Preferential Localization of Effector Memory Cells in Nonlymphoid Tissue

See allHide authors and affiliations

Science  23 Mar 2001:
Vol. 291, Issue 5512, pp. 2413-2417
DOI: 10.1126/science.1058867


Many intracellular pathogens infect a broad range of host tissues, but the importance of T cells for immunity in these sites is unclear because most of our understanding of antimicrobial T cell responses comes from analyses of lymphoid tissue. Here, we show that in response to viral or bacterial infection, antigen-specific CD8 T cells migrated to nonlymphoid tissues and were present as long-lived memory cells. Strikingly, CD8 memory T cells isolated from nonlymphoid tissues exhibited effector levels of lytic activity directly ex vivo, in contrast to their splenic counterparts. These results point to the existence of a population of extralymphoid effector memory T cells poised for immediate response to infection.

After encounter with antigen, CD8 T cells differentiate into effector cells, which form a crucial arm of the adaptive immune response against intracellular pathogens through the action of cytokines and cell-mediated cytotoxicity (1, 2). Activation leads to proliferation and a large increase in antigen-specific cytotoxic T lymphocytes (CTLs); this effector population contracts after resolution of the infection, leading to a stable memory population of intermediate frequency (3–9). Although CD8 memory cells are identifiable within secondary lymphoid organs (4, 7, 10, 11), their pattern of migration and their relationship with memory populations in other tissues remains unclear. Previous work demonstrated that T cells with a memory phenotype migrate through peripheral tissues, such as skin and intestine; hence, subsets of memory cells with a migratory preference for a given tissue may exist (9, 12–14). Recent data also indicate that human memory-phenotype CD4 and CD8 T lymphocytes within the blood may be divisible into two subsets on the basis of chemokine receptor expression and effector function (15). Although this theory is intriguing, the population dynamics and the functional status of pathogen-specific memory T cells within different tissues in vivo remain poorly characterized. Moreover, it is unknown whether memory T lymphocyte characteristics measured in lymphoid tissue are indicative of the overall memory population. To address these questions, we compared antigen-specific effector and memory CD8 T cells in lymphoid and nonlymphoid tissues after viral or bacterial infections.

C57Bl/6J mice were infected intravenously with vesicular stomatitis virus (VSV), Indiana serotype. Infection of mice with VSV is transient (16) and has been used as a model of a nonpersistent pathogen (17, 18). At various times after infection, we isolated lymphocytes from lymphoid and nonlymphoid tissues; we then used major histocompatibility complex (MHC) class I tetramers containing the immunodominant peptide epitope from the VSV nucleoprotein and H-2Kb(VSV-N52-59/Kb) (19). MHC class I tetramers containing the appropriate antigenic peptide allow identification of antigen-specific CD8 T cells within an entire lymphocyte population (6, 7, 20, 21). Eight days after infection, VSV-specific CD8 T cells were present in all tissues examined, including spleen, peripheral and mesenteric lymph nodes, peripheral blood, small intestine lamina propria (LP) and intraepithelial lymphocyte (IEL) compartments, lung, fat pad, liver, kidney, bone marrow, and peritoneal cavity (Fig. 1A). Tetramer+ cells in all tissues expressed high levels of CD11a, which is up-regulated by activation (Fig. 1A) and remains high on memory CD8 T cells (22). As a percentage of CD8 T cells, tetramer+cells were more prominent in nonlymphoid tissues than in secondary lymphoid tissues. To determine whether this phenomenon was unique to VSV infection, we infected mice orally with recombinant Listeria monocytogenes expressing ovalbumin (LM- ova) (23) and visualized the response using H-2Kb/SIINFEKL tetramers (where SIINFEKL is the ova peptide, Ser-Ile-Ile-Asn-Phe-Glu-Lys-Leu). This infection resulted in generation of a robust anti-ova CD8 T cell response in all tissues (Table 1), in which high percentages of tetramer+ cells were present in nonlymphoid sites. These data implied that migration of activated CD8 T cells to multiple tissues was independent of the infectious agent and the infection route.

Figure 1

Infection with VSV leads to the appearance of virus-specific CD8 T cells in lymphoid and nonlymphoid tissues. C57Bl/6J mice were infected intravenously with 106plaque-forming units (PFU) of VSV-Indiana, and 8 days (A) or 81 days (B) later, mice were perfused and lymphocytes were isolated from the indicated tissues. The percentage of antigen-specific CD8 T cells was assessed by staining with N52-59/Kb tetramer and antibodies to CD8α and CD11a, followed by fluorescence flow cytometry. Plots shown are gated on CD8α+ lymphocytes; values are mean percentages of tetramer+ cells within the CD8+ T cell population derived from at least four mice. Control staining with a Kb tetramer containing SIINFEKL was negligible. PLN, peripheral lymph nodes; MLN, mesenteric lymph nodes; PBL, peripheral blood lymphocytes; LP, small intestine lamina propria; IEL, small intestine intraepithelial lymphocytes; BM, bone marrow; Perit, peritoneal cavity lymphocytes.

