Circulating Lipoproteins Are a Crucial Component of Host Defense against Invasive Salmonella typhimurium Infection

Background Circulating lipoproteins improve the outcome of severe Gram-negative infections through neutralizing lipopolysaccharides (LPS), thus inhibiting the release of proinflammatory cytokines. Methods/Principal Findings Low density lipoprotein receptor deficient (LDLR−/−) mice, with a 7-fold increase in LDL, are resistant against infection with Salmonella typhimurium (survival 100% vs 5%, p<0.001), and 100 to 1000-fold lower bacterial burden in the organs, compared with LDLR+/+ mice. Protection was not due to differences in cytokine production, phagocytosis, and killing of Salmonella organisms. The differences were caused by the excess of lipoproteins, as hyperlipoproteinemic ApoE−/− mice were also highly resistant to Salmonella infection. Lipoproteins protect against infection by interfering with the binding of Salmonella to host cells, and preventing organ invasion. This leads to an altered biodistribution of the microorganisms during the first hours of infection: after intravenous injection of Salmonella into LDLR+/+ mice, the bacteria invaded the liver and spleen within 30 minutes of infection. In contrast, in LDLR−/− mice, Salmonella remained constrained to the circulation from where they were efficiently cleared, with decreased organ invasion. Conclusions plasma lipoproteins are a potent host defense mechanism against invasive Salmonella infection, by blocking adhesion of Salmonella to the host cells and subsequent tissue invasion.


Introduction
Salmonella infections are a significant cause of morbidity and mortality, despite preventive measures and the availability of antibiotics. A major virulence factor of Salmonella is lipopolysacharide (LPS) [1,2], and Salmonella strains with a reduced LPS expression have a poor growth under stress conditions and are less virulent [2]. In addition, LPS induces proinflammatory cytokines and is essential for internalization of Salmonella by host cells [3]. Interaction of LPS with cellular receptors is essential and therefore, strategies aimed at blocking this interaction may have a therapeutic potential in invasive infections.
Lipoproteins bind and neutralize bacterial LPS, and they prevent the induction of potentially harmful proinflammatory cytokines such as IL-1b and TNFa [4]. In experimental models, administration of lipoproteins protects against endotoxic shock [5][6][7]. Low density lipoprotein receptor deficient (LDLR2/2) mice have a 7 times higher LDL-cholesterol level than control mice. We have shown previously that LDLR2/2 mice survive longer and have lower proinflammatory cytokine concentrations than control mice after LPS challenge, as well as after infection with Klebsiella pneumoniae [8]. In addition, LDL administration can protect against Gram-negative microorganisms through its neutralizing effects on LPS [9].
Although Salmonella is a Gram-negative organism, the clinical picture and inflammatory response in systemic Salmonella infections (e.g., typhoid fever) differs from that in other Gram-negative sepsis [10]. This is most likely due to the behaviour of Salmonella as facultative intracellular pathogens, and to the fact that the pattern of cytokine induction differs from other Gram-negative infections. In mice, TNFa is undetectable in the circulation until several days after S. typhimurium infection, whereas TNFa rises at 1 hour after extracellular Gram-negative infection [11]. The level of cytokinemia during Salmonella infections does not reach the toxic levels seen in endotoxic shock, and inhibition of TNFa during Salmonella infection worsens the outcome [12,13].
Considering these differences in the pathogenesis of Salmonella and extracellular Gram-negative infections, one would envisage either beneficial effects of lipoproteins on the host resistance to Salmonella through blockade of cellular internalization, or deleterious effects by blocking the induction of cytokines by Salmonella LPS that are required for the activation of host defense. In the present study, we investigated the effect of lipoproteins on the outcome of Salmonella infection.

