Capacities of Migrating CD1b+ Lymph Dendritic Cells to Present Salmonella Antigens to Naive T Cells

Dendritic cells (DCs) are well known as professional antigen-presenting cells (APC) able to initiate specific T-cell responses to pathogens in lymph nodes (LN) draining the site of infection. However, the respective contribution of migratory and LN-resident DCs in this process remains unclear. As DC subsets represent important targets for vaccination strategies, more precise knowledge of DC subsets able to present vaccine antigens to T cells efficiently is required. To investigate the capacities of DCs migrating in the lymph (L-DCs) to initiate a specific T-cell response, we used physiologically generated DCs collected from a pseudoafferent lymphatic cannulation model in sheep. The CD1b+ L-DCs were assessed for presenting antigens from the vaccine attenuated strain of Salmonella enterica serovar Abortusovis. CD1b+ L-DCs were able to phagocytose, process and to present efficiently Salmonella antigens to effector/memory T cells in vitro. They were shown to be efficient APC for the priming of allogeneic naive T cells associated with inducing both IFN-γ and IL-4 responses. They were also efficient in presenting Salmonella antigens to autologous naive T cells associated with inducing both IFN-γ and IL-10 responses. The capacities of L-DCs to process and present Salmonella antigens to T cells were investigated in vivo after conjunctival inoculation of Salmonella. The CD1b+ L-DCs collected after inoculation were able to induce the proliferative response of CD4+ T cells suggesting the in vivo capture of Salmonella antigens by the CD1b+ L-DCs, and their potential to present them directly to CD4+ T cells. In this study, CD1b+ L-DCs present potential characteristics of APC to initiate by themselves T cell priming in the LN. They could be used as target cells for driving immune activation in vaccinal strategies.


Introduction
Dendritic cells (DCs) are well known as professional antigenpresenting cells (APC) able to initiate specific T-cell responses to pathogens in lymph nodes (LN) draining the site of infection. However, the respective contribution of migratory and LNresident DCs in this process remains unclear [1]. Moreover, the understanding of this complex process depends on the different DC subsets described. In mice, there are migratory DC subsets including epidermal Langerhans cells (LCs), CD11b + CD103 2 and CD11b 2 CD103 + dermal DCs, LN-resident DCs comprising both CD8a + and CD8a 2 DCs, and inflammatory DCs recruited during infection [2,3]. Studies based on experimental mouse models of cutaneous infection support the role of LCs in antigen (Ag) transport to the peripheral LN but not directly in the induction of pathogen-specific T-cell responses [4,5]. Numerous studies examining the role of migratory and LNresident DCs in the induction of CD8 + T cell-mediated immunity to viruses after cutaneous infection have shown the exclusive cross-presentation of Ag by CD8a + DCs resident in LN, and the role of migratory DCs in delivering and transferring Ag to resident CD8a + DCs [3]. However, these conclusions may not be applicable to the priming of cytotoxic T-lymphocyte (CTL) responses to all viruses since migratory skin DCs have been shown to present lentivirus-derived ovalbumin (OVA) directly to LN CD8 + T cells [6], or at least in cooperation with LN-resident DCs [7]. Moreover, dermal migratory DCs have been shown to play a role in generating CD4 + T-cell responses following subcutaneous (SC) influenza infection [8]. Regarding the LN-resident DCs, CD8a 2 DCs seem to be more efficient than CD8a + DCs at presenting exogenous antigens by MHCII molecules [9]. For Salmonella, the involvement of both CD8a + and CD8a 2 splenic DCs in the priming of T cells was reported in mice, but the involvement of migratory DC subsets was not investigated [10]. The respective contribution of migratory and LN-resident DCs in T-cell priming is thus dependent on the pathogen encountered, but also influenced by the in vivo infection route. Herpes simplex virus (HSV) is rapidly presented by LNresident DCs [11] or dermal DCs [12] to CD4 + and CD8 + T cells after SC injection, whereas HSV Ag is mainly presented by migratory DCs after mucosal administration [11].
Most of these studies were performed in mouse models using diverse experimental strategies. Although these models assessed in detail the diverse functions of migratory DC subsets isolated from tissues, they did not investigate them in the draining lymph directly before their arrival in the LN. As DCs are important targets for vaccination strategies, more precise knowledge of DC subsets able to present vaccine antigens to T cells efficiently is required.
Moreover, attenuated pathogens such as Salmonella can be attractive as a vehicle to deliver Ag to the appropriate DCs involved in a protective immune response.
These questions can be investigated further in large animals, using physiologically generated DCs collected from a pseudoafferent lymphatic cannulation model [13,14]. The ruminant lymph DCs (L-DCs) were originally defined on the expression of the CD1b and CD14 molecules [15]. Several studies have investigated and showed the capacities of L-DCs to acquire soluble antigen in vitro or in vivo and to present it directly and specifically to autologous T cells [14]. More recent studies have described different L-DC subsets including plasmacytoid [16] and CD8a-like DCs [17]. However, a few studies have analyzed the interaction between DCs and Salmonella, and their involvement in the priming of naive T cells in vivo [10]. An in vitro study showed that fewer Salmonellae were taken up by CD1b + L-DCs than by monocytederived macrophages [18]. Moreover, one of our in vivo studies showed that CD1b + L-DCs did not play a major role in Salmonella transport to LN after SC infection of the upper respiratory tract with a live vaccine, despite an increased flow of these cells in the lymph [19]. This study showed the predominant role of neutrophils in the the live vaccine uptake as it was also showed in another study using particulate antigen [20]. This approach can assess directly the ability of migratory L-DC subsets to perform Ag presentation at steady state and under infectious conditions. The aim of this study was to investigate in detail the capacity of sheep CD1b + L-DCs, collected from afferent lymph draining the skin or the head mucosae, to present Salmonella antigens to T cells in vitro and in vivo after mucosal administration. To this end, we used the vaccine-attenuated strain of Salmonella demonstrated to induce protection against abortive salmonellosis in sheep [21], and previously used to assess the capacity of L-DCs to uptake and transport Salmonella to LN [19].

