Abstract

RESULTS

T_EM cells revert to T_CM cells in adoptive transfer systems

We first tested whether the conversion of T_EM cells to T_CM cells occurred when adoptively transferred naive TCR-transgenic T cells were the responding population. For this, we adoptively transferred naive LCMV-specific P14 cells into naive mice, which we then infected with the Armstrong strain of LCMV⁷. The LCMV-specific CD8⁺ T cell response is characterized by an expansion phase, peaking on days 7–8, followed by a contraction phase between days 8 and 30; subsequently, a stable pool of P14 TCR–transgenic memory cells specific for an H-2D^b-restricted epitope of the LCMV glycoprotein, consisting of amino acids 33–41 (KAVYNFATM), is formed^8,9. About 60% of the resulting P14 memory cells were CD62L⁺ (Fig. 1). We sorted LCMV-specific memory CD8⁺ T cells into T_CM and T_EM subsets on the basis of CD62L expression 40 d later and adoptively transferred the cells into naive recipients. At 34 d after transfer of sorted P14 T_EM cells (CD62L^lo) and T_CM cells (CD62L^hi) into separate hosts, a portion (48%) of T_EM cells had converted to T_CM cells (Fig. 1a), whereas transferred CD62L_hi cells retained CD62L expression. We obtained similar results when we sorted P14 cells and transferred them 117 d after infection (data not shown). To determine whether this phenomenon was unique to P14 cells and LCMV infection, we transferred ovalbumin-specific OT-I CD8⁺ T cells and infected the recipients with vesicular stomatitis virus expressing ovalbumin (VSV-OVA)¹⁰. Then, 56 d later we sorted OT-I memory CD8⁺ T cells into T_CM and T_EM subsets on the basis of CD62L expression and adoptively transferred the cells into naive recipients (Fig. 1b). After 34 d, the transferred CD62L^hi T_CM population remained CD62L^hi, whereas approximately half of the CD62L^lo population had converted to CD62L^hi, confirming the ability of OT-I T_EM cells to acquire the phenotype of T_CM cells.

Open in a separate window

Figure 1

T_EM cells convert to T_CM cells when derived from adoptively transferred TCR-transgenic CD8⁺ T cells. (a) Splenocytes from mice transgenic for the TCR specific for an H-2D^b-restricted epitope of the LCMV glycoprotein consisting of amino acids 33–41 (P14; 1 × 10⁶ cells per mouse) were transferred into syngeneic (B6 × B6D2) F1 mice that were infected intraperitoneally 1 d later with LCMV Armstrong (2 × 10⁵ PFU). At 40 d after infection, CD62L^lo and CD62L^hi memory T cells from the spleen were purified by flow cytometry and were adoptively transferred into separate naive mice. After 34 d, CD62L expression by splenic H-2D^b–gp33⁺CD8⁺ T cells of recipients was assessed. Data represent one of three mice per time point. (b) Splenocytes (1 × 10⁵) from CD45.1 OT-I TCR–transgenic mice were transferred into CD45.2 B6 mice, which were infected intravenously 1 d later with VSV-OVA (1 × 10⁶ PFU). At 56 d after infection, CD62L^hi cells from the spleen and CD62L^lo memory T cells from the spleen were purified by flow cytometry and were adoptively transferred into separate naive mice. Between 34 and 40 d, CD62L expression by splenic OVA-specific memory T cells of recipients was assessed. Data are from one representative mouse of three to four mice analyzed per time point from three individual experiments. Vertical axes, relative cell number; numbers to the right or left of the dashed vertical line (which represents negative control staining) indicate the percentage of cells in that area.

