In Vivo Activation of Dendritic Cells and T Cells during Salmonella enterica Serovar Typhimurium Infection

Abstract

The present study was initiated to gain insight into the interaction between splenic dendritic cells (DC) and Salmonella enterica serovar Typhimurium in vivo. Splenic phagocytic cell populations associated with green fluorescent protein (GFP)-expressing bacteria and the bacterium-specific T-cell response were evaluated in mice given S. enterica serovar Typhimurium expressing GFP and ovalbumin. Flow cytometry analysis revealed that GFP-positive splenic DC (CD11c⁺ major histocompatibility complex class II-positive [MHC-II⁺] cells) were present following bacterial administration, and confocal microscopy showed that GFP-expressing bacteria were contained within CD11c⁺ MHC-II⁺ splenocytes. Furthermore, splenic DC and T cells were activated following Salmonella infection. This was shown by increased surface expression of CD86 and CD40 on CD11c⁺ MHC-II⁺ cells and increased CD44 and CD69 expression on CD4⁺ and CD8⁺ T cells. Salmonella-specific gamma interferon (IFN-γ)-producing cells in both of these T-cell subsets, as well as cytolytic effector cells, were also generated in mice given live bacteria. The frequency of Salmonella-specific CD4⁺ T cells producing IFN-γ was greater than that of specific CD8⁺ T cells producing IFN-γ in the same infected animal. This supports the argument that the predominant source of IFN-γ production by cells of the specific immune response is CD4⁺ T cells. Finally, DC that phagocytosed live or heat-killed Salmonella in vitro primed bacterium-specific IFN-γ-producing CD4⁺ and CD8⁺ T cells as well as cytolytic effector cells following administration into naïve mice. Together these data suggest that DC are involved in priming naïve T cells to Salmonella in vivo.

Dendritic cells (DC) are important antigen-presenting cells (APC) involved in initiating and modulating T-cell-mediated immune responses (reviewed in references 2 and 3). DC progenitors arise in the bone marrow, and through transport via the blood, they enter tissues. Murine DC from various tissues and organs share related features such as surface expression of the CD11c p150/90 integrin and constitutive expression of major histocompatibility complex class II molecules (MHC-II) and costimulatory molecules. In general, DC found in peripheral sites such as skin and mucosal surfaces are in an immature stage. That is, they are optimized for capturing and processing antigens but are relatively poor stimulators of naïve T cells (3, 36). Exposure to antigen and inflammatory stimuli initiates a maturation process whereby immature DC become effective activators of T cells and are directed to sites of lymphocyte priming (3, 15, 18, 21, 36).

Although the role of DC in priming naïve T cells to protein antigens is well established (36), a remaining unanswered question relates to the role of this APC relative to other phagocytic APC, such as macrophages (MΦ), in triggering bacterium-specific T cells following bacterial internalization in vivo. Using Salmonella enterica serovar Typhimurium as a model bacterium, it has been shown that both MΦ and immature DC can present antigens processed from this facultative intracellular gram-negative bacterium and induce DC maturation in vitro (33, 37, 39, 44–46). The ability of S. enterica serovar Typhimurium to reside and replicate within phagosomes of phagocytic cells (4, 7, 26) makes this an interesting model to study bacterial interaction with APC in vivo. For example, S. enterica serovar Typhimurium has been found in CD18-expressing cells (34, 42), which include various APC populations (35). The bacterium has also been shown to be associated with CD11c⁺ cells of FLT3-L-treated mice (22) and within CD11c⁺ cells of the subepithelial dome overlying Peyer's patches following administration of bacteria (12). However, despite its association with various phagocytic populations in vivo, and the well-characterized role of T cells in host defense against Salmonella (11, 23, 27, 31, 43), the nature of the APC that primes Salmonella-specific T cells during infection, particularly naïve T cells, is not clear.

The present study addresses the role of DC as APC involved in the specific immune response to S. enterica serovar Typhimurium in vivo. Following a single administration of Salmonella expressing green fluorescent protein (GFP), GFP-positive (GFP⁺) cells among CD11c⁺ MHC-II⁺ splenocytes were apparent, and confocal microscopy showed that bacteria were inside splenic DC (CD11c⁺ MHC-II⁺ cells). In addition, increased surface expression of activation markers on both DC and T cells occurred following a single dose of bacteria, and Salmonella-specific CD4⁺ and CD8⁺ effector T cells were generated. Finally, DC loaded with Salmonella elicited specific effector T cells following injection into naïve hosts. Together these data support a role for DC in eliciting specific anti-Salmonella immunity.

MATERIALS AND METHODS

Mice.

C57BL/6 mice were bred and maintained in the animal facilities at Lund University (Lund, Sweden) and were offered food and water ad libitum. All mice were age matched and used at 8 to 12 weeks of age.

Bacterial strains and culture conditions.

S. enterica serovar Typhimurium χ4550 (SR11 pStSR100⁺ gyrA1816 Δcrp-1 ΔasdA1 Δcya-1) was used as a host for expression of the model antigen ovalbumin (OVA) in the vector pYA3259. pYA3259 is a pUCI8-derived plasmid encoding a truncated form of aspartate β-semialdehyde dehydrogenase (ASD). ASD is an enzyme in the biosynthetic pathway for diaminopimelic acid (DAP), an essential component of the peptidoglycan of the bacterial cell wall; asd mutant bacteria undergo lysis in the absence of DAP. As DAP is not present in mammalian tissues, use of an ΔasdA1-ASD⁺ host-vector pair allows plasmid maintenance in vivo in the absence of antibiotic selection (8, 30).