Ethics statement
The animal experiments were conducted under a license issued by the Direction des Services Vétérinaires of Tours (accreditation B-37-175-3) and were approved by the Regional Centre-Limousin Ethics Committee (CL2006-012).

Sheep and surgery
'Prealpes du sud' ewes (one to four years old) originating from the Unité Commune d'Expérimentation Animale (INRA, Jouy-en-Josas, France) or from the Plateforme d'Infectiologie Expérimentale (PFIE) (INRA, Nouzilly, France), were housed in the PFIE for surgery and Salmonella infection. They were born and raised in salmonellosis-free herds. Prescapular lymph duct cannulation was performed on the right side of the sheep to collect pseudo-afferent lymph from skin as described previously [13], and cervical lymph duct cannulation was performed on the left side of the same sheep, to collect lymph from the head mucosae as described previously [22]. A total of 11 sheep were used for the study, of which eight were successfully cannulated. Sheep received a daily subcutaneous (SC) injection of calcium heparin (300U per kg per day) (CalciparineH, Sanofi-Avantis, France).

Collection of lymph and lymph cell storage
Lymph was collected continuously in 250 ml sterile polypropylene flasks containing 10 ml of buffered saline supplemented with antibiotics (100 IU/ml of penicillin and 100 mg/ml of streptomycin, Sigma-Aldrich, St Louis, MO) and sodium heparin (250 IU/ml) (héparine choayH, Sanofi-Avantis, France) to prevent both bacterial contamination and clotting. Flasks were changed twice a day, the lymph volume was determined and the viable cell number was assessed using the trypan blue dye exclusion test. Lymph cells were spun down at 400 g for five min, step-frozen in fetal calf serum (FCS) containing 10% dimethyl sulfoxide, and kept in liquid nitrogen. All the analyses described hereafter were performed on lymph cells from frozen samples. Over 95% viability was obtained after thawing.

