Dendritic Cell-Induced Memory T Cell Activation in Nonlymphoid Tissues

See allHide authors and affiliations

Science  11 Jan 2008:
Vol. 319, Issue 5860, pp. 198-202
DOI: 10.1126/science.1151869


Secondary lymphoid organs are dominant sites of T cell activation, although many T cells are subsequently retained within peripheral tissues. Currently, these nonlymphoid compartments are viewed as sites only of effector T cell function, without the involvement of renewed induction of immunity via the interactions with professional antigen-presenting cells. We describe a method of reactivation of herpes simplex virus to examine the stimulation of tissue-resident T cells during secondary challenge. The results revealed that memory CD8+ T cell responses can be initiated within peripheral tissues through a tripartite interaction that includes CD4+ T cells and recruited dendritic cells. These findings lend evidence for the existence of a sophisticated T cell response mechanism in extra-lymphoid tissues that can act to control localized infection.

The activation of T cells during localized infection takes place within the draining lymph nodes, where the bulk of T cell priming is thought to occur (1). However, peripheral nonlymphoid tissues harbor a sizable proportion of the overall T cell pool, primarily consisting of long-lived memory T cells (2, 3). During infection, peripheral tissues invariably represent the first point of contact with a wide range of pathogens, with resident T cells considered as critical to local infection control (4). Indeed, it has been suggested that such sites should be viewed as an extension of the secondary lymphoid compartment, and the term effector-lymphoid tissue (ELT) has been used in their description (5). Nevertheless, the contribution of the individual components within the ELT and the nature of the local responses remain poorly defined. Furthermore, it is commonly assumed that peripheral responses occur solely through the elaboration of effector molecules in the absence of new lymphocyte proliferation (6, 7). However, local expansion has been reported in certain situations (8), and it remains possible that dendritic cells (DCs) are involved because these cells are known to aid in memory T cell activation (9). Such DCs might be resident within peripheral tissues (10) or could derive from infiltrating monocytes in response to inflammation (11).

Transplantation of peripheral tissues containing T cells into naïve recipients provides one convenient means of studying local responses, because it permits identification of the memory cells originating from these nonlymphoid sites while allowing recruitment of other cells that may be critical for their optimal activation. We selected sensory dorsal root ganglia (DRGs) from mice that had been infected with herpes simplex virus (HSV) in their flank, because these tissues retain infiltrating T cells well after resolution of acute infection (1214). In addition, these ganglia harbor HSV in a latent form, which is typically reactivated as a consequence of surgical extraction (15), providing an effective means of local virus challenge. Ganglia taken 20 and 40 days after infection had detectable numbers of virus-specific T cells and no infectious virus at the time of extraction (13). Excision and transplantation provided a robust level of virus reactivation over the course of 5 or 6 days (fig. S1). To track virus-specific T cells, we seeded mice (to be used as DRG donors before HSV infection) with low numbers of T cells from animals expressing transgenic T cell receptors specific for the class I–restricted immunodominant peptide of HSV glycoprotein B (gBT-I cells). Within two days of transplantation, the numbers of T cells in the DRGs fell sharply, in most cases to undetectable levels (Fig. 1, A and B) in a fashion similar to that observed recently (16). This decline was followed by a rise in specific T cell numbers, beginning around days 5 or 6 after transplantation (Fig. 1A). The peak in numbers of gBT-I cells at day 9 represented a 17-fold increase over starting cell numbers and at least a 45-fold difference over those found at the day 2 post-transplant nadir (Fig. 1B). Histological comparison of ganglia at days 3 and 9 after transplantation (Fig. 1C) confirmed that these cells were undetectable in ganglia at the earlier time.

Fig. 1.