OVA was cloned into pYA3259 by amplifying a 1.2-kb fragment encoding OVA from the cDNA contained in pOV230 (24). The oligonucleotide primers incorporated an EcoRI site and a Shine-Dalgarno sequence 5′ of the OVA start codon and a BamHI site 3′ of the stop codon. The fragment was cloned into the corresponding sites of pYA3259, resulting in the plasmid pYA3259-OVA. GFP was cloned into pYA3259-OVA by digesting pSK-XhoGFP (40) with Xho and purifying the 0.9-kb fragment encoding GFP after agarose gel electrophoresis. This fragment was subsequently ligated into SalI-digested pYA3259-OVA to create pYA3259-OVA-GFP. S. enterica serovar Typhimurium χ4550 harboring pYA3259, called χ4550; S. enterica serovar Typhimurium χ4550 harboring pYA3259-OVA, called χ4550 OVA; and S. enterica serovar Typhimurium χ4550 harboring pYA3259-OVA-GFP, called χ4550 OVA-GFP, were used in these studies. Bacteria were grown overnight at 37°C with shaking in Luria-Bertani (LB) broth and were quantitated spectrophotometrically by determining the optical density at 600 nm. The bacteria were then centrifuged at 2,300 × g for 5 min and resuspended in Iscove's modified Dulbecco's medium (IMDM) (Life Technologies, Gaithersburg, Md.) without antibiotics. The quantity of live bacteria actually given to mice was determined by viable plate counts. Heat-killed bacteria were prepared by incubating a bacterial suspension at 65°C for 40 min. Loss of bacterial viability was confirmed by plating an aliquot of heat-killed bacteria on LB agar plates.

Immunization of mice.

In experiments where DC interactions with GFP-expressing bacteria were studied, mice were given a single intravenous or intraperitoneal (i.p.) injection of 6 × 10⁸ live or 3 × 10⁹ heat-killed χ4550 OVA-GFP and were sacrificed 4 h later. In experiments where DC and T-cell activation were analyzed, mice were given a single i.p. injection of 10⁶ live or heat-killed χ4550 OVA and were sacrificed 14 days later. In experiments where Salmonella-loaded DC were used (see below), mice were given either 10⁶ DC incubated in medium alone (control DC), 10⁶ DC loaded with live χ4550, or 10⁶ DC loaded with live or heat-killed χ4550 OVA (see below). Mice received two i.p injections 1 week apart and were sacrificed 14 days after the last administration. The spleens from immunized animals were used to isolate DC as described below or were homogenized to analyze effector cell functions. Enriched DC or total splenocytes were used in flow cytometry analysis, and total splenocytes were used in intracellular cytokine staining and cytotoxic T-cell assays. The number of bacteria remaining in the spleen of mice was determined by lysing 10⁷ spleen cells in 0.1% Triton X-100 in phosphate-buffered saline and plating on LB agar plates.

Isolation of splenic DC.

In experiments where splenic DC were purified from mice given GFP-expressing Salmonella, all steps were carried out on ice to avoid ex vivo bacterial uptake. The spleens from immunized animals were homogenized, red blood cells were lysed, and cell suspensions were washed once in Hanks buffered salt solution (HBSS). Then, splenic CD11c-expressing cells were enriched using N418 magnetic beads and MiniMACS columns (Miltenyi Biotech GmbH, Bergisch Gladbach, Germany) following the manufacturer's protocol. Cell suspensions were incubated with N418 magnetic beads for 15 min in phosphate-buffered saline containing 3% fetal calf serum (FCS) and 2 mM EDTA at 4°C. Magnetically labeled cells were then added on a column, and trapped cells were flushed out to be used in flow cytometry or confocal microscopy. The enriched population consisted of approximately 80 to 90% CD11c⁺ cells as determined by the antibody N418 (anti-CD11c) or HL3 (anti-CD11c) in flow cytometry analysis (see below).

Flow cytometry.

Flow cytometry analysis was performed using a FACS Calibur flow cytometer (Becton Dickinson & Co., Mountain View, Calif.). Antibodies from hybridomas 2.4.G2 (anti-FcRγII/III), N418 (anti-CD11c), M5/114 (anti-MHC-II), and GK1.5 (anti-CD4) (reference 37 and references therein), as well as 145.2C11 (anti-CD3 [17]), YTS.169 (anti-CD8α [5]), F4/80 (anti-F4/80 [1]), and RA6-3A2 (anti-B220 [6]), were used. Antibodies were purified from supernatants using γ-bind plus columns (Pharmacia-Biotech, Uppsala, Sweden) and were labeled with biotin (Sigma Chemical Co, St. Louis, Mo.), fluorescein isothiocyanate (FITC) (Sigma), or Cy5 (Amersham Pharmacia Biotech, Uppsala, Sweden). Biotinylated anti-GR-1 as well as phycoerythrin (PE)-labeled anti-CD4, -CD8, -CD11c, -CD40, -CD86, -B220, and -MHC-II antibodies were purchased from Pharmingen (San Diego, Calif.). Streptavidin-allophycocyanin (Pharmingen) was used as the second-step reagent. Dead cells were excluded by staining with 7-amino-actinomycin D (7AAD) (Sigma). Incubations with antibodies or reagents for surface staining were performed for 20 min on ice in HBSS (Life Technologies) containing 3% FCS, 2 mM EDTA, and 0.01% sodium azide.

Confocal microscopy.

For confocal microscopy studies, cells were kept on ice at all possible steps and cold buffers and reagents were used. MACS column-purified CD11c⁺ cells were stained on ice with anti-CD11c-Cy5 and anti-MHC-II-biotin followed by Streptavidin-Alexa Fluor 594 (Molecular Probes, Leiden, The Netherlands). The cells were fixed in 1% paraformaldehyde on ice for 30 min and then at room temperature for another 30 min before being added onto polylysine-treated slides and mounted with ProLong Antifade (Molecular Probes). The cells were inspected by epifluorescence and laser scanning confocal microscopy using MRC-1024 confocal equipment (Bio-Rad Laboratories, Hemel-Hampstead, United Kingdom) attached to a Nikon Eclipse E800 upright microscope (Nikon, Tokyo, Japan) using a band-pass filter (522 ± 32 nm) to detect green fluorescence, another band-pass filter (598 ± 40 nm) to detect red fluorescence, and another band-pass filter (680 ± 32 nm) to detect far-red fluorescence.

