Published online before print November 12, 2008
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
+ dendritic cells but not CD11cintCD11b+ inflammatory cells in vivo via MyD88 and TNFR1Department of Microbiology and Immunology, Göteborg University, Göteborg, Sweden
1 Correspondence: Department of Microbiology and Immunology, Göteborg University, Box 435, SE 405 30 Göteborg, Sweden. E-mail: mary-jo.wick{at}immuno.gu.se
ABSTRACT
Dendritic cells (DCs), whose lifespan influences their ability to stimulate the immune system, are potent APCs that are critical for initiating immunity. Here, we show that oral infection with Salmonella enterica serovar Typhimurium induces death of DCs in the gut-draining lymph nodes. Although CD8
+ DCs were sensitive to Salmonella-induced death, CD8– DCs and in particular recruited CD11cintCD11b+ inflammatory cells, were resistant. Infecting mice deficient for MyD88 revealed that Salmonella-induced death of CD8+ DCs was dependent on this adaptor for TLR signaling. In addition, CD8+ DCs in infected, TNFR1-deficient mice were resistant to Salmonella-induced death. These data, combined with the strict MyD88-dependent production of TNF in Salmonella-infected mice, suggest that MyD88-dependent TNF mediates DC death. As recruited CD11cintCD11b+ cells were resistant to Salmonella-induced death, they could compensate for the infection-induced loss of DCs if they function as APCs. However, in contrast to DCs, CD11cintCD11b+ cells could not present the model antigen OVA expressed in Salmonella to OVA-specific CD4 T cells. These results show that Salmonella induces DC death after oral infection via MyD88 and TNFR1, which could have a negative impact on the initiation of antibacterial immunity.
Key Words: oral infection • mesenteric lymph nodes • Toll-like receptor
INTRODUCTION
Dendritic cells (DCs) are the most potent APCs that have the capacity to activate naïve T cells and induce protective, antimicrobial memory responses. The lifespan of DCs is short [1 2 3 ], and the length of their life influences their capacity to drive an immune response. Thus, an increased lifespan results in DC accumulation, chronic lymphocyte activation, and autoimmunity [4 , 5 ]. On the contrary, a reduction in DC numbers reduces the magnitude of the adaptive-immune response [6 , 7 ]. Therefore, the lifespan of DCs in vivo needs to be regulated tightly.
Several bacteria, including the enteric pathogen Salmonella enterica serovar typhimurium, can modulate the lifespan of phagocytes. For example, Salmonella causes death of CD18+ cells in the liver of infected mice [8 , 9 ]. Moreover, as shown in vitro, Salmonella induces rapid death of phagocytes including DCs via a mechanism dependent on the host enzyme caspase-1 and the Ipaf inflammasome [10 11 12 ]. However, despite the critical role of DCs in inducing antibacterial T cell immunity, little is known about how intracellular bacteria such as Salmonella modulate DC survival in vivo, which is studied here. Pathogen-induced host cell death during infection could serve to limit microbial replication or modulate potentially harmful, prolonged inflammation. Alternatively, it could be a strategy for the bacteria to evade an immune response or promote their spread.
Conventional (CD11chi) DCs are a heterogeneous group of bone marrow-derived APCs. Two major subsets can be defined based on expression of the CD8-chain [13
]: the CD8+ and CD8– subsets, which seem to have some functional specialization in, for example, antigen handling [14
]. An additional CD11c-expressing cell type with some features of DCs, described as TNF/inducible NO synthase (iNOS)-producing DCs, has been identified in the spleen of Listeria-infected mice [15
, 16
]. These CD11cintCD11b+ cells exhibit mixed DC/monocyte characteristics and appear to differentiate from CCR2+ monocytes during inflammatory conditions [15
, 17
18
19
]. The CD11cintCD11b+ cells that are recruited to infected organs produce TNF and iNOS [15
, 16
, 19
, 20
], which are important in host defense against bacteria [21
, 22
]. Indeed, the reduced number of TNF/iNOS-producing DCs in the spleen of Listeria-infected, CCR2-deficient mice is associated with a reduced capacity to clear the infection [15
]. In response to infection, the CD11cintCD11b+ cells express a high level of costimulatory molecules [15
, 23
] and can induce proliferation of allogenic T cells [15
], a feature that resembles conventional DCs. However, unlike conventional DCs [6
, 7
], they are not required to prime naïve T cells in vivo [15
]. The capacity of CD11cintCD11b+ cells to present bacterial antigens has not been studied and is a topic of this investigation.
Microbial infections are sensed by the host via pattern-recognition receptors such as the TLRs, which recognize conserved microbial patterns [24
]. All TLRs, except TLR3, signal via the adaptor protein MyD88 to activate NF-
B and initiate cytokine production [24
]. Indeed, TLR-mediated recognition is crucial to combat invasive bacteria [24
]. In addition, signaling through MyD88 influences DC function during Salmonella infection [25
, 26
]. Here, we have investigated the role of MyD88 in Salmonella-induced DC death.
In this study, we investigated DC death in the gut-draining lymph nodes after oral Salmonella infection. We found that Salmonella infection induces DC death, particularly of the CD8+ subset, which is dependent on MyD88 and TNFR1. In contrast, recruited CD11cintCD11b+ cells were resistant to infection-induced death. Finally, DCs, but not recruited CD11cintCD11b+ cells, presented a Salmonella antigen on MHC-II to specific T cells. The selective killing of CD8+ DCs may allow Salmonella to interfere with the host immune response during infection.
