Report

Resident memory CD8 T cells trigger protective innate and adaptive immune responses

See allHide authors and affiliations

Science  03 Oct 2014:
Vol. 346, Issue 6205, pp. 98-101
DOI: 10.1126/science.1254536

Resident memory T cells sound the alarm

Immunological memory protects against reinfection. Resident memory T cells (TRM) are long-lived and remain in the tissues where they first encountered a pathogen (see the Perspective by Carbone and Gebhardt). Schenkel et al. and Ariotti et al. found that CD8+ TRM cells act like first responders in the female reproductive tissue or the skin of mice upon antigen reencounter. By secreting inflammatory proteins, TRM cells rapidly activated local immune cells to respond, so much so that they protected against infection with an unrelated pathogen. Iijima and Iwasaki found that CD4+ TRM cells protected mice against reinfection with intravaginal herpes simplex virus 2.

Science, this issue p. 98, p. 101, p. 93; see also p. 40

Abstract

The pathogen recognition theory dictates that, upon viral infection, the innate immune system first detects microbial products and then responds by providing instructions to adaptive CD8 T cells. Here, we show in mice that tissue resident memory CD8 T cells (TRM cells), non-recirculating cells located at common sites of infection, can achieve near-sterilizing immunity against viral infections by reversing this flow of information. Upon antigen resensitization within the mouse female reproductive mucosae, CD8+ TRM cells secrete cytokines that trigger rapid adaptive and innate immune responses, including local humoral responses, maturation of local dendritic cells, and activation of natural killer cells. This provided near-sterilizing immunity against an antigenically unrelated viral infection. Thus, CD8+ TRM cells rapidly trigger an antiviral state by amplifying receptor-derived signals from previously encountered pathogens.

CD8 T cells control viral infections. To be licensed with effector functions, naïve CD8 T cells must first be activated by specialized members of the innate immune system that have been alarmed by danger signals in the form of recognition of broadly conserved microbial associated molecular patterns [such as double-stranded RNA or lipopolysaccharide (LPS)] (1). Paradigmatically, CD8+ T cells act very locally by contacting infected host cells for both the detection of pathogen-associated peptides presented by major histocompatibility complex class I (MHC-I) and target-cell-specific delivery of toxic effector molecules (2). After clearance of infection, memory T cells remain. Central, effector, and resident memory T cell subsets (TCM, TEM, and TRM, respectively) occupy different anatomic niches, where they fulfill distinct roles in protective immunity (3, 4). The most recently described subset, TRM, comprises non-recirculating memory T cells that remain positioned at common portals of reinfection, including barrier tissues such as the mucosae and skin (510). When present at the site of reinfection, TRM cells accelerate protection against homologous reinfections, although the mechanism remains obscure (1015). We recently demonstrated that reactivation of TRM cells results in bystander recruitment of recirculating memory T cells to the site of anamnestic Ag exposure in a manner dependent on the cytokine interferon-γ (IFN-γ) (16). This suggested that TRM cells serve a sentinel and alarm function in addition to more prototypical CD8 T cell functions.

We wanted to further understand the mechanism by which CD8+ TRM cells communicated local pathogen-associated antigen reencounter to recirculating memory CD8 T cells located outside the tissue. OT-I or P14 chimeric mice, which contained ovalbumin (OVA)–derived SIINFEKL (E, Glu; F, Phe; I, Ile; K, Lys; L, Leu; N, Asn; S, Ser) or gp33 peptide–specific CD8+ TRM cells, respectively, were established in the female reproductive tract (FRT) as follows: Naïve OT-I or P14 CD8 T cells were transferred intravenously to naïve C57Bl/6J mice, and mice were infected 1 day later with recombinant vaccinia virus expressing OVA (VV-OVA) or lymphocytic choriomeningitis virus (LCMV), respectively (fig. S1, A and B). FRT TRM cells were then reactivated locally by depositing cognate SIINFEKL or gp33 peptide into the cervical lumen (transcervical, t.c.), as described (17). We found that TRM reactivation with either peptide or recombinant vaccinia virus expressing cognate peptide induced expression of the cell adhesion molecule VCAM-1 (vascular cell adhesion molecule–1) on local vascular endothelium (Fig. 1, A to C). IFN-γ is produced by reactivated TRM cells in vivo (16). We found that exogenous t.c. deposition of IFN-γ into naïve mice was sufficient to induce local VCAM-1 expression (fig. S2). Further experiments revealed that when OT-I Ifng−/− TRM cells were reactivated in wild-type mice, VCAM-1 was no longer up-regulated, demonstrating that TRM cells induce local VCAM-1 in a cell-autonomous manner by secreting IFN-γ upon antigen resensitization (Fig. 1, A and B). Additionally, when circulating P14 memory CD8 T cells were depleted (while preserving TRM in the FRT) by injecting 1 μg HIS51 (a Thy1.1-depleting antibody) (16), peptide reactivation resulted in similar levels of VCAM-1 on endothelial cells, suggesting that VCAM-1 expression is driven by local CD8 T cells (Fig. 1D).

