ABSTRACT
A major Salmonella component involved in cellular activation is the lipopolysaccharide (LPS) molecule which can act as a dendritic cell (DC) stimulator. The structure of the lipid A domain of the LPS molecule dictates its immunostimulatory capacity with various cell types. In this study, the role of lipid A as an integral component of Salmonella in stimulating murine DCs was studied by using a Salmonella enterica serovar Typhimurium lpxM mutant with defective lipid A. This study revealed that a mutation in lpxM did not significantly affect the ability of bacteria to activate DCs. Although the lpxM mutant less tumor necrosis factor alpha, interleukin-1β, and inducible nitric oxide synthase than the parental strain, this was only seen at lower multiplicities of infection (MOIs). Both strains upregulated surface molecule expression on DCs and augmented the T-cell-stimulating capacity of these cells in an MOI-independent manner. Thus, the lpxM mutation did not appear to affect the stimulatory capacity of the Salmonella mutant.
The existence of different stages of dendritic cell (DC) maturation is well documented (2). The maturation states are characterized by defined phenotypic and functional properties. Transition from the immature state to the mature state is influenced by many factors, such as viral (4) or bacterial infection (36, 37, 44), bacterial components such as CpG motifs in DNA (3, 42) and lipopolysaccharide (LPS) (10, 50), and inflammatory cytokines (39). DCs mature upon encountering Salmonella, as defined by secretion of cytokines and upregulation of major histocompatibility complex class II and costimulatory molecules (22, 29, 43). LPS is likely to be a key bacterial component responsible for this maturation. Purified LPS molecules from Salmonella and Escherichia coli, as well as synthetic lipid A molecules, are capable of inducing DC maturation (11, 15, 19, 43, 51).
LPS is the major component of the outer leaflet of the outer membrane of gram-negative bacteria. This glycolipid consists of three structural and functional domains: lipid A, core, and O antigen (13, 34). LPS is a potent immunostimulatory molecule, and this activity is primarily associated with the lipid A domain (reviewed in reference 1). Lipid A is highly conserved, which presumably reflects its specific role in outer membrane structure, in most gram-negative bacteria, including Salmonella. Lipids A of E. coli and Salmonella enterica serovar Typhimurium have a β(1,6)-linked d-glucosamine disaccharide backbone which is phosphorylated and has four primary hydroxy fatty acyl substitutions at positions 2, 3, 2′, and 3′. The 3-hydroxyl groups of the 2′ and 3′ fatty acids are further replaced by laurate and myristate residues, respectively. Removal of these secondary acyl groups from Salmonella LPS results in reduced endotoxic activity of the molecule (33).
The immunostimulatory effects of lipid A and LPS from pathogenic bacteria can be inhibited by naturally occurring LPS antagonists, such as the penta-acyl lipid A forms of LPS from Rhodobacter sphaeroides and Rhodobacter capsulatus (27), or more potent synthetic antagonists, such as E5531, whose structure is based on the proposed structure of lipid A from R. capsulatus (5, 23).
E. coli (40, 41), Haemophilus influenzae (26), Salmonella (21, 28), and Shigella flexneri (8) strains which have functional mutations in either the lpxL gene (also known as htrB or waaM) or the lpxM gene (also known as msbB or waaN) have been generated. These genes are responsible for encoding the lauroyl and myristoyl transferases that catalyze the secondary acylation reactions. The mutants synthesize structurally altered lipid A, which results in LPS with reduced toxicity.
lpxM mutants do not add the secondary myristoyl fatty acid to the lipid A domain, and this mutation allows bacteria to grow at 37°C, unlike the lpxL gene deletion, which creates temperature-sensitive mutants (18). Somerville et al. (40) described the first E. coli lpxM mutant. This mutant has a reduced ability to induce tumor necrosis factor alpha (TNF-α) and E-selectin expression in adherent monocytes and on endothelial cells, respectively. Similarly, the serovar Typhimurium lpxM mutants described by Khan et al. (21) and Low et al. (28) show decreased endotoxicity compared to the wild type. The serovar Typhimurium C5 lpxM strain has growth kinetics similar to those of the C5 wild-type parental strain in vivo, growing to very high levels in mouse organs, but its ability to kill the mice is reduced, demonstrating that lipid A has a role in lethality. This mutant also induces lower levels of TNF-α and interleukin-1β (IL-1β) secretion and inducible nitric oxide synthase (iNOS) induction than the parental strain when J774 macrophages are infected (21). The lpxM mutant of strain 14028 described by Low et al. (28), which is capable of accumulating in tumors and suppressing the growth of tumor tissue, is less virulent and induces less TNF-α than the wild type.