MATERIALS AND METHODS
Mice
C57BL/6 mice were purchased from Charles River Laboratories (Sulzfeld, Germany). MyD88–/– (kindly provided by Shizou Akira, Osaka University, Osaka, Japan), TNFR1–/–, IFN-
–/–, and OT-II mice, all on a C57BL/6 background, were bred at the Experimental Biomedicine Animal Facility of Göteborg University (Sweden). Mice were used between 8 and 12 weeks of age and were provided food and water ad libitum. Experiments were performed following protocols approved by the government animal ethical committee and institutional guidelines.
Bacterial strains and infection of mice
Virulent wild-type S. typhimurium SL1344 and its derivative SMO22 expressing enhanced GFP [27
] were used. For antigen-presentation assays, strain 
4550, expressing or not expressing OVA, was used. 4550 contains deletions in the genes encoding adenylate cyclase and the cAMP receptor that reduce its virulence. SL1344 and SMO22 were grown in static cultures of Miller’s Luria-Bertoni (LB) broth supplemented with 100 µg/ml streptomycin (SL1344) and 50 µg/ml kanamycin (SMO22) overnight at 37°C. 4550 was grown in aerated cultures of Miller’s LB broth at constant agitation overnight at 37°C. Bacterial concentration was estimated spectrophotometrically. Mice received 0.1 ml 1% NaHCO3 intragastrically followed by bacteria in a volume of 0.2 ml. The actual bacterial dose was determined by plating serial dilutions of the bacterial suspension on LB agar plates. Mice received a bacterial dose between 0.7 and 6.3 x 109 bacteria orally. The bacterial load in mesenteric lymph nodes (MLN) was determined by plating serial dilutions of the cell suspension on LB agar plates.
Cell preparation, DC enrichment, and cell sorting
For analysis of DCs, a single-cell suspension was prepared by incubating MLN in 400 µg/ml Liberase (Roche Diagnostics, Nutley, NJ, USA) and 100 U/ml LPS-binding polymyxin-B (Sigma-Aldrich, St. Louis, MO, USA) in HBSS (Invitrogen Life Technologies, Carlsbad, CA, USA) at 37°C for 30 min, followed by pipetting the digested organs. Cells were prepared from individual mice except where CD11c-expressing cells were enriched from single-cell suspensions of MLN pooled from two to 15 mice using N418 magnetic beads (Miltenyi Biotec, Auburn, CA, USA) and an AutoMACS (Miltenyi Biotec; see 
Figs. 4
and 5
). MACS-enriched DC were stained with anti-CD11c-FITC, CD8-allophycocyanin, CD11b-allophycocyanin-Cy7, and 7-aminoactinomycin D (7AAD) before sorting viable CD11chigh CD8+ and CD8– conventional DCs as well as CD11cintCD11b+ cells. Sorting was performed on a FACSAria cell sorter (BD Biosciences, San Jose, CA, USA) in PBS at low pressure using a 100-µm nozzle. Cells were typically >90% pure. CD4 T cells were enriched from the spleen of OT-II mice by negative selection using the CD4 T cell isolation kit from Miltenyi Biotec and an AutoMACS (Miltenyi Biotec). Enriched CD4 T cells were washed in PBS and labeled with 5 µM CFSE (Molecular Probes, Eugene, OR, USA) in PBS for 8 min at room temperature.
 View larger version (40K): [in a new window] |
Figure 1. MyD88 modulates the frequency of DCs and CD11cintCD11b+ cells in the MLN during Salmonella infection. (A) C57BL/6 mice were orally infected with Salmonella, and 3 days after infection, MLN cells were stained directly ex vivo with 7AAD and antibodies to CD11c, CD8, CD11b, and MHC-II to identify the conventional DC subsets, which were identified as CD11chiMHC-IIhiCD8+ (CD8+ DCs) and CD11chiMHC-IIhiCD8– (CD8– DCs). (B) The R2 gate contains the inflammatory CD11cintCD11b+ cells [23
]. The CD11c–CD11b+ cells within gate R1 are monocytes/macrophages and neutrophils, and the CD11chi cells within gate R3 are conventional DCs [23
]. (C) Expression of MHC-II, CD80, iNOS, and TNF by the three populations gated in B, isolated from the MLN of C57BL/6 mice 5 days after infection with  108 Salmonella. Open histograms show isotype controls. Geometric mean fluorescent intensities (MFI) are shown in the histograms. The frequencies of iNOS- or TNF-positive cells are shown in the dot plots. (D) The relative frequency of conventional CD8+ and CD8– DCs, as well as CD11cintCD11b+ cells, in the MLN of C57BL/6 and MyD88–/– mice on Days 0, 1, 2, and 3 after oral Salmonella infection. (E) The absolute number of the cell populations in the MLN of C57BL/6 and MyD88–/– mice on Days 0, 1, 2, and 3 after oral Salmonella infection. (F) Total cells in the MLN of C57BL/6 and MyD88–/– mice on Days 0, 1, 2, and 3 after infection. (G) Bacterial load in the MLN on Days 0, 1, 2, and 3 after infection. For all panels, the data were obtained from six to 16 mice in two to seven independent experiments per time-point and mouse strain. The error bars indicate the SEM.