Fig. 1 TRM reactivation induces memory CD8 T cell and B cell recruitment through IFN-γ–dependent VCAM-1 up-regulation.

(A) VCAM-1 mean fluorescence intensity (MFI) on CD31+ vascular endothelium in the FRT 12 hours after OT-I Ifng−/− or OT-I Ifng+/+ TRM cells were reactivated by SIINFEKL peptide deposited transcervically. (B) Representative images of VCAM-1 (red) and CD31 (blue) staining. Twelve hours after t.c. challenge, VCAM-1 expression was quantified on FRT vascular endothelium after (C) P14 immune chimeras were transcervically challenged with either VV-gp33 or VV-OVA, left untreated, or (D) when P14 immune chimeras were injected with Thy1.1-depleting antibody or left untreated 5 days before t.c. deposition of gp33 peptide. (E) OT-I CD8 T cells were transferred intravenously into P14 immune chimeras that were treated with antibodies against VCAM-1, CD49d, or isotype control and were then challenged with gp33 transcervically.OT-I T cells per coronal section were enumerated 48 hours later. (F) B cells in the FRT were enumerated 12 and 48 hours after t.c. gp33 peptide challenge of P14 immune chimeras. (G) Representative images. DAPI, 4′,6-diamidino-2-phenylindole. B cells in the FRT were also enumerated in (H) P14 immune chimeras that were transcervically challenged with either VV-gp33 or VV-OVA or left untreated in (I) LCMV immune mice that never received P14 cells, and (J) in P14 immune chimeras that were injected with Thy1.1-depleting antibody 5 days before t.c. gp33 peptide challenge. (K) B cells within the FRT of OT-I Ifng−/− or OT-I Ifng+/+ immune chimeras were enumerated 48 hours after t.c. SIINFEKL challenge. (L) B cells in the FRT were quantified 48 hours after P14 immune chimeras were challenged transcervically with gp33 peptide in the presence of VCAM-1 or CD49d blocking antibodies. Each experiment shown includes three to six mice, and data are representative of two or three independent experiments. Scale bars indicates 20 μm. NS, not significant; *P < 0.05, **P < 0.01, ***P < 0.001; unpaired two-tailed t test; error bars indicate mean ± SEM.

VCAM-1 is the ligand for α4β1 integrin, which plays a role in lymphocyte migration (18). Bystander OT-I CD8 T cell recruitment in response to gp33-specific TRM reactivation was inhibited when either VCAM-1 or α4β1 (CD49d) was blocked with neutralizing antibodies (Fig. 1E). Because CD8+ TRM cells communicate antigen sensitization to recirculating CD8 T cells through an IFN-γ–VCAM-1 axis, we asked whether other lymphocyte lineages might also be recruited. B cells accumulated within the FRT within 12 hours of local CD8+ TRM reactivation and increased >100-fold by 48 hours (Fig. 1, F and G). Similar results were observed when recombinant gp33-expressing vaccinia virus was used to reactivate TRM instead of peptide (Fig. 1H) and in LCMV immune mice that did not receive a transfer of P14 CD8 T cells (Fig. 1I and fig. S1C). Moreover, depletion of recirculating P14 CD8 T cells with a Thy1.1-depleting antibody (which preserves P14 TRM within the FRT) did not impair B cell recruitment upon t.c. gp33 peptide challenge, demonstrating that B cell recruitment was driven by locally reactivated CD8 T cells (Fig. 1J). However, B cell recruitment was critically dependent on local CD8 T cell–derived IFN-γ and VCAM-1 induction (Fig. 1, K and L).

