Cross-talk between iNKT cells and CD8 T cells in the spleen requires the IL-4/CCL17 axis for the generation of short-lived effector cells

M. Valentea,b,c,1,2, Y. Dölena,b,1, E. van Dinthera,b, L. Vimeuxd, M. Falletc, V. Feuilletd,1, and C. G. Figdora,b,1,2

aDepartment of Tumor Immunology, Radboud University Medical Center, Radboud Institute for Molecular Life Sciences, 6525 GA Nijmegen, The Netherlands; bRadboud University Medical Center, Oncode Institute, 6525 XZ Nijmegen, The Netherlands; cCentre d’Immunologie de Marseille-Luminy, Aix Marseille University, CNRS, INSERM, 13007 Marseille, France; and dUniversité de Paris, Institut Cochin, INSERM, U1016, CNRS, UMR8104, F-75014 Paris, France

Edited by Barry R. Bloom, Harvard T. H. Chan School of Public Health, Boston, MA, and approved November 5, 2019 (received for review August 21, 2019) Mounting an effective immune response relies critically on the with the Vβ2, Vβ7, or Vβ8 chain in mice (13). iNKT cells spe- coordinated interactions between adaptive and innate compart- cifically recognize lipid such as α-galactosylceramide ments. How and where immune cells from these different compart- (α-GalCer) presented by CD1d molecules expressed by DCs, ments interact is still poorly understood. Here, we demonstrate that , and B cells. Upon activation, they rapidly secrete the cross-talk between invariant natural killer T cells (iNKT) and large amounts of and induce the subsequent activation + CD8 T cells in the spleen, essential for initiating productive immune of different cell types, including DCs, NK cells, and conventional responses, is biphasic and occurs at 2 distinct sites. Codelivery of anti- T cells. In the spleen, iNKT are mainly localized in the red pulp gen and adjuvant to -presenting cells results in: 1) initial short- (RP) and, upon immunization with α-GalCer, they are recruited + lived interactions (0 to 6 h), between CD8 T cells, dendritic cells (DCs), to the marginal zone (MZ), where they get activated by MZ and iNKT cells recruited outside the white pulp; 2) followed by long- + macrophages (MZM) and DCs (14, 15). Previous studies, in- lasting contacts (12 to 24 h) between iNKT cells, DCs, and CD8 T cluding our own work (16, 17), have pointed out that efficient cells occurring in a 3-way interaction profile within the white pulp. iNKT help requires to both cell types from Both CXCR3 and CCR4 are essential to orchestrate this highly dy- the same APC. However, it is still unclear where and when namic process and play nonredundant in memory generation. + iNKT cells and CD8 T cells meet, and whether this process IMMUNOLOGY AND INFLAMMATION While CXCR3 promotes memory T cells, CCR4 supports short-lived occurs concomitantly or sequentially. effector cell generation. We believe our work provides insights into In this study, we aimed at elucidating the spatiotemporal ki- + the initiation of T cell responses in the spleen and their consequences netics of the cross-talk between iNKT cells and CD8 Tcellsin for T cell differentiation. the spleen. We exploited nanovaccines that can efficiently code- liver antigen and adjuvant to APCs. With different molecular, iNKT cells | CD8 T cell priming | CCL17 | CD8 T cell memory | spatiotemporal kinetics Significance + he CD8 T cells are essential for tumor eradication by rec- – Efficient iNKT help require antigen presentation to both the Tognizing peptide MHC-I complexes at the cell surface. In CD8+ T cell and the helper cell interacting with the same DCs. the lymph node (LN), T cell priming occurs in 3 distinct phases However, it remains unclear where and when iNKT cells, DCs, + (1). Briefly, at early stages T cells undergo multiple transient and CD8 T cells meet, but it is rather accepted that these in- encounters with dendritic cells (DCs), followed by long-lasting teractions should occur sequentially. Here, we show that contacts concomitant with production. Then, T cells iNKT cell help occurs in 2 spatiotemporal distinct phases and resume their rapid migration and start proliferating. The estab- involves rather concomitant interactions. Indeed, at 24 h, iNKT lishment of extensive contacts with DCs appears to be essential + cells are massively recruited in the white pulp, and the majority for mounting high-quality CD8 T cell responses, in particular + + of activated CD8 T cells are forming concomitant long-lasting regarding generation (2, 3). CD8 T cell priming interactions with iNKT cells and DCs. This study illustrates the depends on efficient cross-presentation, allowing exogenous importance of simultaneous delivery of antigen and adjuvant antigens to be processed into peptides and subsequently pre- and should be considered in the design of new vaccines or sented by MHC-I molecules. Among conventional DCs (cDC), immunotherapies. cDC1 express XCR1 and CLEC9a surface markers, and are con- sidered to be the major antigen-presenting cell (APC) involved in Author contributions: M.V., Y.D., and C.G.F. designed research; M.V. and Y.D. performed this process in vivo (4). The apparent superior ability of cDC1 research; E.v.D., L.V., M.F., and V.F. contributed new reagents/analytic tools; M.V. and depends on an optimal intracellular machinery adapted for this Y.D. analyzed data; and M.V., V.F., and C.G.F. wrote the paper. pathway (5, 6). Furthermore, their critical positioning within sec- The authors declare no competing interest. ondary lymphoid organs to get direct access to antigens (7, 8), This article is a PNAS Direct Submission. combined with their ability to produce high amounts of IL-12 (9), Published under the PNAS license. explain why cDC1 are so efficient in cross-priming. In addition, Data deposition: All data supporting the findings of this study have been deposited to Fig- + splenic CD169 metallophilic macrophages (MM) or their equiva- share repositories under the DOIs: 10.6084/m9.figshare.10304804.v1; 10.6084/m9.figshare. lent in the LN have been proposed to fulfill this role as well (10, 11). 10305071.v1; 10.6084/m9.figshare.10321715.v1; 10.6084/m9.figshare.10321835.v1; 10.6084/ Next to DCs, helper cells are also important to promote ef- m9.figshare.10322021.v1; 10.6084/m9.figshare.10324433.v1; 10.6084/m9.figshare.10324487. + v1; 10.6084/m9.figshare.10324748.v1; 10.6084/m9.figshare.10338341.v1;and10.6084/ fective cellular immunity by enhancing CD8 T cell clonal ex- m9.figshare.10344524.v1. + pansion, differentiation, and survival (12). Besides CD4 T cells, 1M.V., Y.D., V.F., and C.G.F. contributed equally to this work. invariant natural killer T (iNKT) cells can achieve this task. 2To whom correspondence may be addressed. Email: [email protected] or carl. iNKT cells represent a specialized subset of innate immune cells [email protected]. characterized by the expression of a restricted αβ T cell antigen This article contains supporting information online at https://www.pnas.org/lookup/suppl/ (TCR) repertoire composed of the canonical variable doi:10.1073/pnas.1913491116/-/DCSupplemental. α-region 14/joining α-region 18 (Vα14-Jα18) chain in association

www.pnas.org/cgi/doi/10.1073/pnas.1913491116 PNAS Latest Articles | 1of12 Downloaded by guest on September 23, 2021 functional, and imaging techniques, including live imaging on with minimal amounts of antigen and adjuvant and we have + explanted organs, we reveal that iNKT–CD8 T cell interactions demonstrated that vaccination greatly relies on the codelivery, are biphasic and concomitant in vivo. Early after nanovaccine suggesting that efficient iNKT help requires antigen presentation + + administration, both CD8 T cells and iNKT cells are recruited to to both CD8 T cells and iNKT cells from the same APC (17). areas close to the MZ, where they both interact with DCs or Importantly, we have previously shown that the vaccination ef- + CD169 MM, supporting a concomitant antigen presentation. At ficiency depends solely on the help from iNKT cells and not from + + + later stages of activation, CD8 T cells are recruited in the white CD4 T cells to generate potent OVA-specific CD8 T cells pulp (WP) and initiate long-lasting contacts with DCs and iNKT (17). As very limited information is available on how 200-nm- cells. By combining transcriptomic and microscopic approaches, sized PLGA-based nanovaccines (carrying both antigen and ad- we show that networks involved in this process appear juvant) enter the spleen and which cells are the primary target, to encompass the CXCR3 and CCR4 pathways. These 2 chemo- we first examined the capture of fluorescently labeled PLGA kine pathways play nonredundant roles in the differentiation of nanovaccines in the spleen at early (2 and 6 h) and late time + memory CD8 T cells or short-lived effector cells (SLECs). We points (24 h) following intravenous injection. The capture was believe that this study provides insights in the early stages of rapid and transient, as we were able to detect nanovaccines in + CD8 T cell activation that have important consequences for the the spleen within 2 h (Fig. 1 A and B), whereas at 24 h, fluo- generation of responses and memory formation. rescent nanovaccines were hardly visible (Fig. 1 A and B). This Moreover, our findings highlight the importance of codelivering was also confirmed by flow cytometry (SI Appendix, Fig. S1); + the antigen and the adjuvant to get an efficient iNKT cell help. however, with this technique, CD169 MM could not be assessed + for the quantification. As expected, CD68 macrophages and + Results CD11c DCs formed the main subsets involved in the nano- CD169+ Macrophages and DCs Rapidly Capture Nanovaccine at the vaccine uptake (∼70% and 30%, respectively, at 6 h) (Fig. 1 C MZ. To ensure efficient codelivery of the antigen (ovalbumin, and D). Interestingly, nanovaccines were excluded from the WP, OVA) and the iNKT cell agonist (α-Galcer), we decided to but accumulated at the MZ (Fig. 1A). In accordance, at 6 h coencapsulate these 2 compounds within polylactic-co-glycolic about 50% of all nanovaccines detected in the spleen were taken + + acid (PLGA) nanovaccines. With this approach, we could work up by CD169 macrophages, followed by CLEC9a DCs (cDC1)

