Progressive CD4⁺ central–memory T cell decline results in CD4⁺ effector–memory insufficiency and overt disease in chronic SIV infection

SIVmac239 and related CCR5-tropic SIVs are highly pathogenic for Indian origin RMs, inducing rapid disease progression (AIDS in <200 d) in ∼25% of infected animals and normal progression (AIDS in 1–3 yr) in the majority of the remainder (27). Fig. 1 A shows plasma viral loads (pvls) over the course of infection of 14 SIVmac239-infected RMs that did not receive antiretroviral therapy (ART) and were followed to an AIDS endpoint (4 rapid progressors and 10 normal progressors). As reported previously (15), rapid progressors manifest 1–2 log higher pvls than RMs destined for >350 d survival. However, among the latter group, plateau-phase pvls did not distinguish between RMs with considerably different survival periods. Indeed, when the entire study group was included in a correlation analysis of plateau-phase pvls versus time to endpoint, a significant association was observed (by either parametric or nonparametric analysis); however, when the four rapid progressors were excluded from the analysis, this association was no longer apparent (Fig. 1 B). Moreover, in all the normal progressors, pvls remained steady through the development of overt disease. Thus, within the range of viral replication manifested by typical SIV-infected normal progressors, the time to disease was not closely linked to mean pvls over time nor to any late acceleration of viral replication, observations suggesting that the rate of disease progression in these RMs was largely determined by their ability to handle the consequences of chronic viral replication, rather than by differences in viral replication itself.

A major consequence of acute CCR5-tropic SIV infection is destruction of CD4⁺ memory T cells, particularly CD4⁺ T_EM cells in extra-lymphoid effector sites, and we have previously shown that a strong immunologic correlate of rapid progression is the failure to follow this initial CD4⁺ T_EM cell depletion with a sustained increase in the fraction of proliferating CD4⁺ memory T cells (15). This increase provides “targets” for SIV (12, 20), but also contributes to the continuous production of new CD4⁺ T_EM cells that limits extra-lymphoid tissue deficits and prevents immunological collapse (15). As it is possible that the same mechanisms extend to chronic progression, we asked first whether CD4⁺ T_EM cell depletion was a specific and uniform feature of SIVmac239-mediated AIDS (Fig. 2). Because almost all bronchoalveolar lavage (BAL) T cells manifest the phenotypic signature, CCR7⁻ and/or CCR5⁺, of T_EM cells (18, 23), the fate of CD4⁺ T cells in this site provides insight into the extra-lymphoid CD4⁺ T_EM cell compartment. As shown in Fig. 2 A, analysis of 13 normal progressors and 4 rapid progressors demonstrated that overt disease in both settings was uniformly associated with a striking (>99%) decline in CD4⁺ T cell representation among BAL T_EM cells. Because total lymphocyte recovery from BAL in later stages of infection was comparable to or reduced from preinfection (unpublished data), this decline predominantly represents CD4⁺ T_EM cell depletion, rather than CD8⁺ T_EM cell expansion. Importantly, the representation of CD4⁺ T cells in BAL closely correlated with that of small intestinal lamina propria, another T_EM cell–populated effector compartment (18), suggesting that the BAL compartment was an accurate indicator of systemic effector site CD4⁺ T_EM cell representation (Fig. 2 B).

Decline in the absolute number of circulating total CD4⁺ memory T cells was also highly significant in both rapid and normal progression, albeit less dramatic than the loss of CD4⁺ T_EM cells in BAL (Fig. 2 A). In contrast, the numbers of CD4⁺ naive T cells in blood were not a strong correlate of AIDS as they were only variably and modestly diminished in chronic AIDS and, as described previously (15), were actually increased in rapid progression. Because of this CD4⁺ naive T cell heterogeneity, the absolute numbers of total CD4⁺ T cells in blood, the standard marker in disease progression in HIV infection (1), were significantly reduced only in the chronically SIV-infected RMs, and even in this setting, the extent of this loss was highly variable. The numbers of circulating CD8⁺ T cells, both memory and naive, also varied in overt AIDS but tended to be normal or elevated (Fig. 2 A). Cross-sectional analysis of the overall naive and memory T cell compartments of peripheral LNs (PLNs) in AIDS-stage normal progressors versus uninfected RMs revealed a statistically significant decline in CD4⁺ representation in the total memory, but not the naive, compartments (Fig. 2 C), providing further evidence that systemic CD4⁺ memory, but not naive, T cell depletion is a uniform T cell population correlate of AIDS.

Collectively, these data indicate that BAL CD4⁺ T_EM cell depletion is the most profound T cell population abnormality in AIDS (P < 0.0001), but as severe depletion has already occurred in acute infection, the resulting CD4⁺ T_EM cell deficit might not be expected to play a determining role in the much later onset of chronic-phase disease. However, as shown in Fig. 3, CD4⁺ T cell frequencies in BAL initially rebounded after acute depletion, but after day 100, they progressively declined in all normal progressors examined (with a mean population half-life of ∼107 d; see legend to Fig. 3). This observation strongly supports the hypothesis that CD4⁺ T_EM cell depletion constitutes a primary mechanism of immunodeficiency, with overt disease ensuing when the representation of these cells in extra-lymphoid effector sites declines below a critical level. At the time of AIDS onset in our cohort, the average (± SEM) fraction of CD4⁺ T cells within the BAL T cell compartment was 0.42 ± 0.08%, a value that provides a substantive estimate for this critical threshold.