Stable T_EM cell generation from endogenous CD8⁺ T cells

To determine whether the conversion of T_EM cells to T_CM cells was also characteristic of polyclonal populations of CD8⁺ memory T cells, we generated memory cells by infecting naive C57BL/6 (B6) mice with LCMV. In this setting, an LCMV-specific CD8⁺ memory cell population, consisting of 40% CD62L^lo and 60% CD62L^hi cells, developed in the spleen 111 d after infection (data not shown), at which time we sorted memory cells from spleen or liver into CD62L^lo and CD62L^hi populations and adoptively transferred them into naive mice. Unexpectedly, at 34 d after transfer, the gp33-specific T_EM (CD62L^lo) population remained uniformly CD62L^lo in both the spleen and liver, indicating that the endogenous LCMV-specific T_EM cells did not give rise to T_CM cells (Fig. 2a). Similarly, CD62L^lo CD4⁺ T cells transferred from LCMV-immune mice into naive recipients remained CD62L^lo, although we did not have the means to track LCMV-specific CD4⁺ T cells in this case (data not shown). We also noted a lack of conversion from T_EM to T_CM when we sorted endogenous OVA-specific memory CD8⁺ T cells from VSV-OVA-infected mice into CD62L^hi and CD62L^lo subsets and adoptively transferred them into naive recipients. At 34 d after transfer of CD62L^lo memory cells from spleen or liver, very few CD62L^hi OVA-specific memory cells could be detected in either the spleen or liver of recipients (Fig. 2b). In addition, CD62L^lo secondary memory cells generated by challenge of VSV-OVA-infected mice with Listeria monocytogenes producing OVA (LM-OVA) also did not re-express CD62L after transfer (Fig. 2c). These results suggested that unlike CD62L^lo CD8⁺ T cells generated in the TCR adoptive transfer model (Fig. 1), T_EM cells derived from endogenous CD8⁺ T cells represented a phenotypically stable population.

Open in a separate window

Figure 2

Endogenous CD8⁺ T cells generate distinct T_CM and T_EM lineages in response to infection. (a) CD45.2 B6 mice were infected intraperitoneally with LCMV Armstrong (2 × 10⁵ PFU) and, 111 d after infection, CD62L^lo CD8⁺ memory T cells from the spleen or liver were purified by flow cytometry and were transferred into naive CD45.1 B6 mice. At 34 d after transfer, CD62L expression was assessed in tetramer-positive memory CD8⁺ T cells (boxes indicate the population analyzed). (b) CD45.2 B6 mice were infected intravenously with 1 × 10⁶ PFU of VSV-OVA and, 90 d later, purified CD62L^lo and CD62L^hi OVA-specific memory CD8⁺ T cells were transferred into CD45.1 B6 mice. Then, 34 d later, CD62L expression was assessed in the OVA-specific memory CD8 T cells. (c) CD45.2 B6 mice originally infected intravenously with 1 × 10³ CFU of LM-OVA were allowed to ‘rest’ for 132 d and were reinfected intravenously with 1 × 10⁶ PFU of VSV-OVA. At 75 d after reinfection, CD62L^lo and CD62L^hi memory T cells from the spleen or liver were purified by flow cytometry and were transferred into CD45.1 B6 recipients; 34 d later,CD62L expression was assessed in the transferred tetramer-positive population. Data represent three to four mice per time point from three to four individual experiments. Vertical axes, relative cell number; numbers above bracketed lines indicate the percentage of cells expressing CD62L in that area. Boxes (b,c) indicate the gates used for cell sorting; numbers beneath boxes indicate the percentages of the initial populations within the sort gates.

To allow long-term tracking of transferred T_EM cells to further address whether interconversion occurs, we transferred purified T_EM cells and then ‘recalled’ these cells with infection. At 107 d after infection, CD62L⁺ cells were still not evident (Supplementary Fig. 1 online). Although conversion could ensue at later times and/or secondary infection could further delay conversion, given the lifespan of the mouse, it becomes difficult to see the utility of such a system. Instead, we believe that separate T_CM and T_EM lineages are generated in the primary response and that T_CM cells are generally sequestered in lymph nodes and recirculate preferentially through lymph nodes, where they participate in secondary responses.

Precursor frequency governs memory lineage commitment

One possible explanation for the discrepancies we noted between TCR-transgenic and endogenous T cells was the fact that the naive precursor frequency was much greater in the former than in the latter. To test whether precursor frequency affected lineage commitment, we adoptively transferred varying numbers of OT-I T cells into B6 mice, which we then infected with VSV-OVA. At 42 d after infection, we analyzed CD62L expression in transferred transgenic cells from various tissues. Although the total number of OVA-specific memory T cells in the spleen was relatively constant (including both transgenic and endogenous cells; data not shown), the absolute numbers of T_EM cells and T_CM cells differed depending on the number of OT-I T cells transferred (Fig. 3a). Thus, at 42 d after infection of recipients of 1 × 10⁷, 1 × 10⁴ or 5 × 10² OT-I T cells, the percentage of splenic T_CM (CD62L^hi) cells was 71%, 27% or 5%, respectively. We noted the same trend in the lung and liver (Fig. 3a) and obtained similar results for CCR7 expression (data not shown). At the lowest dose of 5 × 10² cells, only the memory cell population in the lymph nodes contained an appreciable population of CD62L^hi cells. This phenotypic distribution among memory cells more closely resembled that derived from an endogenous response (Fig. 3b), which is likely to be initiated from a precursor pool of a few hundred cells¹¹.