+ Fig. 1. Nanovaccines are rapidly captured at the MZ by CD169 macrophages and DCs. NP containing OVA, α-Galcer, or TLR-L (Poly IC + R848) and ATTO 647 (fluorescent dye) were intravenously administered in mice. At 0, 2, 6, or 24 h, mice were killed, and spleens were isolated and fixed in order to perform cryosections (10 μm) for subsequent immunofluorescent staining. (A) Representative confocal pictures at the different time points. The WP was drawn in orange based on CD169 staining. (B) Quantification of fluorescent intensities (FI) from ATTO 647-coupled NP. (C) Cryosections were stained with different cell markers, including CD11c (DC), CLEC9a (cDC1 DC), CD68 (pan macrophages), CD169 (metallophillic macrophages), and SIGNR1 (MZMs). (D) ATTO 647 FI colocalizing with the different cell types were divided by the ATTO 647 FI obtained from the whole section. Quantification at 2- and 6-h time points are depicted. Results are derived from 2 independent experiments involving at least 10 sections per condition. Statistical analysis by 1-way ANOVA test:*P < 0.05, ****P < 0.0001; mean ± SEM. (Scale bars, 50 μm.)

2of12 | www.pnas.org/cgi/doi/10.1073/pnas.1913491116 Valente et al. Downloaded by guest on September 23, 2021 + and SIGNR1 macrophages (Fig. 1 C and D). Importantly, the (CD69 expression) and the nature of adjuvants carried by the nature of the adjuvant encapsulated in the PLGA nanovaccine nanovaccine. Performing a nonsupervised clustering, we observed + did not affect capture by distinct APC subsets, since nano- that OT-I CD8 T cells were clustered together, and depended vaccines containing TLR-L (PolyIC+R-848) instead of α-Galcer more on the type of adjuvant used than on their activation status exhibited the same uptake pattern (Fig. 1D). Thus, nanovaccines (Fig. 2A). This was also true for DCs, whose cytokine/chemokine are captured by both DCs and macrophages located in the MZ profiles demonstrated a clustering based on the type of adjuvant during the first hours after intravenous administration. rather than on the cDC subset (Fig. 2B). A closer look at the data showed that different chemokine pathways were triggered by the α-Galcer–Carrying Nanovaccines Induce CCL17 and CXCL9 Production + different types of adjuvants (SI Appendix,TableS1). We thus by CD8 T Cells and DCs. We next evaluated whether the iNKT cell adjuvant α-Galcer could induce the production of specific sets of decided to explore this phenomenon in more detail by performing when compared to more common adjuvants such as qRT-PCR on the different cell subtypes. CCR4 ligands (CCL17 TLR-L. To this end, mice were vaccinated and 6 h later different and CCL22) and CXCR3 ligands (CXCL9 and CXCL10) were + − α C D cell subsets were sorted by flow cytometry (CD69 and CD69 specifically induced by -Galcer nanovaccine (Fig. 2 and ). In + + OVA-specific CD8 T cells [OT-I] and cDC subsets, XCR1 DC particular, the chemokines CCL17 and CXCL9 were up-regulated + [cDC1] and CD11b DC [cDC2]) (SI Appendix, Fig. S2). Sub- by α-Galcer in OT-I T cells and cDCs. In sharp contrast, CCL3 sequently, mRNA expression of multiple cytokines, chemokines, and CCL4 expression, both CCR5 ligands, were specifically in- and their receptors was measured on sorted populations. Exploiting duced in activated OT-I T cells of mice that received TLR-L– microarrays, we assessed the expression profile of chemokines and carrying nanovaccines (Fig. 2 C and D). Finally, IL-12 pro- + cytokines in OT-I CD8 T cells, based on their activation state duction, which is important for the initiation of cytotoxic responses, IMMUNOLOGY AND INFLAMMATION

+ Fig. 2. iNKT–CD8 cross-talk induces specific sets of chemokines in different cell subsets. CD8 OT-I yeti T cells were isolated, labeled with CTV dye, and adoptively transferred prior to vaccination. One day later, nanovaccines containing OVA, α-Galcer, or TLR-L (Poly IC + R848) were intravenously administered in mice. At 6 h, mice were killed, spleens harvested, and OT-I T cells or endogenous DC subsets were FACS-sorted. Cells were lysed in TRIzol, total RNA was isolated by choloroform extraction, and mRNA was preamplified and used in subsequent qRT-PCR. (A and B) Clustergram obtained from a array focusing on chemokine and for OT-I cells (A) or from a gene array on chemokines and cytokines for DC subsets (B). n = 2. (C–E) Results from different manual qRT-PCR performed from mRNA of OT-I T cells or DC subsets. n > 12. Statistical analysis by 1-way ANOVA test: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; mean ± SEM.

Valente et al. PNAS Latest Articles | 3of12 Downloaded by guest on September 23, 2021 + couldbedetectedinXCR1 DC only after administration of were initially detected, they were also found in the WP at later nanovaccine containing α-Galcer but not with TLR-L (Fig. 2E). stages (24 h). We showed by flow cytometry that both cDC1 and Altogether, these results suggest that iNKT cell help through cDC2 produce these chemokines (SI Appendix, Fig. S3 A and B), α-Galcer stimulation has a major impact on splenic chemokine confirming our qPCR results. While CXCL9 appeared to be networks by specific induction of CCL17 and CXCL9 expression in expressed by multiple cell types, CCL17 expression was detected multiple immune cell types. on fewer cells (Fig. 3 A and D), mainly restricted to cDC pop- ulation (SI Appendix, Fig. S3 C and D). In conclusion, CXCL9 CXCL9 and CCL17 Expression Patterns Are Dynamic over Time in the and CCL17 expression is highly dynamic. Importantly, at the Different Spleen Compartments. Since we found in various cell early stages, these molecules are produced in the MZ and RP types that iNKT cells specifically induce the expression of CCL17 and only start to be expressed in the WP at later stages. This may and CXCL9 at mRNA levels, we next sought their level have important consequences for the migratory behavior of distribution within the tissue by confocal microscopy. We con- the cells. firmed the induction of CXCL9 protein expression upon α-Galcer + CD8 T Cell Localization in the Spleen Is Biphasic during Early Stages administration, which is increased over time (Fig. 3 A and B). of Activation. Following the cues of T cell-attracting chemokines, CCL17 protein expression was also induced 6 h after α-Galcer we assessed whether T cells were following a similar path. The nanovaccine injection, although its level remained stable during + D E localization of antigen-specific OT-I CD8 T cells was tracked 24 h (Fig. 3 and ). While expression of both chemokines was over time by confocal microscopy. As expected, OT-I T cell be- restricted to the MZ and RP at early stages (6 h) where nanovaccines havior was also highly dynamic early after nanovaccine adminis- tration in accordance with the chemokine profiles (Fig. 4A). Interestingly, within the first 6 h, almost half of OT-I T cells ac- cumulated at the MZ and in the RP where CXCL9 and CCL17 were detected exclusively (Figs. 3 A–D and 4B). At this stage, OT-I T cells started to get activated, as evidenced by CD69 expression (Fig. 4 C and D). In addition, we evaluated the distance between + recruited OT-I T cells and DCs or CD169 macrophages, and we + + found that they were closer to CD11c DC and CLEC9a DC + than to CD169 macrophages (SI Appendix, Fig. S4A). This sug- + gests that CD8 T cells were recruited toward DCs rather than to + CD169 macrophages during this time frame. Later, from 12 h on, when chemokines became expressed in the WP (Fig. 4 A, B,and D), OT-I T cells were redirected and during this second phase were mainly found in the WP forming stable T–T clusters together + + with CD11c DC (Fig. 4 E and F)andCLEc9a DC but not with + CD169 macrophages (SI Appendix,Fig.S4B). At this time, they also started producing IFN-γ and CD69 expression reached a level of ∼90% at 24 h (Fig. 4 C–E). Finally, we questioned the origin of OT-I T cells accumulating outside the WP at 6 h, which can come either from the WP or from the RP. To do so, we compared the location of OT-I T cells that have been injected 1 d before the vaccination or 3 h after (referred to as early migrating OT-I), since it has been shown that 3 h was enough for intravenously injected T cells to home to the WP (18). To accurately quantify this, we compared the ratio of “OT-I/early migrating OT-I” in the WP between control and vaccinated mice. As depicted in SI Appendix, Fig. S4C, we found that the main origin appears to be from the RP, suggesting that T cell trafficking is transiently affected upon nanovaccine administration. To substantiate these findings, we next studied the migratory + behavior of antigen-specific CD8 T cells within various splenic compartments. Since intravital microscopy for the spleen is ex- tremely challenging (19), we opted for an explanted organ ap- proach using perfused thick sections of spleen for live imaging. During early stages after vaccine delivery (2 to 6 h), we observed that OT-I T cells kept their normal high-speed motility of around Fig. 3. CXCL9 and CCL17 are exclusively expressed in the RP and in the MZ at μ A B early stages of activation. Nanovaccines containing OVA and α-Galcer were 7 m/min in the WP as at the steady state (Fig. 5 and and intravenously administered in mice. At 0, 6, and 24 h, mice were killed, spleens Movie S1). In the MZ and the RP, OT-I T cells exhibited a harvested, and fixed in order to perform cryosections (10 μm) for subsequent somewhat slower speed with a mean velocity of 5 μm/min (Fig. 5 C immunofluorescent staining. Cryosections were stained with either anti-CXCL9 and D and Movie S2). This slowing could result from repetitive or anti-CCL17 and anti-CD169 antibodies. (A) Representative im- short encounters with APCs. This notion was supported by the ages of CXCL9 (green) and CD169 (red) costaining at the different time points. finding that in the absence of OVA antigen or with polyclonal 2 + (B) Quantification of CXCL9 FI per square micrometer (FI/μm ) of the whole CD8 T cells, the velocity was slightly but significantly higher in μ 2 section. (C) Quantification of CXCL9 FI/ m in each spleen compartment. (D) those regions during this time frame (Fig. 5 C and D). During the Representative images of CCL17 (green) and CD169 (red) costaining at the different time points. (E) Quantification of CCL17 FI/μm2 for the whole section. second phase (12 to 24 h), when T cells were redirected to the WP, μ 2 OVA-primed T cells significantly slowed down their migratory (F) Quantification of CCL17 FI/ m for each spleen compartment. Results are A B derived from 3 independent experiments. Statistical analysis by 1-way ANOVA behavior (Fig. 5 and ), reflecting the formation of stable test for B and E and by t test for C and F:*P < 0.05, **P < 0.01, ***P < 0.001, conjugates together with DCs (Fig. 4 E and F and Movie S3). ****P < 0.0001; mean ± SEM. (Scale bars, 50 μm.) Altogether, these results demonstrate that antigen-specific

