Homeostatic Turnover of Virus-Specific Memory CD8 T Cells Occurs Stochastically and Is Independent of CD4 T Cell Help

The hallmarks of memory CD8 T cells include not only their ability to respond faster and more efficiently compared with naive cells upon secondary challenge, but also their ability to persist long after the infection has been resolved (1–6). This latter quality of memory cells is essential in providing long-term protective immunity in experimental animals and in humans (7, 8). Due to this importance in host immunity, there has been much interest over the years in understanding the mechanisms by which CD8 T cell memory is maintained (9, 10). Initially, it was argued that this persistence of memory CD8 T cells was due to continual stimulation from small amounts of persisting Ag (11). However, subsequent studies have demonstrated that memory T cell maintenance is independent of Ag, and is instead associated with IL-7– and IL-15–mediated survival and homeostatic turnover of memory CD8 T cells (12–20). More recent experiments have shown that helper CD4 T cells are also important for the continued persistence of memory CD8 T cells (21, 22). Although this role of Th cells in the maintenance of memory CD8 T cells after an acute infection is a fairly recent discovery, for most chronic infections, it has long been established that CD4 T cell help is necessary to prevent the exhaustion and deletion of Ag-specific CD8 T cells (23–26).

Despite the substantial progress made in understanding the mechanisms of memory CD8 T cell maintenance, there still remain questions that have yet to be fully addressed. First, do all memory CD8 T cells undergo homeostatic turnover or is there a subpopulation of cells that do not homeostatically divide? Next, what characterizes the turnover of memory cells; that is, do cells divide after a fixed time or is this turnover stochastic? Additionally, how do CD4 T cells aid (e.g., survival versus homeostatic turnover) in promoting the persistence of memory CD8 T cells, and is the Ag-specificity of these Th cells important? Lastly, do memory CD8 T cells continue to require CD4 T cell help to retain their overall quality (e.g., phenotypic profile and function)?

To more carefully address the above questions, we longitudinally analyzed, within individual mice, both the frequency and the homeostatic turnover of fully functional lymphocytic choriomeningitis virus (LCMV)–specific memory CD8 T cells after their adoptive transfer into naive wild-type (WT) mice. This information allowed us to use mathematical modeling to more rigorously and more accurately quantify the homeostatic turnover and the dynamics of memory CD8 T cell maintenance. We also used naive CD4^−/− and MHC II^−/− mice as recipients to address the requirement for CD4 T cell help in the long-term maintenance and the overall quality of memory CD8 T cells. These animals are ideal models to study the role of CD4 T cells, as they exhibit significant impairment in the generation and maintenance of memory CD8 T cells in the context of both acute and chronic infections, and this impairment has been described to be due to the absence of CD4 T cells in these mice (23–30). This study provides a longitudinal and quantitative analysis of the homeostatic turnover of virus-specific memory CD8 T cells and investigates the requirement of CD4 T cells in this process.

LCMV Armstrong and recombinant vaccinia virus expressing the gp33 epitope of LCMV (VV-gp33) were propagated, titered, and used as previously described (31).

Six to 8-wk-old female C57BL/6 (B6), BALB/c, CD4^−/−, and MHC II^−/− mice were purchased from The Jackson Laboratory (Bar Harbor, ME) or Taconic (Germantown, NY). Thy1.1⁺ P14 transgenic mice with T cells expressing the TCR specific for the D^b gp33-41 epitope of LCMV were obtained from The Jackson Laboratory and backcrossed to B6 mice in our colony (4). LCMV-immune WT P14 chimeric mice were generated by adoptively transferring 1 × 10⁵ naive Thy1.1⁺ P14 CD8 T cells into congenic B6 mice and subsequently infecting these animals with 2 × 10⁵ PFU LCMV-Armstrong. For secondary challenge experiments, 3 × 10⁴ immune Thy1.1⁺ P14 CD8 T cells were adoptively transferred into naive B6 mice and infected with 5 × 10⁶ PFU VV-gp33. All mice were used in accordance with National Institutes of Health (Bethesda, MD) and the Emory University Institutional Animal Care and Use Committee guidelines.

Single-cell suspensions were prepared from the spleen and from the brachial, inguinal, and mesenteric lymph nodes. Bone marrow was obtained by flushing two femurs with cold RPMI 1640. Total number of cells in bone marrow was calculated as follows: no. of cells in two femurs × 7.9 (32). Lymphocytes from blood and liver were obtained as described in Becker et al. (17). Memory P14 CD8 T cells were purified using anti-CD8 MACS magnetic beads and columns (Miltenyi Biotec, Auburn, CA) according to the manufacturer’s instructions. The purity of MACS-purified samples was >90%.