Cell culture conditions and in vitro infections of DC.

DC were cultured from C57BL/6 bone marrow as described elsewhere (39, 40). Briefly, on days 6 to 7 of culture in IMDM supplemented with 5% FCS and granulocyte-macrophage colony-stimulating factor, the CD11c-expressing population was enriched using N418 magnetic beads (Miltenyi Biotech GmbH). The enriched population consisted of approximately 90% MHC-II⁺, CD86⁺, and CD11c⁺ cells as determined by the antibodies M5/114 (anti-I-A^b), GL-1 (anti-CD86), and N418 (anti-CD11c) or HL3 (anti-CD11c) in flow cytometry analysis. Purified DC were resuspended in IMDM (without antibiotics) containing 5% FCS and granulocyte-macrophage colony-stimulating factor and seeded at 10⁶ cells per well in 24-well plates (Ultra Low Cluster plates; Costar Corning Corp., Cambridge Mass.). The cells were then infected with live or heat-killed bacteria at a bacteria/DC ratio of 50:1. The 24-well plates were centrifuged at 270 × g for 5 min. Some wells containing DC were incubated in IMDM containing 5% FCS for use as control DC. DC were incubated for 2 h at 37°C, were washed three times with HBSS, and were incubated for an additional 2 h at 37°C in IMDM containing 5% FCS and gentamicin (50 μg/ml). The cells were washed once in HBSS and were resuspended in HBSS (without antibiotics). They were then used either to inject into mice or to restimulate splenocytes. The effective dose of live bacteria given to mice by administering DC loaded with bacteria was approximately 10⁶ as determined by lysing aliquots of Salmonella-loaded DC in 0.1% Triton X-100 in phosphate-buffered saline and plating on LB agar plates. In some experiments, samples of DC infected with bacteria were transferred to 96-well plates, were fixed in 1% paraformaldehyde, and presentation of OVA(257-264)/K^b and OVA(265-280)/I-A^b complexes processed from χ4550 OVA to CD8OVA or OT4H.2D5 T hybridoma cells, respectively, was quantitated as described previously (37).

Intracellular cytokine staining.

Gamma interferon (IFN-γ) production by splenocytes was determined by intracellular cytokine staining and flow cytometry analysis. Briefly, 2 × 10⁷ spleen cells from individual immunized mice were cultured with DC incubated in medium alone or with DC loaded with heat-killed bacteria in 3 ml of IMDM supplemented with 10% FCS. Incubations were at 37°C and 5% CO₂ for a total of 24 h. The splenocyte to DC ratio was 20:1. Four hours prior to the end of the incubation, brefeldin A was added to a concentration 5 μg/ml. Brefeldin A-treated cells were labeled with anti-T-cell receptor αβ (anti-TCRαβ), and with either anti-CD4 or anti-CD8 antibodies as described above and were fixed in 2% paraformaldehyde for 20 min at room temperature. Fixed cells were permeabilized in HBSS containing 3% FCS (Sigma), 0.5% saponin (Sigma), and 0.05% azide for 20 min at room temperature. Permeabilized cells were stained with anti-IFN-γ–FITC (Pharmingen) or with FITC-labeled isotype-matched control antibodies (Pharmingen) for 30 min at room temperature. Flow cytometry was performed on splenocytes from individual mice stimulated separately.

Cytotoxic T-cell assays.

Cytotoxic T-cell activity in the spleens of mice given χ4550 OVA was determined by coculturing 25 × 10⁶ spleen cells from individual mice with 12.5 × 10⁶ irradiated syngeneic spleen cells in the presence of 100 nM OVA(257-264) peptide in 12 ml of IMDM supplemented with 10% FCS and antibiotics at 37°C and 5% CO₂ for 5 days. Cell-mediated lysis was subsequently measured in standard ⁵¹Cr release assays on splenocytes from individual mice cultured and assayed separately. Briefly, target cells were prepared by loading EL4 cells with 50 μM OVA(257-264) peptide for 60 min at 37°C. Titrated amounts of restimulated effector cells were incubated with 10^{4 51}Cr-labeled target cells for 5 h at 37°C in a total volume of 100 μl. Following the incubation, 25 μl of supernatant was collected and counted in a Wallac 1450 MicroBeta counter (Wallac Oy, Turku, Finland) to quantitate the amount of released ⁵¹Cr. The maximum release of ⁵¹Cr was determined by addition of 1% sodium dodecyl sulfate to ⁵¹Cr-labeled target cells. The mean number of counts per minute was calculated from triplicate wells. The percent specific cytotoxicity was calculated according to the following formula: [(mean cpm test cell release − mean cpm spontaneous release) ÷ (mean cpm maximum release − mean cpm spontaneous release)] × 100, where cpm is counts per minute.

RESULTS

S. enterica serovar Typhimurium association with splenic phagocytic cell subsets during infection.

To assess the splenic phagocytic cell subpopulations that interact with S. enterica serovar Typhimurium upon infection of mice, flow cytometry was performed on splenocytes 4 h after i.p. or intravenous administration of χ4550 OVA-GFP. Several populations of phagocytic cells including CD11c⁺ MHC-II⁺ cells were GFP⁺, suggesting that DC associated with the bacteria in vivo using either route of administration (Fig. 1a and Table 1). CD11c⁺ MHC-II⁺ cells were also GFP⁺ following administration of heat-killed χ4550 OVA-GFP, although the percentage of GFP⁺ cells was consistently lower in mice given heat-killed bacteria (Fig. 1a). This could be partly due to the somewhat lower GFP fluorescence detected after heat killing the bacteria (Fig. 1b).

