UvA-DARE (Digital Academic Repository)

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

To settle or wander

Transcriptional regulation and recall responses of tissue-resident memory T cells Behr, F.M.

Publication date 2020

Document Version Other version License Other

Link to publication

Citation for published version (APA):

Behr, F. M. (2020). To settle or wander: Transcriptional regulation and recall responses of tissue-resident memory T cells.

General rights

It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations

If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

Chapter 9

Limited contribution of circulating memory CD8 ⁺ T cells to resident

memory at mucosal sites after reinfection

Felix M. Behr

, Ammarina Beumer-Chuwonpad

, Natasja A.M.

Kragten

, Thomas H. Wesselink

, Regina Stark

, and Klaas P.J.M. van Gisbergen

1 Department of Hematopoiesis, Sanquin Research and Landsteiner Laboratory, Amsterdam UMC, University of Amsterdam, Amsterdam, The Netherlands;

2 Department of Experimental Immunology, Amsterdam UMC, University of Amsterdam, Amsterdam, The Netherlands;

3 BIH Center for Regenerative Therapies, Charité Universitätsmedizin Berlin, Berlin, Germany.

Manuscript in revision

Abstract

Tissue-resident memory CD8⁺ T cells (T_RM) permanently localize to barrier tissues, where they mediate protection against reinvading pathogens. Circulating memory cells, including central memory (T_CM) and effector memory CD8⁺ T cells (T_EM), also contribute to recall responses in mucosa, but their potential to differentiate into mucosal T_RM remains unclear.

Here we employed adoptive transfer and reinfection models to specifically assess secondary responses of T_CM and T_EM at mucosal sites upon reinfection. Donor T_CM and T_EM exhibited robust systemic recall responses, but only limited accumulation in the small intestine, consistent with reduced expression of tissue-homing and -retention molecules. Upon pathogen clearance, T_CM and T_EM readily gave rise to secondary T_EM. T_CM also formed secondary central memory in lymphoid tissues and tissue-resident memory in internal tissues, e.g. the liver. Both T_CM and T_EM failed to substantially contribute to resident mucosal memory in the small intestine, while activated intestinal T_RM, but not liver T_RM, efficiently reformed CD103⁺ T_RM. Our findings demonstrate that circulating T_CM and T_EM are limited in their potential to generate mucosal resident memory upon reinfection. This may pose important implications on cell therapy and vaccination strategies employing memory CD8⁺ T cells for protection at mucosal sites.

9

Introduction

The adaptive immune system and in particular CD8⁺ T cells are crucial to combat acute infection with intracellular pathogens. Naïve pathogen-specific CD8⁺ T cells rapidly proliferate and differentiate into effector CD8⁺ T cells after primary pathogen encounter. These effector CD8⁺ T cells establish viral clearance through direct elimination of infected cells and recruitment and activation of other immune cells. Upon clearance of the infection, the effector CD8⁺ T cell pool contracts, giving rise to long-lived memory CD8⁺ T cells, which confer improved protection against subsequent reinfection. The memory CD8⁺ T cell population consists of different subsets with distinct migratory and functional characteristics, including central memory (T_CM), effector memory (T_EM), and tissue-resident memory CD8⁺ T (T_RM) cells ^1,2. T_CM cells maintain the capacity to survey secondary lymphoid organs (SLOs) via the expression of lymph node homing molecules, e.g. L-selectin (CD62L) and CCR7, while populations of T_EM cells either patrol peripheral tissues or are largely restricted to the circulation ^3,4. Contrary to these circulating memory CD8⁺ T cells, T_RM cells are absent from the blood, and instead permanently localize to peripheral tissues. Suppression of tissue exit receptors such as S1PR1 and CCR7 ^5–7 is essential for the establishment of tissue residency of T_RM cells.

The majority of T_RM cells across tissues express CD69, which in turn assists in suppression of tissue exit through the internalization and degradation of S1PR1 ^8,9. T_RM cells maintain tissue residency through the expression of adhesion molecules, including CD49a and, in mucosal tissues, CD103, which bind to collagen in the extracellular matrix and E-cadherin on epithelial cells, respectively ^8,10,11. Given their localization at portals of pathogen entry, T_RM cells mount immediate protective responses following secondary pathogen encounter, including production of pro-inflammatory cytokines ^12,13.

Many pathogens are re-encountered multiple times during a life time, suggesting that continued T_RM-driven protection against recurrent pathogen exposure requires regeneration of T_RM cells from the memory CD8⁺ T cell compartment. It has been proposed that T_CM cells establish the repopulation of the memory T cell pool upon pathogen re-encounter, given their stem-like properties, which include self-renewal capacity and oligopotency. Indeed, upon antigen reencounter in SLOs, T_CM cells mount strong proliferative responses, giving rise to large amounts of secondary effector CD8⁺ T cells ^3,14. Moreover, T_CM cells readily yield the full spectrum of circulating memory CD8⁺ T cells, including T_CM and T_EM cells, following reactivation during secondary pathogen encounter ^15–17. In contrast, reactivated T_EM cells only exhibit limited proliferative expansion after antigen reencounter, and instead rapidly engage effector functions ^3,14. Furthermore, T_EM cells appear more terminally differentiated than T_CM cells and primarily give rise to secondary T_EM cells, but not T_CM cells, following antigen reencounter ^4,15,16. Currently, it is incompletely understood how T_CM and T_EM cells contribute to the formation of tissue-resident memory populations after reinfection. Recent reports suggest that T_CM cells retain the potential to generate T_RM cells in the skin, albeit with substantially decreased efficiency compared to naïve CD8⁺ T cells ^18,19. In particular, T_CM cells appear impaired in the formation of CD103⁺ T_RM cells in the skin relative to naïve T cells ¹⁹. These findings appear out of line with the oligopotent capacity of T_CM cells to establish repopulation of other memory lineages after secondary pathogen encounter. In particular, it remains unclear how T_CM cells and other memory CD8⁺ T cell subsets contribute to T_RM formation at tissue sites other than skin.

