|Home | About | Journals | Submit | Contact Us | Français|
“This is an author-produced version of a manuscript accepted for publication in The Journal of Immunology (The JI). The American Association of Immunologists, Inc. (AAI), publisher of The JI, holds the copyright to this manuscript. This version of the manuscript has not yet been copyedited or subjected to editorial proofreading by The JI; hence, it may differ from the final version published in The JI (online and in print). AAI (The JI) is not liable for errors or omissions in this author-produced version of the manuscript or in any version derived from it by the U.S. National Institutes of Health or any other third party. The final, citable version of record can be found at www.jimmunol.org.”
Exogenous dendritic cells display restricted trafficking when injected in vivo and stimulate CD8 T cell responses that are localized to a small number of lymphoid compartments. By examining these responses in the presence and absence of FTY720, a drug that causes sequestration of T cells in lymph nodes, we demonstrate that a significant fraction of divided CD8 T cells redistribute into antigen-free lymph nodes within 3 days of activation. Despite variation in the level of expression of CD62L, redistribution of these cells is CD62L-dependent. Redistributed CD8 T cells exhibit characteristics of differentiated effectors. However, when re-isolated from antigen-free lymph nodes 3 days after activation and transferred into naïve mice, they persist for at least 3 weeks and expand upon antigen challenge. Thus, CD8 T cells that redistribute to antigen-free lymph nodes 3 days after immunization contain memory precursors. We suggest that this redistribution process represents an important mechanism for establishment of lymph node resident central memory, and that redistribution to antigen-free nodes is an additional characteristic to be added to those that distinguish memory precursors from terminal effectors.
Over the last several years, much has been done to elucidate the processes of CD8 T cell activation and differentiation in vivo. The interactions between antigen-specific T cells and antigen-presenting dendritic cells (DC)4 during the early stages of T cell activation have been nicely detailed using two-photon microscopy (1-3). Similarly, tracking the entry of soluble antigen into individual lymph nodes (LN) after subcutaneous (s.c.) injection has provided insight into multiple stages of antigen presentation by LN-resident and skin-derived DC (4). In addition, the distribution of activated effectors into peripheral non-lymphoid tissues at later times during the primary response has also been evaluated in detail (5-9). However, the understanding of intermediate stages of T cell differentiation, and their progression within individual lymphoid compartments is incomplete. Many previous studies have utilized pathogens that activate systemic responses, and the ensuing T cell responses have been examined only in the spleen (10-14). Because the spleen is both a T cell activation site and a repository for effectors that may have been activated in other sites, these analyses do not allow a clear understanding of the relationship of the site of CD8 T cell activation to subsequent differentiation and migration patterns. Several other studies have utilized routes of viral infection that resulted in initially localized antigen presentation in LN and examined CD8 T cells exclusively in draining LN or spleen (6, 15-18). A few of these studies have shown that activated CD8 T cells are present in both draining and presumptive non-draining LN (19, 20). However, these studies reached conflicting conclusions about whether activated cells in presumptive non-draining nodes have migrated there, or been activated there at a later stage in the response due to dissemination of the virus (15, 21, 22).
A second area of uncertainty, intertwined with the first, is the relationship of effector cell differentiation to that of memory cells. The differentiation of a stable memory population after viral infection has been reported to require at least 40 days (23). However, it has been argued that this is an artifact of using a large number of adoptively transferred T cells and that memory populations are established much sooner (24). In keeping with this, memory cell precursors have been identified in draining LN as early as 3.5 days after response initiation, based on their survival after adoptive transfer (18), and as early as 7 days based on re-expression of IL-7Rα (CD127) (25). It has also recently been suggested that memory and effector lineages are produced by asymmetric cell division that is evident in the first daughter cells arising from a newly activated T cell (26). These observations highlight uncertainty over whether memory T cells differentiate as a distinct population (27, 28), or represent early stage effectors programmed to survive (24, 25, 29, 30). This issue has been further complicated by the identification of memory cell subsets (6, 31). Central memory (TCM) cells are localized to LN, and are generally considered to express CCR7 and high levels of CD62L. In contrast, effector memory (TEM) cells are localized to peripheral tissue and are generally considered negative for these two markers. Different studies suggest either that one of these populations gives rise to the other (24, 25, 29, 30) or that they arise independently (27, 28). These models also have not addressed the question of whether precursors of TCM cells exit and then recirculate among LN, or complete their differentiation in the LN in which they were initially activated.