FIG. 1.

Splenic DC associate with GFP-expressing Salmonella. (a) Mice were given either 6 × 10⁸ live or 3 × 10⁹ heat-killed χ4550 OVA-GFP as indicated. Four hours later, splenocytes from two mice were pooled, splenic DC were purified on a MACS column and stained with 7AAD, PE-conjugated anti-CD11c, and biotinylated anti-MHC-II followed by streptavidin-allophycocyanin. The R1 region marked in the histogram indicates the CD11c gate set to analyze MHC-II expression shown in the adjacent contour plot. The contour plot shows the MHC-II expression on the gated CD11c⁺ splenocytes (R1). The region R2 indicates the CD11c⁺ MHC-II⁺ cell population that was analyzed for GFP fluorescence and is shown in the three dot plots. The number in each dot plot represents the percentage of CD11c⁺ MHC-II⁺ cells (defined by gates R1 plus R2) that are also GFP⁺. A total of 200,000 CD11c⁺ MHC-II⁺ cells were analyzed for GFP fluorescence. The panel marked control is DC purified from untreated mice. Mice immunized with either live or heat-killed χ4550 OVA resulted in detection of 0.01% GFP⁺ cells within R2, i.e., the same level as that in control mice. The results are representative of four independent experiments. (b) The histogram shows the GFP fluorescence of χ4550 OVA (dotted line), live χ4550 OVA-GFP (thin line), and heat-killed χ4550 OVA-GFP (thick line). Flow cytometry analysis on live or heat-killed χ4550 OVA-GFP was performed after 4 h in IMDM containing 10% mouse serum at 37°C.

TABLE 1.

GFP-positive cells in different cell populations in mice infected with S. enterica serovar Typhimurium expressing GFP

Cell populationa	% GFP-positive cells after immunizationb with indicated no. of bacteria
	Live		HK
	5 × 107	5 × 108	5 × 108	2.5 × 109
CD11c− CD3+	0.00	0.04	0.00	0.01
CD11c− B220+	0.02	0.80	0.01	0.08
CD11c− GR-1+	0.71	7.46	0.51	3.30
CD11c− F480+	0.50	5.20	0.15	1.01
CD11c+ MHC-II+	0.39	2.76	0.09	0.57

^aThe number of GFP⁺ cells (as defined in the dot plots of Fig. 1) in nonimmunized mice or in mice given either live or heat-killed χ4550 OVA (not expressing GFP) was 0.01%. 

^bThe number of live or heat-killed (HK) χ4550 OVA-GFP administered i.p. as indicated is shown. Similar results were obtained using intravenous administration of bacteria. The data are from splenocytes pooled from two mice for each group and are representative of three independent experiments. 

Two findings support the argument that the GFP fluorescence of splenocytes was due to uptake of χ4550 OVA-GFP by phagocytic cells rather than nonspecific binding of the bacteria to the cell surface. First, fluorescent confocal microscopy showed that GFP⁺ bacteria were indeed contained within CD11c⁺ MHC-II⁺ cells (Fig. 2). Second, further characterization of the GFP⁺ cell populations revealed that little, if any, GFP fluorescence was associated with nonphagocytic cells such as CD11c⁻ CD3⁺ cells (T lymphocytes) (Table 1). Furthermore, GFP fluorescence associated with CD11c⁻ B220⁺ B lymphocytes was only apparent at a very high dose of administered bacteria (Table 1). Within other CD11c⁻ cell populations, namely, non-DC populations, significant GFP fluorescence was associated with cells that stained positive for GR-1 and F4/80. The former surface marker on CD11c⁻ cells is associated with granulocytes (16) while MΦ are contained within the CD11c⁻ population staining positive for F4/80 (1). Thus, S. enterica serovar typhimurium associates with several phagocytic splenic cell populations, with the largest percentage of GFP⁺ cells being consistently found within the CD11c⁻ GR1⁺ population, followed by GFP fluorescence associated with CD11c⁻ F4/80⁺ cells and then CD11c⁺ MHC-II⁺ cells (Table 1).

FIG. 2.

GFP⁺ Salmonella is contained within CD11c⁺ MHC-II⁺ splenocytes. Four hours following i.p. administration of 5 × 10⁸ χ4550 OVA-GFP cells, splenocytes from two mice were pooled, and splenic CD11c⁺ DC were purified on a MACS column and stained with Cy5-conjugated anti-CD11c (blue) and biotinylated anti MHC-II followed by streptavidin-Alexa Flour 594 (red). Cells were subsequently fixed with paraformaldehyde, placed on a polylysine-coated microscope slide, and inspected by epifluorescence and laser confocal microscopy. The left panel shows a section through a CD11c⁺ MHC-II⁺ cell containing two GFP⁺ bacteria. The right panels show the images taken when the cell was cut along the indicated a, b, or c axis and confirms that the GFP⁺ bacteria remain confined within the CD11c⁺ MHC-II⁺ cell surface.

DC and T-cell activation in response to Salmonella infection.

As Salmonella is contained within splenic DC in vivo (Fig. 2), and as these cells are the most efficient APC type capable of stimulating naïve T cells (2, 3, 36), we investigated the activation state of both DC and T cells in infected mice. Mice were given a single dose of live χ4550 OVA, and 14 days later splenic DC were analyzed for surface molecule expression. Flow cytometry revealed that the CD11c⁺ MHC-II⁺ cells had increased expression of CD86 and CD40 (Fig. 3) as well as a small but consistent increase in CD80 expression (data not shown). No effect was apparent on DC from mice that received heat-killed χ4550 OVA (Fig. 3).

FIG. 3.

