ cells, these oligo- or monoclonal populations are prevalent within the CD28-negative subset (196) and the CD57-positive subset, which often overlaps with the CD28-negatives (199). It is interesting to note that it is this CD28-negative, CD8-positive subpopulation which was identified many years ago as containing suppressor cells (200). These data may explain the observation that alterations in proportions of different T cell subsets may also be more marked in CD8⁺ than in CD4⁺ cells of aged humans (201). However, this phenomenon may not be absolutely limited to CD8 cells. Recent data from Thorbeke´s group suggest that although "forbidden" CD4 clones are not present in 24 month-old mice (usually the uppermost limit in mouse aging studies), they do appear in those few mice reaching 30 months of age (202). It was suggested that these potentially self-reactive CD4 cells were derived extra-thymically because thymectomy increased rather than decreased their numbers. There is a considerable literature on extra-thymic T cell development in mice, and evidence for increased development of self-reactive extra-thymic, but not thymic-derived, T cells with age (203). The possibility of enhancing extra-thymic development with factors such as oncostatin-M may offer the opportunity for manipulation of this pathway (204). Moreover, the realization that mature thymus-derived T cells can re-acquire sensitivity to positive and negative selection outside the thymus, in germinal centers (205), indicates in theory the generation and selection of T cells may be effected even the absence of a functional thymus.

There are clearly more CD4⁺ CD45RO⁺ (and in mice, CD44^hi) "memory" cells and less CD45RA⁺ (and, in mice, CD44^lo) "naive" cells in PBMC from elderly individuals. Whole blood analyses confirm the relative paucity of CD45RA+ cells in both the CD4 and CD8 populations of the elderly (129). If the CD45RO⁺ cells represent memory cells, and if exportation of naive CD45RA⁺ cells from the thymus decreases with age, then an accumulation of CD45RO⁺ memory cells would be expected in elderly donors. This would be coupled with a predicted reduced ability to respond to new antigens, and a retained ability to respond to recall antigens, as long as the memory cells remained present. Certainly, the proportion of RA⁺ cells decreases and RO⁺ cells increases with increasing age (206). However, in the oldest old, decreases in memory cell phenotype RO⁺ cells have also been recorded (160) and in whole blood analyses, a relative decrease of CD45RO⁺ cells may also be seen in the CD8 but not in the CD4 population (129). However, in exceptional individuals (healthy centenarians) the decrease of RA⁺ cells, especially in the CD8 subset, may be markedly less than in the ordinary old population (207). The meaning of this finding is unclear, because functional tests were not performed, and it is known from other studies that the CD4⁺ cells responsible for the increase of RO⁺ elements express lower levels of CD45RO than do young CD45RO⁺ cells (208). Whether this is related to their impaired function (114,115) is not yet known (208). More subtle analyses may reveal further differences in surface phenotype and function, which remain to be collated and understood. R. Miller´s group has shown in mice, for example, that aging leads to an increase in the proportion of splenic cells expressing high activity P-glycoprotein (Pgp^hi) and therefore able to extrude rhodamine 123. Pgp is known also to participate in the transportation and secretion of cytokines including IL 2, IL 4 and IFN-g (209). Moreover, mAb against Pgp block IL 2 release in PHA-activated T cells, demonstrating a critical role for this molecule in T cell function (210). Despite this, high Pgp expression in old T cells is linked to dysfunctional status. Possibly the Pgp itself is dysfunctional and over-expression represents an attempted compensation. There is an age-associated increase in the expression of MHC class I molecules on these high-Pgp CD4 memory cells. In spite of this, the levels of TAP1 decrease in old mice (TAP1 transporter is usually required for class I peptide loading). The Pgp and TAP1 molecules are related but whether Pgp is taking over the function of TAP1 in old cells and increasing class I remains an open question (211). The failure of Pgp^hi cells to respond to TCR-mediated stimulation cannot be overcome by CD28 signalling, PMA, IL 4 or IL 12 or combinations of these, and may therefore be considered "anergic". Not only do Pgp^hi cells fail to secrete IL 2 but also show impaired IL 5 and IL 10 production and proliferation. Oddly, however, their ability to secrete IFN-g increases with age (212). However, all these data were gathered exclusively in murine systems and it is not known whether human cells behave similarly.

The question of whether antigen-independent functional changes in naive T cells can occur has also been addressed. As discussed above, differentiation of T cells to memory cells coupled with age-related changes in memory cell characteristics may be responsible for much of the altered functional phenotype of the aged individual. Linton et al . looked at TCR-transgenic mice with T cell specificity for pigeon cytochrome C antigen, for which they believed they had good evidence for lack of cross-reactivity with environmental antigens. They found that in aged animals, the TCR-transgenic CD4+ cells were decreased in number and in antigen responsiveness but that they maintained a naive cell phenotype. They concluded that the defects observed were therefore due to aging of the naive cells per se and not to environmental stimulatory influences (213). Such findings are clearly consistent with several studies showing different patterns of cytokine production by young and old cells despite posession of the same "naive" phenotype (214).