|
 View larger version (20K): [in a new window] |
Figure 2. Salmonella infection induces DC death in MLN. C57BL/6 mice were orally infected with Salmonella, and at different time-points after infection, MLN cells were stained directly ex vivo with antibodies as described for Figure 1
and also with Annexin-V-PE and 7AAD. Cells were gated into conventional CD8+ and CD8– DCs and CD11cintCD11b+ cells as shown in Figure 1
, except that 7AAD+ cells were included. Cells were then analyzed for 7AAD and Annexin-V (see Fig. 3A
). (A) The percent of Annexin-V+7AAD+ cells on Days 0, 1, 2, and 3 after oral Salmonella infection. Data are from eight to 16 mice examined in two to four experiments for Days 0, 2, and 3, and Day 1 data are from four mice examined in a single experiment. (B) TUNEL immunohistochemistry of MLN sections from naïve and infected (Day 3) C57BL/6 mice. As a negative control, the TdT enzyme was omitted. Data are representative of four experiments. Original magnification is x400. (C) TUNEL-CD11c double staining of MLN sections from an infected C57BL/6 mouse (Day 3). Arrowheads indicate clusters of CD11c+TUNEL+ cells. As a negative control, the TdT enzyme and primary antibody were omitted. Data are representative of three independent experiments. Original magnification, x200.
|
 View larger version (35K): [in a new window] |
Figure 3. Salmonella-induced DC death is dependent on MyD88 and TNFR1. C57BL/6, MyD88–/–, and TNFR1–/– mice were orally infected with Salmonella, and at different time-points after infection, MLN cells were stained directly ex vivo with the antibodies described for Figure 1
to identify the conventional DC subsets and CD11cintCD11b+ cells and with Annexin-V-PE and 7AAD to identify dead cells. (A) Representative dot plots from the infected mice (Day 3) in B and C showing Annexin-V versus 7AAD staining of the indicated cell populations, gated as in Figure 1
except that 7AAD+ cells were included. The numbers indicate the percent of Annexin-V+7AAD+ DCs. (B) The mean percent of Annexin-V+7AAD+ cells of infected and naïve C57BL/6, MyD88–/–, and TNFR1–/– mice at the indicated time-points. Bar graphs show the mean and SEM of six to eight mice. Data were obtained in two independent experiments. NS, Not significant. (C) Bacterial load in the MLN of the mice shown in B. Circles represent individual mice, and bars are the mean. (D) C57BL/6 and MyD88–/– mice were orally infected with Salmonella, and after 3 days, MLN cells were seeded in a 24-well plate (2x106/well) and cultured for 24 h in the presence or absence of Salmonella lysate. The TNF content in the supernatants was analyzed by ELISA. ND, Not detected. The data are from two to four mice per strain and time-point examined in two independent experiments.
|
 View larger version (44K): [in a new window] |
Figure 4. The frequency of GFP+ DCs is increased in the absence of MyD88 or TNFR1. Mice were orally infected with Salmonella expressing GFP. At the indicated time-points, CD11c-expressing cells from the MLN from C57BL/6, MyD88–/–, TNFR1–/–, or IFN-–/– mice were enriched using MACS, stained with anti-CD11c, CD8, CD11b, and 7AAD, and gated into cell populations as in Figure 1
. (A) Representative dot plots of the mice shown in B, indicating fluorescence of GFP versus CD11b on CD11cintCD11b+ cells and GFP versus CD8 on DC subsets. The number represents the frequency of GFP+ cells among each subset. eGFP, Enhanced GFP. (B) The mean percent of GFP+ DC subsets in the MLN from C57BL/6, MyD88–/–, TNFR1–/–, or IFN-–/– mice 3 days after infection. The MLN from two to 15 mice of the same strain were pooled in each experiment. Data are from two to three independent experiments for each mouse strain. Error bars indicate the SEM. Infection with the parent bacterial strain lacking GFP resulted in <0.027% events in the gate. (C) The bacterial load in the single-cell suspensions of the pooled MLN from the mice shown in A divided with the number of animals, to allow a comparison between different experiments, is shown. Each triangle represents one experiment.
|
 View larger version (42K): [in a new window] |
Figure 5. CD11cintCD11b+ cells cannot present a Salmonella antigen to specific CD4 T cells. (A) Conventional CD8– DCs or CD11cintCD11b+ cells (0.5x105) were sorted from the MLN of infected mice (SL1344, Day 3), and 1 x 105 CD8– DCs were sorted from the MLN of naïve C57BL/6 mice. Sufficient numbers of CD11cintCD11b+ cells for analysis could not be obtained from naïve mice. Similarly, sufficient numbers of conventional CD8+ DCs could not be obtained from naïve or infected mice. Sorted cells were pulsed for 2 h with Salmonella 4550 expressing or not expressing OVA at a 7:1 bacteria:cell ratio. Cells were then washed and resuspended in medium containing 50 µg/ml gentamicin, and CFSE-labeled OT-II T cells were added to the wells at a 1:2 DC:T cell ratio. After 3.5 days, proliferation of the OT-II T cells was assessed by flow cytometry. The purity of sorted, conventional DCs and CD11cintCD11b+ cells was >90%. Data are representative of two independent experiments. (B) Sorted CD8+ or CD8– DCs (1x105) or CD11cintCD11b+ cells isolated from the MLN or spleen of infected mice (SL1344, Day 3) were pulsed for 2 h with 1 µM OVA(323–339) peptide or not pulsed as indicated. Proliferation of CFSE-labeled OT-II T cells was assessed after 3.5 days. The purity of CD8– DCs and CD11cintCD11b+ cells was >90%, whereas the purity of splenic and MLN CD8+ DCs was 70% and 85%, respectively.