Recently, it has been demonstrated that the capacity to form CD103⁺ T_RM cells is imprinted in

naïve CD8⁺ T cells by migratory dendritic cells (DCs) in the lymph nodes ²⁰. These migratory DCs present the cytokine TGF-β to naïve CD8⁺ T cells, which instructs CD103 expression on a fraction of naïve CD8⁺ T cells. Only naïve CD8⁺ T cells that have recently experienced TGF-β triggering in the lymph nodes appear to have the capacity to form CD103⁺ T_RM cells. These findings suggest that commitment to a tissue-resident memory fate is predetermined in naïve CD8⁺ T cells and has already been established prior to antigen encounter and CD8 T cell differentiation. We considered the possibility that, similar as in naïve CD8⁺ T cells, accessibility of the CD103-encoding Itgae locus plays an instrumental role in the capacity of circulating memory CD8⁺ T cells to form (CD103⁺) T_RM cells after antigen encounter.

Therefore, in this study, we investigated the potential of circulating memory CD8⁺ T cell populations to establish T_RM cells following reinfection. We found that circulating memory CD8⁺ T cells, in contrast to naïve CD8⁺ T cells, retained the Itgae locus in a closed state, which was refractory to TGF-β signaling. In line with these findings, reactivated T_CM and T_EM cells exhibited strongly impaired capacity in the generation of CD103⁺ T_RM cells in the small intestine. In contrast, mucosal T_RM cells themselves, but not T_RM cells from other sites, efficiently re-formed CD103⁺ T_RM cells upon reactivation. Hence, we propose that resident rather than circulating memory CD8⁺ T cells are the dominant population that repopulates CD103⁺ T_RM cells after reinfection. These findings may have implications for vaccination and cellular therapy strategies that employ circulating memory CD8⁺ T cells to improve immune protection at mucosal sites.

Results

Circulating memory CD8

T cells lack expression of CD103

Naïve CD8⁺ T (T_N) cells partially express the integrin CD103, whose expression on T_N cells is reversibly induced by α_V integrin–expressing migratory dendritic cells (DC) in lymph nodes

20. In order to assess CD103 expression by effector and memory CD8⁺ T cells, mice were infected with lymphocytic choriomeningitis virus (LCMV) and effector and memory CD8⁺ T cell populations specific for the immunodominant gp33 epitope of LCMV (D^b GP33⁺) were analyzed at >30 days after infection (Supplementary Fig. 1A). Transcripts of Itgae (encoding CD103) were high in naïve T (T_N) cells, but minimal in splenic central memory (T_CM) and effector memory CD8⁺ T (T_EM) cells (Fig. 1A). Concordantly, ca. 50 % of T_N cells from the spleen expressed CD103 at the protein level (Fig. 1B,C). In contrast, CD103 protein was virtually absent on virus-specific (D^b GP33⁺) effector CD8⁺ T (T_Eff) cells and lowly expressed on both total and virus-specific T_EM and T_CM cells from the spleen (Fig. 1B,C, and Supplementary Fig. 1B,C). As previously reported ¹², tissue-specific CD103 expression was observed in the T_RM compartment. T_RM cells from the liver expressed low amounts of CD103, both at the transcriptional and the protein level, while the majority of T_RM cells from the intraepithelial lymphocyte fraction of the small intestine (SI IEL) expressed CD103 (Fig. 1A,D,E). Thus, these data indicate that CD103 expression is largely limited to T_N and mucosal T_RM cells at resting state.

9

Circulating memory CD8

T cells do not upregulate CD103 after TGF-β stimulation

Expression of CD103 on activated CD8⁺ T cells is induced by TGF-β ^21,22, a cytokine that is abundantly expressed in various tissues, including the small intestine ²³. We therefore investigated whether the differential expression of CD103 by T_N, T_EM and T_CM cells might result from their ability to respond to TGF-β signaling. To examine the responsiveness of activated CD8⁺ T cell subsets to TGF-β, we isolated T_N cells from naïve mice and T_EM and T_CM cells from LCMV-immune mice and activated them in vitro in the presence or absence of TGF-β. T cell receptor (TCR) triggering resulted in a uniform activation of the isolated T cell subsets, as determined by upregulation of CD69 (Supplementary Fig. 2A,B). Both total and virus- specific (D^b GP33⁺) circulating memory CD8⁺ T cells did not upregulate CD103 expression upon TCR triggering (Fig. 2A,B and Supplementary Fig. 2C,D). In fact, activation of T_N cells in the absence of exogenous TGF-β led to reduced CD103 expression. As expected, expression of CD103 was strongly induced on T_N cells following TCR triggering in the presence of TGF-β (Fig. 2A,B and Supplementary Fig. 2C,D). In contrast, TCR stimulation in the presence of TGF-β only induced a modest increase of CD103 expression on T_EM and T_CM cells (Fig. 2A,B and Supplementary Fig. 2C,D). These findings suggest differential responsiveness of T_N and circulating memory T cells to TGF-β signaling following activation.

Figure 1 | CD103 is expressed by naïve and intestinal tissue-resident memory CD8⁺ T cells, but not by circulating effector and memory CD8⁺ T cells. (A-E) Mice were infected with LCMV and virus-specific (D^b GP33⁺) effector (T_Eff), CX3CR1⁺ CD62L^- effector memory (T_EM), CX3CR1^- CD62L⁺ central memory (T_CM), and CD69⁺ CD62L^- tissue-resident memory CD8⁺ T (T_RM) cells were analyzed at 8 days and >30 days p.i., respectively. Naïve CD44^lo CD62L⁺ CD8⁺ T (T_N) cells were isolated from naïve mice. (A) Expression [log2 (reads per kilobase million)] of Itgae is shown by T_N, T_CM and T_EM cells from spleen, and T_RM cells from liver and the intraepithelial lymphocyte fraction of the small intestine (SI IEL). (B) Representative flow cytometry plots show expression of CD103 by TN cells and D^b GP33⁺ T_Eff, T_EM and T_CM cells from spleen. (C) The frequency of CD103 expression on T_N cells and D^b GP33⁺ T_Eff, T_EM, and T_CM cells in spleen was quantified. (D) Representative flow cytometry plots show expression of CD103 by D^b GP33⁺ T_RM cells from liver and SI IEL. (E) The frequency of CD103 expression on D^b GP33⁺ liver and SI IEL T_RM cells was quantified. Data from one experiment, taken from ⁷ (A), and combined data from two independent experiments (n = 6-12) (C,E). Symbols represent individual mice; bars represent the mean. Error bars represent mean ± SD. Two-tailed unpaired t-test,