Splenic DC are activated in mice given a single dose of S. enterica serovar Typhimurium. Mice were given 2.5 × 10⁶ live or heat-killed χ4550 OVA i.p. Two weeks later, splenocytes from individual mice were stained with 7AAD, FITC-conjugated anti-MHC-II, Cy5-conjugated anti-CD11c, and either PE-conjugated anti-CD86 or anti-CD40 and were analyzed by four-color flow cytometry. The histograms show the CD86 and CD40 expression on gated CD11c⁺ MHC-II⁺ splenocytes (R2 as in Fig. 1a) from mice given either live (thick line) or heat-killed (dotted line) χ4550 OVA. The thin line represents the fluorescence intensity of the gated CD11c⁺ MHC-II⁺ splenocytes from untreated mice. The staining of cells from an infected mouse using an isotype-matched control antibody is indicated. A similar level of staining using an isotype control antibody was also observed on splenocytes from mice immunized with heat-killed bacteria or naïve mice (data not shown). A total of 10,000 CD11c⁺ MHC-II⁺ cells were analyzed for CD86 or CD40 expression. The results shown are from individual mice and are representative of four independent experiments with a total of six mice per group.

Surface expression of the activation markers CD44 and CD69 was also increased on splenic T cells from Salmonella-infected mice, where more cells expressing high levels of CD44 and CD69 were present in both CD4⁺ and CD8⁺ cell populations following a single administration of live bacteria (Fig. 4). In contrast, little, if any, effect on CD44 and CD69 expression on splenic T cells was apparent in mice that received heat-killed bacteria. Together these data show that a single dose of live but not heat-killed S. enterica serovar Typhimurium results in a significant increase in costimulatory molecule expression on splenic DC and increased expression of CD44 and CD69 activation markers on CD4⁺ and CD8⁺ splenic T cells.

FIG. 4.

CD44 and CD69 surface expression is enhanced on CD4⁺ and CD8⁺ splenic T cells from mice given a single dose of χ4550 OVA. Mice were given 2.5 × 10⁶ live or heat-killed χ4550 OVA i.p. Two weeks later, splenocytes from individual mice were stained with 7AAD, Allophycocyanin-conjugated TCRαβ, and either FITC-conjugated anti-CD4 or anti-CD8. Cells were subsequently stained with either PE-conjugated anti-CD44 or anti-CD69. The top two histograms show CD44 and CD69 expression on gated TCRαβ⁺ CD4⁺ cells, while the bottom histograms show CD44 and CD69 expression on gated TCRαβ⁺ CD8⁺ cells from the spleens of mice given either live (thick line) or heat-killed (dotted line) χ4550 OVA. The x axes represent log fluorescence intensity, while the y axes represent the number of gated events as indicated. CD44 and CD69 expression on gated CD4⁺ and CD8⁺ T cells from untreated mice (thin line) is shown for comparison. A total of 10,000 gated T cells were analyzed for CD44 or CD69 expression. The results shown are from individual mice and are representative of seven independent experiments with a total of nine mice per group. Appropriate isotype subclass control antibodies showed no significant staining on gated CD4⁺ and CD8⁺ splenocytes.

IFN-γ-producing CD4⁺ and CD8⁺ T cells are elicited in Salmonella-infected mice.

T cells are a critical component of the specific immune response to S. enterica serovar Typhimurium, and bacterium-specific CD4⁺ and CD8⁺ effector T cells are elicited following infection (reviewed in references 27 and 44). Although previous studies have enumerated Salmonella-specific IFN-γ-producing CD4⁺ T cells by enzyme-linked immunosorbent spot assay following a 2- to 4-day restimulation of splenocytes from infected mice (25, 32, 41), neither the frequency of Salmonella-specific IFN-γ-producing CD4⁺ T cells following a brief ex vivo restimulation nor the frequency of CD8⁺ T cells that specifically contribute to IFN-γ production in infected mice have been quantitated. Thus, flow cytometry analysis was performed on splenocytes from mice given χ4550 OVA. Upon restimulation for 24 h, the cells were stained for surface expression of TCRαβ and either CD4 or CD8 as well as intracellular IFN-γ. These data revealed that Salmonella-specific IFN-γ-producing CD4⁺ as well as CD8⁺ T cells were generated in mice given χ4550 OVA once (Fig. 5). In contrast, no Salmonella-specific IFN-γ-producing CD4⁺ T cells (<0.1%) and few if any IFN-γ-producing CD8⁺ T cells (0.2%) were apparent in mice receiving heat-killed bacteria once (Fig. 5) or twice (one week apart; data not shown).

FIG. 5.

Both CD4⁺ and CD8⁺ Salmonella-specific IFN-γ-secreting T cells are elicited in mice given χ4550 OVA. Mice were given either 10⁶ live or heat-killed χ4550 OVA cells or were left untreated (control) as indicated. Two weeks later, splenocytes from individual mice were restimulated in vitro with DC loaded with heat-killed χ4550 OVA for 24 h and were subsequently analyzed for intracellular IFN-γ. Restimulated splenocytes were stained with 7AAD, allophycocyanin-conjugated anti-TCRαβ, and either PE-conjugated anti-CD4 and anti-IFN-γ–FITC or PE-conjugated anti-CD8 and anti-IFN-γ–FITC. Dot plots show IFN-γ-producing CD4⁺ cells (left) and CD8⁺ T cells (right). The number in each dot plot represents the percentage of CD4⁺ or CD8⁺ T cells staining positive for intracellular IFN-γ in the indicated gates. Splenocytes from untreated (control) mice restimulated in vitro with DC loaded with heat-killed χ4550 OVA had ≤0.1% of gated cells that stained positive for IFN-γ. Likewise, splenocytes from χ4550 OVA-infected mice restimulated in vitro with DC in medium alone stained positive for IFN-γ at a low level (≤ 0.8% of CD4⁺ or 0.2% CD8⁺ gated T cells). FITC-labeled isotype-matched control antibodies stained less than 0.1% of the CD4⁺ or CD8⁺ T cells (not shown). A total of 30,000 gated T cells were analyzed. The results shown are from individual mice and are representative of four independent experiments with a total of seven mice per group.