Unlike pathogens, exogenous bone marrow-derived DC (BMDC) localization is highly restricted in vivo, and depends on the injection route (32-39). Thus, BMDC enable immune responses in individual lymphoid compartments to be studied without the uncertainty of antigen localization and persistence. To date, several studies have examined the characteristics of CD8 immune responses that develop after BMDC immunization and have noted differences based on route of injection (32, 37-42). However, few have examined the early temporal phases and spatial distributions of these responses, and how these relate to the characteristics of the resulting memory cells.
In the present work, we have used the localized trafficking of BMDC to examine the activation and differentiation of CD8 T cells in individual lymphoid compartments over time. Furthermore, we have confined these activated cells to lymphoid tissue by using the drug FTY720 (43). Using these techniques, we have identified a population of extensively divided cells that rapidly redistribute into demonstrably Ag-free LN in a CD62L-dependent manner. These redistributed cells exhibit effector characteristics, and also contain memory precursors. The existence of this population is not currently appreciated in models of primary immune responses or memory establishment. We suggest that redistribution of activated cells to Ag-free LN early during the primary response represents an important mechanism for establishment of LN resident TCM cells, and an important characteristic of central memory precursor cells that is not included in current models.
C57BL/6, OT-I RAG1-/-, and C57BL/6 Thy-1.1 mice were obtained from Charles River, Taconic, and The Jackson Laboratories, respectively. OT-I Thy1.1 mice were first generation crosses of OT-I RAG1-/- and C57BL/6 Thy-1.1 mice. All animals were maintained in pathogen-free facilities. Recombinant vaccinia virus expressing OVA (vaccinia-OVA) was a kind gift from Dr. J. Yewdell (National Institute of Allergy and Infectious Diseases). Mice were immunized with 1×106 PFU virus i.v. via the dorsal tail vein. All protocols were approved by the Institutional Animal Care and Use Committee.
Single cell suspensions from spleen and pooled LN of OT-I RAG1-/- and OT-I Thy1.1 mice were enriched for CD8 T cells by negative selection (StemCell Technologies). Preparations were consistently 97-99% CD8+ by flow cytometry. CD8-enriched OT-I Thy1.1 cells were over 90% H-2Kb-ova257 tetramer positive, at least 85% CD44lo and CD62Lhi, greater than 98% CD69lo, and less than 2% α4hi. In some cases, cells were labeled with 4μM CFSE (Molecular Probes) in phosphate buffered saline for 15min at 37°C prior to injection. Unless otherwise stated, 4 × 106 CD8-enriched OT-I cells were injected i.v. into the dorsal tail vein of sex-matched recipients. In experiments involving transfer of redistributed OT-I cells, 2×104 electronically sorted OT-I cells were injected into Thy1-mismatched naive recipients together with 5×106 CFSE-labeled Thy1 host-matched T cells.
CD40L-activated BMDC were generated as previously described (44). Prior to injection, BMDC were pulsed with 10μM ova257 (corresponding to residues 257-264 of chicken ovalbumin) for 1h at 37°C in the presence of 10μg/ml human β2-microglobulin (Calbiochem). Mice were immunized with 1×105 BMDC in 200μl into the dorsal tail vein, the peritoneal cavity, or the scapular fold.
Mice were injected i.p. with 1mg/kg FTY720 (a generous gift of Dr. V. Brinkmann [Novartis Pharma AG]) in 200μl 24h or 72h after BMDC injection and then every 24h until harvest. Purified anti-CD62L (MEL-14, American Type Culture Collection)(45) or rat IgG (Sigma) were administered i.v. at 100μg per mouse 6h after BMDC injection and then every 24h until harvest.