Phenotypic analysis of lymph DC cells (L-DCs) by flow cytometry
Lymph cells were thawed, washed once in HANK'S Balanced Salts (HBSS) medium supplemented with 2.5% FCS, 20 IU/ml of penicillin and 20 mg/ml of streptomycin (HBSS-FCS) and adjusted to 5610 6 cells/ml. Labelling was performed in 96 round-bottom well microplates (BD Falcon TM , Becton Dickinson, Franklin Lakes, NJ) using 5610 5 cells per well. Cells were incubated with 50 ml of antibodies diluted at optimal concentrations in RPMI 1640 medium (Life Technologies, Cergy Pontoise, France) supplemented with 10% FCS, 2 mM L-glutamine, 1mM sodium pyruvate, and 50 mM ß-mercaptoethanol, 20 IU/ml of penicillin and 20 mg/ml of streptomycin (complete medium) for 20 min on ice with gentle stirring, and washed three times with HBSS-FCS after each step.
Cells were first incubated with one of the primary or appropriate isotype control monoclonal antibodies (mAbs) at equivalent concentrations to primary mAbs (Table 1) and then with R-Phycoerythrin (RPE)-conjugated F(ab') 2 fragment goat anti-mouse (GAM) IgG (1:200) (Jackson ImmunoResearch, Suffolk, UK). Cells were further incubated with anti-CD1b mAb ( Table 1) followed by FITC-conjugated GAM IgG2a (1:200)(Caltag Laboratories, Burlingame, CA). After washes, cells were resuspended and fixed in 100 ml of 1% paraformaldehyde in buffered saline. Thirty to sixty thousand events were analysed with a FACSCalibur TM (Becton Dickinson) using the CellQuestPro TM software analysis programme (Becton Dickinson).
Cells were analysed in a cell population gated on the basis of forward and scattered angles. The CD1b + L-DCs were selected with appropriate gating and analysed for the expression of different markers.
L-DC subsets sorting using fluorescence-activated cell sorting Lymph cells were quickly thawed and washed once in HBSS-FCS. To deplete lymphocytes, cells were first incubated with a mixture of mAbs including anti-ruminant CD4 (17D1), CD8, cd TCR, CD45R (2 mg/ml of each mAb for 1610 8 cells) (Table 1) for 20 min on ice with gentle stirring, washed three times with HBSS-FCS followed by RPE-conjugated GAM IgG labelling. Cells were further incubated with anti-CD1b mAb, followed by FITCconjugated GAM IgG2a. After three washes with HBSS-FCS, cells were resuspended in RPMI medium without FCS before sorting. Appropriate IgG1 and IgG2a isotypes (Table 1) were used at equivalent concentrations to the primary mAbs to produce controls. CD1b + L-DCs were sorted using a fluorescence-activated cell sorter (MoFloH, DakoCytomation, 4850 Innovation Drive, Fort Collins, CO) after gating on a population negative for lymphocyte markers and positive for CD1b. The proportion of the CD1b + L-DCs subset was enriched from 1% in the lymph to 96% with an average purity of over 98%.

RNA extractions and reverse transcriptase PCR (RT-PCR)
Messenger RNA was extracted from sorted CD1b + L-DCs using the DynabeadsH mRNA DIRECT TM Micro Kit (Invitrogen Dynal AS, Oslo, Norway). The mRNA was processed for reverse transcription with MuMLV reverse transcriptase (25U/ml) (Eurogentec, Liège, Belgium) and Oligo-dT (133 pmole/ml) (Eurogentec). The reaction was maintained for 90 min at 37uC and then heat-inactivated at 85uC for 10 min. The generated cDNA was then analysed for the presence of sequences encoding for DC-SIGN, CD103, CCR7, glyceraldehydes-3-phosphate dehydrogenase (GAPDH) and hypoxanthine-guanine phosphoribosyltransferase (HPRT) by either end-point or real-time quantitative PCR. The cDNA control was produced as described above from ovine spleen cells stimulated by both LPS and Concanavalin A. Primers (Table 2) were designed using Clone Manager 9 (Scientific & Educational Software, Cary, NC) and purchased from Eurogentec. DNA was amplified with REDTaq TM DNA Polymerase (1U/mL) (D5684, Sigma-Aldrich) for 34 cycles at the appropriate annealing temperature for 60 s ( Table 2) and at 72uC for 60 s. PCR products were run on 1% agarose gel with ethidium bromide staining.
Quantitative real-time PCR (qPCR) was carried out using diluted cDNA, in duplicate, with the IQ SYBR Green Supermix (Bio-Rad, Hercules, CA) following the manufacturer's recommendations. Cycling conditions were 95uC for five min, followed by 39 cycles with denaturation at 95uC for 10 s and the appropriate annealing temperature (Table 2) for 15 s. PCR reactions were run on a Bio-Rad iCycler iQ (Bio-Rad). The specificity of the qPCR reactions was assessed by analysing the melting curves of the products and size verification. Samples were normalized internally using the cycle quantification (Cq) of GAPDH and HPRT simultaneously as references in each sample. Cq values were extracted with the qPCR instrument software and subsequently imported into qbase PLUS (http://www.qbaseplus.com) for quality control and generation of the standard curves. Relative quantities were calculated using the qBase quantification model which enables PCR efficiency correction, multiple reference assay normalization, proper error propagation and, if necessary, interrun calibration [23].