** p < 0.01; *** p < 0.001.

Figure 2 | Murine and human circulating memory CD8⁺ T cells exhibit limited CD103 expression upon activation in the presence of TGF-β. (A,B) Naïve (TN), effector memory (T_EM) and central memory CD8⁺ T (T_CM) cells were isolated from mice at >30 days post LCMV infection, and activated in vitro with anti-CD3/anti-CD28 antibodies in the presence or absence of TGF-β for 3 days. (A) Representative histograms show expression of CD103 on activated T_N, T_CM and T_EM cells. (B) Frequencies of CD103⁺ cells were quantified for the indicated T cell subsets. (C,D) TN, T_EM, T_CM and CD45RA⁺ T_EM (T_EMRA) cells were isolated from the blood of healthy human volunteers and activated in vitro with anti-CD3/antiCD28 antibodies in the presence or absence of TGF-β for 3 days. (C) Representative histograms show expression of CD103 on activated T_N, T_CM, T_EM and T_EMRA cells (D) Frequencies of CD103⁺ cells were quantified for the indicated T cell subsets. (E,F) Normalized chromatin accessibility near the Itgae locus is shown for (E) naïve OT-I T cells and splenic circulating memory OT-I T (Tcirc) cells developing after L.m.-OVA infection ²⁴ or (F) endogenous naïve T cells and m45-specific memory CD8⁺ T (T_circ) cells from spleen at day 35 post mCMV infection

25. Rectangles mark areas with binding domains for Smad and Runx transcription factors ²⁰. Combined data from two independent experiments (n = 4-9) (A-D), and data from published data sets ^24,25. Symbols represent individual mice; bars represent the mean. Two-way ANOVA with Tukey’s multiple comparisons test, * p < 0.05; *** p < 0.001.

9

Figure 3 | T_CM and T_EM cells exhibit reduced capacity to accumulate in the small intestine following viral rechallenge. (A) Graphic scheme depicts settings of rechallenge experiments with adoptively transferred TCM and T_EM cells. In brief, congenically marked T_CM and T_EM cells were FACS-purified from spleen of LCMV-immune mice,

In order to evaluate whether the observed responses of naïve and memory CD8⁺ T cells to TGF-β stimulation were conserved in humans, we next examined CD103 expression by human CD8⁺ T cell subsets following stimulation with TGF-β. To this end, naïve, T_EM, T_CM and CD45RA⁺ T_EM (T_EMRA) cells were isolated from the blood of healthy volunteers and activated in vitro in the absence or presence of TGF-β. The human T cell subsets did not upregulate CD103 following TCR triggering alone (Fig. 2C,D). Activation in the presence of TGF-β resulted in CD103 expression on the vast majority of activated naïve T cells, but led to only a modest increase in CD103 expression on the activated memory subsets. Taken together, these data indicate that activated murine and human circulating memory T cells are impaired in their ability to upregulate CD103 in response to TGF-β signaling.

Distinct epigenetic imprints may underlie the differential expression of CD103 by activated T_N cells and circulating memory T cells in response to TGF-β. In order to evaluate the chromatin accessibility of the Itgae locus (encoding CD103), we examined publicly available ATAC-seq (Assay for Transposase-Accessible Chromatin using sequencing) data of naïve and circulating memory T (T_circ) cells from the spleen in two different murine infection models ^24,25. Chromatin accessibility around the Itgae locus was higher in naïve T (T_N) cells compared to pathogen-specific T_circ cells developing in the Listeria and the mCMV infection model (Fig. 5E,F). Importantly, chromatin accessibility in the Itgae locus of T_circ cells was specifically decreased in regions, which have previously been reported to contain binding motifs for Runx and Smad transcription factors ²⁰. Runx and Smad transcription factors are key targets of TGF-β signaling ^26,27. Therefore, differential epigenetic imprinting of the Itgae locus may impact the TGF-β driven regulation of CD103 expression on activated naïve and circulating memory T cells.

Circulating memory CD8

T cells are inefficient at forming effector responses in the small intestine

Given the importance of an accessible CD103 locus for naïve T cells to form T_RM cells ²⁰, we next investigated the ability of T_EM and T_CM cells in forming secondary responses at different tissue sites. To this end, we isolated CX3CR1⁺ T_EM and CD62L⁺ T_CM cells from LCMV immune mice at >30 days after LCMV infection (Supplementary Fig. 3A,B). Congenically marked T_EM and T_CM cells containing equals amounts of D^b GP33⁺ cells were then co-transferred into naïve recipient mice for analysis of their tissue distribution at two weeks after transfer (Supplementary Fig. 3C). At this timepoint, donor T_CM cells were present in blood, spleen

and co-transferred into naïve recipients, which were challenged with LCMV 14 days later. The offspring of the virus-specific (D^b GP33⁺) donor T cells was analyzed at 8 days p.i. (B) The number of T_CM- and T_EM-derived D^b GP33⁺ cells (corrected for input ratio at time of transfer) was determined in spleen and liver at the indicated time points before and after LCMV challenge. (C) Representative flow cytometry plots show expression of the congenic markers CD45.1 and CD45.2 to identify donor T_EM-derived (purple), donor T_CM-derived (blue) and host T_N-derived (grey) cells within the D^b GP33⁺ T cell population in the indicated tissues. (D) Frequencies of donor and host-derived cells within the D^b GP33⁺ T cell population were quantified. (E-J) Representative histograms show expression of (E) Ki-67, (G) α4β₇ and (I) CCR9 by TN-, T_CM- and T_EM-derived D^b GP33⁺ effector cells in indicated tissues. CD44^low CD62L⁺ CD8⁺ T cells are displayed as a reference population. (F,H,J) The frequency of (F) Ki-67 expression, (H) α4β₇ expressionand (J) CCR9 expression within the D^b GP33⁺ offspring of donor T_EM and T_CM populations and host T_N populations was quantified. Data is representative of two independent experiments (n = 3-4). Symbols represent the mean (B) or individual mice (D,F,H,J); bars represent the mean. Error bars represent mean ± SEM. Two-way ANOVA with Tukey’s multiple comparisons test, ** p < 0.01; *** p < 0.001.