In human, most T cells are CD7+, but the frequency of CD7-negative cells increases with age (215) and although isolated T cell clones retain stable expression of their CD7-positive or negative phenotype (216), repeated stimulation and propagation of uncloned lines results in accumulation of CD7-negative cells in the CD4 but not CD8 subset (217). Increased proportions of CD7-negative cells are found in situations of chronic antigenic stimulation in vivo, eg. in RA (218) and in kidney transplant recipients (219). Such CD7-negative cells show low proliferative responses to CD3-stimulation, low IL 2 secretion but high IL 4 and IL 10 secretion (220). These results suggest that loss of CD7 expression may be age-associated, but the fact that long-term cultured T cell clones retain high CD7 levels imply that factors other than merely the number of PD undergone are critical for CD7 expression.

Thus far, CD28 is perhaps the closest to a biomarker of aging found for human lymphocytes. Both in vivo and in vitro, the proportion of CD28⁺ cells decreases with age. In monoclonal populations, the density of expression of CD28 decreases with age (221). Effros et al . were the first to observe a decreasing percentage of CD8 cells carrying CD28 in the elderly, paralleling their observations in T cell lines aging in culture (168). Others have confirmed that particularly the CD8 subset shows progressively decreasing CD28 expression with age (222). There is a correlation between shortening of telomeric repeats and age of the donor, which is not confounded by differences in white blood cell count (223). Moreover, telomere lengths (TL) in the CD28-negative cells were less than in the CD28⁺ cells from the same donors, implying that the former had undergone more rounds of cell division than the latter (224) (see section 7). This type of proliferative senescence may therefore be responsible for the commonly observed accumulation of CD28-negative oligoclonal populations in elderly people (196). Although originally described only in CD8 cells, the number of individuals with such clonal expansions in both CD4 and CD8 cells was very similar (ca. 70% of individuals over 65); moreover, these expansions were stable over a two-year observation period (225).

In diseases with chronic antigenic stimulation, further circumstantial evidence in favour of the hypothesis of proliferative senescence indicated by downregulated CD28 expression can be garnered. To give some examples: the percentage of CD28⁺cells decreases during Chagasic progression (226); both CD4 and CD8 cells show decreased CD28 expression in chronic B lymphocytic leukemia (227); in rheumatoid arthritis, the percentage of CD4 cells carrying CD28 is reduced (228) and in both RA patients and normal controls the CD4⁺ CD28-negative cells show TCRVB oligoclonality (229); in Crohn´s Disease, the ability of CD28 to mediate costimulation of CD4 cells is compromized (230). These examples suffice to illustrate the range of situations in which T cell proliferative senescence may play a role in modulating immune responses independently of the age of the host. The effects of this kind of "clonal exhaustion" in the elderly may simply be more noticeable than in the young because of thymic involution preventing effective generation of naive T cells and because T cells present in the old may already have undergone many rounds of division.

8.1 Longevity of naive and memory cells

An important question raised here concerns the longevity of naive and memory cells. There is some evidence that memory cells are not quiescent long-lived cells, but represent T cell clones in a constant dynamic state of activation. Beverly´s group used a method for measuring intermitotic time to investigate longevity of CD45RA and RO cells in cancer patients following radiotherapy. Irradiation induced dicentric chromosomal lesions which can be visualized cytogenetically. They found small numbers of CD45RA⁺ cells with these lesions up to 10 years after irradiation, consistent with the belief that naive T cells can be very long lived. However, CD45RO+ cells with such lesions had all disappeared by one year, suggesting that they had all attempted unsuccessfully to divide, consistent with memory cells being in cycle (158). A later study extended these data to conclude that proliferation rates of naive cells were 8x lower than proliferation rates of memory cells. They estimated that, on average, naive cells divided once every 3.5 years, whereas memory cells divided every 22 weeks (231). However, as these authors themselves pointed out, there are some problems with these data, viz. they were dealing with cancer patients, whose T cells and immune status were not normal; irradiation can have indirect effects like induction of lymphopenia which further disrupts the system; possibly some cells with dicentric lesions can nonetheless divide; and the proportion of cells dying without attempting to divide is unknown. All these factors could lead to underestimates of memory cell longevity (232). A direct approach to whether memory cells continue to cycle in vivo has been taken using transgenic mice (233). Naive cells deliberately activated with specific antigen were transfered into athymic hosts in the absence of antigen and found to continue to cycle slowly for extended periods. Naive cells, on the other hand, did not begin to proliferate when transferred into the these recipients. The mechanism for the maintenance of slow cycling of the memory cells is unknown. It may be that unlike naive cells, memory cells do not require antigen to survive and proliferate, but only the self MHC molecule (ie. have a lower functional activation threshold, ref. 234). However, it is clear from these experiments that proliferative senescence could play a role in the eventual loss of the memory cells. The above data in man and mouse have to be reconciled with ealier results in the mouse showing that after thymectomy it is the naive population which disappears rapidly, whereas the memory cells are long-lived (235). The reason for this apparent discrepancy is unknown.