4of12 | www.pnas.org/cgi/doi/10.1073/pnas.1913491116 Valente et al. Downloaded by guest on September 23, 2021 Fig. 4. OT-I T cells are transiently recruited outside the WP during the early phases of activation. CD8+ OT-I yeti T cells were isolated, labeled with CTV dye, and adoptively transferred prior vaccination. One day later, nanovaccines containing OVA, and α-Galcer were intravenously administered in mice. At different time points, mice were killed, spleens harvested, and fixed in order to perform cryosections for subsequent immunofluorescent staining. Cryosections were stained with either anti-CD169 or anti-CD11c antibodies. Localization of OT-I T cells was evaluated by confocal microscopy. (A) Representative images from 0-, 6-, and 24-h time points. OT-I are represented in white and CD169 staining in red. (Scale bars, 50 μm.) (B) Quantification of OT-I localization in the WP at the

different time points. Pool from 3 independent experiments. (C and D) CD69 expression on CTV-labeled OT-I T cells at 0, 6, and 24 h analyzed by flow IMMUNOLOGY AND INFLAMMATION cytometry. (C) Representative flow plots. (D) Pool from 3 independent experiments. (E) Representative images of OT-I DC clusters present at 12- and 24-h postvaccination. OT-I are depicted in white, CD11c signal in red and IFN-γ signal derived from YFP (yeti) in green. (Scale bars, 25 μm.) (F) Quantification of OT-I clusters (2 OT-I cells in contact or more) at the different time points. Statistical analysis by 1-way ANOVA test: ***P < 0.001, ****P < 0.0001; mean ± SEM.

+ CD8 T cells exhibit a biphasic behavior, with a first transient regulated at the mRNA level but was not detectable at the cell accumulation at the MZ and the RP early after nanovaccine surface (Fig. 6A and SI Appendix, Fig. S3B). Perhaps engagement administration, where they interact shortly with DCs, and at later of the receptor by its ligand induces its internalization or pre- + stages with the recruitment of CD8 T cells in the WP, with long- vents its recognition by the when the epitope is shielded. lasting contacts involving multicellular clusters with DC. At 24 h, the proportion of OT-I T cells expressing CXCR3 strongly increased and CCR4 expression was detected on a + CCR4 and CXCR3 Are Essential in Early CD8 T Cell Activation. Since fraction of activated CD69 OT-I T cells (Fig. 6A). We then our results point out that both CCL17 and CXCL9 are important evaluated the contribution of these chemokine receptors in vivo chemokines in our vaccination strategy, we decided to evaluate for early T cell activation and recruitment in the WP. In- the expression of their respective receptors, CCR4 and CXCR3, terestingly, we found that blocking each of them impacted early on OT-I T cells. T cell activation monitored by CD69 expression 6 h after vac- At the steady state, a fraction of naive OT-I T cells already cination (Fig. 6B). Furthermore, we saw that the recruitment of harbors CXCR3 at the cell surface, thus rendering these cells re- OT-I T cells in the WP was negatively impacted by CXCR3 or sponsive to CXCL9 at very early stages. In contrast, CCR4 could CCR4 blockade, as depicted in Fig. 6C. Altogether, these results not be detected at the cell surface of naive cells and became only show that the expression of CXCR3 and CCR4 on antigen- up-regulated after vaccination (Fig. 6A). In order to reveal which specific T cells is triggered upon vaccination and participate in factors might induce its expressiononthecellsurfaceovertime,we their activation and recruitment. cultured in vitro purified OT-I T cells in the presence of IL-4 combined or not with OT-I–specific peptide (SIINFEKL) to trig- iNKT Cross-Talk with CD8+ T Cells Involves Concomitant Interactions + ger T cell activation and monitored CCR4 expression by flow in 2 Distinct Phases. Since CD8 T cells were rapidly recruited to cytometry. We found that only the combination of IL-4 and the MZ and the RP after nanovaccine administration, we hy- SIINFEKL induced a strong expression of CCR4, while cells pothesized that they could encounter already activated iNKT cells exposed to either of them alone only induced a mild expression in these areas during this first phase. Therefore, we first examined (SI Appendix, Fig. S5A). CCR4 expression was rapidly induced whether iNKT cells were activated within the first 6 h upon since it could already be detected from 6 h on, although still nanovaccine administration (SI Appendix,Fig.S6A). As expected, higher expression levels were seen at 24 h. Coming back to we found that most of the iNKT cells were indeed activated in vivo vaccination setting, we were able to detect surface ex- according to their CD69 expression. pression of CXCR3 on OT-I T cells at early stages of vaccination We then attempted to directly visualize interactions between (6 h). However, the levels of mRNA expression and protein were OT-I T cells and iNKT cells by confocal microscopy. Adoptive lower when compared to the control (Fig. 6A and SI Appendix, transfer of highly purified and fluorescently labeled iNKT cells + Fig. S5B). Nonetheless, the vast majority of detected CXCR3 was not an appropriate method since it did not result in sufficient OT-I T cells expressed CD69, suggesting an important role of numbers in the spleen for obtaining solid data. A better approach this chemokine receptor to recruit antigen-specific T cells in the was to stain endogenous iNKT cells with a CD1d–α-Galcer early stages of activation. At 6 h, the expression of CCR4 was up- dextramer and to image OT-I T cell/iNKT cell interactions in fixed