Open in a separate window

Figure 3

Initial naive CD8⁺ T cell precursor frequency ‘imprints’ memory lineage commitment during the primary response. (a) OT-I splenocytes (1 × 10⁷, 1 × 10⁴ or 5 × 10²) were transferred into B6 mice and 1 d later the mice were infected with 1 × 10⁶ PFU VSV-OVA. At 42 d after infection, lymphocytes were isolated from various tissues and CD62L was measured on the OT-I T cells. (b) B6 mice were infected with 1 × 10⁶ PFU VSV-OVA and 42 d after infection, CD62L was measured on the tetramer-positive memory CD8⁺ T cells. Data represent five mice (1 × 10⁷ and 1 × 10⁴ OT-I) and six mice (5 × 10² OT-I) from two individual experiments. Vertical axes, relative cell number; numbers above bracketed lines indicate the percentage of OT-I cells in that area. PLN, peripheral lymph node; MLN, mesenteric lymph node.

To test whether altering the ratio of antigen-presenting cells to CD8⁺ T cells during the primary response affected memory lineage ‘choice’, we artificially increased dendritic cell numbers by treatment with the ligand for the cytokine receptor Flt3. Administration of Flt3 ligand resulted in an increase of about five- to sevenfold in dendritic cells in the spleen and lymph nodes (data not shown). We transferred 1 × 10⁵ OT-I cells into mice treated with Flt3 ligand that we then infected with VSV-OVA. Then, 67 d later, we assessed the expression of CD62L by OT-I cells. In control mice, 60% of the memory cells expressed CD62L, compared with only 7% of memory cells in mice treated with Flt3 ligand (Supplementary Fig. 2 online). Thus, increasing dendritic cell numbers enhanced T_EM cell generation. Although the precise mechanism that resulted in enhancement of T_EM cell generation has yet to be defined, an increase in dendritic cells should result in a situation in which T cell competition for antigen is decreased, thereby allowing more complete T cell activation. These data further support our contention that T cell precursor frequency and competition for resources are important physiological regulators of memory CD8⁺ T cell lineage commitment.

Memory lineage commitment during the primary response

The results reported above led us to investigate whether lineage commitment after transfer of low numbers of OT-I cells mimicked our observations with endogenous memory cells; that is, whether CD62L^lo memory cells from these mice would maintain their phenotype after adoptive transfer. We transferred 5 × 10² OT-I cells into naive mice and infected the recipients with VSV-OVA. Then, 8 d later, we isolated CD62L^lo effector cells and transferred them into naive mice; we used early transfer to test whether commitment to a particular lineage occurred early during the immune response. These cells remained CD62L^lo when examined 1 month later (Fig. 4a). Hence, when generated from low numbers of precursors, CD8⁺ T_EM cells acquired a stable CD62L^lo phenotype as early as 8 d after activation. This was not dependent on transfer early after activation, however, as secondary CD62L^lo memory cells, derived from 5 × 10² OT-I cells after infection with VSV-OVA and rechallenge with LM-OVA, did not convert to T_CM cells (Fig. 4b). Thus, despite the fact that the precursor frequency of antigen-specific cells was substantially increased after secondary infection, the resulting T_EM cells did not undergo conversion to T_CM cells, suggesting that the lineage ‘choice’ was established in the primary response.