|
RII/III mAb (2.4G2). The cells were then stained with antibodies conjugated to FITC, PE, allophycocyanin, PE-Cy7, allophycocyanin-Cy7, or Pacific Orange. 7AAD (Sigma-Aldrich) was used for dead cell exclusion. Conjugated antibodies against CD11c (HL3), CD8 (53-6.7), CD11b (M1/70), Ly5.II (104), TCR (H57-597), Ly5.1 (A20), CD80 (16-10A1), and MHC-II (M5/114.15.2) were from BD Biosciences. Anti-CD4 (GK1.5) was from our own production and was conjugated to Pacific Orange. To detect apoptotic DCs, cells were stained with Annexin V-PE and 7AAD using the Annexin-V kit from BD Biosciences. Cells (2x106) were stained in a volume of 100 µl, and 7AAD was diluted at 1:20 and Annexin-V-PE at 1:10. To stain for intracellular iNOS and TNF, samples were incubated with Brefeldin A (5 µg/ml) for 4 h, directly stained for surface followed by fixation in 2% formaldehyde in PBS for 20 min at room temperature. After permeabilizing the cells with HBSS containing 0.5% saponin and 0.5% BSA (Sigma-Aldrich), a buffer that was subsequently used throughout the staining procedure, samples were stained with anti-iNOS antibody (M-19, Santa Cruz Biotechnology, Santa Cruz, CA, USA) or anti-TNF FITC (MP6-XT22, BD Biosciences) for 30 min at room temperature. The iNOS antibody was detected by staining with allophycocyanin-conjugated anti-rabbit IgG (Santa Cruz Biotechnology) for 30 min at room temperature. Samples were acquired on an LSR-II flow cytometer (BD Biosciences) using DIVA software and were analyzed using FlowJo software (Tree Star, Ashland, OR, USA).
Immunohistochemistry
MLN were embedded in TissueTek OCT compound and snap-frozen in liquid nitrogen. Frozen sections (6 µm) were prepared using a cryostat (Leica, Germany) and were stored at –70°C until used for TUNEL single staining or TUNEL-CD11c double staining. TUNEL single staining was performed using the ApopTag® Plus fluorescein in situ apoptosis detection kit according to the manufacturer’s protocol (Chemicon International, El Segundo, CA, USA). Briefly, sections were fixed in 1% paraformaldehyde and postfixed in ethanol and acetic acid at –20°C. Sections were then incubated at 37°C with TdT enzyme to add digoxigenin-labeled nucleotides to DNA nicks followed by staining with antidigoxigenin-fluorescein mAb. The sections were mounted with DakoCytomation fluorescent medium and visualized using a Leica LSC microscope. As a negative control, the TdT enzyme was omitted.
The postfixation step of the TUNEL assay took away reactivity to the anti-CD11c antibody HL3 (data not shown). Therefore, to perform TUNEL-CD11c double stainings, sections were first stained for CD11c, and then the TUNEL assay was performed. Sections were fixed in 1% paraformaldehyde for 10 min at room temperature. After washing in PBS, the tissue was blocked in normal horse serum (5%) for 15 min in a humidified chamber. Endogenous biotin sites were blocked using the biotin/avidin blocking kit (Vector Laboratories, Burlingame, CA, USA). Sections were then stained with biotinylated anti-CD11c mAb (HL3, BD Biosciences) for 60 min, followed by washing in PBS. The CD11c was visualized by adding streptavidin-conjugated Alexa-594 (Molecular Probes) followed by washing in PBS. After this, the TUNEL assay was performed as outlined above, starting at the postfixation step. As a negative control for the CD11c staining, sections were stained with secondary antibody only.
Stimulation of MLN cells with Salmonella lysate
Cells (2x106) from MLN were seeded in a 24-well plate and cultured in complete RPMI medium for 24 h in the presence or absence of Salmonella lysate (1:200). The supernatants were stored at –20°C until assayed for TNF by an ELISA set (BD Biosciences).
Antigen-presentation assay
Depending on the recovery, 0.5–1 x 105 DCs or CD11cintCD11b+ cells, sorted from the MLN or spleen of pooled naïve or infected mice (SL1344, Day 3), were seeded in 96-well round-bottom plates in RPMI medium containing 5% FCS. Cells were cultured with 4550-OVA-GFP or 4550 at a 1:7 or 1:1 DC:bacteria ratio or were pulsed for 2 h with 1 µM OVA(323–339) peptide. Cells were centrifuged at 320 g to facilitate bacterial contact with the cells. After 2 h, the cells were washed four times, and CFSE-labeled, OVA-specific OT-II T cells were added in RPMI medium containing 5% FCS and 50 µg/ml gentamicin at a 1:2 DC:T cell ratio. After 3.5 days, proliferation of the OT-II T cells was assessed by determining CFSE dilution by flow cytometry.
Statistics
Statistical analysis was performed using the Mann-Whitney U test for unpaired samples (two-tailed).
RESULTS
MyD88-dependent modulation of DC numbers during Salmonella infection
To investigate DC survival during Salmonella infection, we first analyzed early changes in the percent and absolute number of DC subsets in the draining lymph node of orally infected mice. The role of proinflammatory signaling via MyD88 for changes in DC numbers during infection was also studied.
Conventional CD8+ and CD8– DCs were identified based on high expression of CD11c and differential expression of the CD8-chain (Fig. 1A
). The inflammatory CD11cintCD11b+ cells that accumulate at sites of infection and inflammation [15
, 16
, 19
, 23
] were gated as R2 in Figure 1B
. They expressed an intermediate level of MHC-II as well as a high level of CD80 (higher than CD11chi conventional DCs within the R3 gate [23
]), and a large fraction produced iNOS and TNF (Fig. 1C)
. The CD11c–CD11b+ cells within the R1 gate are monocytes/macrophages and neutrophils that express low levels of MHC-II as well as CD80 and produce iNOS and TNF (Fig. 1C
; see ref. [23
] for additional phenotyping). The CD11cintCD11b+ cells thus share several properties with the "TNF/iNOS-producing DCs" that are recruited during infection or inflammation [15
, 16
, 19
, 28
].