9

and liver of recipient mice, while donor T_EM cells were mainly detected in blood and liver (Supplementary Fig. 3D,E). Both donor subsets were absent from the intra-epithelial compartment of the small intestine (SI IEL). The respective phenotypes of donor T_EM and T_CM cells were largely preserved following transfer (Supplementary Fig. 3F,G). Importantly, LCMV- specific (D^b GP33⁺) CD8⁺ T cells were detected in both donor T_EM and T_CM cell populations, but not in host T_N cells (Supplementary Fig. 3H). Thus, donor populations of virus-specific T_EM and T_CM cells were stably maintained for at least two weeks after adoptive transfer.

To examine the capacity of donor T_CM and T_EM cells to establish secondary responses, recipient mice were challenged with LCMV two weeks after adoptive transfer (Fig. 3A). Virus-specific donor T_EM and T_CM cells underwent substantial expansion following viral challenge (Fig.

3B), and formed virus-specific effector responses in blood, spleen, peripheral lymph nodes (pLN), liver and small intestine (SI IEL) (Fig. 3C,D). CD8⁺ T cells from the naïve host (T_N) also contributed to the virus-specific effector response (Fig. 3C,D). T_N- and T_CM-derived effector T cells were present at higher frequencies throughout tissues as compared to effector cells of T_EM origin at day 8 post infection (p.i.), suggesting that T_N and T_CM cells expanded more

Figure 4 | Circulating memory CD8⁺ T cells are impaired in acquiring a mucosal T_RM phenotype in the small intestine early after activation. (A) Representative flow cytometry plots show expression of KLRG1 and CD127 on donor T_EM and T_CM-derived, and host T_N-derived D^b GP33⁺ effector cells in indicated tissues at day 8 p.i. (B) The frequencies of KLRG1^- CD127⁺ cells within the D^b GP33⁺ offspring of donor T_EM and T_CM populations and host T_N populations was quantified. (C) Representative flow cytometry plots show expression of CD69 and CD103 on donor T_EM and T_CM-derived, and host T_N-derived D^b GP33⁺ effector cells in indicated tissues at day 8 p.i. (D) Contribution of CD69^- CD103^-, CD69⁺ CD103^- and CD69⁺ CD103⁺ cells derived from T_CM, T_EM and T_N cells to the total D^b GP33⁺ effector cell population in liver and small intestine (SI IEL) is shown. Data is representative of two independent experiments (n = 3-4). Symbols represent individual mice; bars represent the mean. Error bars represent mean ± SEM. Two-way ANOVA with Tukey’s multiple comparisons test, ** p < 0.01; *** p < 0.001.

vigorously compared to T_EM cells after viral challenge. Interestingly, while T_N and T_CM cells contributed to the virus-specific response to a similar extent in blood, spleen and liver, T_N- derived effector cells dominated the effector response in pLN and in the small intestine (SI IEL) (Fig. 3C,D). The dominant contribution of naïve T cells in the small intestine did not appear to result from differences in local proliferation, as the proliferation associated molecule Ki67 was expressed at similar levels by T_N-, T_EM- and T_CM-derived effector cells (Fig.

3E,F). To determine the impact of migration on the local presence of the effector cells, we next investigated the expression of tissue-homing molecules. Localization of lymphocytes to the small intestine is regulated by the integrin α4β7 and the chemokine receptor CCR9

28. Virus-specific effector cells derived from T_N, T_EM and T_CM cells upregulated expression of α4β7 to a similar degree in the spleen after infection (Fig. 3G,H). In the small intestine, α4β7 expression was minimal on effector T cells derived from donor memory T cell populations, while a substantial fraction of T_N-derived effector T cells expressed α4β7. Furthermore, expression of the chemokine receptor CCR9 was largely restricted to T_N-derived cells within the virus-specific effector population in the intestine-draining mesenteric lymph nodes (mLN) (Fig. 3I,J). Taken together, effector cells arising from T_CM and T_EM cells exhibited impaired accumulation in the small intestine compared to their T_N-derived counterparts after infection, in line with their lower expression of intestine-homing molecules.

Circulating memory CD8

T cells are impaired in acquiring a mucosal T

phenotype in the small intestine early after activation

Effector CD8⁺ T cells with potential to develop into memory T cells can be distinguished from terminal effectors based on their expression of the interleukin-7 receptor α (CD127) and lack of KLRG1 expression ^29,30. We observed that the majority of virus-specific T cells derived from T_N, T_EM and T_CM cells expressed Killer cell lectin like receptor G1 (KLRG1), a molecule associated with effector differentiation ^31,32. In line with elevated expression on primary T_EM cells, effector cells derived of T_EM cells expressed KLRG1 to a higher degree than those derived of T_CM or naïve T cells (Fig. 4A). In contrast, T cells with a memory precursor effector cell (MPEC) phenotype were more prevalent among T_N-derived effector cells, compared to effector cells of T_CM and T_EM origin at day 8 p.i. (Fig. 4B). In the small intestine, but not at other sites, effector T cells of T_N, T_EM and T_CM origin already exhibited substantial upregulation of the tissue residency associated molecule CD69 (Fig. 4C). T_N cells outcompeted T_CM and T_EM cells in the formation of CD69⁺ effector cells in the small intestine (Fig. 4C,D). Interestingly, a substantial fraction of naïve-, but not T_EM- or T_CM-derived CD69⁺ effector cells in the small intestine had also upregulated CD103 expression (Fig. 4C,D). Thus, compared to naïve T cells, circulating memory CD8⁺ T cells appear to have a specific deficit in their ability to form secondary effector cells with a T_RM precursor phenotype in the small intestine.