To investigate the mechanisms underlying BAL CD4⁺ T_EM cell decline, we used short-term BrdU labeling (three doses over 24 h) to compare the dynamics of BAL T_EM cell populations in clinically stable, chronically SIVmac239-infected RMs versus uninfected RMs. Fig. 4 A shows a typical profile of a chronically infected RM. Note that peak labeling exceeds 25% of BAL CD4⁺ T cells, occurs 4 d after the last BrdU dose, and includes cells that are almost entirely postproliferative (Ki-67⁻). Collectively with previous observations (15), these data indicate that a substantial fraction of the BAL CD4⁺ T_EM cell population is replaced daily from recently divided precursors, originating elsewhere, and emigrating to the lung via the blood. The vast majority of labeled cells disappear over the ensuing month, which, in the absence of proliferative dilution, is most likely due to in situ death (and replacement by nonlabeled cells). Both the peak labeling of CD4⁺ T_EM cells in chronically infected RMs and the subsequent reduction in number of these labeled cells were two- to threefold greater than in uninfected RMs (Fig. 4 B), indicating a substantial increase in the fraction of short-lived cells within the BAL population of infected versus uninfected animals. Notably, the peak BrdU labeling and the magnitude of the subsequent decay of CD8⁺ T_EM cells in BAL were similarly elevated in the infected versus uninfected RMs, and effective ART normalized the BrdU labeling profiles of both CD4⁺ and CD8⁺ T_EM cells (Fig. 4 C). As viral cytopathogenicity would affect CD4⁺ T_EM cells much more than CD8⁺ T_EM cells, the essentially identical dynamics and ART response of the two subsets strongly suggest that the accelerated turnover of BAL T_EM cells primarily reflects immune activation-mediated apoptosis, rather than viral-mediated destruction. To further assess whether direct coreceptor-targeted, viral-mediated cell destruction contributes to the rapid decline of BrdU-labeled CD4⁺ T_EM cells in BAL, we compared the frequency at which these cells expressed CCR5 over time to that of labeled CD8⁺ T_EM cells (Fig. 4 D). Direct viral-mediated killing would be expected to preferentially target the CCR5⁺ component of CD4⁺ T_EM cells, resulting in decreasing frequencies of CCR5⁺ cells over time among labeled CD4⁺ T_EM cells, but not CD8⁺ T_EM cells. Importantly, CCR5 expression was largely stable in the labeled CD4⁺ T_EM cells in the BAL of infected RMs, further indicating that if direct viral killing of recently divided CD4⁺ T_EM cells in BAL was occurring, its contribution to the decreased lifespan of these cells in infected RMs must be quite small relative to their overall rate of loss.

The above data indicate that during chronic infection, the diminished BAL CD4⁺ T_EM cell population is largely comprised of relatively short-lived cells that are highly dependent on new memory cell production and emigration for homeostasis, a finding suggesting that CD4⁺ memory T cell production dynamics might play a role in the progressive decline of BAL CD4⁺ T_EM cells in chronically infected RMs. As shown in Fig. 5 A, the proliferating (Ki-67⁺) fraction of CD4⁺ memory T cells in the blood may indeed fall during chronic infection (RMs nos. 20197 and 20408), but this decline is not uniform, coming very late in some RMs (no. 21064) and not at all in others (no. 20502). However, systemic production is more directly related to the overall number of proliferating cells, and we therefore followed the absolute number of proliferating total memory T cells in the blood throughout the course of infection in eight untreated SIVmac239-infected normal progressors and in four RMs with attenuated (clinically nonprogressive) SIVmac239(Δnef) infection (Fig. 5 B). This analysis demonstrated a highly significant progressive loss of proliferating CD4⁺ memory T cells in the normal progressors, but not of proliferating CD8⁺ memory T cells in the same RM. In attenuated SIVmac239(Δnef) infection, the numbers of proliferating CD4⁺ memory T cells showed a very slight decline over 600 d that did not quite achieve statistical significance.