Open in a separate window

Figure 4

Commitment to the T_CM and T_EM lineages occurs during the primary response. (a) CD45.1 OT-I cells (5 × 10²) were transferred into CD45.2 B6 mice and 1 d later the mice were infected with 1 × 10⁶ PFU VSV-OVA. Then, 8 d later, CD62L^lo OT-I cells were purified by flow cytometry and were adoptively transferred into naive B6 mice; 34 d later, the expression of CD62L by splenic OT-I cells was assessed. (b) B6 mice received an initial adoptive transfer of 5 × 10² OT-I cells and were infected with VSV-OVA, then were allowed to ‘rest’ for 114 d and were infected intravenously with 1 × 10³ CFU of LM-OVA. Then, 48 d later, CD62L^lo memory T cells from the spleen or liver were purified by flow cytometry and were transferred into naive recipients; 34 d after transfer, CD62L expression by OT-I cells from the spleen and liver was assessed. (c) OT-I splenocytes (1 × 10⁷, 1 × 10⁴ or 5 × 10²) were transferred into B6 mice and 1 d later mice were infected with 1 × 10⁶ PFU VSV-OVA. At 7 d after infection, lymphocytes were isolated from the peripheral lymph node, mesenteric lymph node and spleen, and CD62L was measured on the OT-I T cells. APC, allophycocyanin. Similar results were obtained in three independent experiments. Vertical axes, relative cell number; numbers above bracketed lines indicate the percentage of cells in that area.

To further substantiate that commitment to the T_CM or T_EM lineages occurred during the primary response, we transferred 1 × 10⁵, 1 × 10⁴ or 5 × 10² OT-I (CD45.1) cells into B6 (CD45.2) recipients that we then infected with VSV-OVA. Then, 7 d later we determined the percentages of CD62L^lo and CD62L^hi OT-I T cells (Fig. 4c). As with the memory populations, increasing the input number of antigen-specific T cells resulted in increasingly larger populations of CD62L^hi cells in the lymph node and spleen. When 5 × 10² OT-I cells were transferred, most effector cells in the spleen lacked CD62L, whereas those populations in lymph nodes contained about one third CD62L^hi cells. Thus, sampling only the spleen, as is often done, would provide misleading results regarding CD62L expression. Because CD62L is used by T cells to enter lymph node, enrichment of CD62L⁺ cells in this tissue might be expected. Our results suggested that precursor frequency determined lineage commitment during the primary response.

Initial precursor frequency influences T_EM cell proliferation

Our results suggested that CD62L^lo CD8⁺ memory cells generated from high or low numbers of naive cells were fundamentally different, with only the latter representing a phenotypically stable population. To investigate this possibility further, we assessed the kinetic activity of these cells in vivo. A study of mouse CD8⁺ memory cells has suggested that T_CM cells have a greater capacity for proliferation than do T_EM cells⁶. Thus, we sought to assess whether precursor frequency during the primary response affected the proliferative capacity of the memory cells generated. We adoptively transferred 1 × 10⁷, 1 × 10⁴ or 5 × 10² OT-I T cells into B6 mice and infected the mice with VSV-OVA 1 d later. We infected the mice with LM-OVA 198 d later and at least 40 d after secondary infection, we provided the mice with bromodeoxyuridine (BrdU) in their drinking water for an additional 40 d. Splenic memory cell populations derived from 1 × 10⁷ or 1 × 10⁴ OT-I cells were about 60% CD62L^hi, whereas only about 25% of those derived from 5 × 10² OT-I cells expressed CD62L (Fig. 5a). At all input doses, more than 75% of the CD62L^hi memory cells had incorporated BrdU. In contrast, the BrdU incorporation of the CD62L^lo cells was highly dependent on the input number of T cells. Thus, 66%, 50% and 27% of CD62L^lo cells incorporated BrdU when 1 × 10⁷, 1 × 10⁴ and 5 × 10² OT-I cells were transferred, respectively. This difference in the proliferative capacity of CD62L^lo cells was even more substantial in the lung, with 74%, 41% and 27% BrdU⁺ cells after transfer of 1 × 10⁷, 1 × 10⁴ or 5 × 10² OT-I cells, respectively (Fig. 5b). We also noted this difference in proliferative capacity in the CD62L^lo population in the peripheral lymph nodes and liver (data not shown). In addition, we also noted this difference in experiments examining proliferation of primary endogenous T_EM cells and T_CM cells and with primary memory cells generated from low numbers of transferred OT-I cells. Thus, in general it seems that T_EM cells have lower proliferative capacity than T_CM cells (data not shown). These data show that the initial precursor frequency of antigen-specific T cells ‘imprinted’ fundamental physiological differences on the memory cells generated and that these differences were maintained even through a secondary recall response.