Animals
Homozygous C57Bl/6J mice lacking low density lipoprotein receptors (LDLR2/2) and their wild-type littermates (C57Bl/6J LDLR+/+) were obtained from Jackson Laboratory (Bar Harbour, ME) [8]. Homozygous apolipoprotein E (ApoE)-deficient mice on a C57Bl/6 background were obtained from the Transgenic Facility of Leiden University Medical Center, Leiden, The Netherlands [14]. Six to eight weeks old littermate LDLR+/+ and LDLR2/2 mice were used, weighing 20-25 grams. The animals were fed standard laboratory chow and housed under specific pathogen free conditions. The experiments were approved by the Ethics Committee for animal experiments at the Radboud University Nijmegen.

Salmonella typhimurium infection
A serum-resistant strain of S. typhimurium (phage type 510) was grown by overnight incubation at 37uC in nutrient broth (BHI Oxoid). Mice were injected i.v. or i.p. with 1610 2 cfu of S. typhimurium. Survival was assessed daily for 21 days in groups of at least 20 animals. On day 1, 3 and 7 after infection, mice were killed by cervical dislocation and blood for cytokines or organs for outgrowth of the microorganisms were collected. For this purpose, the liver and spleen were removed aseptically, and bone marrow was flushed from the femur aseptically with 1 ml of sterile saline. The number of viable Salmonella organisms was determined by plating several dilutions on Brilliant Green agar (BGA) plates. The results were expressed as log cfu per gram of tissue.

Distribution of Salmonella cfu
In a separate experiment, Salmonella cfu (10 5 /mouse) were injected i.v. Distribution was determined in blood, liver and spleen after 30, 60, 120 and 360 minutes by plating serial dilutions of blood and homogenized tissue samples on BGA plates. Groups of 5 mice were used for each time point.

Intracellular killing of S. typhimurium by peritoneal phagocytes
Phagocytosis and intracellular killing of S. typhimurium was assessed in vitro using peritoneal macrophages and PMN of LRLR+/+ and LDLR2/2 mice. Exudate peritoneal neutrophils (PMN) were harvested 4 h after an i.p. injection of 10% proteose peptone, and exudate macrophages 72 h after i.p. injection of proteose peptone. 5610 5 cells in 100 ml of RPMI were dispensed into 96-well flat bottom plates (Costar) and incubated at 37uC and 5% CO 2 . To assess phagocytosis, 1610 5 Salmonella organisms/mL were incubated on the phagocyte monolayers at 37uC in RPMI with 10% serum. After 15 min, supernatants were aspirated and the monolayers were gently washed with medium to remove uningested bacteria. The supernatants were plated on BGA agar (the non-phagocytozed fraction). To assess intracellular killing, the wells containing the cells with phagocytosed bacteria were scraped with a plastic paddle and washed with 200 ml distilled H 2 O to lyse the phagocytes. The number of viable bacteria was determined by plating serial dilutions on BGA plates.

In vitro cytokine production
Resident peritoneal macrophages were harvested from peritoneal cavity, and cells were resuspended in RPMI 1640 containing 1 mM pyruvate, 2 mM L-glutamine and 100 mg gentamicin per ml, and incubated (10 5 /well) in 96-wells microtiter plates (Costar). Heat-killed (30 min, 100uC) S. typhimurium (10 6 microorganisms in 100 mL of RPMI) were added to peritoneal macrophages and incubated at 37uC in 5% CO2. After 24 h, the supernatants were collected and stored at 270uC until assayed. To the macrophages in the monolayer, 200 mL of RPMI was added and the cells were disrupted by three freeze-thaw cycles to determine the cellassociated cytokine contents.

Growth of Salmonella in vitro
To investigate the effect of lipoproteins on microbial growth in vitro, 0.5610 3 cfu S. typhimurium in 0.5 mL BHI were incubated with 0.5 mL of plasma obtained from control C57Bl/6J mice, or from LDLR2/2 mice and ApoE2/2 mice. After 2, 7, 12 and 24 hours, aliquots of 0.1 mL were removed, serial dilutions were plated on BGA agar, and cfu were counted after overnight incubation at 37uC.