Expansion of tissue-resident T cells after HSV reactivation. (A and B) B6 mice seeded with naïve gBT-I.CD45.1+ CD8+ T cells were infected with HSV. DRGs from these mice harboring latent HSV infection were transplanted under the kidney capsule of naïve recipients. Numbers of gBT-I.CD45.1+ CD8+ T cells within each graft were enumerated by flow cytometry. Data points represent individual mice from a single experiment (A) and pooled data from seven experiments (B). Horizontal bars indicate the mean. (C) Sections of the graft on days 3 and 9 after transplantation, stained for CD8 (green) and 4′,6′-diamidino-2-phenylindole (DAPI) (blue nuclear staining). (D) B6 mice were seeded with in vitro activated gBT-I cells and OT-I.CD45.1+ CD8+ T cells 4 days after HSV infection. Latently infected DRGs from these mice were transplanted under the kidney capsule of recipients. Grafts were recovered, and the numbers of gB-specific and ovalbumin-specific CD8+ T cells were enumerated. Error bars represent the mean + SEM with five mice per group. (E and F) B6 mice seeded with naïve gBT-I.CD45.1+ CD8+ T cells were infected with either a thymidine kinase mutant HSV incapable of viral reactivation or the wild-type parental strain. Latently infected DRGs were transplanted under the kidney capsule of recipient mice and, at the indicated times, the number of gBT-I.CD45.1+ CD8+ T cells within the graft was determined. The data in (E) represent the mean + SEM with five mice per group. The decay in the number of donor gBT-I.CD45.1+ CD8+ T cells within the graft after transplantation of ganglia latently infected with a thymidine kinase mutant HSV is shown in (F). The data in (F) represent the mean + SEM with eight mice per group. (G) Donor gBT-I.CD45.1+ CD8+ T cells could not be detected within the spleen or kidney-draining lymph nodes (LN) of transplant recipients.

To verify that the expansion of T cells was antigen-specific, we seeded donor mice with activated gBT-I cells and control ovalbuminspecific T cells (OT-I cells). Because T cell infiltration of infected DRGs is nonspecific, reliant solely on an activated T cell phenotype (13), transplanted DRGs initially contained roughly equal numbers of HSV-specific gBT-I cells and irrelevant OT-I cells. However, only the gBT-I cells responded to reactivation (Fig. 1D). Infection with a thymidine kinase mutant virus, which does not reactivate from neuronal tissue (17), failed to drive any T cell expansion (Fig. 1E), showing that the response required overt infection. In contrast, reactivation was not necessary for the loss of the resident T cells (Fig. 1F), which suggests that these cells were probably dying as a consequence of the surgical trauma. Finally, at all times examined, gBT-I cells were absent from the spleen and renal lymph nodes (Fig. 1G), arguing that the antigen-specific increase in T cell numbers resulted from local proliferation.

Despite our inability to detect circulating T cells, the numerical increase in gBT-I cells within the ganglia could nonetheless have resulted from stimulation within lymphoid tissues, followed by rapid recruitment into infected ganglia. To exclude this possibility, we infected two different groups of mice: one received HSV-specific cells tagged with the CD45.1 allelic marker (gBT-I.CD45.1 T cells) before infection, whereas the second group received CD45.2-bearing gBT-I cells (gBT-I.CD45.2 Tcells). All mice were infected with HSV, and on resolution of primary infection, ganglia from each group were transplanted into distinct regions under the kidney capsule of a C57BL/6 (B6) recipient. The reisolated DRGs largely contained either one or the other T cell subset, showing little cross-migration of T cells between DRGs (Fig. 2, A and B). Such strict segregation would not be evident if T cell expansion occurred in lymphoid organs with subsequent infiltration of DRGs. Thus, the local response is maintained in each tissue, with few of the T cells released into the wider circulation.

Fig. 2.

Local reexpansion of resident memory CD8+ T cells. B6 mice seeded with either gBT-I.CD45.1+ or gBT-I.CD45.2+ CD8+ T cells were infected with HSV. Latent DRGs harboring these different T cells were harvested and placed in different locations under a single kidney capsule. (A) Plots of lymphocytes isolated from the grafts in different locations on day 9 after transplantation. (B) The numbers of gBT-I.CD45.1+ CD8+ T cells within both transplants, at the indicated times. The data represent the mean + SEM with seven mice per group.