Valente et al. PNAS Latest Articles | 5of12 Downloaded by guest on September 23, 2021 phase (24 h), most of them were found in the WP where NKT- NKT clusters could still be observed. By evaluating the interac- tions between NKT cells and OT-I T cells, we were able to reveal close contacts (less than 3 μm) between NKT cells and OT-I T cells together with DCs (Fig. 7 E and F). Indeed, while ∼20% of OT-I cells were found in close contact with NKT cells in the MZ and the RP in the first phase (6 h), almost half of OT-I cells were found in concomitant interactions in the WP in the second phase (24 h) (Fig. 7E). To establish the identity of CXCR6-GFPbright cells, we con- + firmed that they were NOT CD8 T cells by immunostaining, as illustrated in the SI Appendix,Fig.S7B.Ofcourse,wecannot guarantee that the observed contacts are all with-type I NKT cells (iNKT cells), since they represent 60% of the CXCR6-GFPbright cell population (forming the main population), but we assume that in the majority of the quantified contacts, iNKT cells are involved. By live imaging, we quantified the duration of these contacts and found that these interactions were short and independent of the presence of the OVA antigen at the early stages (Fig. 7G and Movie S4). However, at the late stages, we observed long-lasting interactions with about half of the contacts that lasted at least Fig. 5. OT-I T cells form long-lasting contacts with DC in the WP 24-h 30 min (duration of the entire movie shown in Movie S5). To + postvaccination. CD8 OT-I yeti T cells were isolated, labeled with CFR dye, exclude that the observed interactions could only be attributed by and adoptively transferred prior vaccination. The next day, nanovaccines the high density of both OT-I and NKT cells in the WP at the late containing OVA and α-Galcer were intravenously administered in mice. At stages, we randomly simulated the position of NKT cell in these different time points, mice were killed, spleens harvested, and embedded in regions. As shown in Fig. 7H, we found that the distance between a low-melting agarose gel. Thick sections of 500 μm were performed using OT-I and NKT cells was closer in reality than when distributed vibratome and stained with anti-CD169 and anti-CD11c antibodies. Live randomly, supporting unambiguously a true connection between imaging was performed using a spinning-disk microscope equipped with a thermostated chamber and perfused at a rate of 0.8 mL/min with medium NKT and OT-I T cells. Finally, we addressed the role of CXCR3 and CCR4 in the re- bubbled with 95% O2 and 5% CO2. OT-I migration was evaluated on movies lasting 30 min. (A and C) OT-I T cell migration tracks derived from 30-min cruitment of NKT cells in the WP at the late stages and found, movies. Depicted tracks (n > 30) start from the same origin. Axes bars rep- surprisingly, that CXCR3 and CCR4 have a differential role in this resent the scale in microns. (B and D) Quantification of the OT-I T cell mean process (Fig. 7I). Indeed, CXCR3 plays a positive role in this process velocity calculated in micrometers per minute. Results are derived from 4 while, in contrast to T cells, CCR4 negatively impacts this migration. independent experiments and involve at least 2 mice per condition and time Based on these findings, we demonstrate that NKT cells are first point. Statistical analysis by 1-way ANOVA test: **P < 0.01, ****P < 0.0001; recruited in the MZ and in the RP, where they get activated and can ± mean SEM. interact shortly with OT-I T cells. In the second phase, they follow the behavior of OT-I T cells by being recruited toward the WP, tissue sections. As depicted in SI Appendix,Fig.S6B, close inter- where they form concomitant long-lasting interactions with DCs. actions between OT-I T cells and iNKT cells could be observed in + The IL-4/CCL17/CCR4 Axis Regulates CD8 T Cell Differentiation into the MZ and in the RP. By quantifying the distance between OT-I SLECs. We next addressed the functional impact of CXCR3 and + and iNKT cells in these regions, we could demonstrate that these CCR4 on the generation of CD8 T cell memory. In particular, close interactions only occurred from 6 h after vaccination (SI Appendix C we focused on CCR4 since the role of this chemokine receptor in ,Fig.S6 ). Thus, while OT-I T cells accumulated at the this process was not really explored. MZ and in the RP, they could form concomitant interactions First, we aimed to confirm in vivo the link between IL-4 and the together with iNKT cells. CCR4 pathway, as we previously showed in vitro in SI Appendix, Such an approach was unfortunately limited by the rapid down- Fig. S5A. As expected, we found that IL-4 was only detected in the regulation of TCR expression on iNKT cell surface triggered by sera of animals administrated with nanovaccines containing an their activation, rendering the quantification difficult after 6 h. To iNKT cell agonist (Fig. 8A). Maximal production was found in the circumvent this drawback and to further study these interactions early stages (6 h) when OT-I T cells accumulate outside the WP. over prolonged periods of time, we exploited CXCR6-GFP mice, Interestingly, we could detect IL-4 signaling in OT-I T cells by which have been already used in intravital microscopy studies to pSTAT6 staining only during this time frame and not in the late track NKT cells in the liver (20). As depicted in Fig. 7A,inthese B bright stages (Fig. 8 ). We then evaluated the impact of IL-4 blockade mice, CXCR6-GFP cells represent a homogenous population in vivo on the expression of CCR4 on OT-I T cells and CCL17 in the spleen and the vast majority of them are NKT cells (based production in the spleen. We found that IL-4 blockade strongly on coexpression of CD3 and NK1.1). Based on CD1d–α-Galcer reduced CCR4 levels on OT-I T cells (Fig. 8C)andwassuffi- dextramer binding, 62% of this population can be designated as cient to abolish the production of CCL17 in the spleen (Fig. type I NKT cells or iNKT cells (CD1d-restricted NKT cells rec- 8D), thus confirming the key role of IL-4 in the induction of the ognizing α-Galcer) and the remainder of the cells are most likely CCR4 pathway. type II NKT cells, which do not bear a specific marker. Further- We then studied the consequence of either CXCR3 blockade more, upon vaccination, the CXCR6-GFPbright cell population was or IL-4/CCL17 blockade at the time of the vaccination on the not increasing and remained representing NKT cells (SI Appendix, generation of memory precursor cells (MPEC) or SLEC 10 d Fig. S7A). Assessment of NKT cell distribution in these mice postnanovaccines administration. Interestingly, we found that revealed that their localization followed the same biphasic dy- blocking CXCR3 drastically reduced the numbers of antigen- namics as OT-I T cells (Figs. 4B and 7 B and C). During the first specific T cells at day 10 postvaccination (Fig. 8E). This can phase (6 h), they were localized to the MZ and RP where they mainly be explained by the strong reduction of the generation of formed NKT–NKT clusters (Fig. 7D), whereas during the second MPEC in these conditions (Fig. 8F). As a consequence, a higher

6of12 | www.pnas.org/cgi/doi/10.1073/pnas.1913491116 Valente et al. Downloaded by guest on September 23, 2021 IMMUNOLOGY AND INFLAMMATION

+ Fig. 6. CCR4 and CXCR3 expressions increase over time at the cell surface upon vaccination and play a role in early T cell activation. CD8 OT-I T cells were isolated, labeled with CFR dye, and adoptively transferred prior vaccination. The next day, nanovaccines containing OVA and α-Galcer were intravenously administered in mice. At different time points, mice were killed, spleens harvested, stained, and analyzed by flow cytometry. (A) Representative expression of CXCR3, CCR4, and CD69 on OT-I T cells in control or vaccinated mice at 6 and 24 h. Results are representative of 2 independent experiments. (B and C) One day prior to vaccination, anti-CXCR3 blocking antibodies or CCR4 antagonist was injected intraperitoneally. (B) Six hours after nanovaccine administration, spleens were harvested and CD69 expression on CFR-labeled OT-I T cells were evaluated by flow cytometry. Representative experiment (Left). Percentage of inhibition + + is indicated as the percentage of CD69 OT-I T cells in blocking antibody-treated cells relative to the percentage of CD69 OT-I T cells in vehicle-treated cells (Right). Pool from 3 independent experiments for CXCR3 blockade and 2 independent experiments for CCR4 blockade. Statistical analysis by t test: *P < 0.05; mean ± SEM (C) Twenty-four hours after nanovaccine administration, spleens were harvested and fixed in order to perform cryosections for subsequent immunofluorescent staining. Cryosections were stained with anti-CD169 antibody and OT-I density (cells/mm2) in the WP was assessed by confocal microscopy. (Scale bars, 50 μm.) Pool from 2 independent experiments. Statistical analysis by 1-way ANOVA test: **P < 0.01, ***P < 0.001; mean ± SEM.

proportion of SLEC was found. Conversely, blockade of either OT-I T cells (Fig. 9D). More importantly, we saw that only the CCL17 or IL-4 did not negatively impact the numbers of antigen- absence of the phase 1 is enough to nearly abolish the generation specific T cells at day 10, with even a slight tendency to increase of SLEC (Fig. 9E) and that phase 2 did not give any additive (Fig. 8G). Importantly, the blockade of each of these molecules effects. Thus, as predicted, the early cross-talk (phase 1) between reduced the generation of SLEC and increased the proportion of iNKT and CD8 T cells is crucial for the efficiency of the nano- MPEC in a completely opposite fashion to CXCR3 (Fig. 8H). vaccine and the generation of SLECs. Thus, CXCR3 and CCR4 have clear differential roles in the To summarize, we propose a model in which iNKT–CD8 T cell generation of MPEC or SLEC, respectively. cross-talk occurs in 2 spatiotemporal phases and involves simulta- neous interactions between 3 cell types (SI Appendix,Fig.S8).The + Early Cross-Talk between iNKT and CD8 T Cells Impacts the Generation first phase (2 to 6 h postvaccination) occurs in the MZ and the RP of Both MPEC and SLEC. Since the IL-4/CCL17/CCR4 axis is crucial for where recruited OT-I T cells interact shortly with activated the generation of SLEC and IL-4 is only produced during the early iNKT cells and DC. During the second phase (12 to 24 h), OT-I stages (phase 1), we hypothesized that early cross-talk between iNKT T cells, iNKT cells, and DC establish long-lasting contacts forming + and CD8 T cells might be important for the generation of SLEC. a “ménage à trois” in the WP. In this process, CXCR3 and CCR4 To test that, we unsynchronized the delivery of antigen and the play a differential role. The IFN-γ/CXCR3 pathway will be in- iNKT cell agonist by administrating nanovaccines (nanoparticles, volved in the generation memory CD8 T cells, while the IL-4/CCR4 NP) containing OVA at the same time, 10 h after (postphase 1) or pathway will regulate the generation of SLECs. We believe that our 48 h after (postphases 1+2) the injection of NP containing study highlights the importance of iNKT cells as helper cells to α-Galcer, as illustrated in Fig. 9A. We first evaluated the expan- orchestrate the complex dynamic process of cross-priming in the sion of antigen-specific T cells 10 d after antigen administration spleen and its importance for proper memory T cell generation. and we found that both phases are crucial in this process (Fig. 9B). Then, we studied as previously the differentiation of antigen- Discussion specific T cells into MPEC or SLEC at day 10 (Fig. 9C). Inter- Here we report that CD8-iNKT cell cross-talk occurs in 2 spa- estingly, we found that both phases are important for the generation tiotemporal distinct phases in the spleen. During the first 6 h, + of MPEC when we looked at the relative numbers of MPEC following the capture of nanovaccines by DCs and CD169