Single cell suspensions were incubated with anti-CD16/32 (93, eBioscience) to block Fc receptors. PerCP anti-CD8α (53-6.7), PE anti-α4 integrin (R1-2), and PE anti-α4β7 (DATK32) were from BD Biosciences. APC anti-IFN-γ (XMG1.2), APC anti-Thy1.2 (53-2.1) and anti-Thy1.1 (HIS51), PE anti-CD69 (H1.2F3), FITC and PE anti-CD44 (IM7), PE and PE-Cy7 anti-CD62L (MEL-14), and anti-CD127-PE (A7R34) were from eBioscience. APC-conjugated H-2Kb-ova257 tetramer was generated in-house. Cells from mice treated with anti-CD62L in vivo were blocked with 10% normal goat serum before incubation with PE goat F(ab)’2 anti-rat IgG (Jackson ImmunoResearch), followed by purified rat IgG (Jackson ImmunoResearch) and then PerCP-Cy5.5 anti-CD8α and APC anti-Thy1.2. Samples were analyzed on FACSCalibur and FACSCanto instruments (Becton Dickinson) using Flow Jo software (Treestar). Electronic sorting was conducted on a Becton Dickinson FACSVantage SE Turbo Sorter with DIVA Option.
Lymphoid cells from mice 5 d after i.v. immunization were assessed for cytokine production by incubation for 5h at a ratio of 1:1 with LB15.13 stimulator cells that had been pulsed with 100μM ova257 . Medium was supplemented with 50U/ml IL-2 (Chiron Corp.) and 10μg/ml Brefeldin A (Sigma). Lymphoid cells were pretreated with 100μM TAPI-2 (Peptides International) for 1 hour and during the assay to prevent CD62L downregulation as described (46). Cells were fixed and permeabilized using Cytofix/Cytoperm (BD Biosciences) and stained for intracellular IFNγ.
We previously demonstrated that peptide-pulsed BMDC infiltrate a limited subset of lymphoid organs based on injection route, resulting in localized T cell activation in specific lymphoid compartments (39). CD8+ OT-I TCR transgenic T cells labeled with 5-carboxyfluorescein diacetate succinimidyl ester (CFSE) were adoptively transferred into mice, and ova257 peptide-pulsed CD40L-activated BMDC were injected by either intraperitoneal (i.p.), i.v. or s.c routes 18-36h later. Twenty-two hours after BMDC injection, activated OT-I cells were identified based on expression of CD69. CD69+ OT-I cells were localized to spleen and mediastinal LN after i.v. immunization, to mesenteric LN and mediastinal LN after i.p. immunization, and to axillary/brachial LN after s.c immunization (Figure 1 and data not shown). S.c. immunization also led to activation of cells in spleen, although this was highly variable in different experiments (data not shown). By 72-96 h after immunization, CFSE dilution revealed a substantial number of divided OT-I cells in these “priming compartments” (Figure 2). However, extensively divided OT-I cells were also present in all other LN examined at this time. Greater than 90% of OT-1 cells remained undivided in the same compartments of mice that had received either no BMDC or unpulsed BMDC (data not shown), demonstrating that divided cells in these other LN had been activated by antigen.
The presence of divided OT-I cells in LN other than those in which initial BMDC priming occurred was not due to delayed migration of BMDC to these sites: we previously demonstrated that OT-I cells injected up to 48 hours after BMDC still underwent activation only in the previously identified priming compartments (39). Thus, the presence of divided OT-I cells in LN non-contiguous with the priming compartments appeared to represent the dissemination of activated T cells. To directly test this hypothesis, we confined OT-I cells to the compartment in which they were activated by treating mice with FTY720, a pharmacological analog of sphingosine 1-phosphate that inhibits T lymphocyte egress from LN by binding to S1P1 receptors (43). In the priming compartments of immunized mice treated with FTY720, the CFSE dilution profiles showed an accentuated accumulation of more extensively divided cells compared to controls (Figure 2), and a 3-4 fold increase in the average number of activated OT-I cells in priming LN (data not shown). However, FTY720 treatment reduced the presence of divided OT-I cells in non-contiguous LN of these same mice by 60-95% in different experiments. These data demonstrate that divided OT-I cells exited priming compartments and redistributed to Ag-free LN by 3 days after immunization.