Salmonella-loaded DC generate bacteria-specific effector T cells in recipient mice.

As it previously has been shown that immature DC can internalize and present Salmonella-derived antigens on MHC-I and MHC-II in vitro (39, 45), we addressed the question of whether these DC loaded with Salmonella in vitro could prime bacterium-specific T cells in vivo. Flow cytometry analysis of splenocytes harvested from mice given DC loaded with χ4550 OVA showed that both CD4⁺ and CD8⁺ cells produced Salmonella-specific IFN-γ following in vitro restimulation (Fig. 6a). The CD4⁺ population consisted of a higher frequency of Salmonella-specific IFN-γ-producing cells than the CD8⁺ population. In addition, splenocytes from mice that received DC loaded with bacteria in the presence of cytochalasin D generated undetectable levels of specific IFN-γ-producing CD4⁺ or CD8⁺ T cells (data not shown). This demonstrates that bacteria contained within the DC, rather than bacteria adhering to the outer surface of the cells, were responsible for eliciting Salmonella-specific T cells. Specific cytolytic activity was also elicited in recipient mice. That is, splenocytes from mice given DC loaded with χ4550 OVA and restimulated with irradiated syngeneic spleen cells in the presence of the OVA(257-264) peptide lysed EL4 target cells pulsed with the same peptide (Fig. 6b). In contrast, little if any lysis of OVA(257-264)-loaded target cells was apparent when splenocytes from mice receiving control DC or DC loaded with χ4550 not expressing OVA were restimulated and used as effector cells in ⁵¹Cr release assays (Fig. 6b). Interestingly, we consistently observed that the percent specific lysis elicited in mice given DC loaded with χ4550 OVA was higher than that observed in mice given χ4550 OVA as free bacteria. This occurred despite the fact that both administrations contained similar amounts of live bacteria (∼10⁶) (Fig. 6b). In addition, the spleens from mice that received either type of administration had similar numbers of CFU at the time of sacrifice (see below) and never exceeded a difference of twofold.

FIG. 6.

DC loaded with live Salmonella induce bacterium-specific IFN-γ-producing CD4⁺ and CD8⁺ T cells and cytotoxic effector cells. (a) Mice were given DC loaded with χ4550 OVA or control DC as indicated on two occasions 1 week apart. Two weeks after the last administration, splenocytes from individual mice were restimulated in vitro with DC loaded with heat-killed χ4550 OVA for 24 h and were subsequently analyzed for intracellular IFN-γ. Restimulated splenocytes were stained with 7AAD, biotinylated anti-CD4 followed by streptavidin-allophycocyanin, PE-conjugated anti-CD8, and anti-IFN-γ–FITC. Dot plots of flow cytometry analysis of CD4⁺ (left) and CD8⁺ (right) are shown. The number in each dot plot is the percentage of CD4⁺ or CD8⁺ cells staining positive for intracellular IFN-γ in the indicated gates. Splenocytes from mice given control DC restimulated in vitro with DC only stained positive for IFN-γ at a low level (0.1% of CD4⁺ or CD8⁺ gated cells). Likewise, splenocytes from mice given DC loaded with χ4550 OVA restimulated in vitro with DC only stained positive for IFN-γ at a low level (≤0.9% of CD4⁺ or 0.1% CD8⁺ gated cells). FITC-labeled isotype-matched control antibodies stained less than 0.1% of the cells in the indicated gates. A total of 30,000 gated CD4⁺ or CD8⁺ cells were analyzed. (b) Mice were given either DC loaded with live χ4550 OVA (DC + χ4550 OVA), DC loaded with live χ4550 not expressing OVA (DC + χ4550), control DC (DC), or free χ4550 OVA as indicated. Two weeks after the second injection, splenocytes from individual mice were restimulated in vitro with OVA(257-264) peptide and were analyzed for cytotoxic activity in ⁵¹Cr release assays. The percent lysis of either OVA(257-264)-loaded (left panels) or unloaded (right panels) EL4 target cells is shown. The results shown are from individual mice and are representative of four independent experiments with a total of four mice per group.

As the DC used in these studies were loaded with live bacteria, it was possible that live bacteria released from the DC contributed to generating the observed effector T cells. Despite the fact that the spleens of mice contained few viable bacteria at the time of sacrifice (∼100 to 1,000 total bacteria), we performed experiments in which mice were injected with DC loaded with heat-killed χ4550 OVA to eliminate any contribution of live bacteria in eliciting Salmonella-specific T cells. Despite the absence of viable bacteria in the DC used for the injections (as tested by viable plate counts on the heat-killed bacteria and an aliquot of the injected DC; data not shown), Salmonella-specific IFN-γ-producing CD4⁺ and CD8⁺ cells were generated (Fig. 7a). In addition, OVA(257-264)-specific cytolytic effector cells were also generated in mice given DC loaded with heat-killed χ4550 OVA (Fig. 7b). Thus, DC loaded with χ4550 OVA elicited bacterium-specific CD4⁺ and CD8⁺ effector T cells in recipient mice, and this did not require that the DC contain live bacteria.

FIG. 7.