Effect of lipoproteins on interaction of Salmonella with monocytes and endothelial cells
To assess the effect of lipoproteins on the production of cytokines, S. typhimurium LPS (10 ng/mL; Sigma) and heat-killed (30 min, 100uC) S. typhimurium (10 7 organisms/mL) were preincubated with lipoprotein-depleted plasma (LPDP) or isolated LDL [16] at various concentrations for 60 min, before being added to the macrophages of LDLR+/+ mice (10 5 /well). The production of TNFa after 24 h stimulation was measured as described above, and expressed as relative TNF production compared to controls in LPDP.
To assess the effect of lipoproteins on the attachment of Salmonella to vascular endothelial cells, S. typhimurium were resuspended to 6610 8 /ml in 0.01 mg/ml FITC (Fluka) in 0.05 M carbonate-bicarbonate buffer (pH 9.5). After incubation for 15 min at room temperature in the dark, FITC-labeled Salmonella cells were washed twice in PBS containing 1% BSA and subsequently incubated with isolated 1.1 mmol/L LDL for 4 hours, or with LPDP as a negative control. The human endothelial cell line (HMEC-1) (CDC Atlanta, GA) was cultured in MCDB131 medium supplemented with 10% fetal calf serum, EGF (10 ng/ml), hydrocortison (1 mg/ml), and glutamine at 37uC and 5% CO 2 ,. 1610 5 HMEC-1 cells were trypsinized and incubated for 1 hour with 3610 9 FITC-labeled S. typhimurium, which were preincubated with either LDL or LPDP. After incubation, the non-bound Salmonella was thoroughly washed off, after which the cells were fixated with 2% paraformaldehyde in PBS and analyzed for binding of FITC-labeled Salmonella by flow cytometry using the FACScalibur (BD Biosciences).

Statistical analysis
Survival of groups of mice was compared by the Kaplan-Meyer log-rank test. Differences in concentrations of cytokines and in organ counts of the microorganisms were analyzed by the Mann-Whitney U test. Differences were considered significant at P,.05. All the experiments were at least performed in duplicate.

Outcome of Salmonella infection in LDLR2/2 mice
The total cholesterol concentrations were significantly higher in the uninfected LDLR2/2 mice than in their wild-type littermates (9.661.1 mmol/L vs. 2.360.5 mmol/L). After i.v. infection with 10 2 cfu of S. typhimurium, only 5% of the LDLR2/2 mice died, whereas the mortality of control LDLR+/+ was 100% within 12 days of infection (P,.001; Fig. 1). A similar difference in mortality was apparent when mice were infected intraperitoneally with S. typhimurium (10% mortality in LDLR2/2 mice, vs. 100% mortality in control LDLR+/+ mice, p,0.01). The reduced mortality to infection in LDLR2/2 mice was accompanied by a markedly reduced bacterial load in the organs (Fig. 2). On day 7 of infection, the differences between the control and LDLR2/2 mice approached 10,000-fold (P,.001; Fig. 2).

Intracellular killing of Salmonella by cells from LDLR+/+ and LDLR2/2 mice in vitro
The numbers of Salmonella CFU phagocytized by neutrophils and macrophages of LDLR2/2 and LDLR+/+ mice were similar (Fig. 3A). In addition, the intracellular killing assay demonstrated that neutrophils and macrophages of LDLR2/2 mice and LDLR+/+ mice did not differ in their ability to kill S. typhimurium intracellularly (Fig. 3B). The killing rate did not differ when lipoprotein-rich serum of LDLR2/2 mice was coincubated with LDLR+/+ control macrophages, and likewise, serum from LDLR+/+ mice did not affect the killing of Salmonella by LDLR2/ 2 macrophages (not shown).