T

cells have greater potential to re-form circulating memory CD8

T cell subsets than T

cells

We next analyzed how the skewing in localization and phenotype of effector cells derived of endogenous naïve and donor memory populations affects the memory response (> 30 days)

9

Figure 5 | T_CM cells yield the full spectrum of circulating memory CD8+ T cells, while T_EM cells primarily re-form themselves after reactivation. (A) Graphic scheme depicts settings of rechallenge experiments with adoptively transferred T_CM and T_EM cells. In brief, congenically marked T_CM and T_EM cells were FACS-purified from spleen of LCMV-immune mice, and co-transferred into naïve recipients, which were challenged with LCMV 14 days later.

The offspring of the virus-specific (D^b GP33⁺) donor T cells was analyzed at >30 days p.i. (B) Representative flow cytometry plots show expression of the congenic markers CD45.1 and CD45.2 to identify donor T_EM-derived (purple), donor T_CM-derived (blue) and host T_N-derived (grey) cells within the D^b GP33⁺ T cell population in the indicated tissues. (C) Frequencies of donor and host-derived cells within the D^b GP33⁺ T cell population were quantified. (D-G) Representative flow cytometry plots show expression of (D) CD62L and CD44, and (F) KLRG1 and CX3CR1 on T_N-, T_CM- and T_EM-derived D^b GP33⁺ memory cells in indicated tissues. (E,G) The frequency of (E) CD44 and CD62L co-expression, and (G) KLRG1 and CX3CR1 co-expression within the D^b GP33⁺ offspring of donor T_EM and T_CM populations and host T_N populations was quantified. Combined data from two independent experiments (n = 4).

Symbols represent individual mice; bars represent the mean. Two-way ANOVA with Tukey’s multiple comparisons test, ** p < 0.01; *** p < 0.001.

following viral challenge of recipient mice (Fig. 5A). Naïve, T_CM and T_EM cells formed systemic memory responses throughout blood, spleen, pLN, liver and small intestine (SI IEL), albeit at different magnitudes (Fig. 5B,C). Although the memory populations in these organs had contracted compared to the effector phase (Fig. 3), the relative sizes of the virus-specific

memory cells of T_N, T_CM and T_EM origin remained constant (Fig. 5B,C). Compared to T_N and T_CM cells, T_EM cells exhibited only limited contribution to the memory pool throughout tissues (Fig. 5B,C). Moreover, similar to the effector phase, the virus-specific population in the pLN and small intestine primarily originated from naïve T cells, rather than from T_CM and T_EM cells, in the memory phase (Fig. 5B,C). Previous research has demonstrated the capacity of T_CM cells to re-form T_CM cells and differentiate into T_EM cells during secondary responses ^15–17. In line with these findings, T_CM cells, but not T_EM cells, gave rise to secondary central memory T cells in pLN following viral challenge (Fig. 5D,E). However, in line with their preferential location in pLN, T_N-derived memory T cells more readily acquired a central memory phenotype (CD44⁺ CD62L⁺) in spleen and pLN than T_CM cells (Fig. 5D,E). Furthermore, naïve, T_CM and T_EM cells all displayed potential to form T_EM cells co-expressing KLRG1 and CX3CR1 after viral challenge (Fig. 5F,G). T_EM cells almost exclusively gave rise to secondary KLRG1⁺ CX3CR1⁺ T_EM cells, consistent with their limited differentiation potential ^4,15. Taken together, these data suggest that naïve and T_CM cells retain the potential to form both T_CM and T_EM cells in the circulation, while the capacity of T_EM cells appears limited to the reformation of T_EM cells.

Circulating memory CD8

T cells are impaired in forming T

cells at mucosal sites

In addition to the formation of circulating memory T cells, acute infection with LCMV also drives the generation of T_RM cells in liver, kidney, salivary glands and small intestine

2,3334,35. In order to assess the potential of T_N, T_EM and T_CM cells to form T_RM cells, we next examined their contribution to local T_RM populations in peripheral tissues. A fraction of virus-specific memory T cells of T_N, T_EM and T_CM origin in the liver expressed the tissue- residency molecule CD69 in the memory phase after viral challenge (Fig. 6A). T_N and T_CM cells contributed to a comparable extent to the virus-specific CD69⁺ T_RM population of the liver, while the contribution of T_EM cells was minimal (Fig. 6B). Similar to liver T_RM cells, the CD69⁺ T_RM population in the kidney largely originated from T_N and T_CM cells rather than from T_EM cells (Fig. 6C,D). Thus, T_CM in contrast to T_EM cells, maintain the potential to form T_RM cells in internal tissues.

T_RM cells at mucosal sites, in contrast to their counterparts in internal tissues, frequently express the integrin CD103 ¹². Therefore, we next evaluated the capacity of circulating memory T cells to form CD103⁺ T_RM cells at mucosal sites. In the salivary glands, virus-specific memory T cells arising from T_N, T_CM and T_EM cells expressed CD69 after viral challenge (Fig.

6E), but only those memory T cells of naïve origin partially co-expressed CD103 (Fig. 6E,F). In the small intestine (SI IEL), T_EM cells largely failed to form secondary T_RM cells, while T_CM cells were able to establish a population of secondary T_RM cells that partially co-expressed CD69 and CD103 (Fig. 6G,H). However, the vast majority of CD103⁺ T_RM cells in the small intestine was derived from naïve T cells, which almost uniformly acquired co-expression of CD69 and CD103 at this site after viral challenge (Fig. 6G,H). Thus, we conclude that, compared to naïve T cells, T_EM cells have limited potential to form T_RM cells in internal and mucosal tissues, while T_CM cells were specifically compromised in their ability to form CD103⁺ T_RM cells at mucosal sites.