It was possible that the redistribution of activated T cells from priming to Ag-free lymphoid compartments is due to the adoptive transfer of large numbers (4×106) of T cells, resulting in the “spillover” of activated T cells into other LN. Thus, we compared responses elicited by i.v. immunization with BMDC in mice transferred with as few as 4×102 OT-I T cells. We also examined the endogenous CD8 T cell response in mice that did not receive any transferred cells. By day 7 after immunization, redistribution of CFSE-diluted cells into axillary/brachial LN had occurred at all adoptive transfer numbers tested (Figure 3A). Similar data were also obtained after immunization via the i.p. and s.c routes (data not shown). Redistribution of activated ova257 +H-2Kb tetramer positive CD8 cells was also evident in the endogenous response (Figure 3B, C). These results demonstrate that the rapid redistribution of activated T cells into Ag-free LN is not simply due to “spillover” from priming compartments based on limitations of space and large numbers of activated cells, but is also characteristic of immune responses based on relatively low numbers of T cells.
After immunization by different routes, a significant fraction of divided OT-I cells in priming compartments of mice that received 4×106 OT-I cells had downregulated CD62L expression by 72h, consistent with a pre-effector or effector phenotype (Figure 3D and data not shown). However, about 1/3 of these divided cells remained CD62Lhi despite having undergone as many as 6-7 cell divisions. Although the divided cells in Ag-free LN had undergone a comparable number of cell divisions, the CD62Lhi fraction was even greater: 62-76% (Figure 3D and data not shown). Similar data were also obtained after immunization via the i.p. and s.c routes (data not shown). In keeping with previous work (24, 47), we found that the percentage of cells that remained CD62Lhi after activation was directly correlated with the number of adoptively transferred cells (Figure 3A, C and data not shown). However, in all cases, as well as in the endogenous response, the representation of CD62Lhi divided cells was enhanced in Ag-free axillary LN (Figure 3A, C). This again demonstrates that redistribution is not due to the elevated percentage of activated CD62Lhi cells that develop in animals transferred with large numbers of OT-I cells.
The above results suggested that CD62L was responsible for the entry of divided OT-I cells into Ag-free LN. To test this, i.v. immunized animals were treated with anti-CD62L antibody beginning 6h after BMDC injection, and continuing every 24 hours until harvest 3 days post-immunization. This resulted in the accumulation of more extensively divided OT-I cells in the mediastinal LN priming compartment (Figure 4A), and the ratio of divided to undivided cells increased from about 5:1 to roughly 150:1. This likely reflects anti-CD62L blockade of entry of naïve cells coupled with continued division of activated OT-I cells already in the LN. In contrast, CD62L blockade substantially inhibited the redistribution of divided OT-I cells into Ag-free peripheral LN (Figure 4A). While there was seemingly little effect of anti-CD62L treatment on the redistribution of divided OT-I cells into mesenteric LN, the majority of these cells in anti-CD62L treated mice expressed α4β7 integrin, while those in control mice were predominantly negative (Figure 4B). No such enrichment for α4β7+ OT-I cells was evident in other Ag-free LN. Expression of this integrin is known to suffice for entry of naïve or memory T cells into mesenteric LN (48). Thus, our results show that activated cells that redistribute into most Ag-free LN do so by a CD62L-dependent mechanism.
The CD62L-dependent redistribution of CD62Lhi divided OT-I cells was not surprising. Importantly, however, the level of CD62L expression on CD62Llo cells was significantly above background (Figure 4C). Thus, we were interested to know whether CD62Llo cells, which predominated at lower adoptive transfer numbers and in the endogenous response, redistributed by this same mechanism. To test whether CD62Llo cells were enriched among the divided OT-I cells that continued to redistribute in anti-CD62L treated mice, we used fluorescently-labeled anti-rat IgG to detect bound anti-CD62L antibody. Using this approach, we could clearly distinguish CD62Lhi and CD62Llo populations in all lymphoid organs. We found that there was no enrichment of CD62Llo divided OT-I cells in Ag-free axillary/brachial or cervical LN in anti-CD62L treated mice relative to the populations in untreated mice (Figure 4A, D). Thus, the presence of divided CD62Lhi and CD62Llo cells in Ag-free peripheral LN is equally dependent on CD62L. This also establishes that, in contrast to redistribution into mesenteric LN, CD62Llo cells have no method to enter peripheral LN independent of CD62L. This result suggests that either the level of expression on CD62Llo cells suffices for LN entry or these cells enter LN as CD62Lhi cells and downregulate it thereafter.