DC loaded with heat-killed Salmonella induce bacterium-specific IFN-γ-producing and cytotoxic T cells. (a) Mice were immunized with DC loaded with heat-killed χ4550 OVA on two occasions 1 week apart. Splenocytes from individual mice were restimulated with DC loaded with heat-killed χ4550 OVA (top two dot plots) or with DC only (bottom two dot plots). The cells were then stained with 7AAD and allophycocyanin-conjugated anti-TCRαβ and either PE-conjugated anti-CD4 or anti-CD8 followed by anti-IFN-γ–FITC. Dot plots of flow cytometry analysis of TCRαβ⁺ CD4⁺ (left) and TCRαβ⁺ CD8⁺ (right) are shown. The number in each dot plot shows the percentage of cells staining positive for intracellular IFN-γ. FITC-labeled isotype-matched control antibodies stained less than 0.1% of the T cells. A total of 80,000 T cells were analyzed. (b) Mice were given DC loaded with either live (DC + χ4550 OVA) or heat-killed (DC + heat-killed χ4550 OVA) χ4550 OVA or with live χ4550 not expressing OVA (DC + χ4550) as indicated. Cytotoxic activity was measured as in Fig. 6b. The results shown are from individual mice and are representative of four independent experiments with a total of four mice per group.

DISCUSSION

The importance of peripheral DC in capturing exogenous soluble antigens and migrating to draining lymph nodes where they prime naïve T cells is well established (15, 18, 21, 36). In contrast, the role of immature DC in capturing intact microbes encountered in the periphery and their subsequent migration to draining lymph nodes for T-cell priming is not well characterized, with the exception of cutaneous infection with Leishmania major (28, 29). Thus, the APC type(s) that phagocytose gram-negative bacteria, such as S. enterica serovar Typhimurium, and prime Salmonella-specific T cells in secondary lymphoid organs in vivo is not clear.

In the present study we show that the three major phagocytic cell types, neutrophils, macrophages, and DC, all associate with GFP-expressing Salmonella in vivo. These phagocytic cells may be involved in controlling bacterial replication and/or producing proinflammatory cytokines (37, 38). As the latter may influence the subsequent recruitment and activation of cells of the innate as well as the adaptive immune response, it is clear that the interplay between phagocytic cells harboring bacteria is complex. However, DC have a superior capacity to stimulate naïve T cells (36). This, combined with our data showing that splenic CD11c⁺ MHC-II⁺ DC internalize Salmonella in vivo, suggests that DC may be involved in priming bacterium-specific T cells. Furthermore, we show that splenic DC have increased surface expression of CD40 and CD86 following administration of live Salmonella, which augments their capacity to productively interact with T cells. The activation of DC following bacterial administration was likely not due to a general effect of lipopolysaccharide. This is supported by data showing that the surface expression of CD40 and CD86 on splenic CD11c⁻ B220⁺ B cells, which are polyclonally activated in mice injected with lipopolysaccharide isolated from Salmonella (13), was identical in naïve mice and in mice that received χ4550 14 days earlier (data not shown).

Infection with live S. enterica serovar Typhimurium resulted in broad activation of CD4⁺ and CD8⁺ T cells and generation of Salmonella-specific IFN-γ-producing T cells in both of these subsets. Although the specificity of the broadly activated T cells is not clear, it is likely that not all of these cells are bacterium specific; at least some of the increased CD44 and CD69 expression may reflect bystander activation. Generation of bacterium-specific IFN-γ-producing CD4⁺ and CD8⁺ T cells is consistent with the critical role of TCRαβ⁺ cells, particularly the CD4⁺ subset, and IFN-γ in immunity to Salmonella (reviewed in reference 27). Our data revealed that the frequency of specific IFN-γ-producing CD4⁺ T cells was significantly greater than that of CD8⁺ cells in mice given a single dose of live Salmonella. This is the first quantitation of the frequency of Salmonella-specific CD4⁺ versus CD8⁺ cells producing IFN-γ during infection of the same animal and supports the argument that the predominant source of IFN-γ production by cells of the specific immune response is CD4⁺ T cells. Although Salmonella-specific IFN-γ-producing CD8⁺ T cells are elicited following infection with live bacteria, the contribution of IFN-γ production by specific CD8⁺ T cells appears to be minor relative to that produced by CD4⁺ T cells. This suggests that the importance of the elicited specific CD8⁺ T cells may be in their cytolytic function rather than IFN-γ production.

Studies in β₂m^−/− mice showed that clearance of an avirulent S. enterica serovar Typhimurium strain occurred independently of CD8⁺ T cells (11), while CD8⁺ T cells were important in clearing a virulent strain (20). The present study demonstrates that cytolytic CD8⁺ T cells are elicited following infection with S. enterica serovar Typhimurium, supporting previous reports (reviewed in reference 44). It has also been shown that cytolytic CD8⁺ T-cell activity restricted to the nonclassical MHC-I molecule Qa-1 is generated in mice primed with an avirulent strain and then challenged with a virulent strain (20). However, in preliminary experiments, we were unable to demonstrate Salmonella-specific IFN-γ-producing CD8⁺ splenocytes from C57BL/6 mice given a single dose of χ4550 following restimulation with DC from MHC-mismatched (BALB/c) mice loaded with χ4550. This suggests that the IFN-γ-producing CD8⁺ cells analyzed in the present study are restricted to classical MHC-I molecules rather than class Ib molecules shared by C57BL/6 and BALB/c mice.

Poor immunity in mice receiving heat-killed S. enterica serovar Typhimurium has been documented for nearly 30 years (10). However, only limited data are available on the defects in the specific immune response that could account for the poor immunogenicity of heat-killed bacteria (9). Our data show that heat-killed bacteria associate with DC and other phagocytic cells in vivo. However, administration of 5 to 10 times more heat-killed than live bacteria was required to observe a quantatively similar association of bacteria by phagocytic cells. In addition, immunization with 10⁶ heat-killed bacteria did not result in up regulation of CD86 or CD40 on splenic DC analyzed 2 weeks after bacterial administration. This is in contrast to the increased expression of these costimulatory molecules on splenic DC from mice given the same dose of live bacteria. Furthermore, an increase in the total number of CD11c⁺ splenocytes was present in mice infected with live Salmonella (14), while no such increase was present in mice given heat-killed bacteria (data not shown).