In C57BL/6 mice, the percent of conventional CD8+ DCs in the MLN dropped 2 days after infection and decreased further at Day 3 (Fig. 1D)
. In contrast, the percentage of CD8+ DCs in infected MyD88–/– mice remained relatively constant and was significantly higher than the frequency in C57BL/6 mice 3 days after infection (P=0.006). Consistent with the drop in the percentage of CD8+ DCs in infected C57BL/6 mice, the absolute number of CD8+ DCs dropped 40% 2 days after infection (Fig. 1E)
. However, at 3 days after infection, the absolute number of CD8+ DCs was close to that in naïve C57BL/6 mice (Fig. 1E)
. The increase in the absolute number of CD8+ DCs between Days 2 and 3 after infection is likely a result of the recruitment of new DCs to the MLN at this time-point. The absolute number of CD8+ DCs was greater in MyD88–/– compared with C57BL/6 mice 3 days after infection. However, this difference did not reach statistical significance. In contrast to CD8+ DCs, the frequency of CD8– DCs remained constant in C57BL/6 and MyD88–/– mice, whereas the absolute number increased Days 1–3 after infection (Fig. 1 D and E)
.
CD11cintCD11b+ cells were recruited to the infected lymph node and appeared in the MLN of Salmonella-infected C57BL/6 mice 2 days after infection (Fig. 1 B, D, and E ). The relative frequency and absolute number of CD11cintCD11b+ cells were increased greatly after 3 days, as was the total cellularity in the MLN (Fig. 1F) . In sharp contrast, no increase in CD11cintCD11b+ cells was found in the MLN of infected MyD88–/– mice (Fig. 1 D and E) , which also exhibited a smaller increase in total MLN cellularity [1.3 times compared with 1.8 times in C57BL/6 mice (Fig. 1F) ].
Thus, Salmonella infection induces a MyD88-dependent reduction in the percentage of CD8+ DCs in the MLN. In addition, the recruitment of CD11cintCD11b+ cells to the MLN is dependent on MyD88. The selective reduction in frequency but not absolute number of CD8+ DCs 3 days after infection could be due to two possibilities. First, CD8+ DCs are recruited but also killed, and the total number is unvaried, while the percentage is reduced as a result of influx of other cells such as the CD11cintCD11b+ cells (Fig. 1 D and E)
. Second, CD8+ DCs are not recruited, and given the enlargement of the lymph node as a result of the infection, the percentage of CD8+ DCs is reduced.
Salmonella infection induces death of CD8+ DCs in vivo
To determine whether Salmonella infection influences DC survival, we analyzed the frequency of dead, conventional DCs and CD11cintCD11b+ cells in the MLN of infected C57BL/6 mice using 7AAD and Annexin-V as markers of cell death. We defined cells that were positive for 7AAD and Annexin-V as dead. The transient Annexin-V+7AAD– state of early apoptosis was not possible to reliably detect for two reasons. First, the seeding of bacteria to the MLN after oral infection is nonsynchronous compared with in vitro or i.v. infection. Second, the staining protocol with Annexin-V resulted in a high background and did not allow a good separation of the Annexin-V+ cells unless 7AAD was used as an additional marker (data not shown). Two days after infection, the frequency of dead, conventional CD8+ DCs increased dramatically to 40% and remained high after 3 days (Fig. 2A
). Strikingly, CD8– DCs and CD11cintCD11b+ cells were much less sensitive to Salmonella-induced death. Although the frequency of dead CD8– DCs increased to 15% after 3 days, the CD11cintCD11b+ cells were completely resistant to Salmonella-induced death (Fig. 2A)
. As 7AAD can only penetrate dead cells, and the 7AAD/Annexin-V staining was carried out at +4°C, the observed staining cannot be caused by CD8+ DCs endocytosing apoptotic cells.
We next investigated Salmonella-induced cell death in situ using TUNEL immunohistochemistry. The MLN of naïve mice showed a few scattered TUNEL+ cells (Fig. 2B) . In sharp contrast, the MLN of infected C57BL/6 mice (3 days) exhibited more densely distributed TUNEL+ cells. In addition, we frequently observed clusters of dead cells of varying size in the MLN of infected mice (Fig. 2B) . CD11c-TUNEL double staining was performed to identify dead DCs in situ. Indeed, a fraction of the TUNEL-positive cells (nuclei) was also positive for CD11c (Fig. 2C) .
Thus, using Annexin-V-7AAD staining and TUNEL immunohistochemistry, we have shown that Salmonella induces death of DCs in the gut-draining MLN after oral infection. The CD8+ DCs were particularly sensitive to infection-induced death, whereas the CD11cintCD11b+ cells were resistant.
Salmonella-induced DC death is dependent on MyD88 and TNFR1
We speculated that signaling through MyD88 might contribute to Salmonella-induced DC death in vivo, as the drop in the percentage of CD8+ DCs after Salmonella infection of wild-type mice depended on MyD88 (Fig. 1A)
. To identify the mechanism of Salmonella-induced DC death in vivo, C57BL/6 and MyD88–/– mice were orally infected with Salmonella, and after 3 days, cell death was analyzed by flow cytometry using Annexin-V and 7AAD staining. The death of CD8+ and CD8– DCs in the MLN of Salmonella-infected mice was completely dependent on MyD88, as infected MyD88–/– mice exhibited no increase in the frequency of Annexin-V+7AAD+ DCs compared with naïve mice (Fig. 3 A and B
). Thus, Salmonella-induced DC death in the draining lymph node is dependent on MyD88.