In this study, we investigated the influence of a defective lipid A structure on the maturation of murine DCs when they encounter salmonellae.
MATERIALS AND METHODS
Mice.BALB/c and C57BL/6 mice (Charles River Laboratories, Margate, United Kingdom) that were 6 to 10 weeks old were used for all experiments.
Cell culture conditions.The DC line tsDC (48) was cultured in Iscove's modified Dulbecco's medium supplemented with 5% fetal calf serum (FCS), 2 mM l-glutamine, penicillin (100 U/ml), streptomycin (100 μg/ml), and 50 μM β-mercaptoethanol for variable periods of time at 33°C in the presence of 5% CO2. The J774A.1 macrophage cell line (35) was cultured in Dulbecco's modified Eagle's medium supplemented with 10% FCS, 2 mM l-glutamine, penicillin (100 U/ml), and streptomycin (100 μg/ml) at 37°C in the presence of 5% CO2.
Propagation of DCs from bone marrow.Femoral bone marrow cells were suspended in RPMI 1640 containing penicillin (100 U/ml), streptomycin (100 μg/ml), and 50 μM β-mercaptoethanol (RPMI) supplemented with supernatants containing 5% FCS, 1% mouse granulocyte-macrophage colony-stimulating factor, and 1% mouse IL-4 (see below) at a concentration of 106 cells/ml and incubated at 37°C in the presence of 5% CO2. The medium was replaced every 2 to 3 days, and the cells were used between days 6 and 8 of culture.
Supernatants containing murine granulocyte-macrophage colony-stimulating factor (52) and murine IL-4 (20) were a gift from R. Bujdoso, Department of Clinical Veterinary Medicine, University of Cambridge, Cambridge, United Kingdom.
Bacteria and culture conditions.Serovar Typhimurium strain SL3261, an aroA mutant (16), and the lipid A-deficient mutant SL3261 lpxM were used in this study. SL3261 lpxM was constructed by P22 transduction from the C5 lpxM mutant as previously described (21). SL3261 was grown in Luria-Bertani (LB) broth, and SL3261 lpxM was grown in LB broth supplemented with kanamycin (50 μg/ml). Both strains were grown overnight without shaking and then washed once with tissue culture medium. The optical densities at 600 nm of bacterial cultures resuspended in tissue culture medium were used to estimate bacterial numbers. Counting of viable bacteria was performed on LB agar. Bacteria were heat killed by incubation at 70°C for 30 min.
Stimulation of macrophages and DCs.DCs or macrophages were mixed with live bacteria for 2 h, the cells were washed three times with phosphate-buffered saline (PBS), and then medium containing 15 μg of gentamicin per ml was added and the cells were incubated for a further 2 or 22 h. When heat-killed bacteria and LPS (1 μg/ml; serovar Typhimurium LPS; Sigma, Poole, United Kingdom) were used, the cells were incubated continuously for 4 or 24 h.
Measurement of cytokine and iNOS induction.A total of 105 J774 macrophages or tsDC cells were stimulated as described above in quadruplicate, and the culture supernatants were collected after 4 h for measurement of TNF-α and after 24 h for measurement of IL-1β and nitrite and stored at −70°C before they were used. Cytokine levels were determined by using mouse TNF-α and IL-1β DuoSet enzyme-linked immunosorbent assay (ELISA) systems (R&D Systems Europe Ltd., Abingdon, United Kingdom). The activity of iNOS in the cells was measured by determining the accumulation of nitrite by the Griess assay as described by Green et al. (12).
Detection of cell surface markers by flow cytometry.Expression of cell surface markers was analyzed with a FACSCalibur flow cytometer by using CellQuest, version 3.3 (Becton Dickinson Immunocytometry Systems, Oxford, United Kingdom). Monoclonal antibodies against CD11c, CD80, CD86, CD40, and major histocompatibility complex class II and isotype-matched control antibodies were purchased from BD PharMingen, Oxford, United Kingdom. Phycoerythrin conjugated to streptavidin (streptavidinPE) was obtained from Serotec, Oxford, United Kingdom.