Significant T-cell activation was also not apparent following administration of killed bacteria. That is, CD44 and CD69 expression were not increased on CD4⁺ or CD8⁺ T cells, and IFN-γ-producing Salmonella-specific CD4⁺ T cells were not detectable at time points when T-cell activation peaked in mice given the same amount of live bacteria. However, very low, albeit reproducible, levels of bacterium-specific CD8⁺ cells producing IFN-γ were apparent following administration of heat-killed χ4550. The poor ability of heat-killed Salmonella to elicit specific T cells is not due to an inherent inability of DC to process heat-killed S. enterica serovar Typhimurium for peptide presentation on either MHC-I or MHC-II (37, 46). Lower capacity to access the spleen, an inability to recruit and activate DC and/or an inferior ability to elicit a significant number of specific IFN-γ producing T cells, particularly CD4⁺ T cells, may contribute to the inferior capacity of heat-killed bacteria to elicit anti-Salmonella immunity.

DC contain Salmonella during infection (Fig. 2), and immature DC can internalize and process live or heat-killed S. enterica serovar Typhimurium for presentation of Salmonella-encoded antigens on MHC-I and MHC-II (37, 39, 45, 46). We therefore assessed the role of DC in priming naïve CD4⁺ and CD8⁺ T cells to Salmonella antigens in vivo by evaluating bacterium-specific effector T cells in mice given Salmonella-loaded DC. DC loaded with heat-killed bacteria were used to eliminate any contribution of viable bacteria residing in the host in eliciting Salmonella-specific T cells. DC harboring either live or killed S. enterica serovar Typhimurium primed both CD4⁺ and CD8⁺ T cells specific for bacterial antigens when administered to naïve mice. This occurred despite the fact that one-third fewer bacteria were present in DC loaded with heat-killed bacteria compared to DC loaded with live bacteria as measured by flow cytometry following uptake of GFP-expressing bacteria. Interestingly, administering DC loaded with heat-killed bacteria elicited a higher frequency of specific CD8⁺ T cells producing IFN-γ than CD4⁺ T cells producing this cytokine in the same animal. In contrast, the frequency of bacterium-specific IFN-γ-producing CD4⁺ T cells in a mouse given DC loaded with live bacteria was consistently greater than the frequency of specific CD8⁺ T cells producing IFN-γ in the same host. These data are consistent with the trend observed in mice given either live or heat-killed bacteria not contained within DC.

Our data also show that the frequency of IFN-γ-producing CD4⁺ or CD8⁺ T cells elicited was greater in response to administration of free live bacteria compared to live bacteria contained within DC. Administration of DC loaded with live bacteria, however, resulted in a higher level of cytolytic activity compared to that observed in mice given free live bacteria. This may reflect that the two assays measure distinct functions of T cells. In addition, the frequency of IFN-γ-producing CD8⁺ T cells (Figs. 5, 6a, and 7a) reflects the total bacterium-specific cell population, while the cytolytic activity measures OVA(257-264)/K^b-specific T cells (Fig. 6b and 7b). Furthermore, the percentage of specific IFN-γ-producing CD8⁺ T cells that have cytolytic activity is not known. The difference in the efficiency of eliciting specific T cells by the two immunization strategies (free live bacteria compared to live bacteria contained within DC) is not likely due to a different quantity of bacteria administered or persistence of the bacteria in vivo. That is, a similar amount of bacteria was given in each type of administration (∼10⁶), and similar numbers of bacteria were recovered from the spleens of animals in both groups at the time of sacrifice (100 to 1,000 total bacteria per spleen, never exceeding a twofold difference between the groups).

We found that the optimal method for restimulating splenocytes from infected mice was using bone marrow-derived DC preloaded with heat-killed bacteria. Thus, the data quantitating the frequency of IFN-γ-producing T cells in response to Salmonella infection (Figs 5, 6a, and 7a) using this restimulation protocol measured reactivity to natural Salmonella antigens as well as recombinant OVA. In other experiments, splenocytes from infected mice were restimulated with lysates prepared from χ4550 or χ4550 OVA. This resulted in a lower but similar frequency of CD8⁺ and CD4⁺ T cells producing IFN-γ. However, within either T-cell subset, a similar frequency of IFN-γ-producing cells was detected regardless of whether the restimulating lysate contained OVA (data not shown). These data, combined with our inability to detect OVA(265-280)/I-A^b-specific T cells in experiments where splenocytes from infected mice were restimulated with OVA(265-280) peptide, suggest that the majority of the IFN-γ-producing T cells detected recognize natural Salmonella antigens.

The observed difference in T-cell priming efficiency in mice given free bacteria versus those administered within DC could, for example, result from different trafficking of injected DC versus free bacteria from the peritoneal cavity to secondary lymphoid organs. Alternatively, a different APC population presenting the bacteria administered free or contained within DC could contribute to the observed difference. The latter point raises the question of whether direct or indirect priming of host T cells occurs in mice given Salmonella-loaded DC. Preliminary data showed that Salmonella-specific IFN-γ-producing CD8⁺ T cells were elicited in C57BL/6 mice given DC from TAP1⁻ β₂m^−/− double-knockout mice loaded with heat-killed χ4550 OVA (Yrlid et al., unpublished data). As these DC are unable to directly present Salmonella antigens to host CD8⁺ T cells (19), the data demonstrate that direct priming by the immunizing DC is not required to elicit CD8⁺ effector cells. However, the relative contribution of direct versus indirect priming of T cells by Salmonella-loaded DC remains to be clarified. Taken together the data in the present study support a role for dendritic cells as an APC important in triggering a specific immune response during Salmonella infection.