The TLR adaptor MyD88 controls the production of many proinflammatory cytokines, including TNF [29
], which can induce apoptosis upon binding to the receptor TNFR1 [30
, 31
]. As Salmonella infection induces TNF production in the MLN [20
, 23
], we investigated whether the MyD88-dependent DC death during Salmonella infection was mediated via TNFR1 signaling. To this end, TNFR1–/– mice were orally infected with Salmonella, and the frequency of dead DCs in the MLN was determined by Annexin-V and 7AAD staining. Indeed, the frequency of dead CD8+ DCs in the MLN of infected TNFR1–/– mice was reduced significantly compared with infected C57BL/6 mice (Fig. 3B)
. In contrast, the frequency of CD8– DCs in infected TNFR1–/– mice was not reduced significantly compared with infected C57BL/6 mice (Fig. 3B)
.
Consistent with our results showing that Salmonella-induced death of CD8+ DCs is dependent on MyD88 and TNFR1, the production of TNF by MLN cells after oral Salmonella infection was abrogated completely in MyD88–/– mice, even when ex vivo cultures of MLN cells were supplemented with Salmonella lysate (Fig. 3D)
. Thus, Salmonella-induced death of CD8+ and CD8– DCs is dependent on MyD88. Furthermore, death of CD8+, but not CD8–, DCs is dependent on TNFR1 in infected mice and is likely mediated via the MyD88-dependent production of TNF.
Higher frequency of GFP+ DCs in the absence of MyD88
Next, we investigated how MyD88 and TNFR1 influence the frequency of conventional DCs and CD11cintCD11b+ cells directly associated with bacteria in vivo. To this end, C57BL/6 as well as MyD88–/– and TNFR1–/– mice were orally infected with Salmonella expressing GFP. The frequency of DCs directly associated with Salmonella was then assessed by flow cytometry. The MLN from several mice were pooled to detect a sufficient number of GFP+ DCs. The infected MyD88–/– and TNFR1–/– mice had a substantially higher frequency of Salmonella-associated (GFP+) DCs and CD11cintCD11b+ cells compared with C57BL/6 mice (Fig. 4 A and B
). This is a result of the >1-log higher bacterial load in the MLN of the knockout mice (Fig. 4C)
that results from defects in bactericidal activity of phagocytes in these strains [26
, 32
]. Therefore, we compared the frequency of GFP+ DCs in MyD88–/– and TNFR1–/– mice with that in infected IFN-–/– mice that have a similar bacterial burden in the MLN (Fig. 4C)
and reduced bactericidal activity [33
]. Importantly, the production of proinflammatory cytokines such as TNF is unimpaired in Salmonella-infected IFN-–/– mice [20
]. The single-cell suspensions of the MLN from infected MyD88–/–, TNFR1–/–, or IFN-–/– mice had a similar bacterial load after infection (Fig. 4C)
.
Despite a similar bacterial load, MyD88–/– mice exhibited a higher frequency of Salmonella-containing GFP+ CD8+ and CD8– DCs 3 days after infection compared with IFN-–/– mice (Fig. 4B)
. The higher frequency of GFP+ CD8+ DCs in MyD88–/– mice was partially dependent on TNFR1 (Fig. 4B)
. In contrast, the higher frequency of GFP+ CD8– DCs in MyD88–/– mice compared with IFN-–/– mice was not dependent on TNFR1 (Fig. 4B)
. This is consistent with the data in Figure 3
showing that only the death of CD8+ DCs and not CD8– DCs was dependent on TNFR1 signaling during Salmonella infection. These results indicate that Salmonella-containing DCs may be depleted in the MLN during the infection in a MyD88-dependent manner.
Inflammatory CD11cintCD11b+ cells have a poor antigen-presentation capacity
Next, we wanted to assess the capacity of the conventional DC subsets and CD11cintCD11b+ cells to present a Salmonella-encoded antigen directly ex vivo after oral infection. However, numerous pilot experiments could not detect proliferation of OT-II cells after coincubation with DCs from mice infected 1–3 days earlier with Salmonella expressing OVA. This could be a result of the infection-induced programming of DCs to undergo death, the low frequency of infected, conventional DCs (Fig. 4
; ref. [23
]), or the inhibition of antigen presentation by Salmonella [34
35
36
]. We thus took an alternate approach and analyzed the capacity of the CD11cintCD11b+ cells to present a Salmonella-encoded antigen in vitro.
Given the resistance of CD11cintCD11b+ cells to infection-induced death (Figs. 2
and 3)
, they could compensate for the death of DCs that occurs during Salmonella infection if they could present Salmonella antigens to T cells. We thus purified CD11cintCD11b+ cells from the MLN of infected C57BL/6 mice by cell sorting. Given the rarity of CD11cintCD11b+ cells in naive mice (Fig. 1 D and E
; ref. [23
]), sufficient numbers of these cells could only be obtained from infected mice. Similarly, a sufficient number of conventional CD8+ DCs for analysis of antigen-presentation capacity could not be obtained. Thus, for comparison, conventional CD8– DCs were sorted from the MLN of infected as well as naïve C57BL/6 mice. The cells were then incubated for 2 h with Salmonella expressing or lacking OVA and cocultured with OVA-specific CD4+ OT-II T cells. CD8– DCs isolated from infected or naïve mice induced proliferation of OT-II cells when pulsed with Salmonella expressing OVA (Fig. 5A
). However, CD11cintCD11b+ cells isolated from the MLN did not support OT-II proliferation after incubation with OVA-expressing Salmonella (Fig. 5A)
.