To determine whether CD62Lhi cells downregulated this molecule after redistribution into peripheral LN, we treated i.v. immunized mice with FTY720 starting on day 3 and every 24 hours thereafter until harvest on day 9. By delaying FTY720 treatment, we allowed initial redistribution to occur, but then trapped redistributed cells and prevented movement of cells between compartments. Using mice transferred with either 4 × 106 or 4 × 104 OT-I cells, we found that the ratios of divided CD62Lhi to CD62Llo OT-I cells in Ag-free LN of FTY720 treated mice were not different than the ratios in untreated mice (Figure 5A, B). This result demonstrates that neither population expands or contracts in relation to the other during the 6 days after redistribution occurs. Consistent with this, in the absence of FTY720 treatment the relative percentages of CD62Lhi and CD62Llo cells in Ag-free LN of i.v immunized animals do not change between 7 and 40 days after immunization (Figure 5C). Similar data were also obtained after immunization via the i.p. route (data not shown). We cannot rigorously exclude the possibility that conversion of CD62Lhi cells occurs, but is balanced by an equivalent conversion of CD62Llo cells to a CD62Lhi phenotype. Nonetheless, these data collectively indicate that both CD62Lhi and CD62Llo CD8 T cells are long-term residents of LN.
Most models of CD8 T cell differentiation envision that cells with effector activity will migrate from lymphoid sites to peripheral non-lymphoid tissues, blood, and spleen. Thus, it was of interest to determine the effector status of the divided CD62Lhi and CD62Llo OT-I cells that had redistributed into Ag-free LN. Therefore, we assessed their ability to produce IFN-γ after a short ex vivo peptide restimulation. To avoid the short-term reduction in CD62L expression that follows antigen stimulation, we incubated cells with TAPI-2, an inhibitor of the TACE sheddase (46), during stimulation. TAPI-2 treatment had no effect on IFN-γ production (not shown). Compared with divided cells in spleen as representative of fully differentiated effectors, we found no significant difference in the percentage of divided cells in Ag-free LN making IFN-γ, or in the MFI of IFN-γ expression (Figure 6A-F). Interestingly, in both Ag-free LN and spleen, the representation of effector cells in the CD62Lhi subset was equivalent to or higher than that of the CD62Llo subset, depending on the number of cells that were adoptively transferred. Divided cells in Ag-free LN and spleen also uniformly downregulated CD127 by 72 hours post-immunization (Figure 6G), and upregulated CD44 and α4 integrin to similar extents by 6 days after i.v. immunization (data not shown). Similar data were also obtained after immunization via the i.p. and s.c routes (data not shown). Thus, divided OT-I cells that had redistributed to peripheral LN were well-differentiated.
The foregoing results established that the divided OT-I cells that had redistributed to peripheral LN persisted for at least 9 days, but also exhibited characteristics of well-differentiated effector cells. Thus, we were interested in whether these cells were terminal, or were also a source of memory. In keeping with the latter possibility, we found a significant increase in the percentage of divided OT-I cells in Ag-free LN that were CD127+ by 7 days after immunization (Figure 7A). Even larger percentages of these cells were CD127+ on day 7 when the adoptive transfer number was reduced (Figure 7B). Although these results were consistent with the idea that the redistributed T cells re-expressed CD127, they did not exclude the possibility that CD127+ cells entered Ag-free LN at a later time point. However, when we trapped redistributed cells in Ag-free LN by initiating FTY720 treatment on day 3 and continuing until day 7, we found that over two-thirds of the divided T cells in Ag-free axillary/brachial and cervical LN were CD127+ (Figure 7C). This result indicates that the divided OT-I cells that redistribute on day 3 posses the ability to upregulate CD127 by day 7, or that a minor subset of redistributed cells that are initially CD127+ expand significantly over this time period. Either mechanism is consistent with the idea that the redistributed cells contain memory CD8 T cell precursors.