Open in a separate window

Figure 5

Proliferative capacity of T_EM cells is regulated by clonal competition. CD45.2 B6 mice received an initial adoptive transfer of 1 × 10⁷, 1 × 10⁴ or 5 × 10² CD45.1 OT-I T cells and were infected with VSV-OVA, then were allowed to ‘rest’ for 198 d and were infected intravenously with 1 × 10³ CFU of LM-OVA. Then, 40 d later, mice were provided with BrdU in their drinking water for 40 d, at which time BrdU incorporation was analyzed in CD62L^lo and CD62L^hi OT-I cells from the spleen (a) and lung (b). Data represent three mice (1 × 10⁷ and 1 × 10⁴ OT-I cells) and five mice (5 × 10² OT-I cells) from two individual experiments. Numbers above bracketed lines indicate the percentage of cells in that area.

Our findings suggested a model of CD8⁺ memory T cell lineage commitment in which high initial naive T cell precursor frequency resulted in the development of T_EM and T_CM cells but also of a transitional CD62L⁻ memory subset with the proliferative capacity of CD62L⁺ memory cells and the ability to re-express CD62L (Supplementary Fig. 3 online). In contrast, in physiological conditions of low initial precursor numbers, distinct T_EM and T_CM lineages developed without the ability to interconvert (Supplementary Fig. 3 online).

DISCUSSION

At present, the idea of memory T cell development is based on the finding that memory T cells can be divided into subsets distinguishable by the expression of homing molecules and chemokine receptors including CD62L and CCR7 (refs. ^1–3). In addition, T_CM (CD62L⁺CCR7⁺) cells lack immediate effector function, thus functionally distinguishing them from T_EM (CD62L⁻CCR7⁻) cells^2,3. Therefore, based on migratory abilities, T_EM cells ‘preferentially’ reside in nonlymphoid tissues, whereas T_CM cells can be found in the spleen and lymph node^1–3, although these phenotypes and locations are by no means absolute¹². Nevertheless, both T_EM cells and T_CM cells are members of a blood-borne pool of memory cells able to circulate to most tissues of the body, depending on the expression of the appropriate homing molecules for entry into a given tissue¹³.

The lineage relationship between T_CM and T_EM cells has been the topic of several reports. For example, based on in vitro studies using human T cells, it was proposed that T_CM cells could provide a continual source of T_EM cells that were considered end-stage cells¹. In contrast, analysis of the TCR repertoires of human memory T cell subsets indicated that the lineages develop independently⁵. To directly test lineage relationships among memory T cell subsets, an experimental model using adoptive transfer of TCR-transgenic CD8⁺ T cells and LCMV infection was used⁶. Transfer of purified CD62L⁺ or CD62L⁻ memory cells demonstrated that T_EM cells are able to convert to T_CM cells over several weeks. Our results presented here have corroborated those findings with transfer of LCMV- or OVA-specific TCR-transgenic T cells. However, conversion of T_EM cells to T_CM cells occurred only in a subset of the T_EM cells, at least in the time frame studied, suggesting that there may be lineage heterogeneity in the T_EM population. Indeed, our results have shown that the ability of T_EM cells to convert to T_CM cells increased with the transfer of larger numbers of naive CD8⁺ T cells. In fact, when we transferred memory cells derived from 5 × 10² naive OT-I cells or when we used endogenous memory cells, we did not note conversion of CD62L⁻ cells to CD62L⁺ cells. Coupled with that result was the finding that the proliferative capacity of CD62L⁻ cells also increased with increasing input numbers of responding CD8⁺ T cells. Those data indicated that with nonphysiologically high numbers of responding CD8⁺ T cells, commitment to the T_EM lineage seemed incomplete. This resulted in the production of a ‘transitional’ T_EM population with the proliferative capacity of a T_CM population and the ability to phenotypically convert to T_CM cells. Thus, the protracted development of memory CD8⁺ T cells proposed using gene expression analysis of responders derived from the transfer of large numbers of TCR-transgenic T cells bears reexamination¹⁴.

Our findings indicated that T_EM and T_CM CD8⁺ cells represented stable memory cell lineages when generated from physiologically low numbers of naive precursors. Based on those data, we propose a model in which naive CD8⁺ T cells develop into effector cells and generate memory cells of two distinct lineages based on CD62L expression. Our data would support the idea that the memory lineage ‘decision’ is made during the primary response, as the early phenotype of responding cells was dictated by the naive precursor frequency. In fact, with low numbers of transferred cells or with assessment of the endogenous response, small numbers of CD62L⁺ cells were present in the lymph node but not the spleen throughout the primary response. The link between precursor frequency and differentiation stage suggested that competition for limiting resources (such as access to antigen-presenting cells¹⁵ or cytokines such as interleukin 7 or 15; refs. ^16–19) may be involved. This hypothesis was further supported by the demonstration that increasing the ratio of antigen-presenting cells to CD8⁺ T cells preferentially drove T_EM cell development.