The overall memory T cell population in blood is composed of three phenotypically distinguishable components:T_CM cells, transitional T_EM cells, and fully differentiated T_EM cells, with both T_EM cell subsets expressing CCR5 and being effector site directed (Fig. 6 A) (18). On average, approximately one third of the proliferating CD4⁺ memory T cell compartment in the blood of uninfected RMs is comprised of CCR5-expressing T_EM cell subsets; however, in SIV-infected RMs, this fraction is highly dependent on the degree of pathogenesis. In attenuated SIV infections or wild-type SIV infections in RMs with spontaneous control, we observed a significant increase in the T_EM cell fraction of proliferating CD4⁺ memory cells in comparison to uninfected RMs, consistent with the expected effect of chronic immune activation in driving effector differentiation (28). In contrast, the T_EM cell fraction was reduced approximately fourfold throughout pathogenic infection (Fig. 6, B and C). CD4⁺ T_EM cells (predominantly transitional type) were also reduced in PLNs in chronic SIVmac239 infection (Fig. 6 D), indicating that the reduction in proliferating CD4⁺ T_EM cells in the blood was not due to retention in secondary lymphoid tissues. Moreover, effective ART results in an immediate burst of proliferating or recently divided transitional CD4⁺ T_EM cells in the PLNs and blood (Fig. 6, D and E), followed later by mature CD4⁺ T_EM cells (Fig. 6 E), suggesting that although rapid CD4⁺ T_EM cell generation and differentiation is continuously driven by infection-associated immune activation, these new T_EM cells and/or their immediate precursors are being destroyed almost as rapidly by direct virus-mediated cytopathogenicity. Importantly, the efficiency of this depletion does not appear to change over the course of chronic infection (compare early vs. late plateau-phase in Fig. 6 C), suggesting that this effect acts as a relatively constant impediment to CD4⁺ T_EM cell production that does not, by itself, explain the progressive decline in CD4 T_EM cell populations in extra-lymphoid sites.

Collectively, these data (an overall decrease in proliferating CD4⁺ memory T cells and stable rates of CD4⁺ T_EM cell differentiation) lead to the conclusion that the progressive decline in CD4⁺ T_EM cell production in chronic SIV infection originates from a progressive decline in proliferating CD4⁺ T_CM cells caused by a decreasing CD4⁺ T_CM cell proliferation rate and/or a progressive decline in overall CD4⁺ T_CM cell numbers (Fig. 2, A and C). Factors contributing to loss of CD4⁺ T_CM cell homeostasis would therefore constitute primary determinants of immunodeficiency development. One such factor might be persistent immune activation with its previously demonstrated effect of increasing T cell proliferation and turnover (12, 34). Indeed, we found that in the chronically SIV-infected RMs studied here, CD4⁺ T_CM cells manifest a higher fraction of proliferating cells with rapid turnover (Fig. 7 A), potentially leading to proliferative senescence and/or inadequate renewal of long-lived cells (11, 29). However, interpreting these factors as contributing to the progressive failure of CD4⁺ T_CM cell homeostasis is not straightforward because CD8⁺ T_CM cell populations, which are relatively stable during chronic infection, manifest similar, although somewhat less severe, changes in proliferation and turnover in infected RMs (Fig. 7 A).

A factor with perhaps greater potential to differentially affect CD4⁺ T_CM cell versus CD8⁺ T_CM cell homeostasis is direct SIV infection of the former subset. True CD4⁺ T_CM cells lack detectable surface expression of the CCR5 coreceptor and thus would be predicted to be protected from infection; however, previous work has suggested that detectable surface expression of CCR5 may not be necessary for a CD4⁺ memory cell to be infected (30). Indeed, as shown in Fig. 7 B, SIV DNA was readily detectable by real-time PCR in sorted, precisely defined (CCR7⁺ CCR5⁻) CD4⁺ T_CM cells from both early plateau-phase RMs and normal progressors with AIDS. SIV DNA was either absent from CD8⁺ T_CM cells or found at significantly lower levels. In normal progressors with AIDS, SIV DNA levels in CD4⁺ T_CM cells were ∼15% of those detected in CD4⁺ (CCR5⁺) transitional T_EM cells, consistent with a less efficient infection of the former subset. Interestingly, the SIV DNA content of CD4⁺ T_CM cells from RMs with early plateau-phase SIVmac239 infection was significantly (threefold) higher than in the RMs with end-stage infection, whereas SIV content of CD4⁺ T_EM cells was not different between these RM groups. This difference in CD4⁺ T_CM cell infection was not related to overall SIV replication, as pvls were not significantly different in the two groups. However, CD4⁺ T_CM cell proliferation (%Ki-67) was significantly higher in the early plateau-phase RMs, suggesting that CD4⁺ T_CM cell infection can be driven by high activation/proliferation to levels approaching those of CD4⁺ T_EM cells with overt CCR5 expression.