Circulating cytokines during Salmonella infection
On day 1, cytokine concentrations in all samples were under the detection limit. No detectable concentrations of IL-1ß (,20 pg/ ml) were found at any time point during the infection. On day 3, IL-1a and TNFa were under the detection limit in LDLR2/2 mice, while TNFa concentrations tended to be slightly higher (45610 pg/ml) in LDLR+/+ mice (n.s.). On day 7, circulating concentrations of IL-1a and TNFa were significantly higher in LDLR+/+ than in LDLR2/2 mice: 95663 pg/ml vs 30610 pg/ml for IL-1a (P,0.02) and 11406290 pg/ml vs 4366 pg/ml for TNFa (P,0.01) (Fig. 4A). These differences were most likely due to the greater amounts of Salmonella in the LDLR+/+ mice, leading to increased cytokine stimulation.  Importantly, the differences in cytokine concentrations were not secondary to an intrinsically deficient cytokine production in the organs of the LDLR2/2 mice: the amounts of TNF and IL-1b mRNA in the spleens of LDLR+/+ and LDLR2/2 mice on day 1 of infection was similar (Fig. 4B). No differences in the expression of IL-1a mRNA were observed either (not shown). In line with this, peritoneal macrophages of LDLR2/2 and LDLR+/+ mice stimulated with heat-killed Salmonella produced similar amounts of TNFa in vitro: unstimulated macrophages, 194691 pg/ml and 186655 pg/ml, macrophages coincubated with heat-killed Salmonella, 5506166 pg/ml and 5206135 pg/ml for LDLR2/2 and LDLR+/+ mice, respectively (P..05).

Resistance of ApoE2/2 mice to Salmonella infection
To investigate whether the increased lipoprotein concentrations in another model of hyperlipoproteinemia can also protect against salmonellosis, hyperlipoproteinemic ApoE2/2 mice (total serum cholesterol, 16.163.7 vs. 1.960.2 mmol/L [14]) were infected i.v. with 10 2 cfu of S. typhimurium. Whereas little differences were apparent on day 1 of infection, the outgrowth of Salmonella on day 3 and 7 after the infection was 100 to 1000-fold less in the liver and spleen of ApoE2/2 mice compared to that in ApoE+/+ control animals (Fig. 5).

Effect of lipoproteins on the growth of S. typhimurium in vitro
To test whether lipoproteins have a direct inhibitory effect on the growth of Salmonella, plasma isolated from control mice (cholesterol concentration 2.3 mmol/L) and hyperlipoproteinemic plasma from either LDLR2/2 or ApoE2/2 mice (cholesterol concentrations, 9.6 and 16.1 mmol/L) was added to the culture. The growth curves of Salmonella were similar in broth with plasma obtained from all mouse strains (Fig. 6A).

Inhibition of the interaction of Salmonella with monocytes and endothelial cells by lipoproteins
Because lipoproteins are known to bind and neutralize LPS, we preincubated Salmonella LPS and heat-killed whole Salmonella bacteria with LDL at various concentrations, and subsequently stimulated normal (LDLR+/+) macrophages for cytokine production. As shown in Fig. 6B, TNFa production induced by S. typhimurium is reduced in the presence of elevated LDL concentrations. To assess the effect of lipoproteins on the attachment of Salmonella to endothelial cells, FITC-labelled S. typhimurium was preincubated with LDL. Their attachment to endothelial cells was significantly reduced compared to that of Salmonella preincubated with lipoprotein-free plasma, as shown by both reduction of the percentage of cells binding Salmonella (97% cells bound LDPD-Salmonella, whereas only 82% cells bound LDL-Salmonella), as well as the mean fluorescence intensity per cell (41% reduction, from 56 to 33 conventional units) (see also Fig. 6C).

Protection against organ invasion by Salmonella in LDLR2/2 mice
To investigate whether hyperlipoproteinemia influences the early organ invasion from the bloodstream by Salmonella, we determined the early distribution of the microorganisms after i.v. injection of 10 5 S. typhimurium cfu. Blood from LDLR2/2 mice contained significantly more Salmonella cfu than that of LDLR+/+ mice 30 minutes after injection (P,0.01), but an efficient elimination of the microorganisms occurred during the next 6 hours (Fig. 7A). The numbers of cfu in the liver and spleen were 70-80% lower in LDLR2/2 mice than those in LDLR+/+ mice at 30 minutes after injection, and remained lower throughout the experiment, the difference between mouse strains being significant at 30, 60, 120 and 360 minutes for the liver (Fig. 7B, P,0.05) and at 30 and 60 minutes for the spleen (Fig. 7C, P,0.05).