Next, we evaluated the capacity of the conventional DC subsets and CD11cintCD11b+ cells isolated from the MLN or spleen of infected mice to induce proliferation of OT-II T cells after a brief pulse with the OVA(323–339) peptide. Peptide-pulsed CD11cintCD11b+ cells induced detectable proliferation of OT-II cells (Fig. 5B)
. However, peptide-pulsed, conventional DCs induced four to eight times more proliferation of OT-II cells compared with CD11cintCD11b+ cells. The CD8+ DCs isolated from the MLN of infected mice were an exception, having a reduced capacity to induce OT-II T cell proliferation compared with their splenic counterpart (Fig. 5B)
. This could be a result of the fact that the DCs in MLN are more programmed to Salmonella-induced death at this time-point compared with DCs in spleen, although all DCs were negative for 7AAD when gated for sorting.
In summary, CD11cintCD11b+ cells recruited to the MLN of infected mice are not able to present a Salmonella-encoded antigen. In addition, CD11cintCD11b+ cells as well as conventional CD8+ DCs from the MLN of infected mice have a poor capacity to induce proliferation of CD4 T cells after a brief peptide pulse.
DISCUSSION
We show that during oral Salmonella infection, CD8+ DCs and to a lesser extent, CD8– DCs in the MLN undergo MyD88-dependent death. The reduced death of CD8+ DCs in infected TNFR1–/– mice combined with the strict MyD88 dependence of TNF production in the MLN in response to Salmonella strongly support that MyD88-induced production of TNF mediates DC death during Salmonella infection. Indeed, CD8+ DCs are more susceptible to apoptosis than CD8– DCs in response to a range of infections [37
38
39
]. Furthermore, CD8+ DCs had a higher frequency of Annexin-V+ 7AAD+ cells than CD8– DCs when analyzed directly ex vivo (Figs. 2
and 3)
. This indicates that the CD8+ DCs may be more sensitive than CD8– DCs and CD11cintCD11b+ cells to different apoptosis-inducing stimuli. Indeed, CD8+ DCs express lower mRNA levels of the antiapoptotic protein Bcl-2 compared with CD8– DCs in the spleen [40
]. CD8+ DCs also have a faster turnover rate than CD8– DCs in vivo and thus, have a shorter lifespan than CD8– DCs, even under steady-state conditions [1
2
3
].
In sharp contrast to conventional DCs, CD11cintCD11b+ cells recruited to infected tissues were resistant to Salmonella-induced death during the early stages of infection. The resistance of CD11cintCD11b+ cells to infection-induced death could be a result of their relatedness with monocytes/macrophages, as autocrine production of TNF induces long-term survival of macrophages after LPS treatment [41
]. Moreover, monocytes and the CD11cintCD11b+ cells are major producers of TNF in the MLN during Salmonella infection (Fig. 1C
; refs. [20
, 23
]) and may thus be responsible for killing CD8+ DCs.
Our data suggest that MyD88-dependent production of TNF, signaling through TNFR1, is responsible for the death of CD8+ DCs during Salmonella infection. TNF is a highly pleiotropic, proinflammatory cytokine that affects cellular proliferation and differentiation, as well as activation of apoptosis [30
, 31
]. Signaling through TNFR1 leads to activation of NF-B via TNFR-associated death domain protein, receptor-interacting protein 1 (RIP-1), and TNFR-associated factor 2 (TRAF-2) [30
]. Signaling through TNFR1 can also initiate apoptosis after formation of a second complex containing Fas-associated death domain protein (FADD), which activates procaspase-8 [42
]. In most cases, signaling through TNFR1 does not induce apoptosis. Resistance to TNFR1-FADD-mediated apoptosis is dependent on TRAF-2, RIP-1, and the induction of NF-B-dependent antiapoptotic proteins [43
]. Thus, TNF can only induce apoptosis in cells with low levels of antiapoptotic proteins. Interestingly, Salmonella can inhibit the antiapoptotic NF-B pathway via a virulence protein, AvrA [44
]. Therefore, Salmonella may induce apoptosis in infected cells by inhibiting the NF-B-mediated up-regulation of antiapoptotic molecules in response to TNF. Considering that AvrA-mediated cell death can be expected to be limited to infected cells, and given the low percentage of conventional DCs directly associated with Salmonella (Fig. 4
; ref. [23
]), AvrA-mediated cell death is unlikely to account for the majority of the observed DC death in the MLN during infection. Inhibition of NF-B by AvrA could, however, influence cell death in DCs associated directly with bacteria. Rather, the low, steady-state expression of antiapoptotic proteins in conventional CD8+ DCs could explain their general susceptibility to TNFR1-mediated cell death [40
].
Despite the well-studied effects of caspase-1 on Salmonella-induced phagocyte death in vitro [10 , 11 ], we could not detect an increased survival of MLN DCs in caspase-1-deficient mice 3 days after infection (data not shown). It is possible that caspase-1-mediated cell death acts at another time-point or another organ than investigated here with oral Salmonella infection. Indeed, ileal loop experiments revealed that caspase-1-deficient mice contained fewer TUNEL+ cells in the Peyer’s patches 1 h after Salmonella injection compared with wild-type mice [45 ].