To determine the pathobiologic significance of this CD4⁺ T_CM cell infection, we compared the dynamics of (CCR7⁺ CCR5⁻) CD4⁺ and CD8⁺ T_CM cells during acute SIVmac239 infection and through early plateau-phase ART administration. As illustrated in two representative RMs (Fig. 7 C), absolute numbers of CD4⁺ T_CM cells in the blood decline ∼70% in acute infection, concomitant with a 98–99% depletion of BAL CD4⁺ T_EM cells in the same RM (not depicted). The observations that (a) CD8⁺ T_CM cells show only a modest, very transient depletion during this period (Fig. 7 C), and (b) that CD4 representation in the PLN T_CM cell compartment declines to a similar degree as blood CD4⁺ T_CM cell counts (not depicted) suggest that the CD4⁺ T_CM cell depletion primarily reflects direct viral-mediated killing of CD4⁺ T_CM cells, rather than a redistribution effect. Increased proliferation rates stabilize circulating CD4⁺ T_CM cell numbers in plateau-phase infection, but continued viral-mediated killing is demonstrated by the observation that ART administration in plateau-phase results in a very rapid and lineage-specific increase in CD4⁺ T_CM cell counts and in the fraction of proliferating CD4⁺ T_CM cells (Fig. 7 C). A similar increase in proliferation with ART is observed in PLNs (Fig. 7 D), suggesting that ART is acting immediately to relieve direct viral-mediated destruction of rapidly proliferating CD4⁺ T_CM cells. Thus, despite undetectable CCR5 expression, CD4⁺ T_CM cells, particularly proliferating cells, are SIV infectable and directly killed by virus.

In the past few years, it has been widely documented that primary infection with pathogenic, CCR5-tropic HIV and SIV, though generally asymptomatic, has extensive immunologic consequences (12, 20, 21, 23). The high viral replication of primary infection rapidly destroys the bulk of the body's CD4⁺ memory T cell population, particularly CD4⁺ T_EM cells in mucosal extra-lymphoid effector sites (15, 19, 30, 31), and also initiates a state of persistent immune activation that has been widely implicated in immunodeficiency development (10–12, 21, 24, 25, 32). However, CD4⁺ memory T cells are not equally susceptible to this acute destruction, and in SIV-infected RMs, we have previously demonstrated that spared populations (predominantly CCR5⁻ T_CM cells) undergo a substantial increase in proliferative activity that sustains these populations and provides sufficient T_EM cell production and influx into effector sites to maintain clinical immune competence (15). Chronically SIV-infected RMs and HIV-infected humans establish a new immunologic steady-state associated with high CD4⁺ memory T cell turnover and characterized by a substantially diminished CD4⁺ T_EM cell compartment (11, 12, 15, 22, 33–35). Ultimately, this steady-state is degraded and immunodeficiency ensues, but the quantitative changes in rate-limiting parameters that underlie this degradation in the course of chronic infection, as well as the mechanisms responsible for these changes, have not been clearly defined.

Here, we systematically assessed the relationship between potential determinants of pathogenesis and AIDS onset in chronically SIVmac239-infected RMs. Although it remains possible that there are multiple paths to AIDS in this model, our data strongly support a common “primary” pathway and suggest a surprisingly cohesive mechanistic sequence. Two important virologic observations guided our investigation: first, that within the range of plateau-phase pvls manifested by SIVmac239-infected normal progressors, these pvls did not strongly predict time to disease, and second, that the plateau-phase pvls in these normal progressors remained steady through to disease onset. Regarding the first observation, it is clear that in HIV infection, plateau-phase pvls delineate broad groups of individuals with different rates of disease progression (3), but the predictive value of viral replication is limited (36), and host factors (such as the degree of immune activation) have been demonstrated to exert a major influence on prognosis (24–26). Similarly, although SIVmac239 replication rates predict rapid versus normal versus slow progression in RMs (15, 37–39), they did not account for the range of disease-free survival times exhibited by normal progressors in our study, strongly suggesting that in these RMs, host factors are the primary determinants of chronic-phase disease. The stability of plateau-phase pvls through disease onset in the normal progressors studied here further supports this hypothesis, suggesting that a diminishing ability of host immunity to control viral replication or the emergence of more efficiently replicating viral variants were unlikely to be the critical processes leading to AIDS.

We next established that BAL CD4⁺ T_EM cell depletion was the strongest correlate of disease. Most importantly, having shown that changes in the CD4⁺ T_EM cell representation in BAL closely paralleled changes in the small intestine, we demonstrated a highly consistent, gradual progression in BAL CD4⁺ T_EM cell depletion from the initial post-acute steady-state to disease onset in chronic, progressive infection. The latter observation supports the existence of a threshold in effector site CD4⁺ T_EM cell representation below which the SIV-infected host becomes highly susceptible to opportunistic infection. This interpretation is even more compelling given our previous observation that rapid progression to AIDS was associated with cessation of CD4⁺ T_EM cell production and influx into extra-lymphoid sites (15), and the fact that the opportunistic infections found in our AIDS-afflicted RMs (CMV, cryptosporidia, pneumocystis) most commonly developed in such effector sites (unpublished data). We do not discount the possibility that functional or repertoire alterations of CD4⁺ or CD8⁺ memory T cells or abnormalities in accessory non–T cells might also contribute to immune deficiency, but would postulate that such qualitative abnormalities would serve to facilitate immune deficiency only in the setting of profound CD4⁺ T_EM cell depletion, perhaps affecting the threshold of CD4⁺ T_EM cell depletion at which overt immune deficiency ensues.