Discussion
In the present study, we demonstrate that hyperlipoproteinemic mice are resistant against S. typhimurium infection. The protection was not due to the absence of the LDLR in the knock-out mouse strain, but to a direct effect of hyperlipoproteinemia. The beneficial effect of lipoproteins was exerted by blocking the interaction of Salmonella with host cells, including endothelial cells and monocytes, which led to inhibition of organ invasion. This resulted in an altered distribution of the microorganism to the organs of the host, and increased survival.
It has previously been shown that lipoproteins bind and neutralize LPS, with beneficial effects in Gram-negative infections [5][6][7][8]. As S. typhimurium is an LPS-containing Gram-negative bacterium, and Salmonella LPS plays a crucial role in cytokine stimulation [17] and induction of mortality in vivo [18], an improved survival of LDLR2/2 mice during systemic S. typhimurium infection would have been expected. Indeed, LDLR2/2 mice were less susceptible to S. typhimurium infection, but this was not due to blunted cytokine production, as we observed undetectable or low cytokine circulating concentrations during the infection in the LDLR2/2 and LDLR+/+ mice. In addition, the expression of cytokine mRNA was similar in the organs of LDLR2/2 and LDLR+/+ mice. Thus, the cytokine response is not responsible for the mortality due to systemic S. typhimurium infection in this model, and the major difference is the almost complete absence of Salmonella in the organs of the LDLR2/2 mice.
The low circulating cytokine response during systemic S. typhimurium infection may be attributed to the facultative intracellular nature of the organism. It should be noted that at later time points during the infection, the LDLR+/+ mice Figure 4. In-vivo cytokine production in LDLR+/+ and LDLR2/2 mice infected with S. typhimurium. LDLR2/2 and control (LDLR+/+) C57Bl/6J mice were infected i.v. with 10 2 cfu of S. typhimurium. After 1, 3 or 7 days, groups of 5 mice/group were sacrificed, and circulating concentrations of cytokines in the plasma were measured by specific ELISA. In a separate experiment, TNF and IL-1b mRNA in the spleens of LDLR+/+ and LDLR2/2 mice was assessed by semi-quantitative RT-PCR, and expressed as ratio to GAPDH mRNA expression. Data are presented as means6SD, *p,0.05. doi:10.1371/journal.pone.0004237.g004 Figure 5. Outgrowth of S. typhimurium in the organs of ApoE+/+ and ApoE2/2 mice. Outgrowth of S. typhimurium in the liver and spleen of ApoE2/2 and control (ApoE+/+) C57Bl/6J mice after i.v. injection of 10 2 cfu. Each point represents the mean6SD for at least 10 animals. Significant differences between ApoE2/2 and ApoE+/+ mice were found for both liver and spleen on days 3 and 7 (p,0.01). doi:10.1371/journal.pone.0004237.g005  exhibited a greater cytokine response than did LDLR2/2 mice, and this is likely due to the 1000-fold greater bacterial burden in the control mice, leading to substantial stimulation of the host response.
We hypothesized that the decreased bacterial growth in the LDLR2/2 mice may be a result of enhanced phagocytosis and intracellular killing of Salmonella organisms in these mice. However, this proved not to be the case: phagocytosis and subsequent intracellular killing of the organisms by both neutrophils and macrophages did not differ between LDLR2/2 and LDLR+/+ mice. Another reason for protection could have been the absence of the LDL receptor in the knock-out mice. Toxoplasma gondii is known to use the host LDLR for cholesterol acquisition [19], whereas penetration of cells by Pseudomonas exotoxin A is mediated through LDLR-related protein [20,21]. Thus, use of the LDLR by Salmonella during organ invasion could be envisaged, but was ruled out by the observation that hyperlipoproteinemic ApoE2/2 mice, which have an intact LDLR, also were resistant to Salmonella infection. In addition, LDL affected the adhesion of S. typhimurium to LDLRbearing endothelial cells. This demonstrates that elevated lipoprotein concentrations, and not the lack of LDLR itself, are responsible from the resistance of mice against Salmonella infection.