DCs were stimulated for 24 h as described above or were maintained in medium alone and then incubated with appropriate dilutions of monoclonal antibody for 40 min at 0°C and washed three times with PBA (PBS with 0.1% bovine serum albumin and 0.01% sodium azide). Cells stained with fluorescein isothiocyanate-conjugated antibody were kept on ice. Biotinylated anti-CD11c-stained cells were then incubated with streptavidinPE for 20 min on ice and washed. The cells were fixed with 2% formaldehyde before acquisition.
Mixed leukocyte reaction.tsDC cells or bone marrow-derived DCs (BMDCs) were incubated at 37°C for 24 h either in medium alone or with heat-killed salmonellae (ratio of bacteria to cells, 0.5 or 5). After 24 h, the cells were washed and irradiated (3,000 cGy; Pantak linear accelerator). CD4 splenocytes were isolated from mouse spleens by using CD4 (L3T4) microbeads (Miltenyi Biotec Ltd., Bisley, United Kingdom). The appropriate numbers of irradiated stimulator cells and 105 responder CD4 splenocytes were incubated at 37°C for 72 h, and the supernatants were collected and frozen at −70°C. T-cell cytokine levels in the supernatants were determined by the CTLL-2 proliferation assay. Each test was carried out in quadruplicate.
CTLL-2 bioassay.CTLL-2 (ECACC no. 93042610) cells were routinely maintained in complete medium with 2 ng of murine IL-2/ml (R&D Systems Europe Ltd.). Before the bioassay, CTLL-2 cells were cultured overnight in RPMI with 2% FCS without IL-2 at a concentration of 1 × 106 to 1.5 × 106 cells/ml. A total of 104 cells in complete medium were cultured with supernatants from mixed leukocyte reaction mixtures, medium, or IL-2 (2 ng/ml) for 24 h at 37°C in the presence of 5% CO2 and labeled with [3H]thymidine (1 μCi/well; Amersham Biosciences, Little Chalfont, United Kingdom) for the last 16 to 18 h. Cells were harvested onto glass fiber filter mats and counted with a Wallac 1450 Microbeta liquid scintillation counter to determine tritiated thymidine uptake by cells.
Preparation of LPS samples.Large-scale isolation of high-purity LPS was performed by using a modification of the hot phenol-water method of Westphal and Jann (49). Bacteria were grown with shaking overnight in LB broth at 37°C, washed three times, suspended in distilled water, and sonicated. The suspensions were treated sequentially with DNase I (100 μg/ml; Sigma), RNase (100 μg/ml; Sigma), and proteinase K (1 mg/ml; Sigma) and then sonicated again. The resulting suspensions were extracted with phenol heated to 70°C. Aqueous phases were dialyzed (Spectra/Por dialysis tubing; Medicell International Ltd., London, United Kingdom) against distilled water for 2 to 3 days. Particulate material was removed by centrifugation (20,000 × g, 1 h). The supernatants were then lyophilized and suspended in 5 ml of distilled water before a final centrifugation to isolate LPS (80,000 × g, 1 h). The resulting pellets were lyophilized, and these preparations were dissolved in distilled water to a final concentration of 1 mg/ml and stored in aliquots at −20°C. The levels of protein or nucleic acid in LPS samples were assessed by measuring the absorbance at 260 and 280 nm and by a protein assay (BCA assay; Bio-Rad). The molarity of LPS samples was then determined by the Purpald assay (25).
SDS-PAGE analysis of LPS.For rapid analysis of LPS profiles by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE), small-scale preparations of LPS were made by proteinase K treatment of whole-cell lysates for 1 h at 60°C (14). Samples were transferred to 2.9% SDS-70 μM Tris-HCl (pH 6.8)-14.3% glycerol-0.06% bromophenol blue and boiled for 10 min before they were stored at −20°C until they were used. LPS from an equivalent number of cells was loaded into each well of 16% (wt/vol) polyacrylamide gels as described by Laemmli (24). Samples were visualized by silver staining (46).
Statistical analysis.The data below are means ± standard deviations unless indicated otherwise. The Student t test was used to determine statistical significance for two cultures that were treated differently. Differences were considered significant if the P value was ≤0.05.