Lymphocyte subset sorting using immunomagnetic microbeads
Peripheral blood mononuclear cells (PBMC) were obtained from fresh blood diluted 1/2 in phosphate buffered saline without calcium and magnesium (PBS) and layered onto HistopaqueH 1077 (density 1.077, Sigma-Aldrich). After centrifugation at 1600g for 20 min at room temperature, PBMC were harvested at the interface and washed three times in HBSS-FCS. For the positive selection of CD4 + or CD62L + T cells, PBMC were then incubated with anti-CD4 (SBUT4) or anti-CD62L mAb (Table 1), washed three times in PBS supplemented with 0.5% bovine serum albumin (BSA, A7906, Sigma-Aldrich), and further incubated with GAM IgG (H+ L)-coated magnetic microbeads according to the manufacturer's recommendations (GAM IgG MACSH microbeads, 130-048-402, Miltenyi Biotec, Paris, France). After three washes, cells were positively selected with an MS column and octoMACS TM separator (Miltenyi Biotec), and the efficiency of the resulting positive selection was checked by flow cytometry after incubating the sorted cells with RPE-conjugated GAM IgG ( Table 1). The purity of CD4 + T cells and CD62L + T cells were on average over 96% and 90% respectively.

Preparation of Salmonella and infection
The Salmonella enterica serovar Abortusovis Rv-6 strain is a live attenuated vaccinal strain described previously [19] and stored in aliquots at 280uC in 10% glycerol-buffered saline. Aliquots were thawed, washed in buffered saline and the bacterial suspension adjusted to the appropriate concentration for each experiment. For in vivo infection, sheep were inoculated either by the SC route behind each shoulder with 1 ml containing 0.5 to 0.8610 9 Salmonella, or by the conjunctival route with 1.2610 9 Salmonella administered in four drops of 50 ml in the left eye drained by the cannulated cervical duct. The administration of Salmonella by the conjunctival route was carried out after cervical cannulation, and lymph was collected for 72 hours for analysis. For the in vitro experiment with L-DCs, the bacterial suspension was adjusted to the appropriate concentration in complete medium without antibiotics. The GFP-Salmonella Rv-6 strain, obtained by transformation with the plasmid pBRD940 [18], was also used for phagocytosis assays. The purity of the Salmonella culture and the number of viable bacteria were checked by plating serial dilutions on tryptic soy agar (TSA, Difco Laboratories, Detroit, MI).