To directly establish the potential of redistributed cells to seed memory, we sorted divided CD62Lhi and CD62Llo Thy1.1+ OT-I T cells from Ag-free LN 3 days after i.v. immunization. The purity of each of these populations was over 90% as assessed by flow cytometry (data not shown). Twenty thousand cells of each phenotype were transferred into naïve Thy1-mismatched hosts together with a known quantity of CFSE-labeled naïve T cells obtained from a TCR transgenic mouse with an irrelevant specificity (49) and Thy1-matched to the recipient. These cells served as an internal control for the efficiency of transfer in different recipients. Three weeks after transfer we immunized some of the recipients with recombinant vaccinia expressing OVA and analyzed the representation of OT-I cells in LN and spleen 7 days later. Transferred OT-I cells were virtually undetectable in either peripheral LN or spleen of unimmunized animals (Figure 6D). However, they were readily detectable in both compartments of vaccinia challenged animals. Relative to the co-transferred CFSE-labeled control cells, recall OT-I cells were more evident in spleen that LN, as would be expected for day 7 effectors. Nonetheless, both CD62Lhi and CD62Llo transferred cells gave rise to significant numbers of recall cells in LN. Interestingly, the number of recall cells derived from transferred CD62Lhi precursors was about 5-fold higher than from the same number of CD62Llo precursors. Collectively, our results suggest that the activated cells that redistribute into Ag-free LN within 72h after activation contain precursors of memory CD8 cells, and these are enriched in the subset of redistributed cells that are CD62Lhi.
In the current work, we have utilized local immunization with peptide-pulsed, CD40L-activated BMDC to characterize the differentiation of CD8 T cells in individual LN and their migration among LN. Peptide-pulsed BMDC traffic rapidly into lymphoid tissue in a highly restricted pattern after injection (37, 38), and remain localized and active for at least 48h (39). Thus, the use of this immunogen avoids the uncertainty about location and amount of antigen that is inherent in the use of viruses, bacteria, or peptides in adjuvant to study T cell activation in vivo. Using this system, we have demonstrated the existence of a subset of activated CD8 T cells that: a) redistribute into Ag-free LN within 72h via a CD62L-dependent mechanism, b) exhibit the characteristics of fully differentiated effector cells; and; c) contain precursors of memory cells.
An important observation in this study is that CD8 T cells activated in one LN redistribute within 3 days into other LN that are demonstrably free of APC bearing cognate antigen. Most models of CD8 T cell differentiation emphasize its occurrence in lymphoid compartments in which antigen is encountered, followed by dissemination to blood, spleen, and non-lymphoid tissues (5-14). However, models for the differentiation of CD8 T cells in LN specifically have evolved from studies in which differentiated effector cells were identified primarily or exclusively in draining LN or spleen, and presumptive non-draining LN were not evaluated (6, 15-18). It was suggested in one previous study (20) that CD8 T cells activated after intranasal administration of influenza virus redistributed to non-draining LN. However, no conclusive data to support this suggestion was provided, and it was also acknowledged that these results might be due instead to delayed antigen presentation in these initially non-draining compartments. Indeed, using the same virus and route of administration, another group made similar observations and concluded that the response was “broadly systemic” (19). This is in keeping with the demonstration that antigen presentation after viral infection is prolonged (21, 22) and occurs in LN where virus is not detected (15). The suggestion that activated CD8 T cells had redistributed from priming to Ag-free LN was also made in one prior study using BMDC as an immunogen (40). Again, however, no direct demonstration of this phenomenon was provided, and the authors acknowledged the possibility of disseminated antigen presentation due to transfer of peptide-MHC complexes to endogenous DC. More recently, Liu et al (50) used FTY720 administration to provide the first direct demonstration that virus specific cells in skin-draining LN could redistribute to other LN. However, this study did not investigate the mechanism of redistribution, nor the relationship of these cells to either conventional effector cells or memory precursors. Here, we circumvented uncertainty concerning antigen localization by using peptide-pulsed BMDC, and used FTY720 to directly demonstrate that activated OT-I cells redistribute from priming compartments to Ag-free LN. We have extended the observations of Liu et al (50) by showing that redistribution occurs regardless of the number of OT-I cells adoptively transferred, and also during the endogenous response, indicating that it is not simply a consequence of activating large numbers of antigen specific T cells in spatially limited lymphoid compartments. We have also observed FTY720-inhibitable redistribution of activated T cells to LN using vaccinia virus as an immunogen (unpublished data) demonstrating that it is not limited to responses initiated with BMDC. Instead, redistribution of activated CD8 T cells to Ag-free LN is a normal aspect of immune responses whose significance has not been well-understood.