RESULTS
Characterization of the SL3261 lpxM mutant.It has been shown that lpxM mutants can be used in vivo and in vitro successfully to study the role of lipid A. However, as LPS is a vital structural component of gram-negative bacteria, defects in this molecule may lead to secondary mutations. Growth defects and pleiotropy associated with lpxM mutation in Salmonella have been identified recently (32; Emilianus, unpublished data). Therefore, to control for associated secondary effects caused by the mutation, the SL3261 lpxM mutant was complemented with a wild-type lpxM gene provided in trans on a plasmid. A functional copy of the lpxM gene in a 3.1-kb DraI fragment of strain C5 genomic DNA was subcloned into the HindIII site of pBR322 to produce pRR1. Insertion into the HindIII site resulted in inactivation of the promoter region of the tetracycline resistance gene of pBR322. Therefore, a control plasmid for the complementation assays was created in which a 787-bp HindIII fragment from the chloramphenicol acetyl transferase gene was inserted into the HindIII site of pBR322 (pRK1) to inactivate the tetracycline resistance gene promoter. Inactivation of the tetracycline resistance gene was confirmed before this plasmid was used as a vector control for SL3261 lpxM complementation. Then SL3261, the SL3261 lpxM mutant, and the SL3261 lpxM mutant transformed with pRR1 or pRK1 were used to assess the effect of lpxM deletion on the immunostimulatory capacity of bacteria.
The presence of O antigen on the mutants was determined by using a P22 sensitivity assay. Bacteriophage P22 requires O antigen (smooth LPS) for binding to the bacterial surface. SL3261, the lpxM mutant, and all of the derivatives described above were found to be highly susceptible to infection, which showed that they express smooth LPS (data not shown). All the bacteria used were positive for antigens O-4 and O-5 when Salmonella agglutinating antisera were used, and the plasmid stabilities of the complementation plasmid and the vector plasmid were more than 90% after overnight culture (data not shown).
LPS preparations from the parental strain (SL3261) and the mutants derived from this strain were separated on an SDS-PAGE gel (Fig. 1A). There was a clear difference between the banding pattern of SL3261 lpxM LPS and that of the LPS of the parental strain, which produced bands at apparently higher molecular weights. Similar banding patterns were observed for the SL3261 LPS and SL3261 lpxM/pRR1 LPS. The banding patterns of the SL3261 lpxM LPS and SL3261 lpxM/pRK1 LPS (vector-only control) were similar to each other. The presence of smooth LPS on all strains was confirmed by the presence of the O-antigen ladder in all tracks (Fig. 1A).
Complementation of serovar Typhimurium strain SL3261 lpxM. Complementation was assessed by LPS production and the ability to induce TNF-α secretion from J774 macrophages. (A) LPS from SL3261 and mutants of this strain were electrophoresed on an SDS-PAGE gel and then silver stained. (B) A total of 2 × 105 J774 macrophages were incubated with heat-killed salmonellae for 4 h at 37°C. TNF-α levels in the culture supernatants were determined by an ELISA. The bars and error bars indicate the means ± standard deviations for four samples.
For characterization of the C5 lpxM mutant (21) we used TNF-α induction from J774 macrophages as a measure of endotoxin activity. This parameter was also used to assess complementation of the lpxM mutant. As expected, SL3261 lpxM was impaired in the ability to induce TNF-α release from J774 macrophages compared to the ability of the parental SL3261 strain; however, a significant difference between SL3261 and SL3261 lpxM was seen only at a low multiplicity of infection (MOI) (≤0.5). As shown in Fig. 1B, the abilities to induce TNF-α secretion from macrophages were similar for SL3261 and SL3261 lpxM/pRR1 and for SL3261 lpxM and SL3261 lpxM/pRK1.
These findings clearly showed that any differences between SL3261 lpxM and SL3261 were due to the lack of the lpxM gene and not to secondary changes. We therefore continued to use SL3261 and mutants of this strain to investigate the interaction of these bacteria with DCs.