These observations indicate that reactivated TRM cells may broadcast detection of pathogen-associated peptides to other cell types through cytokines. Although most microbes are cleared by the innate immune system without priming adaptive responses, professional pathogens require adaptive immunity for clearance. Dendritic cells (DCs) are principal messengers between the innate and adaptive immune systems. Detection of broadly conserved microbial associated molecular patterns instructs DC maturation, which in turn initiates T cell activation (1, 19, 20). Because TRM recognition of peptides indicates reexposure to a pathogen that previously caused sufficient infection to elicit an adaptive immune response, we asked whether CD8+ TRM reactivation could trigger DC maturation. When either transgenic or endogenous cognate-antigen-specific CD8+ TRM cells were established, local exposure to either pathogen-associated peptide or recombinant VV expressing cognate peptide resulted in DC maturation within only 12 hours, as indicated by the induced expression of co-stimulatory molecules CD80, CD86, and CD40, as well as the lymph node homing chemokine receptor CCR7 (Fig. 2, A to C, and fig. S3). Depletion of circulating P14 CD8 T cells did not alter DC maturation after peptide resensitization, suggesting that CD8 TRM activation was sufficient for DC maturation (Fig. 2D). These data indicate that CD8+ TRM ligands also serve as potent inducers of innate immune responses. Intracellular cytokine staining indicated that CD8+ TRM cells expressed the cytokine tumor necrosis factor–α (TNF-α) in vivo within 12 hours of local reactivation (Fig. 2E) and that TNF-α was essential for DC maturation (Fig. 2, F and G).

Fig. 2 TRM reactivation induces DC maturation.

CD86 or CD80 and CCR7 expression was evaluated on CD11c+/MHC-II+ DCs in the FRT 12h after t.c challenge of (A) P14 immune chimeras challenged with gp33 peptide, (B) LCMV immune mice (that never received P14 cells) challenged with gp33 peptide, (C) P14 immune chimeras that were transcervically challenged with either VV-gp33 or VV-OVA, or (D) P14 immune chimeras that were injected intraperitoneally with 1 μg of Thy1.1-depleting antibody 5 days before gp33 challenge. (E) Intracellular TNF-α expression was evaluated in P14 CD8 T cells from the FRT by flow cytometry 12 hours after t.c. gp33 challenge. (F and G) DC phenotype was evaluated as in (A), but the indicated mice were pretreated with TNF-α–blocking antibody. Representative of two or three experiments totaling 6 to 14 mice per group. *P < 0.05, **P < 0.01, ***P < 0.001, unpaired two-tailed t test; error bars indicate mean ± SEM.

Natural killer (NK) cells are innate lymphoid cells that play essential roles in control of viral infections (21, 22). They have advantages over the adaptive immune system because they are constitutively abundant and broadly distributed (21). When activated, they kill host cells showing signs of stress but lack highly specific antigen receptors for detecting pathogen-associated peptides (22). Thus, viral infections also require the presence of CD8 T cells for clearance. TRM reactivation with either peptide or recombinant VV expressing cognate peptide resulted in granzyme B up-regulation among NK cells within the FRT within 12 hours, and this event also occurred when circulating memory P14 CD8 T cells were depleted (Fig. 3, A to C). Contemporaneously, local memory CD8 T cells of irrelevant specificities also increased expression of granzyme B (Fig. 3A). These data suggested that, in the event of local TRM reactivation, NK cells and memory CD8 T cells receive signals to become poised for cytolysis of infected host cells (2). TRM cells expressed the cytokine interleukin (IL)–2 in vivo within 12 hours of local reactivation (Fig. 3D), and blocking IL-2 receptor β (IL-2Rβ) abrogated granzyme B up-regulation on both NK cells and bystander memory CD8 T cells (Fig. 3, E and F).

Fig. 3 TRM reactivation induces NK cell activation.

(A) P14 immune chimeras were challenged transcervically with gp33 to reactivate TRM cells. Twelve hours later, intracellular granzyme B expression was evaluated in bystander CD8 T cells (P14 CD8 T cells were excluded from analysis), and NK cells isolated from the FRT (gray line, without gp33 challenge; black, gp33 challenge). Intracellular granzyme B expression within NK cells isolated from the FRT was evaluated 12 hours after (B) P14 immune chimeras were transcervically challenged with either VV-gp33 or VV-OVA or left untreated or (C) P14 immune chimeras were previously treated with Thy1.1-depleting antibody before t.c. gp33 peptide challenge. (D) Intracellular IL-2 expression by P14 CD8 T cells isolated from the FRT of P14 immune chimeras was evaluated 12 hours after t.c. gp33 peptide challenge. Intracellular granzyme B expression by (E) NK cells and (F) bystander CD8 T cells isolated from the FRT of P14 chimeras 12 hours after gp33 peptide challenge when mice were pretreated with IL-2Rβ–blocking antibody. n = 3, representative of three experiments. *P < 0.05, **P < 0.01, ***P < 0.001,unpaired two-tailed t test, mean ± SEM. GMFI, geometric mean fluorescence intensity.