Our work has also demonstrated that redistribution of activated CD8 T cells to most peripheral Ag-free LN is mediated by CD62L, in keeping with its already well-known involvement in the entry of naïve T cells into LN. In separate work (39), we established that a small population of activated CD8 T cells expressing α4β7 redistribute from mesenteric to mediastinal LN in a process mediated by that integrin. However, in that instance, both LN were priming compartments. In the present study, we observed that activated α4β7+ cells redistribute to Ag-free mesenteric LN in a CD62L-independent manner. Thus, redistribution of activated cells to Ag-free LN during the primary immune response is a general phenomenon that can be mediated by distinct mechanisms. Interestingly, regardless of the number of cells initially transferred, redistributed CD8 cells included subsets expressing high and low levels of CD62L,and the fraction of CD62Lhi cells was higher in cells than had redistributed than in priming LN. However, the representation of CD62Lhi and CD62Llo cells in Ag-free LN was equally susceptible to anti-CD62L blockade. The relative frequencies of these populations also remained stable after redistribution, suggesting that they do not interconvert. However, it remains possible that a portion of redistributing cells downregulate CD62L very rapidly after LN entry, and also that this process is subsequently balanced by upregulation, leading to the appearance that CD62Lhi and CD62Llo cells are stable and separate populations. Alternatively, CD62Llo expression, which is measurably above that of CD62Lneg cells, may be sufficient for LN entry. Regardless of the exact mechanism, we and others (24, 50) have also observed that previously activated CD8 cells with a reduced level of CD62L expression persist in LN long term.
Recent work has led to significant concerns about how accurately the processes of differentiation and memory are represented in animals transferred with large numbers of CD8 T cells. In particular, it was demonstrated that the fraction of activated CD8 cells that retain CD62Lhi expression after infection with an antigen expressing pathogen is higher in animals transferred with relatively high numbers of cells (24, 47). It was also shown that long-term CD62Llo memory cells rescued from such animals reacquired a CD62Lhi phenotype after adoptive transfer, leading to their designation as “transitional memory cells” (24). The development of transitional memory cells was diminished by increasing the number of endogenous APC, suggesting that an elevated ratio of antigen-specific T cells to APC results in compromised or altered activation. Whether activated early stage CD62Llo cells show a similar reacquisition of CD62L is unknown. However, asymmetrical division of a single T cell has been shown to produce memory and effector lineage cells that are CD62Lhi and CD62Llo, respectively, as early as the first cell division (51). Thus, while adoptive transfer may alter the fraction of CD62Lhi and CD62Llo expressing cells, both are physiologically relevant offspring ot the CD8 T cell activation process. Larger adoptive transfer numbers were also associated with a higher fraction of activated early stage cells that retained CD127 (47), although a second study using a distinct immunogen observed different kinetics of CD127 loss and re-expression, and a more complex relationship to adoptive transfer number (52).