Endocytosis and phagocytosis assays
FITC-conjugated ovalbumin (FITC-OVA, 0-23020, Molecular ProbesH, Invitrogen) was used to examine the endocytic ability of the L-DCs. Lymph cells were adjusted to 1610 6 cells/ml of complete medium and incubated at 37uC or 4uC for one hour with 200 mL of FITC-OVA (50 mg/ml) [24]. Controls included cells incubated at 37uC and 4uC without FITC-OVA. For phagocytosis assays, lymph cells were infected by Salmonella (100 bacteria/cell) for 30 min at 37uC. Cells were then washed with medium supplemented with gentamicin (G1397, Sigma-Aldrich) once at 200 mg/ml and twice at 50 mg/ml to remove and kill any remaining extracellular bacteria. Cells were then washed with HBSS-FCS and first incubated with anti-CD1b mAb followed by RPE-conjugated GAM IgG (Table 1). After washes, cells were incubated with a mixture of mAbs including anti-ruminant CD4, CD8, cd TCR, CD45R labelled with the Alexa FluorH 647 using the monoclonal antibody labelling kit according to the manufacturer's recommendations (A-20186, Molecular ProbesH, Invitrogen). After washes, cells were resuspended and fixed in 100 ml of 1% paraformaldehyde in buffered saline. Thirty thousand events were analysed using FACS. The CD1b + L-DCs were selected with appropriate gating, and the proportion of FITC labelled cells was analysed.

Antigen presentation assays by L-DCs
To assess the allogeneic reaction, sorted CD1b + L-DCs were resuspended in complete medium and seeded in triplicates (100 ml/well) at different ratios in round-bottom plates (Falcon 3077, Becton Dickinson) for 24h at 37uC. Purified allogeneic CD4 + or CD62L + T cells (1610 5 /100 ml/well) were then added to CD1b + L-DCs. After 72h, 150 ml of supernatant were harvested and frozen at -80uC for cytokine detection, and 150 ml of fresh complete medium were added. The co-cultures were further incubated for 72h and proliferation was assessed by [ 3 H]thymidine incorporation (1 mCi/3.7610 4 Bq, NEN Research Products, Paris, France) for the last eight hours of culture followed by scintillation counting (Packard 1600TR meter, Meriden, CT).
To assess Salmonella presentation, CD1b + L-DCs were resuspended in complete medium without antibiotics, distributed in round-bottom plates (1610 3 /well) and infected by Salmonella as described above. After washes with antibiotics to remove and kill any remaining extracellular bacteria, cells were resuspended in complete medium supplemented with gentamicin (50 mg/ml) (100 ml) and incubated for 24h at 37uC. Autologous purified Tcell subsets (1610 5 cells/well) were added to the DCs (100 ml/well) and co-cultured for six days. After 72h, 150 ml of supernatant was harvested and frozen at -80uC for cytokine detection and 150 ml of fresh complete medium was added. The proliferative response was assessed by [ 3 H]-thymidine incorporation for the last eight hours of culture followed by scintillation counting.
Standard curves were generated for each cytokine assay with recombinant ovine IFN-c, IL-4 and IL-10 (kindly provided by Dr S. Wattegedera, MRI, Scotland). Cytokine concentrations were expressed in IU/ml.

Statistical analyses
In the experiments performed with several sheep, the statistical analysis used was carried out with the package ''nparLD'' designed to perform non-parametric analysis of longitudinal data in factorial experiments [28]. In the case of the repetition of several independent experiments performed with sorted CD1b + L-DCs and T cells sampled at different times in the same sheep, we have defined that results of proliferative response were representative of a biological response when the ratio of antigen stimulated to medium stimulated conditions was .2.