Effect of lpxM mutation on cytokine and iNOS induction by Salmonella in DCs.As shown in Fig. 1B, serovar Typhimurium strain SL3261 and mutants of this strain induced TNF-α secretion from J774 macrophages in an MOI-dependent manner. This phenomenon was also assayed in tsDCs by stimulating the cells with various numbers of heat-killed salmonellae (MOI, 0.05 to 5). Both heat-killed SL3261 and SL3261 lpxM induced TNF-α and IL-1β secretion from, and iNOS activity in, tsDCs in an MOI-dependent manner (Fig. 2). The highest levels of cytokines and nitrite were seen with an MOI of 5, but as previously seen with J774 cells (Fig. 1B), there was no significant difference between the levels induced by the parental strain and the levels induced by the lpxM mutant at this MOI (Fig. 2). At an MOI of 0.5, there was a significant difference between SL3261 and SL3261 lpxM (Fig. 2). However, at an MOI of 0.05 there was very little TNF-α present in the supernatant, while no detectable IL-1β or nitrite was found (data not shown).
Induction of cytokine secretion and iNOS activity by serovar Typhimurium strains SL3261 and SL3261 lpxM in tsDCs: the effect of the ratio of bacteria to cells. DCs were incubated with a given number of heat-killed bacteria for 4 h (TNF-α) or 24 h (IL-1β and iNOS), and the cytokine levels in the culture supernatants were determined by an ELISA. The presence of nitrite in the supernatants was used as a marker for iNOS activity, and the nitrite levels were determined by the Griess reaction. Representative results of three experiments are shown. The bars and error bars indicate the means ± standard deviations for four samples. An asterisk indicates that the value is significantly smaller than the value obtained with SL3261 (P ≤ 0.05).
Preparations of heat-killed bacteria contain disrupted bacteria, and the LPS from these bacteria may be presented to cells free in solution. To determine if there were similar differences in lipid A activity for cellular activation with intact bacteria, live washed bacteria were used at an MOI of 0.5 to activate tsDCs. To ensure that only bacteria which had been taken up by cells were involved in the activation process, unbound live bacteria were washed off and extracellular bacteria were killed with gentamicin after 2 h. When this protocol was used with heat-killed bacteria, there was no detectable cytokine or iNOS induction (data not shown). We therefore had to maintain continuous stimulation of DCs with heat-killed bacteria (without washing) for 4 h (TNF-α) or 24 h (IL-1β and iNOS) to detect changes. As shown in Fig. 3, live salmonellae, as well as heat-killed salmonellae, induced secretion of cytokines (TNF-α and IL-1β) from tsDCs.
Effect of Salmonella viability on cytokine secretion and iNOS activity in tsDCs. DCs were incubated with heat-killed salmonellae for 4 h (TNF-α) or 24 h (IL-1β and iNOS). For live bacterial infections, DCs were incubated with stationary-phase salmonellae for 2 h. Then the unbound bacteria were washed off, and the cells were cultured for a further 2 h (TNF-α) or 22 h (IL-1β and iNOS) in medium containing 15 μg of gentamicin per ml. Supernatants were assayed by an ELISA or a Griess assay. Representative results of three experiments are shown. The bars and error bars indicate the means ± standard deviations for four samples. An asterisk indicates that the value is significantly smaller than the value obtained with SL3261 (P ≤ 0.05).
There was a similar reduction in the ability of the lpxM mutant to induce production of cytokines (TNF-α and IL-1β) in tsDCs when either live bacteria or heat-killed bacteria were used. The ability was restored by supplying the lpxM gene in trans (Fig. 3). When tsDCs were stimulated with heat-killed salmonellae, SL3261 lpxM induced less iNOS activity than SL3261 induced (Fig. 3), and again the wild-type phenotype could be rescued by supplying the lpxM gene in trans. With live bacteria at an MOI of 0.5 there was no detectable nitrite in the supernatants.
Effect of lpxM mutation on induction by Salmonella of cell surface marker expression on DCs.tsDCs were stimulated with either heat-killed or live SL3261 or SL3261 lpxM at an MOI of 5. Bacterial viability had no effect on the results, and SL3261 and the lpxM mutant upregulated costimulatory molecule expression on tsDCs to the same extent (Fig. 4A). However, cytokine induction was MOI dependent, and significant differences between the parental strain and the lpxM mutant were seen only at low MOIs (e.g., 0.5). Therefore, cell surface marker expression was analyzed after stimulation and infection at an MOI of 0.5. At this lower MOI there was very little increase in surface molecule expression on tsDCs, but again there was no apparent difference between stimulation by SL3261 and stimulation by the lpxM mutant (data not shown). Similar results were obtained when BMDCs were stimulated with heat-killed salmonellae (Fig. 4B). As day 7 BMDCs were a mixed cell population, DCs were defined as cells expressing high levels of CD11c.