Theoretically, there are several mechanisms that could account for the beneficial effects of the lipoproteins on Salmonella infection. Firstly, a direct effect of lipoproteins on the growth of Salmonella could be envisaged. In this respect, it is of interest that HDL has been found to be cidal against Trypanosoma cruzi [20]. However, the growth of Salmonella was similar in the plasma of LDLR2/2, ApoE2/2 and control mice, excluding a direct antimicrobial effect of LDL. Secondly, lipoproteins may interact with Salmonella, putatively with its LPS component, and thus block bacterial binding and internalization by host cells. As LPS is crucial for the internalization of Salmonella [22], blocking the interaction between LPS and host cells may prevent subsequent tissue invasion. Indeed, preincubation of Salmonella with LDL led to reduced cytokine production, demonstrating that lipoproteins are able to inhibit the interaction of Salmonella with monocytes. Even more relevant for tissue invasion, preincubation of Salmonella with LDL significantly reduced its attachment to endothelial cells. This protective mechanism in which lipoproteins block Salmonella interaction with endothelial cells by their blockade of LPS represents the same type of mechanism as previously shown by the blockade of MSCRAMMs (microbial surface components recognizing adhesive matrix molecules) of Staphylococcus by naturally occurring antibodies, resulting in the inhibition of staphylococcal adhesion to endothelial cell and reduced tissue invasion [23].
Interaction of Salmonella with host cells likely is an important early step in the pathogenesis of invasive infection. The ability to infect tissue macrophages has been described as an invasive trait of intracellular bacteria such as Salmonella spp. [24], and attachment to endothelial cells is the first step in organ invasion by Salmonella from the bloodstream. The hypothesis that lipoproteins are able to directly modify organ invasion by Salmonella, was tested by assessing early distribution of Salmonella organisms in LDLR2/2 and LDLR+/+ mice after i.v. injection of a large bacterial load. Indeed, we observed that invasion of Salmonella organisms into the liver and spleen of LDLR2/2 mice was markedly lower than that in control mice. The difference in clearance of bacteria between the two strains of mice was already apparent within 30 minutes after injection. In this early phase of infection, the numerical balance between bacterial burden and host defense mechanisms in the tissues will determine the outcome of infection, and the LDLR2/2 mice start with a substantial advantage over control animals. In contrast to the LDLR+/+ mice, the vast majority of Salmonella organisms in the LDLR2/2 remain sequestrated in the circulation, where are eliminated by complement and neutrophils, known to efficiently clear Salmonella from the bloodstream [25]. Blocking tissue invasion by lipoproteins is not unique for Salmonella, as others have reported that VLDL inhibits liver invasion by Plasmodium sporozoites, leading to protection against malaria [26].
The precise molecular interaction between Salmonella and lipoproteins remains to be elucidated, but the bacterial LPS is the most likely candidate to be involved. Salmonella is known to interact with host cell Toll-like receptor (TLR)-4 through its LPS component [17,27], and this signaling mechanism is likely blocked by binding of lipoproteins to the LPS. The exact nature of the lipoprotein particle responsible for interaction with Salmonella has yet to be identified. Phospholipids have been shown to mediate LPS neutralization, but protein components, such as apolipoprotein E, have also been reported to bind LPS [28]. Our findings in the apoE-and LDLR-deficient mice, which are also protected against salmonellosis, however, point to a binding site other than apoE. In addition to TLR4, the macrophage scavenger receptors [29], and the cystic fibrosis transmembrane conductance regulator protein [30] are probably involved in the entry of Salmonella into cells. Whether the interaction of these receptors with Salmonella is also influenced by lipoproteins remains to be elucidated.
In conclusion, plasma lipoproteins appear to be an important host defense mechanism against invasive Salmonella infection. A direct and rapid interaction between lipoproteins and Salmonella in the bloodstream occurs, preventing the invasion of the microorganisms from the bloodstream into the organs. These new insights improve the understanding of the pathogenesis of Salmonella infection, and could ultimately lead to the design of new therapeutic strategies.