Valente et al. PNAS Latest Articles | 7of12 Downloaded by guest on September 23, 2021 Fig. 7. NKT cells make long-lasting contacts together with OT-I T cells and DC at late stages of activation. (A) Single-cell suspension of splenocytes from + CXCR6-GFP mice were stained with CD8, CD3, and NK1.1 antibodies, and Cd1d/α-Galcer dextramer and analyzed by flow cytometry. (B and F) CD8 OT-I T cells were purified, labeled with CFR dye, and adoptively transferred in CXCR6-GFP mice prior to vaccination. One day later, nanovaccines containing OVA and α-Galcer were intravenously administered in CXCR6-GFP mice. At 0, 6, and 24 h, mice were killed, spleens harvested, and fixed in order to perform cryosections (20 μm) for subsequent immunofluorescent staining. Cryosections were stained with either anti-CD169 or anti-CD11c antibodies. (B) Representative images of the CD169 (red) and CXCR6-GFP (green) costaining at the different time points. (Scale bars, 50 μm.) (C) Percentage of the CXCR6hi cell localized in the WP at the different time points. (D) Percentage of the clustering of CXCR6hi cells at the different time points. (E) Percentage of OT-I T cells in close contacts (distance inferior to 3 μm) with CXCR6hi cells in the MZ/RP or in the WP at the different time points. (F) Representative images of OT-I (white), CD11c (red), and CXCR6- + GFP (green) costaining at 24 h in the WP. (Scale bar, 10 μm.) (G) CD8 OT-I T cells were purified, labeled with CFR dye and adoptively transferred in CXCR6-GFP mice prior to vaccination. Sixteen hours later, nanovaccines containing OVA and α-Galcer or only α-Galcer were intravenously administered in CXCR6-GFP mice. Six and 24 h later, mice were killed, spleens harvested, and embedded in a low-melting agarose gel. Thick sections of 500 μmwereperformedusinga vibratome. and stained with anti-CD169 and anti-CD11c antibodies. Live imaging was performed using a spinning-disk microscope equipped with a thermostated

chamber and perfused at a rate of 0.8 mL/min with medium bubbled with 95% O2 and 5% CO2. OT-I and NKT contacts were evaluated on movies lasting 30 min. (H) From analyzed images for the 24-h time point, the positions of NKT and OT-I T cells in the WP were extracted. For each picture, 20 runs of simulation were performed in order to randomly change the location of NKT cells. The closest distance between OT-I and NKT cells was measured for every OT-I T cell and the mean of the closest distance between these 2 cell types was calculated and then compared to the 1 from the original picture. (I) One day prior to vaccination, anti- CXCR3 blocking antibodies or CCR4 antagonist was injected intraperitoneally into CXCR6-GFP mice. Then, nanovaccines containing OVA and α-Galcer were in- travenously administered, and spleens were harvested 24 h after vaccination, and fixed in order to perform cryosections (20 μm) for subsequent immunofluo- rescent staining with anti-CD169 antibody. NKT cell density (cells/mm2) in the WP was calculated. Results involve at least 2 mice per condition and time point. Statistical analysis by 1-way ANOVA test for C, D, G,andI,andbyt test for E and H:*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; mean ± SEM.

+ macrophages at the MZ and in the RP, CD8 T cells and As previously described, α-Galcer adjuvant exploits an alter- + + iNKT cells accumulate in these regions. Recruited CD8 T cells native pathway for CD8 T cell priming, by privileging CCR4 interact with APCs and concomitantly with iNKT cells. Sub- ligands instead of CCR5 ones, in contrast to other adjuvants, + sequently, from 12 h after vaccine administration, CD8 T cells such as TLR-L (16). Our work confirms and extends this ob- are redirected to the WP where they form long-lasting clusters servation. Next to CCR4/CCL17, we identified the CXCL9/ with DC and iNKT cells. We found that CXCR3 and CCR4 are CXCR3 pathway as equally important for cross-priming in the the key chemokine receptors involved in this complex choreog- context of α-Galcer. Although CXCL9 has been associated with + raphy and these molecules, respectively, promote the generation early CD8 T cell memory responses (21), our results pinpoint + of MPEC and SLEC. that, in the first place, a fraction of naive CD8 T cells already

8of12 | www.pnas.org/cgi/doi/10.1073/pnas.1913491116 Valente et al. Downloaded by guest on September 23, 2021 express the receptor CXCR3, and moreover, its expression can be increased overtime, as others have described (22). Second, its ligand CXCL9 can be strongly induced not only during recall responses, as has been described (21, 23, 24), but also during primary responses when α-Galcer is used as an adjuvant, as we currently show. Importantly, we report also that CCL17 and CXCL9 expression patterns are very dynamic and follow DC and + CD8 T cell redistribution. These 2 chemokines are produced outside the WP during the first phase and then partly relocalized into this compartment at later stages. This likely can be explained by DCs that produce CCL17 and CXCL9, while mi- grating from the RP and MZ to the WP, driven by CCR7 up- regulation (25). In this study, we show that these 2 chemokine pathways play a differential role in promoting T cell memory. In + particular, the role of CCR4 in the CD8 T cell memory gen- eration has never been really evaluated, to our knowledge. In this study, we found that CXCR3 promotes memory T cell genera- tion, whereas CCR4 promote the T cell differentiation into SLECs. Importantly, we obtained similar results with IL-4 block- ade, suggesting that CCR4 might be important to recruit antigen- specific T cells close to IL-4–producing cells in order to promote the generation of SLECs. Conversely, CXCR3 might be important for the T cell recruitment close to IFN-γ–producing cells, pro- + moting thus the generation of memory CD8 Tcells. A remaining question is, which cells are responsible for the cross-priming? We saw that nanovaccines were rapidly captured + at the MZ and in the RP, mainly by CD169 macrophages and DCs present in these areas. This was not surprising since these cells have previously been shown to capture a wide array of IMMUNOLOGY AND INFLAMMATION antigen types (24, 26–29). cDC1 are considered to be the main cell type involved in cross-priming in vivo (30, 31). However, + under certain circumstances CD169 MM have been shown to fulfill this role as well (10, 11). At this time, it is not clear whether these cells directly cross-prime or transfer the antigen to cDC1 (32). We were not able to address this question in our + study since CD169 isolation was not satisfactory. Indeed, as described by Gray et al. (33), CD169 molecules ap- pear to be acquired by a vast array of cell types after organ di- gestion both in the LN (25, 34) and in the spleen (35), rendering + the purification of pure CD169 MM unrealizable. We have observed similar features, and thus we could not rely on the apparent CD169 staining by flow cytometry. This is an important open issue in particular because numerous studies have analyzed this population based on CD169 expression. Nonetheless, our Fig. 8. IL-4/CCL17 axis promotes the generation of short-lived effector cells. results suggest that cDC1 are primarily responsible for cross- (A) nanovaccines containing OVA and α-Galcer or TLR-L were intravenously priming. This notion is supported by the finding that, in early administered in mice. At different time points, mice were killed, sera were + + stages, CD8 T cells are recruited much closer to cDC1 than collected, and ELISA for IL-4 was performed. (B) CD8 OT-I T cells were iso- + CD169 MM, and at later stages, long-lasting contacts involve lated, labeled with CFR, and adoptively transferred prior to vaccination. The + + next day, nanovaccines containing OVA and α-Galcer were intravenously DCs but not CD169 MM in the WP. However, CD169 MM administered in mice. At 6 or 24 h, mice were killed, spleens harvested, might represent a nonnegligible source for chemokines, such as stained, and analyzed by flow cytometry. Representative expression of CXCR3 ligands. pSTAT6 on OT-I T cells in control or vaccinated mice at 6 and 24 h. Results are Another important question is, how is cross-talk is orches- + + representative of 2 independent experiments. (C and D) CD8 OT-I T cells trated between CD8 T cells and iNKT cells? Does it involve + were isolated, labeled with CFR, and adoptively transferred prior to vacci- direct contact between 3 different cell types (DCs, CD8 T cells, nation. One day prior to vaccination, isotype or anti–IL-4 blocking antibodies were injected intraperitoneally. The next day, nanovaccines containing OVA and iNKT cells) or sequential spatiotemporal interactions? Our and α-Galcer were intravenously administered in mice and spleens from the data support the idea that concomitant interactions can occur 24-h time point were harvested. (C) Expression of CCR4 on OT-I T cells was throughout the 2 stages moving from the RP toward the WP. evaluated by flow cytometry. (D) Expression of CCL17 in spleen cryosections Although we cannot formally exclude that sequential interac- (20 μm) was evaluated by confocal microscopy. (Scale bars, 50 μm.) (E–H) tions may happen as well, we show that half of the OT-I T cells + One-thousand CD8 OT-I T cells were adoptively transferred prior to vacci- are in close contacts with iNKT cells during late stages, sug- nation into CD45.1 mice. One day prior to vaccination, isotype, anti-CXCR3 gesting that direct concomitant help might represent the pre- (E and F), or anti-CCL17 or IL-4 antibodies (G and H) were injected in- dominant phenomenon. This is in line with what is known for α + traperitoneally. The next day, nanovaccines containing OVA and -Galcer CD4 T cell help where concomitant interactions occur between were intravenously administered in mice and 10 d later spleens were har- + + vested and analyzed by flow cytometry. MPEC were defined as CD127hi CD8 and CD4 T cells on the same DC during late stages (36). − + KLGR1 cells and SLEC as CD127lo KLGR1 cells. Pool from 2 independent However, in the first phase, only a fraction of OT-I T cells were experiments with at least 5 mice per condition. Statistical analysis by 1-way found in close proximity to iNKT cells and this raises the ques- ANOVA test for A and t test for C–H:*P < 0.05, **P < 0.01, ***P < 0.001; tion of the importance of these early concomitant interactions. mean ± SEM. Nevertheless, we show in this study that the early cross-talk is