Results
Phenotyping and uptake capacities of migrating CD1b + L-DC at steady state As two subsets of CD1b + L-DCs were found to phagocytose Salmonella in vitro, CD1b + CD14 hi and CD1b + CD14 lo [17], the phenotype of CD1b + L-DCs was characterized on cells in our possession collected from a pseudo-afferent duct draining the skin and to complete further a previous study [29]. CD1b + L-DCs were all found positive for DC-specific markers, i.e., CD11c, MHCII, CD205, CD44 and costimulatory molecules CD40, CD80, CD86, and expressed strongly these molecules (Fig. 1B, 1C). The expression of CD14 on CD1b + L-DCs was positive and relatively homogeneous (Fig. 1C), in contrast to a previous study performed in sheep [18], possibly due to the different sheep breeds used. The clone VPM65 used may recognize a specific CD14 isoform expressed on sheep L-DCs, as clones CAM36 and TUK4 did not label sheep L-DCs [19,22]. The expression of the other CD26 and SIRP-a markers on the majority of the cells demonstrated the presence of both CD26 + and SIRP-a + L-DC subsets in CD1b + L-DCs. Moreover, CD1b + L-DCs expressed CCR7 mRNA as migratory DCs, and CD103 mRNA (Fig. 1E) possibly as dermal DCs, which have been described as CD11c + CD11b lo CD103 + in mice [30], and related to the CD26 + L-DC subset which has functional similarities with CD8a-like DCs [17]. In contrast, expression of CD11b and CD206 (Mannose receptor) was observed on less than 20% of the CD1b + L-DCs, and the antibody to human CD209/DC-SIGN did not react with cells, whereas intestinal ovine DCs did [31], and despite the expression of DC-SIGN mRNA (Fig. 1E).
We then investigated the ability of these cells to capture antigens by measuring the endocytic uptake of soluble antigens as FITC-OVA. More than 50% of CD1b + L-DCs were able to endocytose soluble antigens (Fig. 1G). The capacity of CD1b + L-DCs to phagocytose Salmonella was also assessed using GFP-Salmonella and showed that within the CD1b + L-DC subset, 62% of the cells were fluorescent, whereas only 15% of CD1b 2 L-DCs were fluorescent (Fig. 1H).
Overall, CD1b + L-DCs have phenotypic features of mature DCs, including those of several conventional DC subsets, and functional abilities to uptake soluble antigens and Salmonella.
CD1b + L-DCs are able to present Salmonella antigens to specific effector/memory CD4 + T cells The ability of CD1b + L-DCs to present bacterial antigens to specific T cells, induced in vivo by SC infection of sheep by Salmonella, was also investigated. To this end, we studied the kinetics of blood T-cell activation following infection. CD4 + and CD4 2 T cells were isolated from the blood of three sheep at different times between one and nine weeks after infection to assess their capacity to be activated by CD1b + L-DCs primed in vitro with Salmonella (data not shown). The variation in the multiplicity of infection determined the optimal ratio to 100 Salmonella per DC to induce the highest proliferative response. The induction of the highest specific proliferative response of CD4 + T cells to CD1b + L-DCs primed with Salmonella was observed 21 days after infection, and no proliferative response was observed for CD4 2 T cells ( Fig. 2A). Under these conditions, CD1b + L-DCs primed in vitro with Salmonella were able to activate a high proliferative response (ratio Salmonella/Medium: 21.6) of CD4 + T cells and a lower response of CD4 2 T cells (ratio Salmonella/Medium: 4.9) isolated from non-vaccinated sheep (Fig. 2B). The priming of CD1b + L-DCs with inactivated Salmonella induced a lower proliferative response of CD4 + T cells than with live Salmonella (data not shown). Thus, in vitro, CD1b + L-DCs alone are able to phagocytose, process and present efficiently Salmonella antigens to effector/ memory CD4 + T cells.
CD1b + L-DCs are efficient APC for the priming of naive T cells The CD1b + L-DCs were then tested for their ability to prime the naive T-cell response. To this end, first an in vitro model of allogeneic response was used. Four independent experiments were performed with different sheep and showed that CD1b + L-DCs were able to induce a proliferative response of blood CD4 + T cells with an optimal ratio of 20 effector cells per DC (Fig. 3A). This proliferative response was associated with the production of cytokines in the supernatants of CD4 + T cell/CD1b + L-DC cocultures. IFN-c and IL-4 in a lesser extent, production was detected in supernatants but IL-10 was not (data not shown), demonstrating the predominance of the induced Th1-response (Fig. 3B). CD1b + L-DCs were then tested for their ability to stimulate the priming of allogeneic naive T cells. The naive T cells were sorted by the expression of L-selectin (CD62L), the homing receptor which allows their recruitment into organized lymphoid tissues such as LN via high endothelium venules [32]. CD1b + L-DCs were tested in two allogeneic experiments performed in one sheep with CD62L + T cells. One of these experiments is represented in figure 3C showing the intense proliferative response of CD62L + T cells associated mainly with IFN-c production (Fig. 3D). Thus, in vitro, CD1b + L-DCs alone are able to induce an efficient allogeneic naive T-cell response.
To evaluate the potential of CD1b + L-DCs to present bacterial antigens to naive T cells, we tested in vitro the capacity of CD1b + L-DCs to process and present Salmonella to autologous CD62L + T cells. One of the two experiments performed in one sheep is shown in figure 4A and shows a high proliferative response of autologous CD62 + T cells (ratio Salmonella/Medium: 12) in co-culture with CD1b + L-DCs infected by live Salmonella. This response was associated with the production of both IFN-c and IL-10, but not IL-4 (Fig. 4B). The CD1b + DCs are thus able to present Salmonella antigens to naive T cells efficiently.