9

Figure 6 | Circulating memory CD8⁺ T cells are impaired in T_RM formation at mucosal sites. (A-H) The memory offspring of virus-specific (D^b GP33⁺) T_N cells and donor T_CM and T_EM cells was determined in (A,B) liver, (C,D) kidney, (E,F) salivary glands and (G,H) SI IEL at >30 days p.i. (A,C,E,G) Representative flow cytometry plots show expression of CD69 and CD103 on T_N-, T_CM- and T_EM-derived D^b GP33⁺ memory cells. (B,D,F,H) Contribution of CD69^- CD103^-, CD69⁺ CD103^- and CD69⁺ CD103⁺ cells derived from T_CM, T_EM and T_N cells to the total D^b GP33⁺ memory cell population is shown. Combined data from two independent experiments (n = 4). Bars represent the mean, error bars mean ± SEM.

Mucosal T

cells maintain capacity to re-form CD103

T

cells after reactivation

The limited potential of circulating memory T cells to form CD103⁺ T_RM cells spurred us to determine the potential of T_RM cells subsets to re-form CD103+ T_RM cells. To this end, we isolated internal T_RM cells from the liver and mucosal T_RM cells from the SI IEL of LCMV immune mice at >30 days after LCMV infection. The secondary response of congenically marked liver

T_RM and SI IEL T_RM cells containing equal amounts of D^b GP33⁺ cells was monitored after co-transfer into naïve recipient mice, which were challenged with LCMV two weeks after transfer (Fig. 7A). Virus-specific memory T cells originating from donor liver and SI IEL T_RM cells were present at comparable frequencies in the spleen in the memory phase after viral challenge, while the donor memory population in the small intestine was primarily derived from SI IEL T_RM cells (Fig. 7B,C). Secondary memory T cells in the small intestine originating from SI IEL T_RM also readily upregulated CD69 and CD103 (Fig. 7D,E). In contrast, liver T_RM cells only established few secondary CD103⁺ T_RM cells in the small intestine (Fig. 7D,E). Taken together, these data indicate that the re-formation of CD103⁺ T_RM cells at mucosal sites occurs efficiently from pre-existing mucosal T_RM cells.

Figure 7 | Intestinal T_RM cells efficiently re-form mucosal T_RM cells in the small intestine after viral rechallenge. (A) Graphic scheme depicts settings of rechallenge experiments with adoptively transferred T_RM cells from the liver and small intestine (SI IEL). In brief, congenically marked liver and SI IEL T_RM cells were FACS-purified from LCMV-immune mice, and co-transferred into naïve recipients, which were challenged with LCMV 14 days later. The offspring of the virus-specific (D^b GP33⁺) donor T cells was analyzed at >30 days p.i. (B) Representative flow cytometry plots show expression of the congenic markers CD45.1 and CD45.2 to identify donor liver T_RM-derived (turquoise) and IEL T_RM- derived cells (green) within the D^b GP33⁺ memory T cell population in indicated tissues. (C) Frequencies of donor- derived cells within the D^b GP33⁺ memory T cell population were quantified. (D) Representative flow cytometry plots show expression of CD69 and CD103 on liver and IEL T_RM-derived D^b GP33⁺ memory cells in small intestine (SI IEL). (E) Contribution of CD69^- CD103^-, CD69⁺ CD103^- and CD69⁺ CD103⁺ cells derived from liver and IEL T_RM cells to the total D^b GP33⁺ memory cell population in the small intestine (SI IEL). Combined data from two independent experiments (n = 10). Symbols represent individual mice; bars represent the mean. Error bars represent mean ± SEM. Two-tailed paired t-test, * p < 0.05.

9

Discussion

T_RM cells play a critical role in host defense against recurrent pathogens ^12,13. In this paper, we examined how secondary pathogen exposure shapes the T_RM compartment. In particular, we have determined which CD8⁺ T cell populations have the capacity to replenish local T_RM populations after reinfection of a host with pre-existing immunological memory. Using adoptive transfer and rechallenge approaches, we found that mucosal T_RM cells efficiently re-form CD103⁺ T_RM cells after pathogen reencounter, underlining the repopulation capacity of the T_RM compartment. In contrast, circulating memory CD8⁺ T cells did not substantially contribute to the formation of mucosal CD103⁺ T_RM cells after reactivation, despite efficient capacity to repopulate T_CM and T_EM cells, as well as T_RM cells at non-mucosal sites.

Naïve and central memory CD8⁺ T cells share the capacity to survey secondary lymphoid organs (SLO) for antigens and mount robust proliferative responses following activation

3,14,36. Responses of T_N and T_CM cells are thus likely shaped by competition between the naïve and the memory compartment, in particular in hosts with sustained thymic output. In the present study, the responses of donor T_CM cells and endogenous T_N cells were compared after viral challenge. An uninfected mouse contains an estimated 100-200 epitope-specific (D^b GP33⁺) T_N cells ³⁷. We transferred approximately 20 times more epitope-specific memory CD8⁺ T cells in our settings. Hence, virus-specific donor T_CM cells were probably more abundant than their naïve counterparts prior to viral challenge, even if not all T_CM cells were maintained following transfer. Despite this possible difference in precursor frequencies, we observed that activated T_N cells established effector and memory responses of equal magnitude as T_CM cells in circulation, spleen and liver, and even outcompeted T_CM cells in peripheral lymph nodes and the small intestine following viral challenge. Previously, it has been demonstrated that differential activation by migratory DCs allows for the induction of naïve responses even when large numbers of competing memory CD8⁺ T cells are present ³⁸. Furthermore, on a per cell basis, T_N cells exhibit a more profound expansion ability than their circulating memory counterparts during effector responses ^39,40. These properties of T_N cells may contribute to their superior responses throughout tissues after viral challenge in the presence of competing T_CM and T_EM cells.