Valente et al. PNAS Latest Articles | 9of12 Downloaded by guest on September 23, 2021 + Fig. 9. Early cross-talk between iNKT and OT-I T cells is crucial for the generation of short-lived effector cells. CD8 OT-I T cells were isolated and 1.000 cells were adoptively transferred prior to vaccination. The next day, nanovaccines containing α-Galcer were intravenously administered in mice. At the same time 0 h, 10 h later (+10 h), or 48 h later (+48 h) nanovaccines containing OVA were intravenously delivered. Ten days after NP OVA administration, spleens were − + harvested and analyzed by flow cytometry. MPEC were defined as CD127hi KLGR1 cells and SLEC as CD127lo KLGR1 cells. (A) Diagram of the experimental set up. (B) Evaluation of OT-I T cell counts relative to the CD8+ T cell population. (C) Representative staining of MPEC and SLEC by flow cytometry at day 10. (D and E) Percentage (Left) or relative counts (Right)ofMPEC(D) or SLEC (E). Statistical analysis by t test: *P < 0.05; mean ± SEM.

necessary for the efficiency of the nanovaccine and the genera- CD62L (MEL-14), anti-CD69 (H1.2F3), anti-CD127 (A7R34), anti–KLGR-1 tion of short-lived effector cells. It can be hypothesized that the (MAFA), anti-mPDCA1 (129C1), anti-Vα2 (B20.1), anti-NK1.1 (PK136), anti- + CD8 T cells in close contacts with iNKT cells might receive B220 (RA3-6B2), anti-CD11b (M1/70), anti-CD11c (N418), anti-F4/80 (BM8), additional signals that will impact their subsequent differentia- anti–MHC-II (M5/114.15.2), anti-CCR4 (2G12), anti-CXCR3 (CXCR3-173), anti- β tion into effector or memory cells. Indeed, local inflammation XCR1(ZET), anti-CD3 (145-2C11), anti-CD5 (53-7.3), and anti-TCR (H57-597), β cues can clearly influence this process and skew it toward effector all from Biolegend; anti-V 5.1/5.2 (MR9-4) from eBioscience. The following antibodies were used for microscopy: anti-CD169 (3D6.112), cell generation (37). In particular, it is tempting to speculate that anti-CD11c (N418), anti-CD68 (FA-11), and anti-CXCL9 (MIG-2F5.5), all from these cells are instructed by IL-4 produced by iNKT cells at early Biolegend; anti-mCLEC9a (polyclonal sheep IgG) and anti-mCCL17 (polyclonal stages, and become subsequently an additional source of this cy- goat IgG), both from R&D systems; anti-CD3 (polyclonal rabbit) from Abcam; tokine during late stages in the WP, where they could influence anti-SIGNR1 from AbD Serotech. the differentiation program of neighboring cells. CD1d-α-GalCer dextramer was obtained from Immudex. Corresponding In summary, our study sheds on the role of iNKT cells in secondary antibodies were bought either from Life Technogies (donkey anti- T cell priming and shows a biphasic process of iNKT–CD8 cross- goat) or from Jackson Immunoresearch Laboratories [Goat anti-hamster, Fab′ talk at 2 distinct locations involving direct concomitant interac- (2) donkey anti-rat, donkey anti-rabbit, donkey anti-sheep]. tions. Additionally, we found that the IL-4/CCL17/CCR4 axis PLGA (Resomer RG 502 H, lactide/glycolide molar ratio 48:52 to 52:48) promotes the generation of SLECs. We believe that these im- was purchased from Boehringer Ingelheim. Solvents for PLGA preparation portant mechanistic cues may contribute to future vaccine design (dichloromethane) were obtained from Merck. Polyvinyl alcohol (PVA) was and highlight the importance of codelivering an iNKT cell ago- obtained from Sigma. R848 was from Enzo Life Sciences, poly I:C from nist together with the antigen. Sigma-Aldrich, and endotoxin-free OVA from Hyglos. α-GalCer was pur- chased from Funakoshi. DMSO 99.9%, Triton-X100, and Tween 20 was Experimental Procedures obtained from Sigma. Atto647N was obtained from Atto-tec. RPMI me- Mice. All animal work was done in accordance with Institutional Animal Care dium 1640 was from Life Technologies. Celltrace violet (CTV) and Celltrace and Use Guidelines of the Central Animal Laboratory (Nijmegen, The Neth- far red (CFR) were purchased from Life Technologies. SIINFEKL peptide was erlands). All mice were housed in specific pathogen-free conditions before use. obtained from Anaspec. BSA was bought from Roche. SuperScriptII, RNase − − Wild-type C57/BL6J mice were from the Charles River, OT-I yeti or Rag1 / OT-I OUT (RNase inhibitor), First Strand Buffer, Second Strand Buffer, DNA mice were bred in house. CXCR6-GFP mice were kindly provided by the Rachel polymerase I (), 0.1M DTT, RNase H (E. coli) were obtained Golub laboratory, Pasteur Institute, Paris, France, and were maintained as from Invitrogen. AMPure XP beads and RNAClean XP beads were obtained heterozygous at the animal facility of the Cochin Institute Paris, France. from Beckman Coulter.

Antibodies and Reagents. The following antibodies were used for flow cytometry: PLGA Synthesis and Characterization. PLGA nanovaccines were manufactured anti–CD8α (53-6.7), anti-CD44 (IM7), anti-CD45.1 (A20), anti-CD45.2 (104), anti- and characterized as described previously (17).

10 of 12 | www.pnas.org/cgi/doi/10.1073/pnas.1913491116 Valente et al. Downloaded by guest on September 23, 2021 Cell Suspension and Flow Cytometry. Spleens were harvested and digested RNAClean XP beads, resuspended in nuclease-free water, and stored at with DNase I (Thermofisher) and Collagenase III (Worthington) at 37 °C for −80 °C prior use. 30 min. Organs were then mechanically digested and passed over 100-μm qRT-PCR. cDNA was produced by a reverse-transcriptase reaction using ran- cell strainers (Corning). Red blood cell lysis was performed with 2 mL of 1× dom hexamer amplification for 1 h at 42 °C and the enzyme was subsquently ammonium chloride solution for 5 min at room temperature. Cell suspension heat-inactivated 10 min at 70 °C. cDNA was then used for gene arrays was then stained with fluorescently labeled antibody in PBA (PBS 1% BSA (mouse Cytokines & Chemokines RT2 Profiler PCR Array for cDC or mouse 0.01% sodium azide) for 20 min at 4 °C in presence of CD16/CD32 blocking Chemokines & Chemokine Receptors RT2 Profiler for OT-I T cells, Qiagen), or (24G2 BD Pharmingen). For iNKT staining, cells were incubated 10 min with with manual qPCR with SYBR green reaction. Samples were read on Bio-Rad fluorescently labeled CD1d-αGalCer dextramer at room temperature in PBA CFX96. Cell cluster analysis was performed with Qiagen analysis software. prior incubation with fluorescently labeled antibodies. Cells were analyzed Three housekeeping genes were used for manual qPCR with OT-I T cells: with FACS verse, FACS LSRII, or FACS Fortessa (BD). Results were analyzed β2m, GADPH and PBGD. β2m was omitted with DC subsets since maturation with FlowJo (Tree Star). affected its expression. Primer list. CCL17 Fwd: TGGTATAAGACCTCAGTGGAGTGTTC; CCL17 Rev: + and Vaccination Experiments. OT-I CD8 Tcellswere GCTTGCCCTGGACAGTCAGA; CCL22 Fwd: GAGTTCTTCTGGACCTCAAATCC; CCL22 isolated from spleen and peripheral LNs by a negative selection (CD8α Rev: TCTCGGTTCTTGACGGTTATCA; CCR4 Fwd: TGCACCAAGGAAGGTATCAAGG; isolation kit mouse Miltenyi) with added biotinylated anti-CD44 antibodies CCR4 Rev: GTACACGTCCGTCATGGACTT; CCL3 Fwd: TGTACCATGACACTCTGCAAC; according to Gerner et al. (38). Purity was routinely superior to 95%. OT-I cells CCL3 Rev: CAACGATGAATTGGCGTGGAA; CCL4 Fwd: GCCCTCTCTCTCCTCTTGCT; were labeled with either CTV (Life Technologies) or CFR (Life Technologies) CCL4 Rev: GAGGTCAGAGCCCATTG; CCR5 Fwd: ATGGATTTTCAAGGGTCAGTTCC; with a dye concentration of 5 μM following the manufacturer’s procedure. CCR5 Rev: CTGAGCCGCAATTTGTTTCAC; CXCL9 Fwd: CTTTTCCTCTTGGGCATCAT; 6 + Next, 1 to 2 × 10 dye-labeled OT-I CD8 T cells were adoptively transferred CXCL9 Rev: GCATCGTGCATTCCTTATCA; CXCL10 Fwd: GCTGCCGTCATTTTCTGC; into congenic wild-type mice, followed by intravenous injection of nano- CXCL10 Rev: TCTCACTGGCCCGTCATC; CXCR3 Fwd: GGTTAGTGAACGTCAAGTGCT; vaccine 1 d later with a dose corresponding to 1 μgOVA,1.5ngα-GalCer, CXCR3 Rev: CCCCATAATCGTAGGGAGAGGT; CCR7 Fwd: TGTACGAGTCGGTGTGCTTC; μ 215 ng R848, and 105 ng poly I:C. For the blocking experiment, 500 ganti- CCR7 Rev: GGTAGGTATCCGTCATGGTCTTG; β2m Fwd: TTCTGGTGCTTGTCTCACTGA; μ CXCR3 blocking antibodies, 300 g of IL-4 or CCL17 antibodies, or their β2m Rev: CAGTATGTTCGGCTTCCCATTC; IL-12p40 Fwd: TGGTTTGCCATCGTTTTGCTG; corresponding isotype (Bioxcell/R&D systems) were injected intraperitoneally IL-12p40 Rev: ACAGGTGAGGTTCACTGTTTCT; IL-12p35 Fwd: CTGTGCCTTGGTAG- μ 1 d prior to vaccination. Alternatively, 2.5 g of CCR4 antagonist (CAS 864289- CATCTATG; IL-12p35 Rev: GCAGAGTCTCGCCATTATGATTC; IL-12p19 Fwd: ATGCTG- 85-0 Cayman) or the vehicle (0,05% DMSO) was injected intraperitoneally GATTGCAGAGCAGTA; IL-12p19 Fwd: ACGGGGCACATTATTTTTAGTCT. 1 d before and at the same time of vaccination. Cryosections and Confocal Microscopy. Spleens were fixed with 4% PFA for 2 h MPEC vs. SLEC Experiment. For the MPEC and SLEC experiments, 1.000 purified + on ice. Successive baths of 10%, 20%, and 30% fructose were performed at OT-I CD8 T cells were intravenously injected into CD45.1 animals. Nano- 4 °C for 48 h. Organs were then embedded in OCT (Sakura), snap-frozen, and IMMUNOLOGY AND INFLAMMATION vaccines were administrated intravenously the following day, and 10 d later, stored at −80 °C. Ten- to 20-μm cryosections were performed with a mi- mice were killed, and spleens harvested and analyzed by flow cytometry. crotome (Microm HM 500 Cryostat) at temperatures between −18 °C and −21 °C. For immunostainings, residual PFA was quenched with PBS 0.1M In Vivo T Cell Homing. For in vivo T cell homing, 500 × 103 purified CTV labeled + Glycine for 5 min at room temperature, sections were then stained overnight OT-I CD8 T cells were intravenously injected into wild-type animals 1 d prior at 4 °C with primary antibodies in PBA 0.1% Triton-X100. After washing with to vaccination. Nanovaccines were administrated intravenously the follow- PBA, sections were then stained in PBA with secondary antibodies 2 h at 4 °C, × 3 + ing day, and 3 h later 500 10 purified CFR-labeled OT-I CD8 T cells washed, and then counterstained with DAPI (Sigma) if needed. Sections (representing “early migrating OT-I cells”) were intravenously injected. were finally washed in PBS and then in milliQ and mounted with a coverslip Three hours later (6 h postvaccination), mice were killed, and spleens were and Prolong Antifade Diamond mounting medium (Life Technologies). harvested and fixed for subsequent immunofluorescence staining. The ratio Confocal microscopy was performed with Olympus FV1000 or Zeiss LSM880 of CTV OT-I/CFROT-I was calculated in the WP on the different images in the with 20× or 40× objectives. Pictures were then analyzed with ImageJ. To control mice and normalized to 1. The ratio of CTV OT-I/CFR OT-I was cal- analyze NP uptake or chemokine expression, a background threshold was culated in the vaccinated mice and compared to the control. applied to quantify the total fluorescence intensities. Concerning the cal- culation for fluorescence intensities per square micrometer, fluorescence Simulation for OT-I and NKT Cell Distribution and Distance. In order to validate background was subtracted and then pictures were analyzed. Percentage of the significance of possible “contacts” between NKT cells and OT-I cells in the cells per area was performed on the total of cells in the whole section. Cells WP, we developed an R Shiny application using the R package ‘shiny’ version with a minimal distance shorter than 3 μm were considered to be in (https://CRAN.R-project.org/package=shiny) (39). The NKT cells and OT-I cells close contact. positions and the contour of the WP were extracted manually from different images with Image J (NIH). The R program calculates the mean of the closest Thick Sections and Imaging. Spleens were embedded in a solution of 7.5% low distance between the OT-I cells and NKT cells for each OT-I T cell (using nncross: Nearest Neighbors Between Two Patterns from spatstat package). gelling-temperature agarose (type VII-A; Sigma-Aldrich) prepared in PBS. μ In order to simulate the effect of density to the percentage of cells in pos- Slices (500 m) were cut with a vibratome (VT 1000S; Leica) in a bath of ice- μ sible contact, we ran 20 simulations of random positions of NKT in the same cold PBS. Slices were then transferred to 0.4- m organotypic culture inserts region of interest (WP) and calculated the mean of the closest distance (Millicell; Millipore) in 35-mm Petri dishes containing 1 mL RPMI 1640 between OT-I and NKT cells for each analyzed picture. without Phenol red. Live vibratome sections were stained for 15 min at 37 °C with antibodies. All antibodies were diluted in RPMI without Phenol red and μ Gene Array, Primers, and qRT-PCR. used at a concentration of 10 g/mL. Tissue sections were imaged with a RNA isolation. Splenocytes from vaccinated animals were stained with cor- DM500B upright microscope equipped with an upright spinning-disk con- responding antibodies and cell subsets were FACS sorted with BD ARIA III. Cell focal microscope (Leica) in a 37 °C thermostated chamber. For dynamic im- pellets were resuspended in TRIzol and stored at −80 °C. Total RNA was aging, tumor slices were secured with a stainless steel slice anchor (Warner isolated by chloroform extraction and precipitated overnight at −20 °C with Instruments) and perfused at a rate of 0.8 mL/min with a solution of RPMI × × RNase-free isopropanol. without Phenol red, bubbled with 95% O2, and 5% CO2. Next, 10 or 25 mRNA preamplification. mRNA was amplified using polydT primers to amplify water-immersion objectives were used and 30-min movies of different areas polyA-tailed RNA with first an annealing for 2.5 min at 65 °C, then a reverse- were recorded. Velocity was calculated using the plugin Manual Tracking α transcriptase reaction with dNTPs, First Strand buffer, an RNase inhibitor, from ImageJ. For CD1d/ -GalCer staining, thick sections were stained in PBA DTT, and SuperScriptII. The reverse-transcriptase reaction was performed for for 30 min at room temperature, and then 2 h on ice with CD169 and CD11c 1 h at 42 °C, then the reverse-transcriptase was heat-inactivated for 10 min antibodies. Thick sections were then fixed 20 min at room temperature with at 70 °C. Then, a second strand reaction was performed with Second Strand 4% PFA, washed with PBS, and then imaged by confocal microscopy. Buffer, dNTPs, DNA Ligase, DNA polymerase I and RNase H for 2 H at 16 °C. + cDNA was cleaned up using AMPure XP beads. A T7-based in vitro tran- In Vitro Culture. For in vitro culture, 50 × 103 purified OT-I CD8 T cells were scription was performed from cDNA samples using the transcripAID high- cultured in a96U-well plate with a complete medium (RPMI 1640 supple- yield transcription kit (Thermofisher) overnight at 37 °C and then enzymes mented with 10% FCS, nonessential amino acid [Gibco], L-glutamine [Gibco], were inactivated 10 min at 70 °C. Finally, amplified RNA was purified with anti-anti [Gibco], sodium pyruvate [Gibco], Hepes, and β-mercaptoethanol)

Valente et al. PNAS Latest Articles | 11 of 12 Downloaded by guest on September 23, 2021 at 37 °C for 5 or 24 h. Concentrations of SIINFEKL and IL-4 were, respectively, ACKNOWLEDGMENTS. The authors thank Emmanuel Donnadieu, Vincenzo 1 μg/mL and 10 ng/mL. Cerundolo, and Alain Trautmann for stimulating discussions and critical advice concerning the manuscript; and the Cochin Imaging Facility, the microscopy Statistical Analysis. For statistical analysis, 1-way ANOVA or Mann–Whitney center in Radboud University Medical Center, and the imaging core facility t test was used with Graphpad; *P < 0.05, **P < 0.01, ***P < 0.001, and (ImagImm) of the Centre d’Immunologie de Marseille-Luminy. This work was ****P < 0.0001. supported by European Research Council PATHFINDER (269019) and EU H2020 Grant PRECIOUS (686089), by la Ligue contre le Cancer, and by the French Data Availability. All data supporting the findings of this study have been National Research Agency through the “Investments for the Future” program deposited in Figshare. (France-BioImaging, ANR-10-INBS-04).

1. T. R. Mempel, S. E. Henrickson, U. H. Von Andrian, T-cell priming by dendritic cells in 22. M. Kurachi et al., Chemokine receptor CXCR3 facilitates CD8(+) T cell differentiation lymph nodes occurs in three distinct phases. Nature 427, 154–159 (2004). into short-lived effector cells leading to memory degeneration. J. Exp. Med. 208, 2. S. Hugues et al., Distinct T cell dynamics in lymph nodes during the induction of 1605–1620 (2011). – tolerance and immunity. Nat. Immunol. 5, 1235 1242 (2004). 23. J. H. Sung et al., Chemokine guidance of central memory T cells is critical for antiviral + 3. S. E. Henrickson et al., Antigen availability determines CD8 T cell- in- recall responses in lymph nodes. Cell 150, 1249–1263 (2012). – teraction kinetics and memory fate decisions. Immunity 39, 496 507 (2013). 24. Y. O. Alexandre et al., XCR1+ dendritic cells promote memory CD8+ T cell recall upon 4. O. P. Joffre, E. Segura, A. Savina, S. Amigorena, Cross-presentation by dendritic cells. secondary infections with Listeria monocytogenes or certain viruses. J. Exp. Med. 213, Nat. Rev. Immunol. 12, 557–569 (2012). 75–92 (2016). 5. D. Dudziak et al., Differential antigen processing by dendritic cell subsets in vivo. 25. S. Calabro et al., Differential intrasplenic migration of dendritic cell subsets tailors Science 315, 107–111 (2007). – 6. A. Savina et al., The small GTPase Rac2 controls phagosomal alkalinization and an- adaptive immunity. Cell Rep. 16, 2472 2485 (2016). tigen crosspresentation selectively in CD8(+) dendritic cells. Immunity 30, 544–555 26. B. Ravishankar et al., Tolerance to apoptotic cells is regulated by indoleamine 2,3- – (2009). dioxygenase. Proc. Natl. Acad. Sci. U.S.A. 109, 3909 3914 (2012). 7. J. Idoyaga, N. Suda, K. Suda, C. G. Park, R. M. Steinman, Antibody to /CD207 27. S. C. Saunderson, A. C. Dunn, P. R. Crocker, A. D. McLellan, CD169 mediates the localizes large numbers of CD8alpha+ dendritic cells to the marginal zone of mouse capture of exosomes in spleen and lymph node. Blood 123, 208–216 (2014). spleen. Proc. Natl. Acad. Sci. U.S.A. 106, 1524–1529 (2009). 28. T. Iyoda et al., The CD8+ dendritic cell subset selectively endocytoses dying cells in 8. M. Kitano et al., Imaging of the cross-presenting dendritic cell subsets in the skin- culture and in vivo. J. Exp. Med. 195, 1289–1302 (2002). draining lymph node. Proc. Natl. Acad. Sci. U.S.A. 113, 1044–1049 (2016). 29. C. H. Qiu et al., Novel subset of CD8alpha+ dendritic cells localized in the marginal 9. H. Hochrein et al., Differential production of IL-12, IFN-alpha, and IFN-gamma by zone is responsible for tolerance to cell-associated antigens. J. Immunol. 182, 4127– – mouse dendritic cell subsets. J. Immunol. 166, 5448 5455 (2001). 4136 (2009). + 10. C. A. Bernhard, C. Ried, S. Kochanek, T. Brocker, CD169 macrophages are sufficient 30. J. M. den Haan, S. M. Lehar, M. J. Bevan, CD8(+) but not CD8(-) dendritic cells cross- for priming of CTLs with specificities left out by cross-priming dendritic cells. Proc. prime cytotoxic T cells in vivo. J. Exp. Med. 192, 1685–1696 (2000). – Natl. Acad. Sci. U.S.A. 112, 5461 5466 (2015). 31. M. L. Lin et al., Selective suicide of cross-presenting CD8+ dendritic cells by cyto- 11. K. Asano et al., CD169-positive macrophages dominate antitumor immunity by chrome c injection shows functional heterogeneity within this subset. Proc. Natl. crosspresenting dead cell-associated antigens. Immunity 34,85–95 (2011). Acad. Sci. U.S.A. 105, 3029–3034 (2008). 12. F. Castellino, R. N. Germain, Cooperation between CD4+ and CD8+ T cells: When, 32. R. Backer et al., Effective collaboration between marginal metallophilic macrophages where, and how. Annu. Rev. Immunol. 24, 519–540 (2006). + 13. A. Bendelac, P. B. Savage, L. Teyton, The biology of NKT cells. Annu. Rev. Immunol. 25, and CD8 dendritic cells in the generation of cytotoxic T cells. Proc. Natl. Acad. Sci. – 297–336 (2007). U.S.A. 107, 216 221 (2010). 14. T. Globisch et al., Cytokine-dependent regulation of dendritic cell differentiation in 33. E. E. Gray, S. Friend, K. Suzuki, T. G. Phan, J. G. Cyster, Subcapsular sinus macrophage the splenic microenvironment. Eur. J. Immunol. 44, 500–510 (2014). fragmentation and CD169+ bleb acquisition by closely associated IL-17-committed 15. P. Barral, M. D. Sánchez-Niño, N. van Rooijen, V. Cerundolo, F. D. Batista, The location innate-like . PLoS One 7, e38258 (2012). of splenic NKT cells favours their rapid activation by blood-borne antigen. EMBO J. 31, 34. A. Audemard-Verger et al., Macrophages induce long-term trapping of γδ T cells with 2378–2390 (2012). innate-like properties within secondary lymphoid organs in the steady state. J. Im- 16. V. Semmling et al., Alternative cross-priming through CCL17-CCR4-mediated attraction munol. 199, 1998–2007 (2017). – of CTLs toward NKT cell-licensed DCs. Nat. Immunol. 11,313 320 (2010). 35. H. Veninga et al., Antigen targeting reveals splenic CD169+ macrophages as pro- 17. Y. Dölen et al., Co-delivery of PLGA encapsulated invariant NKT cell agonist with moters of germinal center B-cell responses. Eur. J. Immunol. 45, 747–757 (2015). antigenic protein induce strong T cell-mediated antitumor immune responses. 36. S. Eickhoff et al., Robust anti-viral immunity requires multiple distinct T cell-dendritic OncoImmunology 5, e1068493 (2015). cell interactions. Cell 162, 1322–1337 (2015). 18. M. Bajénoff, N. Glaichenhaus, R. N. Germain, Fibroblastic reticular cells guide T 37. N. L. Pham, V. P. Badovinac, J. T. Harty, A default pathway of memory CD8 T cell entry into and migration within the splenic T cell zone. J. Immunol. 181,3947–3954 (2008). differentiation after dendritic cell immunization is deflected by encounter with in- 19. A. P. Benechet, M. Menon, K. M. Khanna, Visualizing T cell migration in situ. Front. – Immunol. 5, 363 (2014). flammatory cytokines during antigen-driven proliferation. J. Immunol. 183, 2337 20. F. Geissmann et al., Intravascular immune surveillance by CXCR6+ NKT cells patrolling 2348 (2009). liver sinusoids. PLoS Biol. 3, e113 (2005). 38. M. Y. Gerner, K. A. Casey, W. Kastenmuller, R. N. Germain, Dendritic cell and antigen 21. W. Kastenmüller et al., Peripheral prepositioning and local CXCL9 chemokine- dispersal landscapes regulate T cell immunity. J. Exp. Med. 214, 3105–3122 (2017). mediated guidance orchestrate rapid memory CD8+ T cell responses in the lymph 39. W. Chang et al., Shiny: Web application framework for R, Version 1.4.0. https:// node. Immunity 38, 502–513 (2013). cran.r-project.org/web/packages/shiny/index.html. Accessed 24 November 2019.

12 of 12 | www.pnas.org/cgi/doi/10.1073/pnas.1913491116 Valente et al. Downloaded by guest on September 23, 2021