LCMV-specific memory CD8 T cells were labeled with CFSE (Molecular Probes, Eugene, OR) by incubating at 5 mM in PBS, quenching with FCS, and washing as described previously (33). Approximately 1 × 10⁶ memory P14 CD8 T cells or 20–50 × 10⁶ total LCMV-immune splenocytes (from B6 or BALB/c) were adoptively transferred i.v. into naive recipient mice.

All Abs were purchased from BD Biosciences (San Diego, CA), except for anti-mouse IL-7Ra, which was purchased from eBioscience (San Diego, CA). Cells were stained for surface proteins and intracellular proteins and cytokines as described previously (34).

Tetramers of D^b containing NP396–404 and gp33–41 and L^d containing NP118–126 to quantify CD8 T cells specific for these LCMV epitopes were prepared as previously described (35).

The mean number of divisions and variance in the number of divisions of memory CD8 T cells from the CFSE data were calculated as follows. If f_n(t) equals the fraction of cells having undergone n divisions at time t, then the mean and variance in the number of divisions are given by

Frequencies rather than absolute numbers were used because the average number of memory CD8 T cells remains constant over time, and thus the measurement of frequencies is more accurate than that of total number of cells.

For a stochastic model of division it can be shown that the frequency of cells with n divisions follows a Poisson distribution (see Ref. 36 for details). If we assume both division and death occur at random (i.e., a random birth-death model), then the number and the frequency of cells with n divisions, x_n and f_n, are given by

where λ and d are the rate constants for division and death (per unit of time). We see that the frequency of cells with n divisions, f_n, follows a Poisson distribution with the mean number of divisions and variance in the number of divisions increasing at rate 2λ.

To quantitatively analyze the homeostatic turnover of fully functional memory CD8 T cells, naive Thy1.1⁺ P14 CD8 T cells (specific for the D^b gp33–41 epitope of LCMV) were adoptively transferred into naive WT congenic (Thy1.2⁺) mice, and then these P14 chimeric mice were infected with the Armstrong strain of LCMV. At >60 d postinfection, most memory P14 CD8 T cells had differentiated into canonical central memory T cells (CD127^high, CD62L^high, Bcl-2^high) that are capable of making rapid recall responses, persisting for extended periods by homeostatic proliferation, and conferring long-term protective immunity (37). These (day >60 postinfection) memory P14 CD8 T cells were labeled in vitro with CFSE and adoptively transferred into naive WT congenic (Thy1.2⁺) recipients. Recipient mice were then serially bled at various time points posttransfer and both the number and the CFSE profiles of the transferred memory cells were longitudinally assessed in individual mice. Consistent with earlier studies demonstrating long-term persistence of memory CD8 T cells (9, 10), we observed that the transferred memory CD8 T cells were stably maintained in WT animals for the entire duration of the experiment (Fig. 1A). At day 1 posttransfer, the percentage of donor memory cells observed in the peripheral blood ranged between 0.14 and 0.26%. Although this percentage fluctuated slightly over time, overall, the number of transferred memory CD8 T cells remained constant as far out as 120 d posttransfer. More specifically, we calculated the mean rate of the loss of donor memory cells to be 0.0005, which was not statistically different from 0 (p = 0.78). This stability in the maintenance of the donor memory cells can be better visualized in Fig. 1C.

In regard to the homeostatic turnover of the transferred memory cells, the initial inspection of the CFSE profiles of these cells at different time points posttransfer suggested that the turnover of memory CD8 T cells occurred slowly but continuously. In support of this observation, the percentage of cells that had undergone at least one round of division increased from ∼60% at day 21 posttransfer to ∼80–90% by 60 d posttransfer (Fig. 1B). Examination of the frequency of undivided cells at different time points revealed that the percentage of undivided cells decreased exponentially with time (Fig. 2A), suggesting that there was a single homogeneous population of memory cells (no separate population of nondividing memory CD8 T cells) and that the recruitment into division was stochastic.