Induction of costimulatory molecule expression by serovar Typhimurium strain SL3261 lpxM on DCs. tsDCs (A) or BMDCs (B) and salmonellae were mixed at a ratio of 1:5 and incubated for 24 h. For live bacterial infections, tsDCs were incubated with stationary-phase salmonellae for 2 h. Then the unbound bacteria were washed off, and the cells were cultured for a further 22 h in medium containing 15 μg of gentamicin per ml. The cells were then washed and stained with isotype-matched control antibodies or antibodies specific for surface markers. For all the markers except CD11c, antibodies directly conjugated to fluorescein isothiocyanate were used. Anti-CD11c conjugated to biotin was visualized with streptavidinPE. The grey histograms indicate the background staining of cells with isotype-matched control antibodies. (The staining profiles for unstimulated and stimulated cells were similar.) The open histograms show cell surface marker expression. The thick solid lines show the staining of unstimulated cells. The thin solid lines and the dotted lines show the responses of the DCs to SL3261 and the SL3261 lpxM mutant, respectively. The data are representative of the data obtained in three experiments that yielded comparable results.
In live or heat-killed bacteria there are various components, such as bacterial DNA, which activate DCs. Therefore, to exclude the possibility that bacterial components other than LPS were responsible for upregulation of surface molecule expression, purified LPS from the bacteria was used in parallel with wild-type LPS purchased from Sigma to stimulate tsDCs. The cellular responses to all three LPS preparations were similar (data not shown). At a concentration of 1 μg/ml, there was no significant difference between LPS preparations from SL3261 and SL3261 lpxM in terms of the ability to upregulate costimulatory molecule expression in tsDCs (Fig. 5). A similar result was obtained with 100 ng of LPS per ml (data not shown).
Upregulation of costimulatory molecules on tsDCs after treatment with purified mutant LPS. DCs were coincubated with purified LPS at a concentration of 1 μg/ml for 24 h. Flow cytometry was performed with these cells and a set of unstimulated cells. The gray histograms indicate staining with isotype control antibodies (there was no difference between unstimulated and stimulated cells). The open histograms show cell surface marker expression. The thick solid lines show the staining in untreated cells. The thin solid lines and the dotted lines show the responses of the DCs to LPS purified from SL3261 and Sl3261 lpxM, respectively. The data are representative of the data obtained in three experiments that yielded comparable results.
To confirm that LPS is the stimulatory molecule, polymyxin inhibition of lipid A binding to cells was used. Polymyxin B is a cationic antimicrobial peptide which binds to the lipid A portion of the LPS molecule and thus inhibits binding of lipid A to the cell (31). Polymyxin B could partially inhibit the upregulation of CD80, CD86, and CD40 expression on BMDCs stimulated by purified LPS (Fig. 6). However, no difference was seen when heat-killed serovar Typhimurium strain SL3261 was used in the presence or absence of polymyxin B to stimulate BMDCs (Fig. 6).
Polymyxin B inhibition of LPS. BMDCs were treated with purified LPS (A) or with heat-killed serovar Typhimurium strain SL3261 (B) in the absence or presence of polymyxin B for 24 h. The gray histograms indicate staining with isotype control antibodies for cells incubated in medium containing polymyxin B. The expression of surface markers on unstimulated cells is indicated by solid thick lines. There was no difference between the staining profiles of BMDCs incubated in medium with polymyxin B and the staining profiles of BMDCs incubated in medium without polymyxin B (data not shown). The thin solid lines and the dotted lines show the staining profiles for stimulated cells incubated in the absence and in the presence of polymyxin B, respectively.