BrdU labeling of chronically SIVmac239-infected RMs was then used to explore the dynamics of pulmonary CD4⁺ T_EM cells and provide insights into the mechanisms underlying their progressive decline. We found that this population is largely comprised of short-lived nonreplicating cells and consequently is highly dependent on the continuous emigration of recently divided T_EM cells from the blood. Notably, pulmonary CD8⁺ T_EM cells manifested similar kinetic properties suggesting that (a) the short life span of the CD4⁺ T_EM cells in BAL was not attributable to direct viral-mediated destruction, but rather was likely a result of chronic immune activation (12, 34), and (b) that the progressive decline in CD4⁺ T_EM cells was determined elsewhere, probably by factors controlling the supply of these cells to extra-lymphoid effector sites. In keeping with the latter hypothesis, we found that absolute numbers of proliferating or recently divided CD4⁺, but not CD8⁺, memory T cells in the blood progressively declined in chronic SIVmac239 infection (but not in attenuated SIVmac239(Δnef) infection). The fraction of these proliferating or recently divided CD4⁺ memory cells exhibiting T_EM cell differentiation (i.e., the immediate precursors to extra-lymphoid site T_EM cells) was considerably diminished in progressive infection by direct virus-mediated destruction (Fig. 6), but, importantly, this fraction did not appear to change over the course of chronic infection. Collectively, these findings strongly suggested that a progressive, systemic decline of the proliferating CD4⁺ T_CM cell compartment resulted in a progressively diminished substrate for CD4⁺ T_EM cell differentiation. This in turn reduced CD4⁺ T_EM cell production and effector site emigration, leading to decreasing effector site CD4⁺ T_EM cell populations and ultimately (when the latter fell below threshold) to AIDS (Fig. 8).

In contrast to a previous report (33), CD4⁺ naive T cell populations were not generally depleted in the progressive infections studied here, as measured by both the absolute numbers of CD4⁺ naive T cells in the blood and the relative representation of CD4⁺ versus CD8⁺ naive T cells in PLNs. Thus, at least in the SIVmac239 AIDS model, CD4⁺ T_CM cell homeostasis can fail in the face of a relatively intact CD4⁺ naive cell compartment. Moreover, studies of thymectomized RMs infected with CXCR4-tropic (naive CD4⁺ T cell–depleting) SIVmac155T3 (15) have demonstrated that CD4⁺ T_CM cell homeostasis can be maintained long-term in the complete absence of a CD4⁺ naive T cell compartment (unpublished data). These data suggest that although naive CD4⁺ T cells are nominally the precursors for CD4⁺ T_CM cells, the naive population is not a major determinant of CD4⁺ T_CM cell homeostasis in chronic SIV infection.

What then is responsible for the progressive decline in CD4⁺ T_CM cell production? BrdU labeling studies demonstrate that both CD4⁺ and CD8⁺ T_CM cell populations manifest a sustained increase in cell turnover in chronic SIV infection, consistent with an immune activation effect (12) and potentially setting up these populations for slow proliferative senescence (10, 29). This increase in turnover, measured in the blood, is slightly more pronounced for CD4⁺ T_CM cells than CD8⁺ T_CM cells, but this difference, by itself, seems unlikely to explain the gross disparity in CD4⁺ versus CD8⁺ T_CM cell stability. However, we also demonstrate that, as opposed to CD8⁺ T_CM cells, CD4⁺ T_CM cells are readily infected by CCR5-tropic SIV in vivo (albeit less efficiently than CCR5⁺ T_EM cells), and are, in fact, continuously destroyed in vivo by a direct, virus replication-dependent mechanism (as revealed by their immediate post-ART kinetics). This killing clearly targets proliferating CD4⁺ T_CM cells in LNs (Fig. 7 D) before their emigration into the blood, therefore making their loss “invisible” to BrdU analysis in the blood (indicating that total CD4⁺ T_CM cell loss rates are underestimated by this analysis). This ongoing, virus-mediated CD4⁺ T_CM cell destruction, especially of the proliferating compartment in LNs, would clearly put CD4⁺ T_CM cell homeostasis under more stress than its CD8⁺ counterpart, potentially leading to imbalance in production versus destruction in the former, but not the latter, subset. Other factors could also contribute to a CD4⁺ T_CM cell homeostatic imbalance, including progressive damage to microenvironments supportive of T_CM cell homeostasis (9, 40) and progressive loss of regeneration capacity due to accelerated “aging-like” disorganization of the immune system (12). Even though such factors would potentially affect both CD4⁺ and CD8⁺ subsets, the effects need not be equal (9, 21).