Table 1

Induction of peripheral tissue memory byListeria or VSV infection. C57Bl/6J mice were infected by gavage with 109 colony-forming units of LM-ova or by intravenous injection with 106 plaque-forming units of VSV, and lymphocytes were isolated from numerous tissues at the indicated time points; the percentage of antigen-specific CD8 T cells was assessed by staining with SIINFEKL/Kb tetramer (LM-ova) or N52-59 tetramer (VSV) and antibodies to CD8α and CD11a, followed by fluorescence flow cytometry. Control staining with the irrelevant tetramer was negligible. Values represent the mean percentage of tetramer+ cells of the CD8+ T cell population ± standard error from three to six mice per time point. PLN, peripheral lymph node; MLN, mesenteric lymph node; PP, Peyer's patch; PBL, peripheral blood lymphocytes; BM, bone marrow; LP, small intestine lamina propria; IEL, small intestine intraepithelial lymphocytes; Perit, peritoneal lavage; ND, not detectable.

View this table:

To test whether localization of memory CD8 T cells reflected the primary response, we analyzed tissues at protracted times after infection. At 81 and 296 days after VSV infection, or at 20 and 59 days after Listeria infection, memory cells were detectable in all nonlymphoid tissues (Fig. 1B and Table 1). Differences were observed when the kinetics of the tissue responses were compared (Fig. 2A). Although the percentage of tetramer+ cells in the spleen, lymph nodes, and blood declined rapidly, the response was prolonged in most other tissues. This finding, and the fact that tissues such as lung, liver, and kidney were perfused before lymphocyte isolation, indicated that the tetramer+ cells detected were not derived from blood contamination. Interestingly, only very small populations of memory cells were present in lymph nodes, indicating a preference for CD8 memory cells to migrate and expand elsewhere. Among nonlymphoid tissues analyzed, the lung, liver, and small intestine LP contributed the majority of tetramer+ cells. When total numbers of antigen-specific T cells per tissue were compared, the cumulative sum of the cell numbers in the lung, liver, and LP was similar to that in the spleen at the peak of the response, as the response declined, and in the memory phase of the response (Fig. 2B). Thus, the number of cells within nonlymphoid tissues represents a substantial portion of the overall response. The kinetics of the response with regard to total cell numbers in each tissue was distinct and did not always reflect the kinetics as calculated from percentages of tetramer+ cells (Fig. 2B). Thus, relative to the spleen, tetramer+ cell numbers in the LP and lung declined gradually while tetramer+ cells in the liver declined rapidly. Nevertheless, a stable population of memory CD8 T cells was detectable in each tissue.

Figure 2

Tissue-specific kinetics of the anti-VSV CD8 T cell response. At the indicated numbers of days after infection, lymphocytes from four to eight mice were isolated and stained with tetramer and antibodies to CD8α and CD11a, then analyzed by fluorescence flow cytometry. At all time points, the total number of viable lymphocytes isolated per tissue was enumerated by trypan blue exclusion and light microscopy. (A) Kinetics of the response based on percentage of CD8+ cells that were tetramer+. Symbols: solid squares, spleen; right-pointing arrowheads, liver; left-pointing arrowheads, lung; triangles, kidney; inverted triangles, peritoneum; stars, PBL; open circles, PLN; solid circles, LP; diamonds, fat pad. (B) Kinetics of the response based on total numbers of tetramer+ cells per tissue. Values were derived by multiplying the percentage of total lymphocytes that were tetramer+ by the total number of lymphocytes isolated from that tissue. Values for “tissues” in the upper left panel (solid circles) were the total tetramer+ cells from lung, liver, and LP. Values represent means ± standard error.

The distinct response kinetics observed in lymphoid versus nonlymphoid tissues prompted us to examine possible functional differences that might exist between the antigen-specific cells in different locations. Twenty days after VSV infection, short-term culture of cells with the N52-59 peptide induced the majority of antigen-specific cells from all tissues tested to produce interferon-γ (IFN-γ), based on percentages of tetramer+ cells (Fig. 3A). Although in some cases the percentage of LP and lung cells responding was less than that observed by tetramer staining, this was not a consistent finding. However, when direct ex vivo antigen-specific lytic assays were performed, striking differences were observed between cells from lymphoid versus nonlymphoid sites. Twenty days after infection with VSV, high lytic activity was mediated by cells from LP, lung, and liver, but not by splenocytes at identical effector-to-target (E:T) ratios (Fig. 3B). At later times after infection, insufficient percentages of tetramer+ cells were present in some tissues to allow a definitive comparison at overlapping E:T ratios. To circumvent this problem, we infected mice with VSV–New Jersey, a second VSV serotype that induces cross-reactive CTLs (24, 25), followed by secondary infection 7 months later with VSV-Indiana. This resulted in the generation of a larger VSV-specific memory population. Seven months after secondary infection, lymphocytes were assayed for ex vivo antigen-specific lytic activity. In agreement with previous reports, memory splenocytes exhibited low levels of lytic activity (4, 9, 26). Lymphoid memory cells were, however, able to rapidly up-regulate lytic activity after antigen challenge (9,27). Nevertheless, at comparable E:T ratios, CD8 T cells from liver, lung, and LP exhibited potent lytic activity (Fig. 3C). Experiments performed at 111 days (28) and 134 days (27) after secondary infection yielded similar results and also showed that the cell isolation procedures were not responsible for induction of lytic activity (28). Also,Listeria-specific memory cells in tertiary tissues were highly lytic (27).