Presentation of Salmonella Ag to CD4 + T cells by migrating DCs primed by mucosal vaccination with the Salmonella Rv-6 strain
To demonstrate further the potential of CD1b + L-DCs to process and present microbial antigens to T cells in vivo, the conjunctival route was used to inoculate the S. Abortusovis Rv-6 strain vaccine and cannulation of the cervical pseudo-afferent lymph was performed to collect CD1b + L-DCs before and after inoculation. The phenotype of cervical CD1b + L-DCs was analysed at steady state for three sheep, and no difference was observed with the cutaneous CD1b + L-DC phenotype, either in terms of the percentage of cells (Fig. 5A), or of MFI (data not shown). The relative expression of CD103 and DC-SIGN mRNA by cervical CD1b + L-DCs did not differ from that of cutaneous CD1b + L-DCs (data not shown). After inoculation of Salmonella by the conjunctival route, the proportion of CD1b + L-DCs did not differ from that before inoculation and the intensity of CD1b expression was similar. After Salmonella inoculation, the different marker expression within the CD1b + L-DCs population did not show any significant variation in the percentage of cells or in the MFI (data not shown).
CD1b + L-DCs were then isolated at different times after inoculation to assess their capacity to present Salmonella Ag by analysing the proliferative response of autologous CD4 + T cells. The results of three independent experiments show a variable proliferative response of CD4 + T cells (ratio 16h/0h: 1.8, 8.5, 1.5 respectively), but no response of CD4 2 T cells, to CD1b + L-DCs collected 16h after inoculation ( fig. 5B). This suggests the in vivo capture of Salmonella antigens by the CD1b + L-DCs, and their capacity to present them directly to CD4 + T cells.