According to our in vitro data on T cell longevity, where TCC survive on average 35 PD and maximally 80 PD, this would put the life expectancy of a memory cell at 35 x 22 = 770 wk = 15 years average; 80 x 22 = 176 wk = 34 years maximum (without taking the primary immune response into account, for which the number of cell divisions required can only be guessed at (see ref. 236)). What might be the source of stimulation of the memory cells, which need to persist in the absence of antigen? Several groups have argued that the solution lies in the hypothesis that antigen does persist somehow. This is certainly true for many viruses, but it is unclear for which other antigens it may apply. For viral infection, there is clear evidence that memory T cells are indeed maintained in a constant activated state even for very long periods. Thus, Rehermann et al . studied patients up to 23 years after clinical and serologic resolution of HBV infection and still found evidence for recently activated HBV-specific CTL (237). This might well eventually allow enough time for "clonal exhaustion" of the originally responding T cell clones, in the absence of the generation of de novo responses (see calculation above). Data from other chronic infections such as HIV may be consistent with the scenario of clonal expansion leading to clonal exhaustion and lack of replacement by new thymic emigrants eventually resulting in diminution of the repertoire and loss of anti-HIV responses (238). CD8 cells of the same CD28-negative phenotype as seen in HIV-uninfected senescent cultures and possessing similarly short telomeres have also been described in young persons with AIDS, leading to the suggestion that replicative senescence of virus-specific T cell clones in vivo might contribute to disease progression (239). A marked decrease in telomere length has also been recorded in CD4, CD8 and B cells from HIV-infected patients with advanced immunodeficiency, supporting the notion of a high turnover of these cells and suggesting that replicative senescence may be involved in the final immunosuppression of these patients (240). There is also an increase in CD4⁺ CD28-negative cells in HIV infection, albeit not so marked as in the CD8 subset (241). Indeed, disease progression in AIDS is reported to be marked by an accumulation of CD28-negative cells unable to secrete IL 2, whereas long-term non-progressor patients maintain CD28 levels and IL 2 secretion capacity (242). The rate of destruction of HIV-infected cells in young and old patients seems similar, leading to the suggestion that the elderly cannot replace CD4 cells as rapidly as the young (243). It has been found that T cell responses in HIV patients are characterized by severe TCRVB biases and clonal expansions in CD4 cells, and that such responses are exaggerated with disease progression (244). Despite this, others have found no evidence for increased CD4 turnover in HIV infection on the basis of lack of truncation of telomere length in CD4 cells during progressive disease, although this was clearly confirmed in CD8 cells (245). However, virally infected cells may conceivably express dysregulated telomerase. Independent evidence for enhanced cycling of CD4 cells in HIV derives from measurements of mutations in the hprt locus, which showed that mutation rates in the CD8 and CD4 cells were similarly high (246), consistent with an increased division rate in both subsets. The cytokine secretion profile of mutant CD4 clones (from healthy or HIV patients) was predominantly Th2-like, whereas the CD8 mutants had the same pattern as wild-type.

For non-viral antigens, and thymus-independent regeneration of T cells in general,it is probable that antigen-driven peripheral expansion commonly occurs (247), implying that a finite proliferative lifespan of the T cells would be of critical import for the functional intergrity of the immune system. Experimental data have implicated a sort of TCR cross-reactivity responsible for maintaining numbers of T cells, eg. in the case of CD4 cells via class II molecules regardless of their peptide loading (248).

8.2 Activation-induced cell death and aging

Further light has been cast on the aging of human CD4⁺ T cells using the mAb PD7/26 specific for the CD45RB isoform. Salmon et al (249) reported that whereas CD45RA expression is lost rapidly after activation of naive cells in vitro, loss of RB expression is gradual, occurring over many cell cycles, and reciprocating the increase in RO expression. The progressive shift from RB^hi+ to RB^lo+ is paralleled by gradual loss of bcl-2 (which protects against apoptosis) and acquisition of fas (which can mediate apoptosis), as well as gradual loss of ability to secrete enough IL 2 to maintain autocrine proliferation (whereas IL 4 secretion remains intact). This eventually results in the cells becoming dependent on exogenous IL 2 not only for growth but for their survival, because without enough IL 2 they undergo apoptosis. Longevity extension for T cell clones might therefore be achievable by upregulation of bcl-2. However, suppression of apoptosis by bcl-2 or bcl-2 family-member bcl-x(L) results in enhanced radiation-induced mutagenesis, consistent with the original isolation of bcl-2 as an oncogene (250).