Figure 3

Virus-specific CD8 memory T cells in peripheral but not lymphoid tissues are constitutively cytolytic. (A) Twenty days after VSV infection, lymphocytes were isolated as described (9, 34) from spleen, liver, lung, or LP and were cultured with Golgiplug (Pharmingen) in the presence or absence of N52-59 peptide for 5 hours at 37°C. Cells were washed, surface-stained with CD8α and CD11a, and permeabilized with PermWash (Pharmingen). After incubation with anti–IFN-γ, cells were washed and analyzed by flow cytometry. Plots shown are gated on CD8α+ lymphocytes; values are the mean percentages of IFN-γ+ or tetramer+ cells ± standard error within the CD8+ T cell population derived from at least four mice. (B) Twenty days after VSV infection, lymphocytes were incubated for 4 to 5 hours with51Cr-labeled untreated EL4 target cells (27) or target cells pulsed with N52-59 peptide. E:T ratio was 200:1 for all tissues. E:T values shown in plots are corrected for the number of tetramer+ cells in each population. Percentage of tetramer+ cells among total lymphocytes: spleen, 0.3%; LP, 1.0%; lung, 1.2%; liver, 1.0%. Data shown are derived from a single spleen (similar results were obtained using spleens from three other mice) and pools of liver, LP, or lung cells from two mice (an equivalent pool from two additional mice gave similar results). (C) As described in (B), except lymphocytes were isolated from mice primed with 106 PFU of VSV–New Jersey, rested >7 months, then infected with VSV-Indiana and rested an additional 224 days. E:T ratio was 300:1 for all tissues. E:T values shown in plots are corrected for the number of tetramer+ cells. Percentages of tetramer+ cells among total lymphocytes: spleen, 1.1%; LP, 0.5%; lung, 1.7%; liver, 1.2%. Data shown are derived from a single spleen and LP (similar results were obtained from these tissues from three other mice), a pool of liver cells from two mice (an equivalent pool from two additional mice gave similar results), and a pool of lung cells from four mice. Spontaneous 51Cr release was <10% and specific lysis was <5% in the absence of peptide. (D) Comparison of lytic activity from primary and memory CD8 T cells. Cells from spleen (squares) or lung (arrowheads), from mice infected with VSV 7 days previously (open symbols) or secondarily 224 days previously (solid symbols), were tested against N52-59peptide–coated 51Cr-labeled EL4 target cells in a 5-hour assay.

We directly compared the lytic activity of primary anti-VSV effectors and memory T cells from spleen and lung. The lytic activity of primary splenic and lung effectors and of lung memory cells was comparable, whereas memory splenocytes had low activity (Fig. 3D). Therefore, migration through multiple nonlymphoid tissues was a hallmark of CD8 effector memory T cells generated through viral or bacterial infection. We also examined the size and phenotype of lymphoid and nonlymphoid memory cells (28). Although memory cells from all tissues were similar in size to naı̈ve cells, phenotypic differences were noted between memory cells in the LP and other tissues, indicating that an intestinal mucosa-specific memory population may exist (28), as we previously suggested (29).

This study demonstrated that bacterial or viral infection resulted in margination of CD8 T cell effectors and memory cells into nonlymphoid tissues. Our results suggest that memory cells either continuously recirculate through peripheral tissues, or permanently reside in them. Previous studies have demonstrated the existence of activated as well as memory phenotype T cells in a limited number of tissues after a variety of infections (6, 9, 30–32), but the extent of CD8 T cell migration or retention during primary and memory phases of the response was not fully appreciated. Perhaps more important, our results demonstrate a functional distinction between tissue-homing or resident memory cells and memory cells located in the secondary lymphoid organs. A previous study suggested that CCR7+CD62L+ central memory cells without immediate effector ability traffic through secondary lymphoid tissue, whereas CCR7CD62L effector memory T cells circulate outside of lymphoid tissue (15). Our data support the existence of functionally distinct memory T cell subsets, but also suggest additional complexity in the system, because CD62L splenic and tissue memory cells had distinct functional abilities. We also cannot rule out the potential effects of persisting antigen on nonlymphoid memory cells, although antigen is not believed to be required for anti-LCMV (lymphocytic choriomeningitis virus) splenic CD8 memory cell survival (33). It will be important to determine whether memory cells undergo tissue-specific regulation and perhaps can be distinguished by the expression of distinct sets of homing molecules and chemokine receptors. Teleologically, the ability of memory cells in tertiary sites to exert immediate cytolytic activity provides a mechanism for improved survival of the organism via rapid containment of pathogens.

  • * To whom correspondence should be addressed. E-mail: llefranc{at}


View Abstract

Navigate This Article