Transplantation of ganglia into bone marrow irradiation chimeras suggested that parenchyma was not the dominant driver of T cell expansion. For these experiments, bm3 animals carrying the non–gBT-I cell–stimulating Kbm3 major histocompatibility complex class I (MHC I) molecule (13) were reconstituted with B6 bone marrow. These B6→bm3 chimeras were then used as donors, because they permit gBT-I cell priming by B6 DCs and their subsequent infiltration into infected ganglia. A robust glycoprotein B (gB)–specific T cell expansion was measured in the transplanted B6→bm3 ganglia (fig. S2), despite infected neurons being of nonpresenting bm3 origin. Antigen-presenting cells (APCs), most likely DCs, were thus responsible for the effective T cell stimulation. Analysis of CD45.2-reactivating ganglia transplanted into CD45.1 recipients showed recruitment of substantial numbers of recipient-derived CD11c+ cells after a 2-day delay (Fig. 3A). These cells expressed a range of DC markers (fig. S3) and were sometimes found in close proximity to CD8+ cells (Fig. 3B).

Fig. 3.

Stimulation of resident memory CD8+ T cells is dependent on recruited APCs. (A) Latently infected DRGs harvested from B6 donor mice were transplanted under the kidney capsule of CD45.1 recipients. The number of donor (CD45.1) and recipient (CD45.1+) DCs (CD11c+MHCII+) was determined at the indicated times after transplantation. Error bars represent the mean + SEM with four mice per group. (B) Section of the graft on day 9 after transplantation, stained for CD8 (green) and CD11c (red). (C) Latent ganglia were grafted into recipient mice containing fluorescein isothiocyanate (FITC)–bead-labeled Gr-1hi monocytes. The level of expression of Gr-1, CD11b, CD11c, and MHC II on the bead-positive cells in the grafts was determined at the indicated times. (D to F) Latent DRGs containing gBT-I.CD45.1+ CD8+ T cells were grafted into recipient mice depleted of circulating monocytes via the administration of clodronate-loaded liposomes (+Clod-liposomes). On day 9 after transplantation, the number of CD11c+MHCII+ cells [(D) and (E)] and gBT-I.CD45.1+ CD8+ T cells (F) in the graft was determined by flow cytometry. The data in (D) and (F) represent the mean + SEM with seven mice per group. *, P = 0.0262 in (D); *, P = 0.0175 in (F). (G) Latent DRGs from HSV-infected mice containing gBT-I.CD45.1+ CD8+ T cells were transplanted into either B6→B6 or bm3→B6 bone marrow chimeras. The number of gBT-I.CD45.1+ CD8+ T cells within the graft was enumerated at the time of transplantation (day 0) or 9 days after transplantation. The data represent the mean + SEM of the combined experiments shown in fig. S5. The difference between groups was determined to be statistically significant with a Mann-Whitney test. ***, P = 0.0002.

DCs recruited into sites of inflammation are derived from a subset of circulating monocytes that develop into DCs once they enter peripheral tissues (1821). To investigate whether such DCs were formed in reactivating DRGs, we took advantage of the finding that Gr1hi monocytes that convert into DCs can be labeled with fluorescent beads after eliminating an interfering noninfiltrating subset via the administration of clodronate-containing liposomes (22). Figure 3C shows that such bead-labeled cells mature in reactivating ganglia to a Gr1lo population containing CD11bhiCD11chi DCs expressing high levels of MHC II molecules. Similar recruitment of blood-derived DCs is seen after primary infection (fig. S4), and these cells are the likely source of the low number of persisting DCs found in the latent ganglia (Fig. 3A). To formally show monocyte-derived DC involvement in T cell expansion, we eliminated their precursors by repeatedly administering clodronate-containing liposomes after transplantation (23). Despite incomplete DC ablation (Fig. 3, D and E), treatment inhibited the level of T cell expansion (Fig. 3F). A contribution from other presenting cells during T cell stimulation may explain the residual expansion, although blood-derived APCs clearly make a substantial contribution to the observed response.