To confirm this stochastic nature of memory CD8 T cell turnover, we subjected the CFSE profiles of donor memory CD8 T cells to further mathematical analysis (see 1Materials and Methods). If all memory cells were capable of division and the recruitment into division were stochastic, then the frequency of cells having undergone different numbers of divisions at a given time would follow a Poisson distribution. A characteristic of the Poisson distribution is that the mean number of divisions and the variance in the number of divisions are the same, and if the division rate does not change over time, they should both increase linearly with time. In Fig. 2B, we show that this was indeed the case, and the mean number of divisions and the variance in the number of divisions both increased linearly with time. As might be expected, the variance in the number of divisions was slightly less than the mean number of divisions because while the time for cells to undergo division was much smaller than the rate of recruitment into division, it was not 0 as we had assumed in the model. In Fig. 2C, we made maximum use of all the CFSE data by fitting the entire dataset for the CFSE distribution in the number of cells over time, in each mouse, to the model for stochastic division. We observed that the rate of division (λ) estimated from the fit to the entire CFSE distribution was indeed in agreement with that estimated from the rate of increase in the mean number of divisions described above. Collectively, these observations implied that the homeostatic turnover of memory CD8 T cells occurred stochastically, where the probability that a memory cell divided did not depend on its previous division history. This stochastic turnover resulted in the mean number of divisions of cells in the population increasing at a rate of ∼0.04 divisions per day, which corresponds to an average rate of division (λ) of 0.02 divisions per day or an intermitotic time (1/λ) of ∼50 d.

Next, we wanted to examine whether the pattern of homeostatic turnover observed for transgenic D^b gp33–41⁺ memory P14 CD8 T cells also held true for populations of endogenous memory CD8 T cells specific for different epitopes of LCMV in two distinct strains of mice. To do this, we infected naive WT C57BL/6 mice and naive WT BALB/c mice with the Armstrong strain of LCMV. At >60 d postinfection, LCMV-infected animals were sacrificed, and the total splenocytes were labeled with CFSE and adoptively transferred into normal naive C57BL/6 or BALB/c recipients. The recipient mice were then sacrificed at different time points posttransfer and the CFSE profiles of different populations of fully functional LCMV-specific memory CD8 T cells (D^b NP396–404⁺ in C57BL/6 and L^d NP118–126⁺ in BALB/c) were assessed.

As observed in Fig. 3B and 3C, the overall pattern of the homeostatic turnover of D^b NP396–404⁺ and L^d NP118–126⁺ memory CD8 T cells closely resembled that observed earlier with memory P14 CD8 T cells. For example, we observed that for both D^b NP396–404⁺ and L^d NP118–126⁺ memory CD8 T cells, the mean number of divisions and the variance in the number of divisions also increased linearly with time. Additionally, the CFSE distribution of these memory CD8 T cells followed a Poisson distribution. These data indicated that the homeostatic turnover of memory cells of different specificities (D^b NP396–404⁺ versus L^d NP118–126⁺) in different strains of mice (C57BL/6 versus BALB/c) was consistent with a stochastic model of cell division. We also observed that the CFSE profile of the total CD44^high memory CD8 T cells was virtually identical to that of D^b NP396–404⁺ memory CD8 T cells at day 21 posttransfer (Fig. 3A), suggesting that all memory CD8 T cells, irrespective of their specificity and mouse strain, exhibit similar homeostatic turnover (slow, continuous, and stochastic recruitment into division).

For most acute infections, CD4 T cell help has been traditionally thought to be important in the development of fully functional memory CD8 T cells (38). However, more recently, it has been proposed that CD4 T cell help is necessary for the long-term persistence of memory CD8 T cells (27). Therefore, we next wanted to assess more quantitatively the role of helper CD4 T cells in the homeostatic proliferation of memory CD8 T cells. Briefly, fully functional memory P14 CD8 T cells generated in the presence of CD4 T cell help were purified from WT LCMV-immune mice (>60 d postinfection) as described earlier. Cells were then labeled with CFSE in vitro and adoptively transferred into either uninfected WT, CD4^−/−, or MHC II^−/− mice. These recipient animals were then serially bled at various time points posttransfer, and both the frequency and the CFSE profiles of the transferred memory cells were longitudinally assessed in individual mice (Fig. 4A).