High expression of the fractalkine receptor CX3CR1 identifies a uniform population of T_EM cells, which is largely confined to the circulation during homeostatic trafficking ⁴. Following pathogen rechallenge, we observed that these T_EM cells underwent limited expansion, as compared to T_N and T_CM cells, and remained primarily confined to tissues with a large circulating compartment. In line with our findings, previous research has demonstrated the restricted proliferative capacity of T_EM cells upon restimulation ^3,14,17. The limited expansion potential of T_EM cells has been associated with a more terminal differentiation state, as indicated by a shorted telomere length ⁴¹. Concordant with an advanced differentiation state, we found that reactivated CX3CR1⁺ T_EM cells were limited to the formation of secondary T_EM cells and did not contribute to re-establishment of the T_CM or T_RM compartment. The limited differentiation potential of T_EM cells after pathogen reencounter has been observed in other models, with T_EM cells primarily giving rise to secondary effector and T_EM cells ^4,15,16. In contrast, T_CM cells in these and our systems readily gave rise to secondary T_CM and T_EM cells following reactivation, corroborating their oligopotent capacity ^15–17. In addition to the generation of secondary circulating memory CD8⁺ T cells, we found that T_CM cells retain the capacity to form T_RM cells at non-mucosal sites, i.e. the liver and kidneys. Thus, unlike

their T_EM counterparts, T_CM cells maintain the potential to regenerate a broad spectrum of circulating and resident memory CD8⁺ T cells.

Despite their oligopotent properties, we observed that T_CM cells were limited in generating CD103⁺ T_RM cells at mucosal sites, such as the small intestine and the salivary glands, after viral challenge. Similar observations have been made previously in the skin, where T_CM cells are impaired in forming resident memory as compared to T_N cells after reactivation, in particular regarding the establishment of CD103⁺ T_RM cells ^18,19. These observations may suggest that T_CM and CD103⁺ T_RM cells develop as separate lineages after primary infection, which have restricted potential to re-form one another after restimulation. Overlapping TCR repertoires suggest that skin T_RM cells and T_CM cells can arise from common naïve CD8⁺ T cell precursors after immunization ⁴². Similar to T_CM cells, T_RM cells have been found to primarily arise from the KLRG1^- CD127⁺ (MPEC) population of effector CD8⁺ T cells ¹¹, although transient KLRG1 expression at early stages is not incompatible with T_RM cell fate ⁴³. However, circulating effector CD8⁺ T cells lose the potential to form intestinal T_RM cells as early as 7 days after infection, suggesting an early lineage separation between circulating memory and resident memory of the intestine ^44,45. These findings indicate that the commitment to T_CM or mucosal T_RM cell fates is established during effector cell differentiation and maintained into the memory phase.

A recent study demonstrated that the capacity of naïve CD8⁺ T cells to form CD103⁺ T_RM cells in the skin is imprinted by migratory dendritic cells (DCs) via TGF-β presentation in the lymph nodes prior to antigen encounter ²⁰. Insufficient TGF-β mediated conditioning of naïve CD8⁺ T cells is associated with decreased chromatin accessibility of the CD103-encoding Itgae locus and results in impaired CD103⁺ T_RM formation during subsequent antigen encounter ²⁰. We found that circulating memory CD8⁺ T cells exhibit restricted chromatin accessibility at the Itgae locus, as compared to T_N cells. Hence, epigenetic imprinting at the Itgae locus may obstruct circulating memory CD8⁺ T cells from forming CD103⁺ T_RM cells after reinfection.

Epigenetic profiling during CD8⁺ T cell differentiation has demonstrated that memory CD8⁺ T cells either maintain the chromatin profile of effector CD8⁺ T cells, such as enhanced accessibility at the loci encoding effector molecules, or re-acquire certain epigenetic states of naïve CD8⁺ T cells, such as enhanced accessibility at loci of Sell (encoding CD62L) and IL7r ^46,47. Similar to sustained accessibility of the Sell locus in the T_CM lineage, sustained accessibility of the Itgae locus may occur specifically in the mucosal T_RM lineage. In fact, we already observed minimal CD103 expression on splenic effector CD8⁺ T cells, while effector cells in the small intestine partially expressed CD103. Environmental factors such as antigen recognition and pro-inflammatory cytokines (e.g. IL-12 and IFN-β) can suppress CD103 expression, suggesting that the potential to generate CD103⁺ T_RM cells may be actively repressed in a fraction of effector cells ^48,49. Thus, commitment to the CD103⁺ T_RM cell fate may require sustained maintenance of an open Itgae locus during differentiation from naïve CD8 T cells.

Following pathogen rechallenge, we found that intestinal T_RM cells, but not liver T_RM cells or circulating memory CD8⁺ T cells, efficiently formed CD103⁺ T_RM in the small intestine.

Similarly, recall responses in mucosal tissues such as the skin and female reproductive tract are dominated by local T_RM populations, with limited contribution of the circulating memory compartment ⁵⁰. The cellular origin of secondary T_RM cells at mucosal sites may thus be primarily restricted to pre-existing T_RM populations. However, the potential of T_CM

9

cells to give rise to T_RM populations in non-lymphoid tissues suggests plasticity to re-form CD8⁺ T cell memory beyond the circulating compartment. This ability of T_CM cells may be particularly relevant in the context of establishing resident memory at novel sites without a pre-existing competing T_RM pool. Given the important role of T_RM cells in host protection, it remains to be determined whether these T_CM-derived T_RM populations provide protection against subsequent pathogen exposure. Overall, our results provide new insights into the differentiation potential of memory CD8⁺ T cells during recall responses, which may have important implications for vaccination and cell therapy strategies aiming to induce T_RM populations in mucosal tissues. In light of our findings, these strategies may benefit from specifically engaging naïve or pre-existing T_RM cells to establish lasting T_RM-driven protective immunity at mucosal sites.

Materials & Methods

Study design

The primary aim of this study was to investigate the capacity of circulating memory CD8⁺ T cells to form T_RM cells at different tissue sites after reactivation. To this end, adoptive co-transfer and rechallenge experiments of T_EM and T_CM cells were performed in an LCMV infection setting. The tissue distribution and acquisition of tissue retention molecules by donor T_EM and T_CM cells was assessed in the effector and memory response after viral rechallenge.