It was possible that the cells recruited from the circulation were driving T cell expansion solely by some indirect means rather than by direct antigen presentation. To exclude this possibility, we transplanted latent ganglia from HSV-infected B6 mice containing gBT-I cells into bm3→B6 bone marrow chimeras that contain DCs unable to stimulate the gBT-I cells. These chimeras suffered from a high degree of variability (fig. S5), although on average, T cell expansion was much reduced in the bm3→B6 recipients as compared with that in the B6→B6 controls (Fig. 3G), which is consistent with a host APC contribution. Again, inhibition was incomplete, which suggests some contribution by donor APCs.

Donor CD4+ T cells, which persist within latent ganglia, also expanded within transplanted grafts after viral reactivation (Fig. 4A). This CD4+ T cell proliferation was significantly reduced after transplantation into MHC II–deficient recipients as compared with such proliferation in B6 controls, implying a further role for blood-recruited APCs in local T cell activation (Fig. 4B). Unexpectedly, these MHC II–deficient recipients also showed attenuated expansion of DRG-resident gB-specific T cells (Fig. 4C), arguing that the local memory CD8+ Tcell activation was itself CD4+ helper T cell–dependent. To confirm this finding, we grafted latent ganglia, taken from donor mice depleted of CD4+ cells just before harvest, into recombination-activating gene (RAG)–deficient mice. Eliminating the CD4+ T cells inhibited CD8+ T cell expansion on secondary antigen encounter (Fig. 4D), although it did not compromise their intrinsic responsiveness because the cells were still able to produce interferon-γ after in vitro stimulation (fig. S6). The tissue-resident CD4+ T cells alone were sufficient to elicit help because the RAG-deficient recipients are otherwise devoid of T cells. It should be noted that primary HSV-specific CD8+ T cell activation is also helper T cell–dependent (24).

Fig. 4.

Stimulation of resident memory CD8+ T cells is dependent on CD4+ T cell help. (A) The numbers of donor CD4+ T cells (CD45.1+CD4+βTCR+) in HSV-infected DRGs after transplanting into B6 recipients. TCR, T cell receptor. The data represent the mean + SEM with five mice per group. (B and C) Latent DRGs from donor mice seeded with gBT-I.CD45.1+ CD8+ T cells were transplanted into either MHC II knockout (KO) or B6 mice. On day 9 after transplantation, flow cytometry was used to enumerate the number of CD4+ T cells (B) and gBT-I.CD45.1+ CD8+ T cells (C). The data represent the mean + SEM. The difference between groups was determined to be statistically significant with a Mann-Whitney test. **, P = 0.0012; ***, P < 0.0001. Horizontal bars in (C) indicate the mean. (D) Latent ganglia harvested from CD4-depleted donor mice containing gBT-I.CD45.1+ CD8+ T cells were grafted into RAG-deficient recipient mice. At the times indicated, the numbers of gBT-I.CD45.1+ CD8+ T cells within the graft were enumerated. The data represent the mean + SEM with six mice per group.

At face value, our results appear to contradict findings arguing that antigen encounter within nonlymphoid tissues does not result in local T cell proliferation (6, 7). However, it may prove that such a lack of proliferation is only seen under certain conditions: for example, during activation of the T cells by parenchymal cells or where the virus is rapidly brought under control by resident T cells (25). As a consequence, we propose that local responses are tailored to the level of inflammation or the extent of infection. In settings with limited inflammation or where presentation of virus antigen is confined to the parenchyma, pure elaboration of effector function results in little overall T cell expansion. If, however, infection progresses and DC recruitment comes into play, then local T cells proliferate in situ. Finally, our results provide insight into HSV immune evasion during reactivation. Virus-specific CD8+ T cells, thought capable of suppressing reactivation (12, 26), are lost from latent ganglia in situations that promote virus recrudescence (16). The natural delay in DC infiltration and local Tcell expansion affords a natural opportunity for transient HSV replication before the virus is once again brought under renewed immune control.

Supporting Online Material

Materials and Methods

Figs. S1 to S6


References and Notes

Stay Connected to Science

Navigate This Article