The Molecular Signature of Tissue Resident Memory CD8 T Cells Isolated from the Brain Linda M. Wakim, Amanda Woodward-Davis, Ruijie Liu, Yifang Hu, Jose Villadangos, Gordon Smyth and Michael J. This information is current as Bevan of September 25, 2021. J Immunol published online 24 August 2012 http://www.jimmunol.org/content/early/2012/08/24/jimmun ol.1201305 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2012/08/24/jimmunol.120130 Material 5.DC1 http://www.jimmunol.org/ Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

by guest on September 25, 2021 *average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2012 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published August 24, 2012, doi:10.4049/jimmunol.1201305 The Journal of Immunology

The Molecular Signature of Tissue Resident Memory CD8 T Cells Isolated from the Brain

Linda M. Wakim,*,†,1 Amanda Woodward-Davis,* Ruijie Liu,‡ Yifang Hu,‡ Jose Villadangos,†,1 Gordon Smyth,‡,x and Michael J. Bevan*

Tissue resident memory (Trm) CD8 T cells represent a newly described memory T cell population. We have previously characterized a population of Trm cells that persists within the brain after acute virus infection. Although capable of providing marked protection against a subsequent local challenge, brain Trm cells do not undergo recall expansion after dissociation from the tissue. Further- more, these Trm cells do not depend on the same survival factors as the circulating memory T cell pool as assessed either in vivo or in vitro. To gain greater insight into this population of cells, we compared the expression profiles of Trm cells isolated from the brain with those of circulating memory T cells isolated from the spleen after an acute virus infection. Trm cells displayed altered

expression of involved in chemotaxis, expressed a distinct set of transcription factors, and overexpressed several inhibitory Downloaded from receptors. Cumulatively, these data indicate that Trm cells are a distinct memory T cell population disconnected from the circulating memory T cell pool and display a unique molecular signature that likely results in optimal survival and function within their local environment. The Journal of Immunology, 2012, 189: 000–000.

lassically, memory CD8 T cells are broadly divided into ficking through the tissue as part of their immunological surveillance. http://www.jimmunol.org/ two subsets termed central memory (Tcm) and effector Recent studies have shown that a proportion of these memory T cells C memory (Tem). Tcm CD8 T cells express the adhesion reside in the tissue and represent a distinct memory T cell pop- molecule L-selectin (CD62L) and the CCR7, ulation (4). These peripherally deposited memory T cells, termed traffic through lymph nodes, spleen, and blood, and represent a tissue resident memory (Trm) cells, are a self-sustaining population long-lived population of memory cells with high proliferative capacity that persist long term within nonlymphoid tissue, commonly at upon rechallenge. Tem CD8 T cells lack CD62L and CCR7, are sites of prior infection. Trm cells have been identified in a variety present in spleen and blood, recirculate through the peripheral of peripheral tissues including the skin and sensory ganglia of tissues, and maintain immediate cytotoxic potential (1, 2). Nonlym-

mice latently infected with HSV (4–6) and in the gut (7), brain (8), by guest on September 25, 2021 phoid tissues house a large proportion of the memory T cell pool (3). lung (9, 10), and salivary glands (11). These cells rapidly acquire ef- Previously, it was thought that these were simply Tem cells traf- fector function upon secondary pathogenic encounter (6), and those in the skin are highly protective against subsequent local infection (4, 12). *Department of Immunology, Howard Hughes Medical Institute, University of Wash- We have recently characterized the memory CD8+ T cell pop- ington, Seattle, WA 98195; †Division of Inflammation, Walter and Eliza Hall Institute of Medical Research, Melbourne, Victoria 3052, Australia; ‡Division of Bioinfor- ulation that persists within the brain after an acute systemic ve- matics, Walter and Eliza Hall Institute of Medical Research, Melbourne, Victoria sicular stomatitis virus (VSV) infection (8). We showed that memory 3052, Australia; and xDepartment of Mathematics and Statistics, University of Mel- bourne, Melbourne, Victoria 3010, Australia T cells persisting within the brain survive without replenishment from the circulation. These cells selectively expressed the integrin 1Current address: Department of Microbiology and Immunology and Department of Biochemistry and Molecular Biology, University of Melbourne, Melbourne, Victoria CD103, the expression of which was dependent on Ag recognition 3010, Australia. within the tissue. Ag persistence was not required for memory Received for publication May 8, 2012. Accepted for publication July 29, 2012. T cell retention within the brain. The memory CD8 T cells iso- This work was supported by the Howard Hughes Medical Institute and by National lated from the brain died rapidly upon isolation from the tissue Institutes of Health Grant U19 AI083019 (to M.J.B.). L.M.W. is supported by an Overseas Biomedical Fellowship from the National Health and Medical Research and failed to undergo recall expansion after adoptive transfer into Council of Australia. the bloodstream of Ag-challenged recipients. Cumulatively, these + The data sets presented in this article have been deposited in the National Center for data showed that memory CD103 CD8 T cells that persist within Biotechnology Information Gene Expression Omnibus database (http://www.ncbi. the brain after an acute virus infection are bona fide Trm cells. nlm.nih.gov/geo) under accession number GSE39152. Recent work by several groups in both mice and human studies Address correspondence and reprint requests to Dr. Michael J. Bevan, Department of provides compelling evidence to support the presence of a locally Immunology, Howard Hughes Medical Institute, University of Washington, 1959 NE Pacific Street, Box 357370, Seattle, WA 98195. E-mail address: [email protected]. confined memory T cell population that is deposited at former sites edu of Ag encounter (13). However, to date there was no molecular The online version of this article contains supplemental material. genetic profile defining these peripherally deposited memory Abbreviations used in this article: 7-AAD, 7-aminoactinomycin D; eomes, eomeso- T cells. Using a previously characterized model of intranasal in- dermin; FDR, false discovery rate; i.c., intracranially; i.n., intranasal(ly); IFITM3, IFN-induced transmembrane 3; LN, lymph node; MDS, multidimensional fection with VSV, we genetically profiled Trm CD8 T cells that scaling; p.i., postinfection; S1P1R, sphingosine-1-phosphate receptor 1; Tcf-1, T cell develop within the brain and compared them to memory T cells that factor 1; Tcm, central memory; Tem, effector memory; Trm, tissue resident memory; are part of the circulating memory T cell pool. Our data clearly VSV, vesicular stomatitis virus. indicate that Trm cells are a distinct memory T cell population with Copyright Ó 2012 by The American Association of Immunologists, Inc. 0022-1767/12/$16.00 a unique molecular signature.

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1201305 2 MOLECULAR SIGNATURE OF BRAIN Trm CELLS

Materials and Methods Viral titer determination Mice The presence of infectious VSV in tissue samples was determined using C57BL/6 and B6.SJL-PtprcaPep3b/BoyJ (CD45.1) mice were obtained standard PFU assays on confluent Vero cell monolayers as described pre- from The Jackson Laboratory (Bar Harbor, ME) and housed in specific viously (8). pathogen-free conditions in the animal facilities at the University of Flow cytometry Washington (Seattle, WA). OT-I TCR transgenic mice congenic for Ly5.1 or deficient in bim and MHC class II-deficient mice were bred Single-cell suspensions were prepared from spleens and lymph nodes (LN) and maintained in the same facilities. All experiments were done in by mechanical disruption. Brains were enzymatically digested for 1 h at 37˚C accordance with the Institutional Animal Care and Use Committee in 3 ml collagenase type 3 (Worthington) (3 mg/ml in RPMI 1640 supple- guidelines of the University of Washington. mented with 2% FCS), and lymphocytes were separated on a Percoll gra- dient. Cells were stained for 25 min on ice with the appropriate mixture of T cell adoptive transfer and infections mAbs and washed with PBS with 1% BSA. The following conjugated mAbs a 4 + were obtained from BD Biosciences or eBioscience: anti-CD8 , CD45.1, Mice received 10 naive OT-I.CD45.1 CD8 T cells via i.v. injection prior CD103. Cells were analyzed on a FACSCanto II using FlowJo software to infection. Cells were sorted from the brain and spleens of mice at the (Tree Star). indicated times postinfection using a FACSAria II, and 3 3 103 cells were transferred into mice either intracranially (i.c.) or i.v. prior to infection. Survival assay Mice were infected intranasally (i.n.) with 5 3 104 PFU or i.p. with 2 3 107 PFU of a recombinant VSV that expresses GFP and a secreted form of Cells were cultured for 18 h in RPMI 1640 supplemented with 2% FCS, 2 3 25 OVA (14). Mice were infected i.c. with 10 CFU of either wild-type or a mM glutamine, 5 10 M 2-mercaptoethanol, and antibiotics and with recombinant Listeria monocytogenes that expresses a secreted form of OVA 10 ng/ml of either recombinant IL-7 or IL-15. Annexin V staining was performed using an annexin V staining (BD Biosciences) following the

(LM-OVA). Growth and quantification of LM-OVA was performed as Downloaded from described previously (15). manufacturer’s instructions. For in vitro stimulation experiments, 20,000 memory OT-I T cells were RNA samples cultured at a 1:1 ratio with SIINFEKL pulsed (1 mM) or unpulsed irra- diated splenocytes. The absolute number of live cells was determined Mice received 104 naive OT-I.CD45.1 CD8+ T cells via i.v. injection prior 60hlater. to i.n. infection with VSV-OVA. CD103+ and CD1032 OT-I cells were http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 1. Resident memory T cells do not undergo recall expansion after dissociation from the tissue in which they reside. Mice were seeded with OT- I.CD45.1 T cells before i.n. infection with VSV-OVA. On day 20 p.i., OT-I cells were recovered from the brain and spleen of mice, sorted into CD103+ and CD1032, and adoptively transferred into naive recipient mice that were challenged with VSV-OVA (i.n.). (A) Representative flow cytometry profile of brain on day 20 p.i. demonstrating the OT-I.CD103+ and OT-I.CD1032 subsets. (B) Schematic diagram representing the experimental setup. (C) The proportion of OT-I of the total CD8+ T cell population in the brain and spleen on day 8 p.i. after i.v. transfer of memory T cells. Data represent the mean + SEM (n =3 to 5), and data are pooled from two independent experiments. (D) Mice were seeded with either bim2/2 OT-I.CD45.1 or wild-type (wt) OT-I T cells before i.n. infection with VSV-OVA. On day 20 p.i., OT-I cells were recovered from the brain and spleen of mice, sorted into CD103+ and CD1032, and adoptively transferred (i.v.) into naive recipient mice that were challenged with VSV-OVA (i.n.). Depicted is the proportion of OT-I of the total CD8+ T cell population in the spleen on day 8 p.i. Data represent the mean + SEM (n = 4–8), and data are pooled from two independent experiments. (E) As described for (C) except memory T cells were transferred intracranially prior to i.n. challenge with VSV-OVA. The Journal of Immunology 3 sorted from the brain and spleens of mice on day 20 postinfection (p.i.) Results using a FACSAria II. Cells were subjected to two successive rounds of Brain Trm cells do not undergo recall expansion after sorting. All samples were maintained at 4˚C for the duration of the sort, and purity was 95–99% for all populations. The treatments and cell sorts dissociation from the tissue in which they reside were repeated for three independent pools of mice. One or two samples Intranasal infection with VSV results in a transient systemic in- were extracted of each cell population from each mouse pool, making 13 fection that is cleared from all organs including lung, spleen, and samples in total. RNA was isolated and hybridized to Affymetrix Mouse Gene 1.0ST GeneChips by the Immunological Genome Project (www. brain by day 8 postinfection. Using a recombinant virus that ex- immgen.org), following standard operating procedures. presses the model Ag OVA (VSV-OVA) in combination with OT-I TCR transgenic cells, we have previously demonstrated that long Microarray analysis after the resolution of the acute VSV-OVA infection, memory OT-I The microarray intensities were background corrected, normalized, and CD8 T cells persist within the brain (8). The memory CD8 T cells summarized according to the robust multichip average algorithm (16) using persisting within the brain can be subdivided into two populations the aroma.affymetrix software package (http://www.aroma-project.org). based on CD103 expression although by day 20 p.i. the vast majority The chip definition file MoGene-1.0-st-v1,r3.cdf was used to define a core of memory OT-I cells persisting within the brain express this set of probe sets with reliable gene annotation. integrin (Fig. 1A). Further characterization of these brain-residing Differential expression analysis was undertaken using the limma + software package for R (17). Genes were filtered as nonexpressed if they CD103 memory T cells revealed that they are Trm cells (8). failed to exceed the bottom 25% of expression values on at least three We have previously shown that brain CD103+ Trm cells undergo arrays. Differential expression was assessed using empirical Bayes poor recall expansion after dissociation from the brain tissue. moderated t-statistics and F-statistics (18). The analysis was adjusted for Specifically, when we sorted CD103+ and CD1032 OT-I CD8+ correlation between samples from the same mouse pool using the du- Downloaded from plicate correlation function. The p values were adjusted to control the T cells from the brains and spleens of mice on day 20 after i.n. false discovery rate (FDR) using the method of Benjamini and Hochberg infection with VSV-OVA and adoptively transferred these T cells (19). i.v. into naive recipient mice and subsequently challenged these http://www.jimmunol.org/ by guest on September 25, 2021 FIGURE 2. Brain memory T cells do not depend on the same survival factors as circulating memory T cells in vitro or in vivo. (A) Mice were seeded with OT- I.CD45.1 T cells before i.n. infection with VSV-OVA. On day 20 p.i., memory OT-I.CD103+ and CD1032 cells were sorted from the brain and spleen of mice and cultured in vitro overnight in the presence of IL-7 or IL-15. Cells were stained with annexin V and 7-AAD, and the percentage of dead cells (annexin V+/2 and 7-AAD+) was determined by flow cytometry. (B) Mice (either B6 or MHC class II KO) were seeded with OT- I.CD45.1 T cells before i.n. infection with VSV-OVA. Graphs depict the per- centage or number of OT-I cells of the total CD8 T cell population in the spleen, lymph node (LN), and brain on day 60 p.i. Bars represent the mean + SEM (n = 9–17). Data are pooled from four independent experiments. **p , 0.005 (Student t test). 4 MOLECULAR SIGNATURE OF BRAIN Trm CELLS animals with VSV-OVA, we observed that all memory populations mice on day 60 p.i. revealed that although memory OT-I within expanded in response to secondary challenge except the brain the LN and spleen from MHC class II KO mice underwent sig- resident CD103+ memory T cells (8) (Fig. 1B, 1C). Delaying nificant decay (Fig. 2B, 2C), consistent with the original studies isolation of the spleen and lung memory T cells until a later (21) we observed no difference in the number of memory T cells memory time point (day 50 p.i.) to allow further maturation of the within the brain when comparingB6toMHCclassII-deficient memory cells failed to restore the proliferation potential of brain mice (Fig. 2D). Thus, brain memory T cells and circulating CD103+ Trm cells (Supplemental Fig. 1A). The lack of recall memory T cells do not depend on the same survival factors for expansion by the CD103+ memory T cell population after isola- their long-term maintenance. Furthermore, these data further tion and adoptive transfer may reflect a defect in survival after support the notion that the memory T cell population within the tissue dissociation. It is noteworthy that we could not rescue the brain exists independently of the circulating memory T cell pool inability of brain CD103+ memory T cells to expand when we because if this population was continually supplemented by cir- repeated the experiments described earlier using OT-I cells defi- culating memory T cells, then we would also expect a decay in cient in the proapoptotic Bcl-2 family member Bim (OT-I.bim2/2) brainmemoryTcellnumbersmirroring the events within the (Fig. 1D), which is known to be critical for controlling normal circulation. homeostasis of memory T cells (20). This suggests that if the in- Brain Trm cells provide protection during a localized infection ability of brain CD103+ memory T cells to mount a secondary recall + response is due to a defect in survival after dissociation from the Brain CD103 Trm cells exhibit survival defects ex vivo although tissue, then this is a bim-independent death. when left in situ, they do provide an overt advantage during a

We next sought to determine if we could enhance the recall Downloaded from response of brain CD103+ memory T cells by reintroducing them back into the microenvironment of the brain. To do this, we sorted CD103+ and CD1032 OT-I CD8+ T cells from the brains and spleens of mice on day 20 p.i. with VSV-OVA and adoptively transferred low numbers of these memory T cells directly into the

brain (i.c.) of naive recipient mice. We subsequently challenged http://www.jimmunol.org/ these animals with VSV-OVA. Even when reseeded back into the brain tissue, brain CD103+ memory T cells failed to mount a recall response (Fig. 1E). This is in contrast to both CD1032 brain and splenic memory T cells, which underwent expansion. Brain CD103+ Trm cells also displayed impaired recall expansion when stimulated ex vivo. We sorted CD103+ and CD1032 OT-I CD8+ T cells from the brains and spleens of mice on day 20 after i.n. infection with VSV-OVA. These cells were cultured in vitro with SIINFEKL peptide-loaded splenocytes. We observed that all by guest on September 25, 2021 memory populations expanded in response to ex vivo stimulation except the brain resident CD103+ memory T cells (Supplemental Fig. 1B). Cumulatively, these data demonstrate that brain CD103+ memory T cells are functionally distinct from brain CD1032 and splenic CD1032 memory T cell populations. Brain memory T cells do not depend on the same survival factors as circulating memory T cells both in vitro and in vivo We previously reported that memory CD8 T cells isolated from the brain die rapidly in vitro (8). We wished to determine whether the addition of cytokines known to be important in memory T cell survival could improve ex vivo survival of these tissue-residing memory T cells. CD103+ and CD1032 OT-I cells were sorted from the brain and spleen of mice on day 20 postinfection with VSV-OVA. The cells were cultured in vitro in the presence of IL-7 or IL-15, and survival was assessed 18 h later by annexin V and 7-aminoactinomycin D (7-AAD) staining. Although IL-7 and IL-15 greatly enhanced the survival of splenic memory T cells, neither cytokine rescued either brain memory T cell population (Fig. 2A). Memory T cells within the brain also do not appear to rely on FIGURE 3. Resident memory T cells within the brain can provide the same survival factors as splenic memory T cells in vivo. We protection against local infection. Mice were seeded with OT-I.CD45.1 demonstrated this by monitoring the persistence of memory T cells before i.n. or i.p. infection with VSV-OVA. On day 20 p.i., mice CD8 T cells within the brain of MHC class II-deficient mice. It were challenged intracranially with LM-OVA. The proportion of OT-I T cells of the total CD8+ T cell population in the (A) spleen and (B) lymph has been previously reported that memory CD8 T cells undergo a node and (C) the number of OT-I CD8 T cells in the brain on day 20 p.i. gradual decay in MHC class II-deficient mice (21). To assess Bars represent the mean + SEM (n = 6). Data are representative of two whether brain resident memory T cells were vulnerable to a independent experiments. (D) Survival (measured as a loss of .20% of similar decay, we adoptively transferred naive OT-I cells into ei- starting weight) of animals primed either i.p. or i.n. with VSV-OVA and ther B6 or MHC class II-deficient mice prior to i.n. infection challenged i.c. with LM-OVA. Shown is the mean + SEM (n = 15). Data with VSV-OVA. Analysis of the brain, LN, and spleen of these are pooled from three independent experiments. The Journal of Immunology 5 localized secondary infection. To demonstrate this, we took ad- Mice primed 20 d earlier with VSV-OVAvia either the i.p. or i.n. vantage of the fact that only local virus infection of the brain route were challenged intracranially with a recombinant Listeria results in the generation of CD103+ Trm cells. Mice seeded with that expresses OVA (LM-OVA) and were monitored for survival low numbers of OT-I.CD45.1 T cells were infected either i.n. or (Fig. 3D). We observed 100% survival among animals primed i.p. with VSV-OVA. Irrespective of the route of infection, we with VSV-OVA i.n. and therefore containing brain CD103+ observed a similar-sized memory OT-I response in the LN and memory T cells. In contrast, by day 3 after challenge with LM- spleen (Fig. 3A, 3B). Unlike i.n. administration of VSV-OVA, i.p. OVA, 40% of mice primed via the i.p. route, and therefore lacking infection does not result in virus infection of the brain (Supple- brain CD103+ memory OT-I T cells, had succumbed to infection. mental Fig. 2). Analysis of the OT-I cell population within This protection is not merely due to presence of innate immune the brains of mice infected with VSV-OVA either i.p. or i.n. mechanisms present in the brain of mice primed i.n. with VSV- 20 d earlier shows that in both cases there are approximately OVA, as when we intracranially challenge these mice with a wild- equivalent numbers of CD1032 OT-I cells. However, only after i.n. type Listeria (not expressing OVA) we failed to observe any immunization, which results in a virus infection of the brain, did protection (Supplemental Fig. 3). Thus, brain CD103+ memory we observe the development and persistence of CD103+ memory T cells are functional in situ and provide enhanced protection during OT-I cells (Fig. 3C). In summary, this model provides us with a a localized infection. system where we have equivalent circulating memory T cell pop- + ulations in the LN and spleen and similar numbers of brain CD1032 Gene expression profile analysis of brain CD8 Trm cells memory T cells. The only difference between the two cohorts of To characterize brain CD103+ Trm cells further, we performed mice is the presence of CD103+ memory OT-I cells in the animals gene array analysis. To define the molecular signature of CD103+ Downloaded from that had been primed intranasally. Trm cells, we compared the transcriptomes of CD103+ and http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 4. Brain resident memory T cells have a different molecular signature compared with circulating memory T cells. (A) An MDS plot depicting the relationship between brain CD103+, brain CD1032, and spleen CD1032 OT-I T cell populations. Distances between samples represent leading fold change, the average log2-fold change between the 500 genes with largest differences. (B) Number of genes differentially expressed between brain resident + 2 and splenic memory T cells (FDR , 0.05). (C) Heat map of expression profiles (normalized log2 intensities) of brain OT-I.CD103 , brain OT-I CD103 , and spleen OT-I.CD1032 memory T cells. Heat map shows the 50 genes with the most significant difference between the cell populations. 6 MOLECULAR SIGNATURE OF BRAIN Trm CELLS

2 2 CD103 OT-I T cells sorted from the brain to CD103 memory Table I. Fold changes (log2) and p values for the top 10 differentially OT-T cells isolated from the spleen of mice 20 d after i.n. infection expressed genes for each subtype comparison with VSV-OVA. Note that there were too few CD103+ memory cells in spleen to include in the analysis. An example of pre-sort and post- ID Gene logFC p Value sort analysis of purified cell subsets is shown in Supplemental Fig. 4. Brain CD103+ versus Brain CD1032 Multidimensional scaling (MDS) was used to display the leading 10378286 Itgae 3.01 1.26E–18 fold-changes between the expression profiles, demonstrating that 10522051 Klf3 22.23 7.71E–16 10385776 Tcf7 22.47 7.21E–15 the three cell populations have distinct profiles and that brain 2 + 2 10439583 Sidt1 1.32 8.58E–14 CD103 OT-I cells are most similar to brain CD103 cells (Fig. 4A). 10351691 Slamf6 22.11 1.10E–13 Furthermore, when comparing the brain populations to the splenic 10407940 Naip3 2.52 9.30E–13 memory T cell pool, it is the brain CD1032 OT-I cells that share 10501586 S1pr1 22.42 9.74E–13 more genetic similarity to the splenic memory T cell population. 10388234 Gsg2 2.03 2.26E–12 10403821 Tcrg-V3 2.01 4.69E–12 Although the two memory OT-I populations isolated from the 10589994 Eomes 21.81 5.72E–12 brain are closely related, their genetic profiles indicate that they do Spleen CD1032 versus Brain CD103+ represent distinct cell populations. We find that there are a total of 10378286 Itgae 25.08 5.72E–24 490 genes that are significantly differentially expressed (FDR , 10439583 Sidt1 3.17 8.14E–22 2 0.05) when comparing the brain CD103+ to brain CD103 10522051 Klf3 3.81 3.01E–21 memory OT-I cells (Fig. 4B). There were more differences in gene 10569017 Ifitm3 23.53 1.04E–19 2 expression comparing either CD103 or CD103+ brain OT-I to the 10351691 Slamf6 3.62 4.55E–19 Downloaded from 10554240 Isg20 23.01 6.38E–19 splenic memory T cell population with 3941 and 5411 genes being 10350733 Rgs16 23.02 9.13E–19 upregulated or downregulated, respectively (Fig. 4B). 10437687 Litaf 22.47 1.77E–18 Fig. 4C shows a heat-map analysis of the 50 most differentially 10360028 Fcgr2b 3.15 1.87E–18 expressed genes when comparing brain CD103+, brain CD1032, 10385776 Tcf7 3.31 2.69E–18 2 and spleen CD103 memory OT-I cells. These data further high- Spleen CD1032 versus Brain CD1032 + 2 light that the gene expression profiles of brain CD103 and CD103 10439583 Sidt1 1.85 2.26E–16 http://www.jimmunol.org/ 2 memory T cells are closely related. However, these two brain- 10346799 Icos 2.09 2.84E–16 10554240 Isg20 22.41 5.40E–16 derived subsets were markedly different from splenic memory 10350733 Rgs16 22.45 5.59E–16 T cells in terms of their gene expression profile. A list of the top 10 10437687 Litaf 22.05 7.70E–16 most differentially expressed genes when comparing the three sub- 10378286 Itgae 22.07 1.06E–15 sets is presented in Table I, and Table II presents a list of genes that 10533729 Vps37b 22.72 3.56E–15 10358389 Rgs2 22.83 3.70E–15 are significantly differentially expressed between the three cell sub- 10454782 Egr1 22.22 3.95E–15 sets and grouped into immune system-related categories of interest. 10530145 Tlr1 2.15 5.57E–15 Brain CD103+ T cells express genes characteristic of activated FC, Fold change. effector cells, including granzyme B and IL-2Ra at higher levels by guest on September 25, 2021 than spleen memory cells. Notably, two inhibitory receptors in the CD28:CTLA-4 family (22) (CTLA-4 and PD-1) are upreg- 2 13). Thus, in addition to a role in circulating T cell egress, S1P1R ulated in brain CD103+ Trm cells compared with brain CD103 2 also may be involved in tissue egress as suggested by earlier reports and splenic CD103 memory T cells. Furthermore, transcripts (26). encoding key regulators of T cell differentiation such as eomeso- These array data indicate that brain CD103+ Trm cells are a dermin (eomes), T cell factor 1 (Tcf-1), lef1, and T-bet are down- distinct memory T cell population displaying a unique molecular regulated in CD103+ Trm cells compared with the other cell signature. subsets. Brain CD103+ Trm cells also express the highest levels among all the subsets of several IFN-stimulated genes including IFN-induced transmembrane protein 3 (IFITM3) in addition to Discussion Irf4 and Isg20. Memory CD103+ CD8 T cells that persist within the brain after an Both CD103+ and CD1032 memory T cell populations isolated acute virus infection are bona fide Trm cells. We base this clas- from the brain share similar expression of several chemokine and sification on several functional and phenotypical characteristics cytokine receptors. Fig. 5 shows a heat-map analysis of genes that including i) they are a self-sustaining T cell population that per- are similarly expressed between the memory T cells isolated from sists independently of the circulating memory T cell pool; ii) they the brain but which differ from the splenic memory T cells. In are resident within the tissue and do not recirculate; iii) they ex- comparison with the splenic memory T cell pool, the memory press the integrin CD103; and iv) they do not survive well after T cell isolated from the brain displayed upregulation of several dissociation from the tissue (8). However, to date there was no chemokine genes including CCL3, CXCL10, CCL4 and down- molecular genetic profile defining these peripherally deposited regulation of others including CX3Cr1 and CCL9. It is noteworthy memory T cells. We have now determined the molecular signature both CXCL10 (23) and CCL3 (24) have been previously linked to of a Trm cell population. the recruitment of T cells into the brain. There are also several Genetic profiling revealed that CD103+ Trm cells express low migration molecules that are differentially expressed between the levels of the transcription factors Tcf-1 and eomes in comparison two brain populations (Fig. 6). These include sphingosine-1- with the other memory T cell subsets analyzed. Both transcription phosphate receptor 1 (S1P1R), CCR7, and CXCR4. Notably, S1P1R factors have been demonstrated to be important in differentiation is involved in regulation of T cell egress from lymph nodes. Ex- and persistence of memory CD8+ T cells (27, 28). Zhou et al. (27) pression of CD69 has been shown to suppress the function of show that Tcf-1 deficiency impaired central memory T cell dif- S1P1R and thereby reduces T cell migration out of the lymph ferentiation, and Tcf-1–deficient memory T cells were progres- node in response to an S1P gradient (25). CD69 expression has sively lost over time. This memory T cell decay was most striking been documented on Trm cells isolated from different tissues (4, when analyzing memory T cells within the spleen, and, notably, The Journal of Immunology 7

Table II. Fold changes (log2) for differentially expressed genes for each subtype comparison

Gene Brain CD103+/Brain CD1032 Spleen CD1032/Brain CD103+ Spleen CD1032/Brain CD1032 Cytokine and Chemokine Receptor Activity IL6ra 20.5 1.6 1.1 IL21r 0.2 21.4 21.2 Cx3Cr1 20.7 2.7 2 IL2ra 1.3 22.2 20.9 IL12rb2 20.2 21.5 21.6 IL4ra 0.1 21.4 21.3 Ccr7 21.3 2.1 0.8 Cytokine and Chemokine Activity Ccl4 0.6 22.2 21.6 Xcl1 20.5 22.2 22.7 Cxcl10 0.1 23.2 23.2 Ccl3 0.4 22.4 22 Tnfsf8 21.1 2.8 1.7 Ccl9 20.7 2 1.3 Tnfsf10 0.7 21.9 21.1 Tnf 0.7 22.4 21.7

Regulation of Transcription, DNA Dependent Downloaded from Klf3 22.2 3.8 1.6 Adam8 0.9 21.7 20.8 Aff3 21.1 2.2 1 Aff3 0.9 23.1 22.2 Bach2 20.7 1.4 0.7 Cd38 0.7 22 21.3 2 2

Cdh1 0.7 2.9 2.2 http://www.jimmunol.org/ Dmrta1 21.2 1.4 0.3 Egr1 0.1 22.3 22.2 Eomes 21.8 2.1 0.3 Fosi2 0.01 22.2 22.2 Irf4 0.7 23.4 22.7 Jun 0.2 22 21.7 Lef1 21.1 1.46 0.4 Litaf 0.4 22.5 22 Nr4a3 20.3 23.1 23.5 Tcf7 22.5 3.3 0.8 by guest on September 25, 2021 Programmed Cell Death Bag3 0.2 22.3 22.1 Cdh1 0.7 22.9 22.2 Dapl1 21.1 3 1.8 Gadd45b 0.3 22 21.7 Gramd4 20.8 1.9 1.1 Traf4 0.2 21.2 21.1 Tnfsf10 0.7 21.9 21.1 Rhob 0.3 23.2 22.9 Pdcd1 0.3 22.3 22 Gzmb 0.5 22.2 21.7 Negative Regulation of Immune System Process Btla 21.4 1.9 21.4 Cd55 20.8 2.36 1.5 Cd86 0.5 21.9 21.4 Ctla4 0.6 22.87 22.2 Dtx1 20.5 2 1.5 Fcgr2b 21.1 3.15 2.1 Tigit 0.3 22.7 22.3

when other organs were assessed (i.e., liver and lung), the decay Jeannet et al. (29) demonstrate that Tcf-1–deficient memory T cells was not as pronounced. This implies that Tcf-1 may only be im- are impaired in their ability to expand upon secondary challenge. portant in the persistence of circulating memory T cells and not Brain CD103+ T cells fail to undergo recall expansion after dis- memory T cells lodged within peripheral tissues. It is noteworthy sociation from the tissue in which they reside, and this lack of that Zhou et al. (27) also performed a transcriptome analysis of expansion may be associated with the Tcf-1 deficiency these cells Tcf1-deficient memory T cells. They show that Tcf1-deficient display. T cells express lower levels of eomes, CCR7, and SELL and Our microarray analysis reveals that brain CD103+ Trm cells higher levels of granzyme A and B compared with wild-type express high levels of several inhibitory receptors including memory T cells. Notably, brain CD103+ memory T cells express CTLA-4 and PD-1 (30). It is interesting to speculate that ex- low levels of Tcf1 and the subsequent genetic profile associated pression of these inhibitory receptors by CD103+ Trm cells is in with Tcf-1 deficiency including low levels of eomes, CCR7, SELL, an effort to maintain peripheral tolerance. Trm cells are a highly andhighlevelsofgranzymeAandB.Furthermore,studiesby sensitive, activated T cell population that resides within a variety 8 MOLECULAR SIGNATURE OF BRAIN Trm CELLS

FIGURE 5. Heat map of brain memory OT-I signa- ture genes. Microarray analysis of brain OT-I.CD103+, brain OT-I CD1032, and spleen OT-I.CD1032 memory T cells presented as a heat map of the 50 genes that Downloaded from have the greatest difference in expression when com- paring brain (both CD103+ and CD1032) to spleen populations (presented as normalized log2 intensities). http://www.jimmunol.org/ by guest on September 25, 2021

of peripheral tissues. Expression of these inhibitory receptors may We have previously reported that the brain CD103+ Trm cells serve as a means to prevent this memory T cell population from represent a T cell pool that has encountered cognate Ag locally accidentally being activated and unnecessarily attacking self. within the brain microenvironment, most likely during the acute Brain CD103+ Trm cells are not exhausted or senescent, and in stages of the infection. This local Ag presentation promoted CD103 this study we show that during a local bacterial infection, these expression and the retention of these memory T cells within the cells are able to provide substantial protection. Furthermore, it has brain tissue (8). This is in contrast to the CD1032 memory T cell been shown that brain CD103+ Trm cells can make cytokines and population, which we suspect represents a T cell population that are cytotoxic (8). Hence, during a localized infection, Trm cells was recruited to the brain during the acute infection but fails to must be released from this inhibitory state as these cells are clearly attain local Ag stimulation. These CD1032 memory T cells do persist functional during a secondary bout of infection. within the brain tissue, albeit poorly compared with their CD103+- Brain CD103+ Trm cells express elevated levels of the anti-viral expressing counterparts. Functional assays highlight further dif- protein IFITM3 (31). The constitutive expression of anti-viral ferences between these two cell populations with brain CD1032 may protect Trm from pathogens that they will un- memory T cells undergoing better ex vivo recall expansion in com- doubtedly encounter as a consequence of being situated within the parison with brain CD103+ Trm cells (8). We have now identified peripheral tissue. Resistance to virus infection is an extremely several hundred genes that are differentially expressed between beneficial attribute for a memory T cell population localized to brain CD103+ and brain CD1032 memory T cells. Hence, this common portals of pathogen entry to possess. This finding further local “programming” of Trm cells not only influences CD103 ex- supports the notion that Trm cells develop characteristics that pression and thus T cell retention but also initiates the expression allow them to better survive and function within their local en- of a series of genes that likely influence T cell survival, function, vironment. It remains to be determined whether Trm cells in other and maintenance within the tissue microenvironment. These data tissues express a similar signature to that of brain Trm cells or highlight that there is vast complexity within the memory T cell whether Trm cells in different tissue microenvironments express pool, and even memory T cells occupying the same niche have unique genetic profiles. strikingly different genetic profiles. The Journal of Immunology 9 Downloaded from http://www.jimmunol.org/

FIGURE 6. Gene expression profile comparing brain CD103+ and brain CD1032 resident memory T cells. Microarray analysis of brain OT-I.CD103+ 2 and brain OT-I CD103 memory T cells presented as a heat map of the 50 genes with the greatest difference in expression (presented as normalized log2 intensities). by guest on September 25, 2021

Gene expression profile analysis indicates that CD103+ Trm 6. Wakim, L. M., J. Waithman, N. van Rooijen, W. R. Heath, and F. R. Carbone. cells are a distinct memory T cell population displaying a unique 2008. Dendritic cell-induced memory T cell activation in nonlymphoid tissues. Science 319: 198–202. molecular signature, which likely results in optimal survival and 7. Masopust, D., D. Choo, V. Vezys, E. J. Wherry, J. Duraiswamy, R. Akondy, function within their local environment. In summary, these studies J. Wang, K. A. Casey, D. L. Barber, K. S. Kawamura, et al. 2010. Dynamic T cell migration program provides resident memory within intestinal epithelium. J. provide a base work with which to begin dissection of the factors Exp. Med. 207: 553–564. that influence Trm cell function, survival, and maintenance. This 8. Wakim, L. M., A. Woodward-Davis, and M. J. Bevan. 2010. Memory T cells knowledge will ultimately open the way for novel vaccination persisting within the brain after local infection show functional adaptations to their tissue of residence. Proc. Natl. Acad. Sci. USA 107: 17872–17879. strategies that promote Trm cell formation. 9. Purwar, R., J. Campbell, G. Murphy, W. G. Richards, R. A. Clark, and T. S. Kupper. 2011. Resident memory T cells (T(RM)) are abundant in human lung: diversity, function, and antigen specificity. PLoS ONE 6: e16245. Acknowledgments 10. Teijaro, J. R., D. Turner, Q. Pham, E. J. Wherry, L. Lefranc¸ois, and D. L. Farber. This work benefited from data assembled by the ImmGen consortium (32). 2011. Cutting edge: tissue-retentive lung memory CD4 T cells mediate optimal protection to respiratory virus infection. J. Immunol. 187: 5510–5514. 11. Hofmann, M., and H. Pircher. 2011. E-cadherin promotes accumulation of Disclosures a unique memory CD8 T-cell population in murine salivary glands. Proc. Natl. The authors have no financial conflicts of interest. Acad. Sci. USA 108: 16741–16746. 12.Mackay,L.K.,A.T.Stock,J.Z.Ma,C.M.Jones,S.J.Kent,S.N.Mueller, W. R. Heath, F. R. Carbone, and T. Gebhardt. 2012. Long-lived epithelial immunity by tissue-resident memory T (TRM) cells in the absence of per- References sisting local antigen presentation. Proc. Natl. Acad. Sci. USA 109: 7037– 1. Masopust, D., V. Vezys, A. L. Marzo, and L. Lefranc¸ois. 2001. Preferential lo- 7042. calization of effector memory cells in nonlymphoid tissue. Science 291: 2413– 13. Ariotti, S., J. B. Haanen, and T. N. Schumacher. 2012. Behavior and function of 2417. tissue-resident memory T cells. Adv. Immunol. 114: 203–216. 2. Sallusto, F., D. Lenig, R. Fo¨rster, M. Lipp, and A. Lanzavecchia. 1999. Two 14. Turner, M. J., E. R. Jellison, E. G. Lingenheld, L. Puddington, and L. Lefranc¸ois. subsets of memory T lymphocytes with distinct homing potentials and effector 2008. Avidity maturation of memory CD8 T cells is limited by self-antigen functions. Nature 401: 708–712. expression. J. Exp. Med. 205: 1859–1868. 3. Marshall, D. R., S. J. Turner, G. T. Belz, S. Wingo, S. Andreansky, 15. Sacks, J. A., and M. J. Bevan. 2008. TRAIL deficiency does not rescue impaired M. Y. Sangster, J. M. Riberdy, T. Liu, M. Tan, and P. C. Doherty. 2001. Mea- CD8+ T cell memory generated in the absence of CD4+ T cell help. J. Immunol. suring the diaspora for virus-specific CD8+ T cells. Proc. Natl. Acad. Sci. USA 180: 4570–4576. 98: 6313–6318. 16. Irizarry, R. A., B. M. Bolstad, F. Collin, L. M. Cope, B. Hobbs, and T. P. Speed. 4. Gebhardt, T., L. M. Wakim, L. Eidsmo, P. C. Reading, W. R. Heath, and 2003. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res. F. R. Carbone. 2009. Memory T cells in nonlymphoid tissue that provide en- 31: e15. hanced local immunity during infection with herpes simplex virus. Nat. Immu- 17. Smyth, G. K. 2005. Limma: Linear Models for Microarray Data. Springer, New nol. 10: 524–530. York. 5. Khanna, K. M., R. H. Bonneau, P. R. Kinchington, and R. L. Hendricks. 2003. 18. Smyth, G. K. 2004. Linear models and empirical bayes methods for assessing Herpes simplex virus-specific memory CD8+ T cells are selectively activated differential expression in microarray experiments. Stat. Appl. Genet. Mol. Biol. and retained in latently infected sensory ganglia. Immunity 18: 593–603. 3: Article 3. 10 MOLECULAR SIGNATURE OF BRAIN Trm CELLS

19. Benjamini, Y., and Y. Hochberg. 1995. Controlling the false discovery rate— phate receptor 1 causes tissue retention by inhibiting the entry of peripheral a practical and powerful approach to multiple testing. J. R. Stat. Soc. Series B tissue T lymphocytes into afferent lymphatics. Nat. Immunol. 9: 42–53. Stat. Methodol. 57: 289–300. 27. Zhou, X., S. Yu, D. M. Zhao, J. T. Harty, V. P. Badovinac, and H. H. Xue. 2010. 20. Wojciechowski, S., P. Tripathi, T. Bourdeau, L. Acero, H. L. Grimes, J. D. Katz, Differentiation and persistence of memory CD8(+) T cells depend on T cell F. D. Finkelman, and D. A. Hildeman. 2007. Bim/Bcl-2 balance is critical for factor 1. Immunity 33: 229–240. maintaining naive and memory T cell homeostasis. J. Exp. Med. 204: 1665– 28. Pearce, E. L., A. C. Mullen, G. A. Martins, C. M. Krawczyk, A. S. Hutchins, 1675. V. P. Zediak, M. Banica, C. B. DiCioccio, D. A. Gross, C. A. Mao, et al. 2003. 21. Sun, J. C., M. A. Williams, and M. J. Bevan. 2004. CD4+ T cells are required for Control of effector CD8+ T cell function by the transcription factor Eomeso- the maintenance, not programming, of memory CD8+ T cells after acute in- dermin. Science 302: 1041–1043. fection. Nat. Immunol. 5: 927–933. 29. Jeannet, G., C. Boudousquie´, N. Gardiol, J. Kang, J. Huelsken, and W. Held. 22. Greenwald, R. J., G. J. Freeman, and A. H. Sharpe. 2005. The B7 family 2010. Essential role of the Wnt pathway effector Tcf-1 for the establishment of revisited. Annu. Rev. Immunol. 23: 515–548. functional CD8 T cell memory. Proc. Natl. Acad. Sci. USA 107: 9777–9782. 23. Klein, R. S., E. Lin, B. Zhang, A. D. Luster, J. Tollett, M. A. Samuel, M. Engle, 30. Watanabe, N., M. Gavrieli, J. R. Sedy, J. Yang, F. Fallarino, S. K. Loftin, and M. S. Diamond. 2005. Neuronal CXCL10 directs CD8+ T-cell recruitment M. A. Hurchla, N. Zimmerman, J. Sim, X. Zang, et al. 2003. BTLA is and control of West Nile virus encephalitis. J. Virol. 79: 11457–11466. a lymphocyte inhibitory receptor with similarities to CTLA-4 and PD-1. Nat. 24. Trifilo, M. J., C. C. Bergmann, W. A. Kuziel, and T. E. Lane. 2003. CC che- Immunol. 4: 670–679. mokine ligand 3 (CCL3) regulates CD8(+)-T-cell effector function and migration 31. Brass, A. L., I. C. Huang, Y. Benita, S. P. John, M. N. Krishnan, E. M. Feeley, following viral infection. J. Virol. 77: 4004–4014. B. J. Ryan, J. L. Weyer, L. van der Weyden, E. Fikrig, et al. 2009. The IFITM 25. Shiow, L. R., D. B. Rosen, N. Brdickova´, Y. Xu, J. An, L. L. Lanier, J. G. Cyster, proteins mediate cellular resistance to influenza A H1N1 virus, West Nile virus, and M. Matloubian. 2006. CD69 acts downstream of interferon-alpha/beta to and dengue virus. Cell 139: 1243–1254. inhibit S1P1 and lymphocyte egress from lymphoid organs. Nature 440: 540–544. 32. Heng, T. S., M. W. Painter; Immunological Genome Project Consortium. 2008. 26. Ledgerwood, L. G., G. Lal, N. Zhang, A. Garin, S. J. Esses, F. Ginhoux, The Immunological Genome Project: networks of gene expression in immune M. Merad, H. Peche, S. A. Lira, Y. Ding, et al. 2008. The sphingosine 1-phos- cells. Nat. Immunol. 9: 1091–1094. Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021