We have demonstrated a direct correlation between adoptive transfer number and CD62Lhi expression on activated early stage cells similar to that described previously (24, 47). However, we found that the percentage of activated cells expressing CD127 on day 7 was elevated, not diminished, when lower numbers of cells were transferred. This difference with previous work is likely ascribable to our use of BMDC as an immunogen. The administration of exogenous DC may reduce the elevated T cell: APC ratio thought to be responsible for altered T cell differentiation in other models using pathogens presented by endogenous DC. Finally, larger adoptive transfer numbers were associated with a smaller fraction of cells expressing granzyme B, but a larger fraction expressing IL-2 and TNF (47). In our hands, we found that the fraction of cells expressing IFN-γ was enhanced at high adoptive transfer numbers. Nonetheless, in both studies, these differences were quantitative rather than qualitative. Thus, while adoptive transfer number can clearly affect the proportions of cells expressing effector or memory markers, it does not obviate the fundamental observation that activated CD8 T cells that have redistributed to a Ag-free LN contain both effectors and memory precursors.
Another important observation in the present study is that the CD8 cells that redistribute to Ag-free LN, regardless of their level of CD62L expression, have characteristics of well-differentiated effectors. Their production of IFN-γ was equivalent to that of “classical” splenic effectors, and they had acquired high level expression of either α4β1 or α4β7 integrin (39), enabling them to home to peripheral non-lymphoid tissue. The biological significance of having CD8 T cells with effector activity in LN where there is no antigen or active infection is not clear. On the other hand, we have also demonstrated that redistributed CD8 cells are not simply terminal effectors that die as the primary response contracts, but contain the precursors of memory cells. Regardless of adoptive transfer number, a significant fraction of redistributed cells express CD127 by day 7, which has been associated with a memory or pre-memory phenotype (25, 53). More directly, redistributed cells rescued from LN 3 days after immunization persisted as memory cells after adoptive transfer into naïve recipients. This time frame for the establishment of memory precursors is consistent with that of other work in which activated cells were recovered from priming LN (18). From this perspective, the observation of effector activity in the cells that redistribute to Ag-free LN is also consistent with other work suggesting that memory cells go through a stage in which they express effector activity (18, 25, 29).
Interestingly, the size of the recall response was approximately 5-fold higher in animals that had received redistributed OT-I cells with a CD62Lhi phenotype than in those receiving and equivalent number of CD62Llo cells. One possibility is that this reflects enhanced survival or homeostatic proliferation of CD62Lhi cells after adoptive transfer (24). However, the ratio of these two populations did not change from day 3 to day 9 in the LN of animals treated with FTY720, nor in the LN of untreated animals up to 40 days after immunization. Alternatively, it is possible that the CD62Lhi cells distribute more efficiently into lymphoid tissue after adoptive transfer, while the CD62Llo cells are more likely to enter peripheral tissue, and that this alters the extent to which they engage with DC after vaccinia infection. This is in keeping with other work on the distribution of adoptively transferred CD62Lhi and CD62Llo cells (6, 54, 55), although the cells used were usually propagated in vitro or isolated from spleen rather than LN. Finally, it is possible that that the CD62Lhi cells have greater capacity to proliferate in response to the vaccinia challenge. This possibility is also consistent with other work (10, 11, 29, 31). Regardless of the exact mechanism, our results suggest that early-stage redistributed CD62Lhi cells are enriched for central memory precursors. CD62Llo cells also harbor memory precursors, although these may represent central, transitional, or effector memory lineages.
High and low level expression of CD62L is often used, irrespective of location, to define TCM and TEM subsets, respectively (6, 31). However, the stable representation of both CD62Lhi and CD62Llo antigen experienced OT-I cells in LN for at least 40 days, and the fact that both populations contain memory precursors that give rise to recall responses in LN after adoptive transfer, indicates that residence in an Ag-free LN is a more inclusive definition of TCM. In relation to this, we also suggest that the rapid redistribution of CD8 T cells to Ag-free LN represents an additional marker of TCM precursors, and is a process that leads to the establishment of systemic memory.
We thank Sarah T. Lewis for generation of H-2Kb-ova257 tetramer and purification of anti-CD62L antibody. We also thank Janet V. Gorman for maintenance of the animal colony and Michael Solga for flow sorting.
1This work was supported by USPHS grant CA78400 (VHE), USPHS Training Grants GM07627 (SSO), AI007496 and the Robert R. Wagner Fellowship (CCB), and USPHS fellowship AI072818 (ARF).
4Abbreviations used in this paper: