Female C57BL/6 J were obtained from Charles River laboratories. F5 TCR transgenic mice, backcrossed on C57BL/6 background (Jubin et al., 2012), were bred and housed under specific pathogen-free conditions in the AniRA-PBES animal facility (Lyon, France). CD8⁺ T cells from these mice (called F5 CD8⁺ T cells hereafter) express a TCR that is specific for the ASNENMDAM epitope (called NP68 peptide hereafter) of influenza A nucleoprotein (aa 366-374). All experimental protocols were approved by the local ethic evaluation committee (CECCAPP, Lyon, France).

CD8⁺ T cell epitope (NP68) was inserted into LDH protein as its expression is maintained when S. aureus become chronic (Szafranska et al., 2014). Partial deletion of the ldh locus and in frame insertion of the NP68 motif in SH1000 strain of S. aureus was performed using pMAD (Arnaud et al., 2004). Chromosomal regions upstream and downstream of the ldh sequence were amplified by PCR (PCR1=nt 227938-229288, PCR2=nt 229289-230966 of S.aureus genome 8325); NP68 epitope (ASNENMDAM) and the four naturally NH2-flanking amino acid residues (GVQI) coding sequences were PCR synthetized. The NP68 coding DNA was cloned between the two ldh fragment onto pMAD, forming pLUG2143. This plasmid was electroporated in E. coli, then into RN4220 recipient strain and transferred to SH1000. Growth at non-permissive temperature (44°C) was followed by several subcultures at 30°C and 37°C to favor double crossing over as previously described (Arnaud et al., 2004), resulting in strain LUG2154. gfp-positive strains LUG2778 and LUG2779 were constructed by electroporating pCN38 into SH1000 and LUG2154, respectively.

LUG2778 and LUG2779 were inoculated into 4 ml of Brain Heart Infusion medium (37g.l^-1 of Brain Heart Infusion Broth from Sigma-Aldrich) and grown overnight at 37°C, 5% CO₂ and 200 rpm. Bacterial numbers were measured by Flow cytometry (BD Accuri C6 cytometer) and by counting Colony Forming Units (CFU) after plating on Columbia agar with 5% sheep blood (Biomerieux) flowed by overnight incubation at 37°C.

To prepare bone marrow-derived dendritic cells, femurs and tibia were harvested from C57BL/6 female mice and bone marrows were flushed. Cell suspensions were filtered through cell strainer 100µm (BD Falcon) and resuspended in RPMI medium supplemented with 10% fetal calf serum (FCS, Gibco), 10mM HEPES (Gibco), 50µM β-mercaptoethanol (Gibco) and 50 µg.ml^-1 gentamycin (Gibco). Cells were plated at 2.10⁶ cells per ml in medium containing 100ng.ml^-1 recombinant human Flt3L (Amgen) in order to generate Flt3L-derived DCs (de Brito et al., 2011). Cells were used after a 7-days incubation at 37°C and 7% of CO₂. The percentage of conventional and plasmacytoid DCs in these cultures are 80% and 20%, respectively (de Brito et al., 2011).

Effector F5 CD8⁺ T cells were generated from F5 transgenic mice. Spleens were harvested and filtered through nylon cell strainer 100µm (BD Falcon). Splenocytes were cultured in DMEM medium supplemented with 10% FCS, 10mM HEPES, 50µM β-mercaptoethanol, 50 µg.ml^-1 gentamycin, 10% supernatant containing murine IL-2 (obtained from the culture supernatants of the mouse myeloma clone X63-Ag8.653 cell lines transfected with mouse IL-2 gene (Karasuyama and Melchers, 1988)) and 30nM NP68 peptide (Proteogenix, sequence: ASNENMDAM). Alternatively, T cells were sorted from spleens (“CD8a⁺ T Cell Isolation Kit, mouse”, Miltenyi) and cultured with pulsed (NP68 peptide, 20nM) and activated (CpG 1826, 2 µg/ml, Invivogen) Flt3L-derived DCs in the presence of 10% supernatant containing murine IL-2. Cells were cultured at 37°C and 7% CO₂ for five days.

The DC2.4 cell line (Shen et al., 1997) was cultured in RPMI medium supplemented with 10% FCS, 10mM HEPES, 50µM β-mercaptoethanol and 50 µg.ml^-1 gentamycin and incubated at 37°C and 5% CO₂.

DCs were seeded in 96 U-well plates (Dutscher Scientific) at 6.10⁴ (DC2.4) or 1.10⁵ (Flt3L-derived DCs) cells per well in RPMI medium supplemented with 10% FCS, 10mM HEPES and 50µM β-mercaptoethanol but no antibiotics. DCs were incubated with infectious S. aureus at the indicated multiplicity of infection (MOI) for 1h, washed with Phosphate Buffer Saline (PBS, Gibco) and further incubated for the indicated time with RPMI medium containing antibiotic (50 µg.ml^-1 gentamycin, Gibco) in order to kill extracellular S. aureus. When indicated, DCs were incubated with equivalent numbers of heat-killed (96°C for 30 min) S. aureus.

Flt3L-derived DCs were first labeled with fixable viability dye eFluor 780 (Invitrogen) 20 min at 4°C in PBS. To minimize non-specific binding of antibodies, samples were incubated with Fc block 2.4G2 supernatant in PBS for 10 min. Then samples were antibody-stained for 30 min at 4°C in PBS, 1% FCS (Hyclone) and 0.9% sodium azide. The following antibodies were used: anti-CD11c (PerCP-Cy5, N418 clone, Biolegend), anti-CD80 (BV605, 16-10A1 clone, BD Biosciences), anti-CD86 (PE-Cy7, GL-1 clone, Biolegend), anti-CD40 (BV421, 3.23 clone, BD Biosciences), anti-B220 (APC, RA3-6B2 clone, BD Biosciences) and anti-CMHI H2Db (PE, ER-HR52 clone, Biorad). After washes, samples were fixed for 20 min with Cytofix/Cytoperm buffer (BD Biosciences). Data were acquired on a LSR Fortessa 4 lasers flow cytometer (BD Biosciences). Data analysis was then performed using FlowJo software (Tree star, version 10).

Culture supernatants were collected and stored at -20°C. Levels of IFN-γ and IL-2 were determined using enzyme-linked immunosorbent assay (ELISA) according to the manufacturer’s recommendations (Biolegend).

F5 effector cells were stained with CellTrace Violet (Invitrogen, according to the manufacturer’s recommendations) and cultured with S. aureus infected DC2.4 at the indicated ratios for 4h. As control, F5 effector cells were incubated with DC2.4 loaded with 20nM of NP68 peptide. Cells were stained with fixable viability dye eFluor 780 (Invitrogen) 20 min at 4°C in PBS and fixed for 20 min with Cytofix/Cytoperm (BD Biosciences). Data were acquired on a LSR Fortessa 4 lasers flow cytometer (BD Biosciences). Data analysis was then performed using FlowJo software (Tree star, version 10).

DC2.4 cells were left to adhere on glass coverslips for 1h before infection with S. aureus as described above. Specimens were fixed with paraformaldehyde 2% (Invitrogen) for 10 min, permeabilized with Triton 0.5% (Sigma) for 15 min, before staining with Phalloidin-TRITC (10 µg.ml^-1, Abcam) for 30 min. EEA-1 was detected using primary antibody (440ng.ml^-1, F.43.1 clone, Invitrogen) and revealed with secondary antibody anti-rabbit conjugated with AF647 (8 µg.ml^-1, Invitrogen). Coverslips were then mounted with Prolong Gold antifade reagent (Invitrogen) containing DAPI (1µg/ml, Invitrogen). Images were acquired with a LSM 980 microscope (Zeiss) and analyzed with ImageJ (version 2.1.0). Images displayed are the Mean Intensity Projection of 10 - 20 slices.

DC2.4 were stained with CellTrace Deep Red (Invitrogen, according to the manufacturer’s recommendations), co-cultured in 96 wells plates with GFP-NP68 S. aureus and imaged using CQ1 microscope (Yokogawa). Effector F5 T cells stained with CellTrace Violet (Invitrogen, according to the manufacturer’s recommendations) were added to the cultures. Images displayed are the Mean Intensity Projection from 10 - 20 slices.

Statistical analyses (t-tests) were performed using Prism software (GraphPad, version 9). Levels of significance are expressed as p-values (* p<0.05, ** p<0.01, *** p<0.001, ****p<0.0001). The bars represent the means ± SD of replicates or of independent experiments.

In order to investigate CD8⁺ T cell immune responses against intracellular S. aureus, we generated SH1000 S. aureus strains that express fluorescent GFP tracer protein and the NP68 epitope. Although it is not as invasive as other S. aureus strains, we chose SH1000 strain as it infects many cell types and activates their cytokine production while preserving cell viability (Strobel et al., 2016).

We first wished to establish whether dendritic cells (DCs) can be infected by S. aureus in vitro. We used murine bone marrow-derived DCs in the presence of FMS-like tyrosine kinase 3 ligand (Flt3L DCs), these cells representing bona fide counterparts of the splenic DC subsets (Naik et al., 2005; Dresch et al., 2012). To follow S. aureus infection and its impact on Flt3L DCs, cells were co-cultured with SH1000 GFP-NP68 at a multiplicity of infection (MOI) of 10. DCs were washed and further cultured with gentamycin, at a dose that does not kill intracellular S. aureus (Mohamed et al., 2014). The viability of Flt3L DCs along with the expression of GFP among CD11c⁺ DCs were monitored 24, 48- or 72-hours post-infection (pi) by flow cytometry. As control, Flt3L DCs were also incubated with heat-killed S. aureus.

Infection of Flt3L DCs with SH1000 S. aureus strain did not induce DCs cell death ( Figure 1A ) and even delayed the decrease in cell viability associated with the spontaneous death of these primary cells. This delay was not due to the infection step as it is also observed when DCs are in contact with heat-killed S. aureus. This delay is rather likely related to Flt3L DCs maturation, illustrated by the increase in MHC-I, CD86 and CD80 expression when cells were cultured with infectious or heat killed S. aureus ( Figure 1B ). One hour of contact with live infectious S. aureus led to 50% of Flt3L DCs containing viable S. aureus as measured one day later ( Figure 1C and data not shown). However, the percentage of infected cells decreased with time as well as the median of fluorescence (MFI) of GFP among cells ( Figures 1C, D ). Flt3L DCs contain two distinct populations, conventional DCs (cDCs, CD11c⁺ B220^-) and plasmacytoid DCs (pDCs, CD11c⁺ B220⁺), both of them being equally infected by S. aureus ( Figures 1F, G ). Taken together, these results show that S. aureus is internalized by primary mouse Flt3L DCs and does not seem to proliferate within these cells under our experimental conditions.

The study was then extended to DC2.4 cell line, an immortalized mouse dendritic cell line that express dendritic cell-specific markers such as DEC-205 and 33D1 and high levels of MHC and co-stimulatory molecules (Shen et al., 1997). As primary DCs, DC2.4 cells have the ability to phagocytose and present exogenous antigens on MHC-I (Shen et al., 1997). DC2.4 were incubated with different MOI of infectious GFP-NP68 expressing SH1000 strain for 1h, washed and further incubated in the presence of gentamycin antibiotic. Figure 2 shows that S. aureus internalization did not affect DC2.4 survival regardless of the MOI ( Figures 2A, C ). The percentage of infection was proportional to the MOI, reaching a maximum at MOI 100 ( Figure 2B ). The viability of infected cells was maintained over a 3 days period ( Figure 2C ), as well as the percentage of GFP-positive DC2.4 ( Figures 2D, E ). The levels of GFP within each DC decreased with time ( Figure 2F ), reflecting the killing of S. aureus and/or the proliferation of DC2.4 ( Supplementary Figure 1 ). Indeed, shortly after the infection, the majority of DC2.4 cells contained several bacteria ( Figure 2G ; Supplementary Video 1 ) which were distributed between daughter cells during proliferation. Internalized bacteria remained alive and a fraction acquired a SCV phenotype ( Supplementary Figure 1 ). Early endosome antigen-1 (EEA-1) is an endosome-specific peripheral membrane protein found in early phagosomes. EEA-1 colocalized with more than 80% of S. aureus at 30 min pi indicating that the internalized bacteria trafficked via the endosomal pathway. Endosomal vacuoles matured further as indicated by the loss of EEA-1 staining ( Figures 2H, I ).

Having established that Flt3L DCs and DC2.4 can be infected by S. aureus, we used these two populations to study the ability of DCs to present S. aureus-derived antigens at their surface on MHC-I molecules to CD8⁺ T cells. Dead heat-killed NP68 expressing S. aureus was used as a positive control as it was likely that phagocytosed dead bacteria would enter the cross-presentation pathway allowing cross-presentation of S. aureus-derived epitopes on MHC-I molecules to CD8⁺ T cells ( Figures 3A, C ). 24-, 48- or 72-hours after DC co-culture with S. aureus, effector F5 CD8⁺ T cells recognizing NP68 epitope were added to the cultures. If the TCR of F5 CD8⁺ T recognize their cognate NP68 epitope presented onto MHC-I proteins, F5 CD8⁺ T cells get stimulated, which can be followed by measurement of cytokine production (Brinza et al., 2016). The production of IFN-γ in the supernatant was measured 16h after the addition of F5 effector cells ( Figure 3 ). F5 T cells produced more IFN-γ when cultured with Flt3L DCs and DC2.4 infected with control S. aureus compared to uninfected DCs. This is also true for the interleukin-2 (IL-2) production ( Supplementary Figure 2 ). This activation, which is not antigen specific, likely reflects an activation of effector CD8⁺ T cells interacting with DCs that have matured in contact with S. aureus ( Figure 1 ; Supplementary Figure 3 ). Nevertheless, the production of IFN-γ was much higher when F5 CD8⁺ T cells interacted with DCs having internalized heat-killed and live S. aureus that express NP68. These results demonstrate that the NP68 epitope expressed by S. aureus is presented at the surface of DCs and is recognized by F5 CD8⁺ T cells.

The decrease of IFN-γ production from Flt3L DCs cultures with time reflects the death of the primary cells. Interestingly, when DC2.4 were loaded with heat-killed NP68 expressing S. aureus, IFN-γ production occurred rapidly and decreased over time, likely reflecting a decrease in antigen cross-presentation due to antigen depletion 72h after engulfment of heat-killed S. aureus. In contrast, there was an increase of IFN-γ production by F5 CD8⁺ T cells over time when they were in contact with DC2.4 that had internalized NP68-expressing live S. aureus. This likely reflects a continuous synthesis and an accumulation of NP68 antigen by live bacteria within DC2.4 ( Figure 3D ).

One of the main functions of cytotoxic CD8⁺ T cells is the specific killing of infected cells. Therefore, we compared the ability of effector F5 T cells to kill DC2.4 infected by S. aureus strain expressing or not the NP68 epitope. It is worth noticing that DC2.4 cells are resistant to CD8⁺ T cell killing compared to other cell types widely used in cytotoxicity assays such as EL4 cells ( Supplementary Figure 4 ). However, EL4 cells could not be used as they do not get infected by S. aureus (data not shown). At the higher CD8⁺ T cells:DCs ratio, F5 CD8⁺ T cells induced killing of DC2.4 infected with a S. aureus strain that did not express NP68. This observation illustrates a non-antigen specific activation of F5 CD8⁺ T cells ( Figure 4A ). However, the cytotoxicity of F5 CD8⁺ T cells toward DC2.4 was significantly increased when S. aureus expressed NP68 antigen ( Figures 4A, B ). Altogether these results indicate that F5 CD8⁺ T cells can recognize and kill in an antigen specific manner cells that have internalized S. aureus bacteria (illustrated in Figure 4C ; in Supplementary Video 1 ).