As shown in Fig. 4B, we observed that the donor memory cells were stably maintained when adoptively transferred into either WT or CD4^−/− recipients, but they slowly declined in number when transferred into MHC II^−/− animals. At day 1 posttransfer, the percentages of donor memory cells in the peripheral blood of WT, CD4^−/−, and MHC II^−/− mice were approximately the same (0.1–0.4%). By 70 d posttransfer, when the animals were sacrificed, the transferred memory cells made up <0.1% of the peripheral blood in MHC II^−/− animals, whereas in WT and CD4^−/− recipient mice, this percentage remained at or around the initial frequency (0.1–0.4%). In support of this observation, when the mean rate of the loss of donor cells was calculated for the different recipient groups, we found that this rate was not significantly different from 0 in both WT (0.0005; 95% CI, −0.004–0.003) and CD4^−/− animals (−0.002; 95% CI, −0.0027–0.007), indicating stable maintenance of the transferred cells in these recipient groups. However, in MHC II^−/− animals, the donor memory CD8 T cells decayed at a mean rate of 0.015 (95% CI, −0.01–0.02), which was significantly different from 0 (p < 0.00001) and was equivalent to a half-life of ∼67 d. This observed trend in memory persistence among the different recipient animals was not unique to the peripheral blood, but was also true for all tissues examined (e.g., spleen, lymph nodes, liver, and bone marrow). As shown in Fig. 4C and 4D, we consistently observed a 5- to 10-fold decrease both in the percentage and in the total number of donor memory CD8 T cells in the tissues of MHC II^−/− animals compared with WT and CD4^−/− mice.

In contrast with the observed loss of memory CD8 T cells in MHC II^−/− recipients, we observed no substantial defect in the homeostatic turnover of the transferred memory CD8 T cells in any of the recipient mice (Fig. 5A). Analysis of the CFSE profiles revealed that the mean number of divisions (Fig. 5B) and the variance in the number of divisions (not shown) also increased linearly with time in WT (mean, 0.04 divisions/day), CD4^−/− (mean, 0.048 divisions/day), and MHC II^−/− (mean, 0.054 divisions/day) mice. Additionally, the distribution of the number of donor cells having undergone different rounds of divisions followed a Poisson distribution in all recipient animals (Fig. 5C). Collectively, these results indicate that memory CD8 T cells do not require CD4 T cell help to undergo homeostatic turnover, in which the recruitment into division is stochastic.

We next examined whether memory CD8 T cells generated in the presence of CD4 T cells continue to require CD4 T cell help to retain their overall quality. To address this, we adoptively transferred fully functional and CFSE-labeled memory P14 CD8 T cells, purified from LCMV immune WT P14 chimeric mice (>60 d postinfection), into either uninfected WT, CD4^−/−, or MHC II^−/− animals. At least 60 d posttransfer, the recipient mice were sacrificed and both the phenotypic profile and the functional quality of the donor memory CD8 T cells were assessed.

In regard to phenotypic expression, we observed no noticeable discrepancy among donor memory CD8 T cells from WT, CD4^−/−, and MHC II^−/− animals (Fig. 6A). For instance, donor memory cells isolated from the spleens of all three groups expressed high levels of CD62L, CD127, CD27, CD44, and CD122 (IL-2/IL-15Rβ)—profiles characteristic of healthy memory CD8 T cells. Additionally, these memory cells expressed negligible levels of markers associated with activation and effector activity (killer cell lectin-like receptor G1, CD25/IL-2Rα, CD69, programmed cell death 1, and granzyme B). This lack of difference in phenotypic profile was also true among donor cells isolated from other peripheral tissues (e.g., liver, lung, bone marrow, peripheral blood, and lymph nodes; data not shown), suggesting that CD4 T cell help was not required for maintaining the phenotypic profile of memory CD8 T cells.

In agreement with the phenotypic data, we also did not observe any significant discrepancy in the functional qualities of the transferred cells in all the recipient animals. When donor CD8 T cells maintained in WT, CD4^−/−, or MHC II^−/− were isolated and restimulated in vitro with LCMV gp33 peptide, they all rapidly produced IFN-γ with ∼94% and 20% of the cells also producing TNF-α and IL-2, respectively (Fig. 6B). Similarly, we observed that the recall capability of the donor memory cells was unaffected by the absence of CD4 T cell help during memory maintenance. This was demonstrated by adoptively transferring equal numbers of donor memory cells, which had been maintained in WT, CD4^−/−, or MHC II^−/− animals for at least 60 d, into normal C57BL/6 mice and then challenging these mice with VV-gp33. Approximately 5 d postchallenge, the animals were sacrificed and the recall response of the donor memory CD8 T cells was analyzed (Fig. 6C). As shown in Fig. 6D, the donor memory cells maintained in WT, CD4^−/−, or MHC II^−/− animals all responded comparably after VV-gp33 challenge. Additionally, we observed similar total numbers of donor cells both in the spleen (Fig. 6E) and in the liver (data not shown). Collectively, these results suggest that although the long-term persistence of memory CD8 T cells may depend on CD4 T cell help (at least in MHC II^−/− animals), the homeostatic turnover and the overall quality (both in phenotypic expression and in recall ability) are unaffected by the absence of helper CD4 T cells during the maintenance phase of CD8 T cell immune response.