These data established that TRM reactivation induced broad activation of innate and adaptive immune components. This observation provided impetus for the hypothesis that TRM may protect the host in part by (i) acting as sensory cells for previously encountered peptides that are associated with intracellular pathogenic infections, (ii) communicating resensitization to abundant nonspecific immune effectors, and thus (iii) triggering an organ-wide antiviral state. To test this hypothesis, we established P14 chimeric mice with gp33 peptide–specific TRM cells within the FRT (Fig. 4). Mice were subsequently challenged with VV-OVA for which memory T cells had not been established. As expected, infection resulted in abundant tissue viral load within 2 days of transcervical exposure in both naïve mice and in mice containing gp33-specific TRM cells. However, if gp33 peptide was included in the inoculum in order to reactivate local TRM cells at the time of viral challenge, we failed to detect infection in 9 of 11 mice, coinciding with a ~104-fold reduction in viral load (Fig. 4A). Similar results were observed when gp33 peptide was delivered 12 hours before viral challenge (Fig. 4B) and in mice that harbored only endogenous gp33-specific TRM cells (i.e., LCMV immune mice that did not receive a transfer of P14 CD8 T cells; Fig. 4C and fig. S1C). These data suggest that TRM reactivation, which was associated with widespread activation of the local immune surveillance network, induced an antiviral state at the site of infection. This near sterilizing protection was completely abrogated when IFN-γ, TNF-α, and IL-2Rβ−dependent cytokines were blocked (Fig. 4A).

Fig. 4 TRM reactivation induces antiviral state.

(A) P14 immune chimeras or control naïve mice were challenged transcervically with 4 × 106 plaque-forming units (PFU) of antigenically unrelated VV-OVA in the presence or absence of gp33-reactivating peptide and/or IFN-γ, TNF-α, and IL-2Rβ–blocking antibodies. Two days later, viral titers were evaluated by plaque assay from homogenized FRT. Data were pooled from two independent experiments totaling 6 to 11 mice per group. (B) As in (A), however, gp33 peptide was delivered 12 hours before viral challenge. Data pooled from two independent experiments totaling eight or nine mice per group. (C) LCMV immune mice (without P14 transfer) were transcervically challenged with 1 × 106 PFU of VV-OVA in the presence or absence of gp33 peptide. Two days after infection, viral titers were evaluated. Data were pooled from two independent experiments totaling six mice per group. *P < 0.05, **P < 0.01,***P < 0.001, unpaired two-tailed t test, mean ± SEM. L.O.D., limit of detection.

CD8 T cells are classically thought to control viral infections by contact-dependent interactions between antigen-specific CD8 T cells and each infected host cell, followed by directed target cell killing (2). These data support an expanded range of functions by which TRM cells mediate protective immunity. Here, we show that TRM cells reverse the flow of information from innate to adaptive immune systems by demonstrating that adaptive TRM sensitization initiates broad local immune activation. In this light, our data suggest that TRM cells could be viewed as a pathogen recognition entity that engages innate immune functions more sensitively than what is accomplished by innate recognition of conserved microbial associated molecular patterns. The TRM functions described here were dependent on IFN-γ, TNF-α, and IL-2Rβ−dependent cytokines, which may help explain why memory CD8 T cell populations that are competent to produce each of these cytokines (referred to as polyfunctional memory CD8 T cells) are often best associated with protective immunity (23).

It may be possible to leverage the functions of TRM described here for therapeutic purpose. For instance, local reactivation of established TRM cells with peptide vaccines could be used to increase local immunity against unrelated pathogens. The potent inflammatory effects of TRM reactivation may also have pathological consequences and may contribute to the observed association between viral infections and exacerbation of tissue-specific autoimmune or inflammatory diseases. Principally, our results indicate how sensitization of relatively small numbers of TRM cells may result in an amplified signal to more-abundant members of the innate immune system in order to trigger an organ-wide antiviral state.

Supplementary Materials

www.sciencemag.org/content/346/6205/98/suppl/DC1

Materials and Methods

Figs. S1 to S3

Reference (24)

References and Notes

  1. See supplementary information on Science Online.
  2. Acknowledgments: The data presented in this manuscript are tabulated in the main paper and in the supplementary materials. This work was funded by NIH grant R01AI084913 (D.M.), NIH Director’s New Innovator Award Program DP2-OD-006467 (D.M.), and NIH grants T32AI007313 and F30DK100159 (J.M.S.).
View Abstract

Navigate This Article