Effect of lpxM mutation on Salmonella-induced modulation of allostimulatory properties of DCs.Another hallmark of DC activation is an increased capacity to activate allogeneic T cells. This property was tested with tsDCs (H-2k) and BMDCs (H-2d) by using allogeneic CD4+ T cells from C57BL/6 (H-2b) mice. CD4 T-cell proliferation levels, as measured by thymidine incorporation, were increased in the presence of bacterially stimulated DCs (data not shown). However, any B-cell contamination may confound this type of assay as LPS is a B-cell mitogen. Therefore, the presence of T-cell cytokines (IL-2 and IL-4) in the supernatants was determined by a CTLL-2 cell proliferation assay. tsDCs stimulated with heat-killed salmonellae exhibited an augmented capacity to stimulate T cells to produce IL-2 and/or IL-4 which was independent of the lpxM mutation even at low MOIs which altered the secretion of cytokines (Fig. 7A).
Changes in alloreactive properties of DCs after treatment with heat-killed salmonellae. tsDCs or BMDCs were incubated at 37°C for 24 h either in medium alone or with heat-killed salmonellae. After 24 h of incubation the cells were washed, irradiated, and cocultured with responder CD4 splenocytes from H-2b for 72 h, and the supernatants were collected. T-cell cytokine levels in the supernatants were determined by the CTLL-2 proliferation assay. Proliferation of the CTLL-2 line in response to IL-2 in the supernatants is expressed as the mean ± standard deviation for incorporated [3H]thymidine for quadruplicate samples. (A) Coculture of 4 × 104 tsDCs and 105 CD4 splenocytes. Ratios of bacteria to cells of 0.5 and 5 were used to stimulate the tsDCs. (B) Preparations containing 105 CD4 splenocytes were mixed with different numbers of BMDCs. Symbols: *, untreated BMDCs; ▴, BMDCs treated with serovar Typhimurium strain SL3261 at an MOI of 5; ▵, BMDCs treated with serovar Typhimurium strain SL3261 at an MOI of 0.5; •, BMDCs treated with serovar Typhimurium strain SL3261 lpxM at an MOI of 5; ○, BMDCs treated with serovar Typhimurium strain SL3261 lpxM at an MOI of 0.5.
A similar experiment was carried out with BMDCs, in which a range of stimulator/effector ratios was used. BMDCs stimulated by heat-killed salmonellae exhibited an increased ability to stimulate allogeneic T cells, and the greatest increase compared to the ability of unstimulated cells was seen with low numbers of BMDCs. This was again lpxM independent and MOI independent (Fig. 7B).
DISCUSSION
Although purified LPS has a clearly defined activity as a DC stimulator (10, 11, 30, 50), a number of bacterial components other than LPS can also activate DCs. In this study we evaluated the role of native lipid A, as a constituent of Salmonella, in the maturation of DCs.
The aim of introducing the lpxM mutation into serovar Typhimurium is to develop less reactogenic bacteria which can be used as vaccines or for other therapeutic purposes, such as anticancer treatment (6, 7, 28, 45). Such organisms could be used with a reduced risk of TNF-α-mediated septic shock. It has been shown that lpxM mutants have an altered ability to stimulate macrophages (21, 28). However, it is important to investigate whether this safer phenotype also affects the immunogenicity of the vector bacteria. As DCs are central to induction of the immune response, we tested the effect of the lpxM mutation on stimulation of DCs.
A non-temperature-sensitive serovar Typhimurium lpxM deletion mutant was an ideal candidate to study both the objectives described above. The fact that it was possible to functionally complement the lpxM mutation with a cloned lpxM gene confirmed that the mutant used did not have major secondary mutations.
The lack of secondary myristoylation of lipid A reduced the induction of secretion of the proinflammatory cytokines TNF-α and IL-1β, as well as iNOS activity in tsDCs, when cells encountered viable or heat-killed salmonellae. Compared to the induction by the parental strain, twofold-lower levels of cytokines and iNOS were induced by the lpxM mutant. This reduction in endotoxic activity reflected that seen on macrophages with similar mutants of E. coli strain K-12 and serovar Typhimurium strains C5 and 14028 (21, 28, 40). Although unbound and extracellular live bacteria were washed off and killed after 2 h, TNF-α induction by treatment of DCs with live bacteria was very efficient. Live Salmonella cells survive in vesicle-bound compartments of infected cells and may continue to produce stimulatory molecules and interact with cellular receptors in these vesicles. There is evidence of Toll-like receptors in intracellular vesicles (17, 47), and these receptors may play important roles in continued signaling with molecules such as LPS after phagocytosis or invasion of the cell.