There has been much interest over the years in understanding the mechanisms and regulations of memory CD8 T cell maintenance, as these cells are required to mount a much more rapid and efficient protective immune response upon secondary exposure to a given pathogen (1, 3, 39, 40). In this study, we longitudinally assessed, in individual mice, both the total numbers and the homeostatic turnover of Ag-specific memory CD8 T cells after adoptive transfer into either naive WT, CD4^−/−, or MHC II^−/− recipients. This allowed us to not only use mathematical modeling to rigorously quantify the homeostatic turnover of memory CD8 T cells, but also to address the role of CD4 T cell help both in this turnover and in the long-term maintenance of the overall quality of memory CD8 T cells.

Our initial analysis of memory CD8 T cells adoptively transferred into WT recipients revealed that these cells were stably maintained and underwent slow homeostatic turnover that was independent of epitope specificity (total CD44^high, D^b gp33–41⁺ P14, D^b NP396–404⁺, and L^d NP118–126⁺) and mouse strain (C57BL/6 and BALB/c). These results are in agreement with the earlier qualitative observations that demonstrated that memory CD8 T cells actively undergo homeostatic turnover and that this turnover is important in the long-term maintenance of memory CD8 T cells (17, 20, 40, 41).

In this study we have done a much needed rigorous quantitative analysis of the homeostatic turnover of Ag-specific memory CD8 T cells. Accordingly, we determined that the recruitment of memory CD8 T cells into division was stochastic rather than deterministic. Had division followed a deterministic model, in which all cells spend the same time between successive divisions, then, at any given time, the number of divisions that all the memory cells had undergone would have differed by at most one division. In contrast, we observed that the variance in the number of divisions increased linearly with time at a rate similar to the increase in the mean number of divisions, which suggested a stochastic model for cell division, in which the probability that a cell divides does not depend on its previous history. We further validated the stochastic nature of memory CD8 T cell turnover by demonstrating that the numbers of divisions followed a Poisson distribution, as predicted by a stochastic model for cell turnover. Lastly, like many other past studies (36, 42, 43), we assumed that death, like division, occurred stochastically (random birth-death model); this allowed us to infer that the mean number of divisions increased at twice the rate of division (λ), as each division would result in the production of two cells having one additional division, while death would kill cells of all divisions equally (36). Based on this assumption, we calculated that the turnover of memory CD8 T cells occurs at a rate of 0.02 divisions per day or an intermitotic time of ∼50 d (1/ λ). Our estimation is comparable to that proposed by Di Rosa and colleagues (43) (intermitotic time of 63 d), who used BrdU to calculate the rate of turnover of memory-phenotype (CD44^high) CD8 T cells in thymectomized naive mice.

The modest difference between the two estimations could be merely due to the different cell labeling techniques or the type of memory cells studied. For instance, our analysis was focused on Ag-specific memory cells of primarily central memory phenotype (e.g., CD62L^high). It has been shown that effector memory CD8 T cells (e.g., CD62L^low) not only further differentiate into central memory cells, but they also undergo less extensive homeostatic turnover compared with CD62L^high memory cells (37, 44–47). Therefore, it may be reasonable to assume that the reduced rate of turnover observed by Di Rosa and colleagues is due to the heterogeneity (mixture of both CD62L^low and CD62L^high) in their memory CD8 T cell population.

It has also been shown that the initial precursor frequency of naive Ag-specific CD8 T cells can influence various aspects of the CD8 T cell response to an infection. Accordingly, various groups have suggested that immune response observed with increased naive precursor frequency may not accurately mirror immune responses under more natural conditions (48, 49). However, Sarkar et al. (45) have subsequently demonstrated that although the initial precursor frequency did influence memory CD8 T cell development, this influence was largely kinetics. They observed that the conversion from CD62L^low to CD62L^high memory CD8 T cells occurred more quickly with increased precursor frequency, but the overall quality of the memory cells generated remained unaltered. Hence, it is likely that the rate of homeostatic turnover of CD62L^high central memory CD8 T cells will be comparable regardless of the initial precursor frequency.