Discussion
This study investigated the potential of migratory DCs to present Salmonella Ag directly to T cells in LN, by challenging the CD1b + L-DCs to present Ag of a Salmonella vaccine strain to specific and naive T cells using an in vitro Ag presentation assay. Although the CD1b + L-DCs display features of mature DCs, they maintain the ability to uptake soluble Ag efficiently and to phagocytose Salmonella. Our results show that CD1b + L-DCs alone are able to present Salmonella antigens to specific autologous effector/memory CD4 + T cells and also to naive T cells associated with a combined IFN-c and IL-10 response. This was also observed for the priming of allogeneic naive T cells associated with inducing both IFN-c and IL-4 responses. The potential of CD1b + L-DCs to present Salmonella Ag in vivo was also shown by collecting CD1b + L-DCs from lymph after conjunctival inoculation of Salmonella and by testing their ability to drive the amplification of autologous CD4 + T cells.
The migrating CD1b + L-DCs comprising a number of DC subsets could originate from LC, dermal or blood-derived monocytes. At steady state, the high percentage of the CD14 + subset in CD1b + L-DCs shows that a large number of L-DCs express the LPS receptor CD14, a monocyte/macrophage-specific molecule, which could testify to the common monocyte-macrophages and DC precursors [33]. The CD26 + L-DC subset which has functional similarities with CD8a-like DCs [17], also represents a high proportion of CD1b + L-DCs which could be the cells producing CD103 mRNA and possibly dermal DCs which have been described as CD11c + CD11b lo CD103 + in mice [30]. In contrast, a low percentage of migrating CD1b + L-DCs express CD11b, associated with weak expression of this molecule on the cell surface, suggesting a small proportion of classical dermal DCs within CD1b + L-DCs. Despite the high expression of MHC II and co-stimulatory molecules, CD1b + L-DCs expressed the endocytic receptor as CD206 and showed a good capacity to endocytose soluble Ag. Moreover, 60% of the CD1b + L-DCs were associated with fluorescent Salmonella, suggesting the high capacity of CD1b + L-DCs to phagocytose our Salmonella vaccine strain. This is in contrast to a previous study showing a limited ability of L-DCs to phagocytose a virulent Salmonella strain [18].
Thus, steady-state migrating CD1b + L-DCs exhibit phenotypic features of both mature and immature DCs keeping the ability, on the one hand, to endocytose soluble antigens and phagocytose Salmonella, and on the other hand, to express constitutively essential molecules involved in T-cell priming.
Our data showed that CD1b + L-DCs were able to present Salmonella antigens to specific effector/memory CD4 + T cells efficiently. In vitro they were also capable of priming autologous naive T cells to Salmonella antigens and of priming allogeneic naive T cells. However, the inactivated vaccine strain amplified CD4 + autologous T cells less efficiently than the live vaccine strain. This is in line with a previous study reporting that live virulent Salmonella Typhimurium induced a greater up-regulation of the costimulatory molecules than killed Salmonella [34]. The T-cell subset activated by Salmonella clearly comprises the CD4 + T cells, and CD1b + L-DCs directed the cytokine responses towards both IFN-c and IL-10. These results are in line with numerous studies carried out in mice on the immunity to Salmonella infection [35] and a few studies performed in sheep [13,36]. Moreover, CD1b + L-DCs were potentially able to induce IFN-c and IL-4 responses in the allogeneic reaction. This suggests the potential of these physiologically derived migrating DCs to modulate their response according to the Ag encountered. As a possible Ag delivery vehicle, this Salmonella vaccine strain may offer a good tool to drive the immune response towards a balanced IFN-c and IL-10 response.
To assess the relevance of our data obtained from an in vitro model of Ag presentation, we challenged L-DCs collected from sheep inoculated with our Salmonella vaccine strain. The phenotype of L-DCs did not change significantly after conjunctival inoculation with the Salmonella vaccine strain, in contrast to the increase observed in CD1b and CD14 expression on L-DCs exposed in vitro to a virulent strain of Salmonella [18]. These differences could be related to the attenuated virulence of the Salmonella vaccine strain, but no change was observed in the flow of L-DCs, in contrast to the recruitment of CD1b + L-DCs observed after SC injection of the same strain of Salmonella vaccine in oral mucosa [19]. This suggests that the inoculation route could modify the traffic of phagocytes in response to the vaccine strain and to set up the appropriate immune response. In the case of the SC injection, the inflammatory signals induced could trigger the rapid recruitment of granulocytes, monocytes and DCs observed in the lymph [19]. On the other hand, conjunctival administration results in a rapid local control of bacteria by non-specific defences, as previously described with the attenuated Rev1 vaccine strain of Brucella melitensis [37], and induces only a minor effect on the lymph traffic with no associated inflammatory response. However, whatever the administration route for the doses used, this vaccine strain is completely cleared in the regional draining LN and is able to induce adaptive protective immunity in pregnant ewes [13]. The CD1b + L-DCs collected after conjunctival administration were able to induce a low, but reproducible, amplification of autologous CD4 + T cells, suggesting that these CD1b + L-DCs had processed in vivo Salmonella Ag and could present them to T cells. This study did not investigate the processing of Salmonella Ag in vivo, but we can postulate that the T cells could phagocytose and process Salmonella directly from the conjunctiva and/or in the lymph, or acquire Ag from Salmonella phagocytosed and killed by granulocytes and monocytes. The main conclusion is that CD1b + L-DCs in our model present potential characteristics of APC to initiate by themselves T-cell priming in the LN, without excluding the involvement of LN-resident DCs. These migrating CD1b + L-DCs are able to uptake, process and present Ag to T cells and to modulate their response according to the Ag encountered. They could be used as target cells for driving immune activation in vaccinal strategies.