The reduction in endotoxic activity was MOI dependent, and a difference between stimulation by the parent and stimulation by the mutant was seen only at a low MOI with both tsDC and J774 cells. To obtain the original data describing lpxM mutation as a way to reduce serovar Typhimurium endotoxicity on macrophages, the workers used an MOI of 0.05 (21). We clearly show here that at MOIs of 5 or more, the lack of the secondary myristoyl group in lipid A had no effect on the endotoxicity of Salmonella. This is similar to results obtained from stimulation of DCs by an LPS-deficient lpxA mutant of Neisseria meningitidis, which showed that at a higher MOI parental and mutant strains induced similar levels of TNF-α and IL-1β secretion, while at a lower MOI reduced levels of cytokine induction by the mutant were seen (9). The reason behind the MOI dependence is not clear. We have not studied binding of the LPS from the lpxM mutant with the cellular LPS receptor and its associated signaling complex. The mutant LPS may only be able to signal weakly compared to wild-type LPS through the complex, but high receptor occupancy caused by high infection ratios may provide enough cumulative signal to activate cells fully. However, multiple stimulatory molecules are present in bacteria; in addition to LPS there are CpG DNA, flagellin, double-stranded RNA, and other molecules, all of which are known to cause signaling in macrophages and DCs which leads to cytokine release. At high infection ratios these other molecules may become more important in causing activation of cells. BMDCs from mice with mutated Toll-like receptor 4, such as C57BL/10ScCr or C3H/HeJ mice, are not sensitive to LPS, but they can mature in response to gram-negative bacteria (38), showing that bacterial components other than LPS are involved in DC activation.
A single fatty acyl chain deletion from the lipid A core apparently had no effect on upregulation of surface molecules on tsDC cells and BMDCs. This was probably not due to involvement of other bacterial components, as purified LPS from wild-type and mutant Salmonella cells induced expression of similar levels of costimulatory molecules on tsDC cells and were inhibited by polymyxin B on BMDCs.
Polymyxin B did not interfere with stimulation of surface molecule expression caused by heat-killed bacteria on BMDCs. As discussed above, other bacterial components are also involved in upregulation of surface molecules on DCs (38), or it may be that LPS from live or heat-killed bacteria was released in a cellular compartment not accessible to polymyxin B but accessible to TLR4. The increased surface molecule expression caused by both of the bacteria used contributed to the augmentation of the allostimulatory properties of tsDC cells and BMDCs.
Activation of the phoPQ two-component regulatory system in Salmonella results in specific modifications of lipid A (e.g., replacement of myristate with 2-hydroxymyristate in the acyloxyacyl moiety at position 3′). Using wild-type and phoP mutant serovar Typhimurium strains, Svensson et al. (43) showed that the lipid A modifications controlled by phoP do not affect the maturation program induced in immature DCs. The data presented here suggest that the absence of a single fatty acyl chain at position 3′ in the lipid A mutant does not result in a major difference between the abilities of parental and mutant serovar Typhimurium strains to activate murine DCs in vitro.
The findings of this study show that although the lpxM mutant was attenuated in the ability to induce cytokine secretion from macrophages and DCs, it did not lose the ability to activate DCs sufficiently to allow them to activate T cells. DCs are critically important in the initiation of an immune response. Therefore, Salmonella lpxM mutants potentially have direct applications in human and veterinary medicine as vaccine and therapeutic agents, as they are able to activate the most important antigen-presenting cell, the DC, but they have a safer phenotype than wild-type bacteria.
ACKNOWLEDGMENTS
Ruwani Kalupahana was supported by a Cambridge Commonwealth Trust scholarship and an Overseas Research Studentship. This work was also supported by the BBSRC.
We thank Shahid Khan, Microscience Ltd., Wakingham, United Kingdom, for providing S. enterica serovar Typhimurium SL3261 lpxM and Brigitta Stockinger, Division of Molecular Immunology, National Institute for Medical Research, London, United Kingdom, for providing the tsDC cell line. We gratefully acknowledge Pietro Mastroeni for helpful comments and discussions.
FOOTNOTES
- Received 21 April 2003.
- Returned for modification 30 May 2003.
- Accepted 29 July 2003.
- Copyright © 2003 American Society for Microbiology