In addition to this quantitative analysis, we have also examined the requirement for CD4 T cell help during the long-term maintenance of memory CD8 T cells, both in their quantity and in their quality. As shown in earlier studies (9, 10, 27), we observed that Ag-specific helper CD4 T cells were not required for memory CD8 T cell maintenance, as memory cells adoptively transferred into naive WT recipients—uninfected and therefore lacking Ag-specific T cells—were stably maintained and retained both their phenotypic profile and functions. Although this demonstrated that Ag-specific CD4 T cells were not required, it did not rule out a nonspecific role for CD4 T cell help. Therefore, we used naive CD4^−/− mice as recipients for our adoptive transfer experiments, since these animals lack CD4 T cells due to the deletion of the CD4 gene (50). Analysis of the CFSE profiles of memory CD8 T cells adoptively transferred into naive CD4^−/− recipients showed that the homeostatic turnover of memory cells was unaffected by the absence of CD4 T cell help. Furthermore, as observed in WT recipient mice, the transferred memory CD8 T cells in CD4^−/− animals were stably maintained in their total number and retained both their phenotypic expression and functions for the entire duration of the experiments, suggesting that CD4 T cell help is not required for the long-term maintenance of memory CD8 T cells, both in their quantity and in their quality.

A potential problem with the use of CD4^−/− mice is that these animals may be contaminated with MHC class II–restricted CD4⁻ T cells that may substitute for the traditional helper CD4 T cells (51–53). Therefore, to circumvent this potential problem, we adoptively transferred memory CD8 T cells into uninfected MHC II^−/− mice, which also lack CD4 T cells (54). We observed that even in these recipient animals, the donor memory CD8 T cells exhibited homeostatic turnover comparable to that observed in WT and CD4^−/− recipients. However, in contrast with the memory cells transferred into WT or CD4^−/− animals, donor memory CD8 T cells slowly declined in their number when transferred into naive MHC II^−/− mice. Interestingly, however, both the phenotype of the donor memory cells and the ability of these cells to respond to secondary challenges remained unchanged, supporting the earlier findings from CD4^−/− animals that the homeostatic turnover and the long-term maintenance of the quality of memory CD8 T cells are independent of CD4 T cell help.

Note that we had also adoptively transferred memory CD8 T cells into naive WT mice that were depleted of their CD4 T cells using anti-CD4 Ab (GK1.5). However, in these animals, we observed substantially increased proliferation of the transferred memory CD8 T cells, compared with even normal WT animals (data not shown). This increase in division closely mirrored the rapid proliferation observed when cells (naive or memory) are adoptively transferred into lymphopenic environments, rather than the slow, continuous homeostatic turnover observed under normal steady-state conditions (33, 55, 56). Therefore, we limited our analysis to WT, CD4^−/−, and MHC II^−/− animals.

It is interesting to speculate on the potential causes for the discrepancy observed in the long-term maintenance of memory CD8 T cells in regard to their total number in CD4^−/− and MHC II^−/− animals. As described earlier, several studies have reported that CD4^−/− mice are contaminated with MHC class II–restricted CD4⁻ T cells that could behave like traditional helper CD4 T cells (51–53); therefore, the absence of impairment in the persistence of memory cells in CD4^−/− animals could be due to the help provided by these nontraditional helper T cells. In contrast, MHC class II molecule is expressed on various cell types (e.g., thymic epithelium, B cells, macrophages, and dendritic cells), and therefore the observed loss of donor memory CD8 T cells in these animals could be more due to the intrinsic quality of these mice rather than merely due to the absence of CD4 T cells (54). Regardless of the reason for the discrepancy in the long-term persistence of the transferred memory cells, the homeostatic turnover and the maintenance of the overall quality of the memory CD8 T cells were comparable in WT, CD4^−/−, and MHC II^−/− animals.

In conclusion, our study provides a quantitative analysis of the homeostatic turnover of memory CD8 T cells and an assessment of the requirement for CD4 T cell help during the memory maintenance phase of CD8 T cell immune response. We show that fully formed memory CD8 T cells are not strictly dependent o CD4 T cells for undergoing homeostatic proliferation and for maintaining their function in the setting of an acute viral infection.

Disclosures The authors have no financial conflicts of interest.