Our data define a complex interplay between cell destruction by direct infection on the one hand and immune activation on the other in AIDS pathogenesis. With regard to the CD4⁺ T_EM cell compartment, direct viral-mediated destruction (either productive infection or virus-triggered apoptosis) is almost certainly responsible for the initial depletion of extra-lymphoid site CD4⁺ T_EM cells in acute infection (15, 30, 31) and for the continuous elimination of ∼70% of the differentiating CD4⁺ T_EM cells arising from T_CM cell precursors in PLNs and blood (this study). In chronic infection, however, the fate of extra-lymphoid site T_EM cells, both CD4⁺ and CD8⁺, appeared to be governed by immune activation with a considerably larger proportion of cells of both subsets possessing shorter life spans (due to activation-induced apoptosis) than those observed in uninfected controls. Interestingly, although infected CD4⁺ T cells can be visualized by in situ hybridization in the extra-lymphoid effector sites of these RMs (unpublished data), we found no functional evidence that this infection further shortened the already brief life span of most CD4⁺ T_EM cells. However, the initial viral-mediated depletion and the immune activation-mediated shortening in the average life span of extra-lymphoid site CD4⁺ T_EM cells render this population highly dependent on a continuous influx of new cells that are derived from the proliferation and differentiation of T_CM cells. Immune activation drives this CD4⁺ T_CM cell proliferation and differentiation, but at the same time it appears to increase the susceptibility of these cells to infection and destruction, reducing their steady-state production. Persistent immune activation is also thought to be responsible for the microenvironmental destruction and aging-like disorganization of the immune system mentioned above (12), and these indirect processes likely combine with the direct viral effects to destabilize CD4⁺ T_CM cell homeostasis, resulting in the slow, progressive decline of this precursor population.

Our model suggests that CD4⁺ T_CM cell decline may “set the clock” in regard to the timing of AIDS onset in RMs and therefore predicts that protection of CD4⁺ T_CM cell homeostasis may be the key to preventing AIDS, a notion supported by the recently reported positive correlation between the numbers of circulating CD4⁺ CD28⁺ T cells (predominantly T_CM cells) with long-term survival in vaccinated SIVmac-challenged RMs that fail to control viremia (41). It is also noteworthy that in the natural apathogenic SIV infections of African monkeys, viral loads are as high or higher than in pathogenic infections of humans and RMs and initial extra-lymphoid CD4⁺ T_EM cell depletion is as extensive, but in the vast majority of these infections, CD4⁺ memory T cell populations in the blood and LNs are maintained and CD4⁺ T_EM cell depletion is not progressive (references 42–46 and Silvestri, G., personal communication). Although the mechanisms underlying the lack of clinical progression in these highly active infections are likely complex and multi-factorial, the fact that preservation of the CD4⁺ T_CM cell compartment is associated with maintenance of a diminished, but stable, CD4⁺ T_EM cell compartment suggests that in these animals, the CD4⁺ T cell system may be able to tolerate the consequences of continuous viral replication because a preserved CD4⁺ T_CM cell compartment can stably maintain CD4⁺ T_EM cell populations at supra-threshold levels. Notably, these infections lack the chronic immune activation response that characterizes pathogenic infection (42–46), in keeping with our above described hypothesis that chronic immune activation undermines CD4⁺ T_CM cell stability.

Progression of typical SIVmac239 infection to AIDS (e.g., normal progression with AIDS onset at 1.5–2 yr after infection) is approximately fivefold accelerated relative to HIV infection, and the plateau-phase pvls in this model are ∼10-fold higher (3), raising the question of whether the mechanisms proposed here for RMs apply to human infection. In arguing that this is indeed likely to be the case, we would submit that the crucial mechanisms that we have identified in SIVmac-infected RMs as driving disease progression have all been documented in human infection, including selective viral targeting of CCR5⁺ CD4⁺ T_EM cells and rapid destruction of this compartment in acute infection (22, 35) (but with the ability to infect CD4⁺ T_CM cells as well [47, 48]), the occurrence of generalized immune activation, and in particular, the effect of this immune activation on the turnover and differentiation of CD4⁺ memory T cells (including T_CM cells) and on prognosis (24–26, 34, 49, 50). Moreover, the clinical manifestations of AIDS in SIVmac-infected RMs are highly analogous to those of HIV-infected humans (27, 51, 52), consistent with a similar immune deficit. The generally lower viral replication rates of HIV-infected people deplete the CD4⁺ T_EM cell compartment less completely than in SIV-infected RMs, and the immune activation-driven production of CD4⁺ T_EM cells is clearly better able to initially keep up with direct and indirect destruction of these T cells, as both the fraction of memory T cells among total CD4⁺ T cells and the fraction of CCR5⁺ CD4⁺ T_EM cells among these memory cells in the blood are typically normal or elevated in HIV infection (35, 50, 53). However, although these differences may allow post-acute phase HIV-infected subjects to maintain their CD4⁺ T_EM cell tissue compartments above the threshold for clinical immunodeficiency for a longer period of time as compared with SIV-infected RMs, the progressive loss of total CD4⁺ T cell counts in the blood that is the hallmark of HIV infection (1), and the extensive tissue depletion of CD4⁺ T cells reported at AIDS (54, 55), clearly indicate that the ability of infected humans to maintain circulating CD4⁺ T_CM cells and their T_EM cell progeny declines with time, just as described herein in SIV-infected RMs. Whether in HIV infection a parallel depletion of CD4⁺ naive T cells is critical for the decline of CD4⁺ T_CM cells, or is only a correlate of the longer time to AIDS, is an open question. Thus, although the extent and/or tempo of CD4⁺ memory T cell depletion and viral replication are exaggerated in SIV infection compared with HIV infection, we believe the evidence strongly suggests that the fundamental pathogenetic processes are similar in these two situations.