The fact that greater numbers of T_CM cells and transitional T_EM cells were produced when antigen-specific T cell frequencies were high fits with the idea that relatively weak signaling favors T_CM cell generation, whereas production of T_EM cells requires strong signaling, a model that has been proposed based on in vitro studies^1,20. In a published study using transfer of TCR-transgenic T cells, low-dose virus infection resulted in what was interpreted as a more rapid conversion of T_EM cells to T_CM cells, whereas high-dose infection preferentially generated CD62L^lo T cells, and the appearance of CD62L^hi memory cells was extended⁶. In that study, the amount of antigen was not measured and only blood was analyzed and no cell sorting or adoptive transfers were done. Thus, conversion was never directly demonstrated; the study only showed that the phenotype of the cells was altered. Reducing the viral dose would probably alter the amount of inflammation induced, which might also affect the outcome of the response. Our results would suggest that these findings may be related to the generation of different ratios of T_CM, T_EM and transitional T_EM cells based on antigen dose and precursor frequency. Thus, at decreasing naive T cell frequencies, which may represent relatively high antigen/T cell ratios, fewer CD62L⁺ cells will be generated. As these cells show greater homeostatic proliferation than CD62L⁻ memory cells^6,21,22, the CD62L⁺ subset would gradually increase over time rather than being generated from CD62L⁻ memory cells.

It remains possible that the proliferative ability of the memory subsets could affect the conversion of T_EM cells to T_CM cells. However, if the reduced proliferative capacity of T_EM cells is responsible for the reduced differentiation from T_EM cells to T_CM cells, then our example with an initial precursor frequency of 1 × 10⁴, in which the CD62L^lo cells also had lower proliferative capacity than the CD62L^hi memory cells, should have resulted in less conversion of T_EM cells to T_CM cells. However, even in the face of lower proliferative capacity than CD62L^hi cells, 50% of the cells converted to the CD62L^hi phenotype. This scenario has also been demonstrated in another study in which 5 × 10⁴ P14 TCR–transgenic T cells were transferred⁶. In that study, although T_EM cells had a reduced proliferative capacity, 50% of the T_EM cells nevertheless underwent reversion to T_CM cells. Therefore, it is unlikely that the reduced proliferative capacity of T_EM cells could explain the reduced differentiation from T_EM cells to T_CM cells using endogenous precursor frequencies. In neither our study here nor in published studies⁶ was the actual number of ‘convertible’ T_EM cells in the overall T_EM population known.

Although the efficiency of conversion may also be regulated by precursor frequency, several pieces of data would suggest otherwise. In our studies, even 40 d after transfer of CD62L^lo cells, we noted little if any conversion. In addition, when nonlymphoid tissues are monitored long term, few T_CM cells are present, particularly in tissues such as the intestinal mucosa. Thus, if conversion occurred in normal circumstances, it would have to occur very slowly and mainly in lymphoid tissues, whereas most T_EM cells are in nonlymphoid tissues. These data indicate that if conversion occurs in physiologically relevant conditions, then the process seems very inefficient. Given the lifespan of a mouse (about 2 years), this hypothetical slow rate of conversion would perhaps have a minimal effect on the overall process of memory development and reactivation.

Although our experiments have identified precursor frequency as one factor controlling memory lineage development, further studies are needed to elucidate the precise molecular mechanisms involved in memory lineage commitment. Nonetheless, the ability to manipulate memory T cell lineage ‘choice’ at will could provide new therapeutic avenues for vaccine development and immunotherapy. Thus, memory development could potentially be intentionally skewed toward T_EM or T_CM cells, depending on the infection type or perhaps location of a tumor. It might be expected that T_CM cells would be more suitable for protection against infectious agents that prime responses preferentially in lymph node, whereas T_EM cells would provide better protection against agents focused in spleen or nonlymphoid tissues. For example, although T_EM cells respond poorly to LCMV infection⁶, T_EM cells generate a robust response to pulmonary virus infection²³. In any case, ongoing delineation of the signals regulating memory development will continue to provide support for rational vaccine design.

METHODS

Supplementary Material

ACKNOWLEDGMENTS

Footnotes

References