The response arising from host naïve CD8⁺ T cells was used as a reference. Finally, the capacity of intestinal and liver T_RM cells to form secondary T_RM cells after adoptive transfer and rechallenge was assessed. Detailed descriptions of the experimental parameters are described below and in the figure legends. Sample sizes, number of experimental replicates, and statistical tests are indicated in the figure legends.

Mice

Wild-type CD45.2 (C57BL/6JRj) and wild-type CD45.1 (B6.SJL-Ptprc^a Pepc^b/BoyJ) mice were purchased from Janvier and from the Jackson Laboratory, respectively. Both lines were crossed to generate CD45.1 x CD45.2 mice. All mice were maintained in the animal facility of the Netherlands cancer institute (NKI) under specific pathogen-free (SPF) conditions. CD45.1 x CD45.2 mice were used as recipients throughout the study. Both female and male mice were used for this study and aged between 8 and 16 weeks at the time of experimentation.

For adoptive transfer experiments, donor and recipient mice were sex-matched. Animals experiments were conducted according to institutional (NKI) and national guidelines.

Human material

Human PBMCs were obtained from fresh heparinized blood or buffy coats of healthy donors using Ficoll-Paque Plus (GE Healthcare) gradient centrifugation. PBMCs were cryopreserved until further analysis. Naïve (T_N), effector memory (T_EM), CD45RA⁺ effector memory (T_EMRA), and central memory CD8+ T (T_CM) cells were isolated using fluorescence-activated cell sorting (FACS) for CCR7 and CD45RA on a FACS Aria III (BD Biosciences) to obtain CCR7⁺ CD45RA⁺ T_N cells, CCR7^- CD45RA^- T_EM cells, CCR7^- CD45RA⁺ T_EMRA cells and CCR7⁺ CD45RA^- T_CM cells.

Blood samples were obtained from anonymized healthy male donors with written informed consent in accordance to guidelines established by the Sanquin Medical Ethical Committee.

In vivo experiments

Adoptive transfer experiments were performed as previously described ⁵¹. In brief, wild-type CD45.1 and CD45.2 mice were infected intraperitoneally with 1 x 10⁵ plaque-forming units (pfu) of LCMV Armstrong. In the memory phase after infection, CD8⁺ TCRγδ^- CD44^high CD69^- CD62L⁺ CX3CR1^- T_CM cells and CD8⁺ TCRγδ^- CD44^high CD69^- CD62L^- CX3CR1⁺ T_EM cells from spleen and CD8⁺ TCRγδ^- CD69⁺ CD62L^- T_RM cells from liver or small intestine intraepithelial lymphocytes (SI IEL) were purified using a FACS Aria III cell sorter (BD Biosciences). To preserve T_RM cell viability, donor mice were injected intraperitoneally with 50 μg of ARTC2.2-

9

blocking nanobody s+16a (BioLegend) 30 min before sacrifice, as published previously ^52,53. The frequency of D^b GP33⁺ cells in purified donor populations was determined to establish a ratio of 1:1 within the D^b GP33⁺ population of co-transferred CD8⁺ T cell subsets. Recipient mice were injected intravenously with 2-4 x 10³ D^b GP33⁺ cells per donor population and infected intraperitoneally with 1 x 10⁵ pfu of LCMV Armstrong two weeks after transfer. At the indicated time points after infection, mice were sacrificed and tissues were collected for analysis of donor CD8⁺ T cell responses.

Isolation of lymphocytes from tissues

Single cell suspensions from spleen, lymph nodes and liver were prepared by mechanical disruption via passing over a 70 μm cell strainer. SI IEL were prepared from the small intestine. After removal of residual fat tissue, Peyer’s patches, and feces, the remaining intestinal tissue was cut into pieces of 0.5 cm and incubated in Hanks’ balanced slat solution (HBSS, Gibco) with 10% fetal calf serum (FCS), 5 mM EDTA, and 1 mM dithiothreitol (DTT) for 30 min at 37°C and vortexted extensively. The IEL fraction was isolated by filtering over a 70 μm cell strainer. To isolate lymphocytes from the kidneys and salivary glands (SG), tissue was cut into pieces and enzymatically digested for 30 min at 37 °C with 375 U ml^-1 Collagenase Type I (Worthington) and 0.15 mg ml^-1 DNase I (Roche, from bovine pancreas, grade II) in RPMI 1640 (supplemented with 10% FCS). The isolated lymphocytes from liver, SI IEL, kidney and SG were purified by density centrifugation on a 66% / 44% Percoll gradient (GE Healthcare). Contaminating erythrocytes were removed using red blood cell lysis buffer (155mM NH₄Cl, 10mM KHCO₃, 0,1mM EDTA).

Flow cytometry

Lymphocytes were labelled with antibodies and tetramers for 25 min at 4 °C. Excess antibodies and tetramers were removed using washing with PBS (supplemented with 0.5%

(v/v) FCS). Antibodies were purchased from BioLegend, eBioscience, or BD Biosciences as indicated in the table below. H-2 D^b KAVYNFATC (GP33) tetramers were kindly provided by R. Arens (Leiden University Medical Center, Leiden). Dead cells were excluded using the live/

dead fixable near-IR dead cell stain kit (Thermo Fisher Scientific). For staining of intracellular molecules, the Foxp3 / Transcription Factor Staining Buffer Set (eBioscience) was used according to the manufacturer’s specifications. Samples were acquired on LSR Fortessa or FACSymphony flow cytometers (BD Biosciences), and data were analyzed using FlowJo V10 software (Tree Star). Cell sorting was performed using Aria III (BD Biosciences).

Antibody Clone Supplier Catalogue N°

CD197 (CCR7) G043H7 BioLegend 353208

CD103 M290 BD Biosciences 563637, 557495

CD103 Ber-ACT8 BioLegend 350206

CD127 A7R34 BioLegend 135027

CD186 (CXCR6) SA051D1 BioLegend 151106, 151109

CD4 GK1.5 BioLegend 100451

CD44 IM7 BD Biosciences 564392