One of the principal enigmas of AIDS pathogenesis, pertinent to both the human disease and the RM model, has been the paradoxical combination of the continuous (often high level) replication of pathogenic virus with long periods of clinical “latency” and delayed, but catastrophic, disease. Our data suggest that this paradox results from a dual nature of CCR5-tropic HIV/SIV infection: an efficient, highly destructive infection of easily replaceable CD4⁺ T_EM cells, and a less efficient/more selective infection and much slower net destruction of precursor CD4⁺ T_CM cells. It is the differential impact of HIV/SIV infection on the CD4⁺ T_CM and T_EM cell compartments that probably determines the typical pattern of chronic AIDS. When the T_CM cell compartment remains intact (e.g., infection of natural hosts), the infection is nonpathogenic; when both compartments are efficiently targeted (e.g., dual-tropic SHIV infection), disease onset is rapid. What may be most crucial for understanding the pathogenesis of typical chronic AIDS is that although the exquisite susceptibility of the CD4⁺ T_EM cell compartment to infection results in the most dramatic changes to the immune system, it is most likely the more covert dynamics of the CD4⁺ T_CM cell compartment that dictate the tempo of disease progression. If these concepts are accurate, countering this second process, the slow erosion of CD4⁺ T_CM cell homeostasis, may prove to be the most effective target of nonvirus-directed AIDS therapies. Interventions capable of preserving or rejuvenating CD4⁺ T_CM cells might well “reset the clock” with respect to disease progression and provide significant increases in AIDS-free survival.

A total of 71 purpose-bred male RMs (Macaca mulatta) of Indian genetic background and free of Cercopithicine herpesvirus 1, D type simian retrovirus, and simian T lymphotrophic virus type 1 infection were used in this study: 37 infected with wild-type SIVmac239, 19 with SIVmac239(Δnef), and 15 uninfected controls (15, 56, 57). 14 SIVmac239-infected and 4 SIVmac239(Δnef)-infected RMs were studied longitudinally for virologic or immunologic parameters or both; the remaining animals were used in one or more cross-sectional analyses. Early infection data (<200 d) on the longitudinally studied SIVmac239- and SIVmac239(Δnef)-infected RMs have been reported (15). SIVmac239 and SIVmac239(Δnef) infections were initiated with intravenous injection of 5-ng equivalents of SIV p27 (2.7–3.0 × 10⁴ infectious centers), as described previously (15). ART consisted of daily subcutaneous injections of 30 mg/kg 9-R-(2-phosphonomethoxypropyl)adenine (tenofovir) and 30 mg/kg β-2',3′-dideoxy-3′-thia-5-fluorocytidine (emtricitabine) (58). BAL and small intestinal biopsies were performed as described previously (15). BrdU (Sigma-Aldrich) was prepared as described previously (15) and administered intravenously in three separate doses of 30 mg/kg body weight over a 24-h period. All RMs were housed at the Oregon National Primate Research Center in accordance with standards of the Center's Animal Care and Use Committee and the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals (59). RMs that developed disease states that were not clinically manageable were killed in accordance with the recommendations of the Panel on Euthanasia of the American Veterinary Medical Association (60). End-stage AIDS was defined by the presence of AIDS-defining opportunistic infections, wasting syndrome unresponsive to therapy, or non-Hodgkin lymphoma (52, 61).

Plasma SIV RNA was assessed using a real-time RT-PCR assay (threshold sensitivity <100 SIV gag RNA copy Eq/ml of plasma; interassay CV ≤25%) (62, 63). Cell-associated SIV DNA was also quantified by real-time PCR (sensitivity = single copy of gag DNA) using parallel quantification of albumin gene copy number to define total cell number in a sample, as described previously (47). When no viral DNA was amplified from a given cell population, we report half the lower limit of detection (47).

PBMCs, PLN cells, BAL cells, and intestinal mucosal cells were obtained and stained for flow cytometric analysis as described previously (15, 17, 64). Six-parameter flow cytometric analysis was performed on a Becton Dickinson FACSCalibur instrument using FITC, PE, peridinin chlorophyl protein-Cy5.5 (PerCP-Cy5.5), and allophycocyanin (APC) as the four fluorescent parameters. Polychromatic (8–12 parameter) flow cytometric analysis was performed on an LSR II Becton Dickinson instrument using Pacific Blue, AmCyan, FITC, PE, PE-Texas Red, PE-Cy7, PerCP-Cy5.5, APC, APC-Cy7, and Alexa 700 as the available fluorescent parameters. Cell sorting was performed on a Becton Dickinson FACS ARIA at 70 PSI using FITC, PE, PE-Cy5, PE-Texas Red, APC, APC-Cy7, and QD-705 as fluorescent parameters. Instrument set-up, data acquisition, and sorting procedures were performed as described previously (47, 65, 66). List mode multiparameter data files were analyzed using the FlowJo software program (version 6.3.1; Tree Star, Inc.).