The first report describing the behavior of CD45RB in vivo in elderly humans confirmed that CD4⁺cells in old donors showed significantly decreased 45RB expression (251). In agreement with the in vitro data, the percentage of human CD4 and CD8 fas+ cells also increases with aging in vivo (252,253). The majority of these express CD45RO, and CD25 and CD69 activation markers (254). Parallel to this, the amount of soluble fas in the blood of elderly donors is significantly increased compared to young donors (255). Moreover, Phelouzat et al . (256) have shown increased CD3-mediated AICD and hence decreased proliferation in the elderly. This was found not to be due to IL 2 deprivation, nor was it associated with decreased bcl-2 expression (256). There is a possibility that increased AICD might be associated with decreased levels of IL 6 because IL 6 has been reported to protect neonatal T cells from AICD (257). However, IL 6 production seems not to be decreased in older individuals, as discussed above. Consistent with the findings of increased susceptibility to AICD ex vivo, T cell clones aged in culture also become increasingly susceptible to AICD; moreover, T cell lines derived from old donors become more susceptible more rapidly than those derived from young donors (258). In mice in vivo, evidence for increased apoptotic death of superantigen-stimulated T cells has been forthcoming (259). In humans, similar phenomena may be reflected in pathological states such as chronic phase HIV infection, where the fraction of apoptotic cells is greatly increased, especially amongst those with an activated (CD45RO+, DR+, Fas+, CD38+) phenotype, suggesting that chronic stimulation leads to clonal exhaustion by increased susceptiblity to AICD (260). Consistent with this is the finding that in vitro anti-oxidant treatment, which can inhibit AICD, can to some extent restore the proliferative defect of HIV-infected CD4 cells (261). Also consistent with the idea of clonal exhaustion, monitoring HIV-infected individuals for strength and breadth of proliferative responses to HIV peptides revealed that patients with weaker responses progressed more slowly than those with higher responses (262). This could be interpreted to mean that stronger proliferative responses, while neuroprotective (262), result in more rapid clonal exhaustion and therefore disease progression.

However, the idea that increased levels of AICD are detrimental to functioning of the immune system must be reconciled with data from several sources suggesting an age-associated increase in resistance to apoptosis on the part of cells from various tissues including lymphocytes. Thus, Zhou et al . (263) generated fas transgenic mice and compared immunological status of young and old transgenics with wild-type littermates. They found that fas expression (like the expression of some other receptors in the same family, eg. TNF-R) and fas-induced apoptosis was decreased in old wt mice, but not old transgenics. In old wt mice, there was an increase in CD44+ fas-negative cells, decrease in autocrine proliferation, decreased IL 2 production and increased IL 4 and IL 10 production. In the transgenics these changes were not found. Even age-related thymic involution was prevented in the fas-transgenics. It was therefore suggested that some of the manifestations of aging on the immune system were related to downregulated apoptosis (263). However, the lifespans of the transgenics were not increased and this seemed to be associated with enhanced production of IL 6 and other factors in these mice (264). How might the transgenic fas expression exert these effects? Perhaps by removing defective cells by apoptosis and making room for fresh cells? More likely may be the alternative function of fas, ie. that of lymphocyte stimulation rather than killing. Thus, low to intermediate fas expression (and ligation) results in apoptosis, whereas high level expression can protect against cell death (265), and thereby result in enhanced responsiveness. Spaulding et al . (266) provided evidence partly consistent with that of Zhou et al . in normal mice, where they demonstrated that T cell apoptosis induced by irradiation, heat shock or CD3-stimulation was reduced in old compared to young mice, unless the former had been maintained on a calorically restricted diet. Polyak et al. 267) also reported higher levels of in vivo and in vitro lymphocyte apoptosis after irradiation in young compared to old mice. Some further supporting data for the concept of decreased apoptosis in aged cells may be found in the report of Lechner et al . (258) who found decreased inducibility of CD95 after CD3-stimulation of old persons´ T cells compared to young. However, as already discussed above, susceptibility to AICD of T cell lines established from old donors was greater than those from young donors (258). Other tissues may also show increasing susceptibility to apoptosis with age, as is the case with hepatocytes (268). Here, caloric restriction also reverses the age-associated effects and reduced apoptosis, the opposite of its effect on lymphocytes.