Three days after oral Salmonella infection, we observed a 50% reduction in the frequency of CD8+ DCs in MLN. However, the absolute number of CD8+ DCs at Day 3 postinfection was not reduced significantly compared with naïve mice, despite the transient reduction in absolute number at Days 1 and 2 (Fig. 1)
. At Day 3, but not at Days 1 and 2, the total cell number in the MLN was doubled. Thus, DC recruitment to MLN is probably masking the DC loss (detected by absolute number) associated with the observed DC death (detected by Annexin-V/7AAD staining) at 3 days postinfection. The reduction in DC frequency 3 days after Salmonella infection was rather low compared with the drop in DC numbers that occur during septicemia [39
]. When we killed Salmonella-infected mice at a later time-point, when the mice appeared moribund, we observed a massive loss of CD8+ DCs in the MLN (90%; data not shown). Thus, in this study, we have investigated early effects of Salmonella-induced DC death in the gut-draining lymph node before the onset of sepsis. Indeed, our goal was to study cell death and antigen presentation in vivo, which are processes that have a narrow window of detection. We thus used a bacterial dose (3x109) that we showed resulted previously in DC maturation early (3 days) after infection [23
]. This allowed analysis of critical events early during infection. Moreover, using tenfold less bacteria delays DC maturation until Day 5 and also results in a 60% reduction in the absolute number of CD8+ DCs after oral infection with 108 Salmonella (data not shown; ref. [23
]). Thus, Salmonella-induced death of CD8+ DCs can also occur at a lower bacterial dose but with a different kinetics.
The infection-induced death of DCs may influence the ability of the host to mount an antibacterial, adaptive-immune response. Alternatively, DC death after infection may be a natural pathway to down-regulate the ensuing immune response after antigen presentation has occurred. We observed that infected MyD88–/– or TNFR1–/– mice had a larger fraction of DCs in the MLN associated with bacteria compared with equally susceptible IFN-–/– mice, which retain the capacity to produce TNF [20
]. These results indicate that Salmonella-containing DCs may be depleted in the MLN during oral infection via a pathway dependent on MyD88 and partially dependent on TNFR1. Accordingly, we found that CD8+ DCs isolated from the MLN of infected wild-type mice had a reduced capacity relative to splenic CD8+ DCs to present a peptide to specific CD4 T cells.
The susceptibility of DCs, particularly the CD8+ subset, to infection-induced cell death compared with the resistance of recruited CD11cintCD11b+ cells led us to examine the capacity of the conventional DC subsets and CD11cintCD11b+ cells from infected mice to present a Salmonella antigen on MHC-II. As discussed in Results, this was not possible to assess ex vivo using DCs from infected mice. However, by infecting purified cells in vitro, we showed that CD11cintCD11b+ cells from the MLN did not induce proliferation of OVA-specific T cells after coculture with OVA-expressing Salmonella. In sharp contrast, CD8– DCs primed OVA-specific CD4 T cells after a 2-h pulse with Salmonella expressing OVA. Thus, when cocultured with Salmonella, CD8– DCs, but not CD11cintCD11b+ inflammatory cells, can induce proliferation of CD4+ OT-II T cells. A substantial fraction of the CD11cintCD11b+ cells in the MLN is associated directly with Salmonella after oral infection (Fig. 4)
, and inflammatory CD11cintCD11b+ cells are much more efficient at internalizing Salmonella than conventional DCs in vitro and in vivo [20
, 23
]. Therefore, the reduced capacity of the CD11cintCD11b+ cells to present a bacterial-encoded antigen is likely not a result of an inability of these cells to take up Salmonella. Moreover, the CD11cintCD11b+ cells had poor capacity to present peptide, which eliminates an antigen-uptake step. Thus, despite the capacity of CD11cintCD11b+ cells to activate allogenic T cells [15
], our results show that they have a poor capacity to present a bacterial-encoded antigen and exogenous peptide to CD4 T cells.
In summary, bacterial infection differentially affects the survival and function of DCs and CD11cintCD11b+ cells, which in turn, can influence the developing adaptive-immune response. Infection-induced DC death was controlled by MyD88 and TNFR1 and could have a negative impact on the initiation of antibacterial T cell responses. Thus, Salmonella may use components of the host immune system to evade adaptive immunity.
ACKNOWLEDGEMENTS
This work was supported by grants from the Swedish Research Council (621-2004-1378 and 621-2007-6536) and by the Sahlgrenska Academy at Göteborg University and was performed at the Mucosal Immunobiology and Vaccine Center (MIVAC) funded by the Swedish Foundation for Strategic Research. We gratefully acknowledge the technical assistance of Stina Lindgren.
Received July 11, 2008; revised September 26, 2008; accepted October 18, 2008.
REFERENCES
-dependent and -independent maturation of dendritic cells and recruited CD11cintCD11b+ cells during oral Salmonella infection J. Immunol. 175,3287-3298β differentially control maturation of bystander but not Salmonella-associated dendritic cells or CD11cintCD11b+ cells during infection Cell. Microbiol. 10,1517-1529[CrossRef][Medline] receptors on dendritic cells J. Immunol. 173,4058-4065 autocrine signaling J. Immunol. 178,3731-3739B activation prevents cell death Cell 87,565-576[CrossRef][Medline] B pathway J. Immunol. 169,2846-2850This article has been cited by other articles:
 |
|
 |
|
  B. Fournier, I. R. Williams, A. T. Gewirtz, and A. S. Neish Toll-Like Receptor 5-Dependent Regulation of Inflammation in Systemic Salmonella enterica Serovar Typhimurium Infection Infect. Immun., September 1, 2009; 77(9): 4121 - 4129. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