Memory and naive T cell subsets were delineated based on CD28, β7-integrin, CCR7, CCR5, and CD95 expression patterns on gated CD4⁺ or CD8⁺ T cells, as described previously (18, 64, 66). In brief, naive cells constitute a uniform cluster of cells with a CD28^moderate, β7-integrin^moderate, CCR7^moderate, CCR5⁻, CD95^low phenotype, which is clearly distinguishable from the phenotypically diverse memory population that is predominantly CD95^high and displays one or more of the following “non-naive” phenotypic features: CD28⁻, CCR7⁻, CCR5⁺, β7-integrin^bright, or β7-integrin^−/dim. All analyses included at least two of these markers (usually CD28 vs. CD95) for memory subset discrimination. The T_CM and T_EM cell components of the memory subset in the blood and LNs were delineated based on the phenotypic criteria shown in Fig. 6 A, with T_EM referring to the combination of transitional and fully differentiated T_EM cells. As we have shown that T cells in effector sites such as BAL and jejunal lamina propria are almost exclusively of the T_EM cell phenotype (18), CD4⁺ and CD8⁺ T cells from these sites are generally referred to as T_EM cells. For determination of CD4 representation within the naive versus memory compartments of LNs, CD4 naive and memory and CD8 naive and memory subsets were individually delineated, and Boolean gating was used to create a total naive cell and a total memory cell gate from which the percent CD4 was determined.

mAbs L200 (CD4; AmCyan, APC-Cy7), SP34-2 (CD3; PerCP-Cy5.5), SK1 (CD8α APC-Cy7, PerCP-Cy5.5, PE-Cy7; unconjugated), CD28.2 (CD28; PE, PerCP-Cy5.5), B56 (Ki-67; FITC, PE), B44 (anti-BrdU; FITC; APC), DX2 (CD95; PE, APC, PE-Cy7), 3A9 (CCR5; PE, APC), and FIB504 (β7-integrin, PE, PE-Cy7) were obtained from BD Biosciences. mAb 2ST8.5h7 (CD8β PE, PE-Cy7), CD28.2 (PE-Texas Red), and streptavidin–PE-Cy7 were obtained from Beckman Coulter. FN-18 (CD3) was produced and purified in house and conjugated to Pacific Blue or Alexa 700 using a conjugation kit from Invitrogen. mAb 150503 (anti-CCR7) was purchased as purified immunoglobulin from R&D Systems, conjugated to biotin using a Pierce Chemical Co. biotinylation kit, and visualized with streptavidin–Pacific Blue (Invitrogen). Purified anti-CD8 mAb was conjugated with QD705 (Invitrogen) using standard protocols (http://drmr.com/abcon).

Linear regression analysis was used to determine the association between (a) the percentages of CD4⁺ T cells in BAL versus small intestinal lamina propria, and (b) percent CD4⁺ T cells in BAL and absolute numbers of memory cells in blood versus time, and both linear regression and Spearman rank analysis was used to evaluate the association between mean plateau-phase viral loads and time to disease. A random slope model was used to estimate the overall decay rate of percent CD4⁺ T cells in BAL and absolute numbers of memory cells in the blood in the study animals over time. A mixed effect model was used to determine whether the size or representation of T cell subsets in the blood or BAL significantly changed over the course of infection (preinfection to terminal disease), and if so, determination of which changes were most closely associated with disease. Unpaired t tests were used to determine the significance of differences in (a) CD4 representation in naive and memory compartments of PLNs, (b) peak and plateau BrdU labeling in SIV-infected versus uninfected monkeys, (c) the representation of cells with T_EM cell differentiation among proliferating total memory T cells in progressive SIV infection versus controlled infection versus uninfected RMs, and (d) the level of SIV DNA content within the CD4⁺ T_CM, CD8⁺ T_CM, and CD4⁺ transitional memory T cell subsets of RMs in early plateau-phase versus end-stage SIV infection. Paired t tests were used to determine the significance of differences in the level of (a) SIV DNA content between the CD4⁺ T_CM versus CD8⁺ T_CM versus CD4⁺ transitional memory T cell subsets, and (b) post-ART proliferation between CD4⁺ and CD8⁺ T_CM cells in PLNs. Statistical analysis was conducted with the SAS program (SAS Institute) and Statview (Abacus Concepts). p-values of <0.05 were considered significant with the Bonferroni adjustment used in situations with multiple comparisons.

The authors would like to thank G. Bocharov and H. Alon for helpful discussions, S. Mongoue and M. Mori for statistical consultation, and M. Miller at Gilead Sciences, Inc. for provision of tenofovir and emtricitabine.

This work was supported by NIH grants RO1-AI054292, P51-RR00163, U42-RR016025, and U24-RR018107, and from National Cancer Institute funds under contract number NO1-CO-124000.

The authors have no conflicting financial interests.