Penetration and Activation of Brain Endothelium by Salmonella enterica Serovar Typhimurium

Abstract

Salmonella meningitis is a serious disease of the central nervous system, common particularly in Africa. Here, we show that Salmonella enterica serovar Typhimurium is able to adhere, invade, and penetrate human brain microvascular endothelial cells (hBMECs), the single-cell layer constituting the blood–brain barrier (BBB). Cellular invasion was dependent on host actin cytoskeleton rearrangements, while expression of a functional type III secretion system was not essential. In addition, Salmonella infection activated a proinflammatory immune response targeting neutrophil signaling and recruitment. Salmonella invasion and immune activation may represent a crucial step in the penetration of the BBB and development of Salmonella meningitis.

Salmonella is a Gram-negative, facultative intracellular pathogen capable of producing a diverse array of human diseases ranging from mild food poisoning to life-threatening systemic infections, including bacteremia, meningitis, septic arthritis, and osteomyelitis. These infections are associated with a significant global burden in both developed and developing countries. In addition, there is an alarming increase in multidrug-resistant Salmonella isolates (up to 90% in some parts of Africa) [1, 2]. Nontyphoidal Salmonella spp are a leading cause of meningitis in Latin America and Africa, which is associated with a very high (up to 60%) mortality [1, 3]. Various serovars of S enterica have been reported to cause meningitis, including Paratyphi, Enteritidis, and Typhimurium [3]. Brain infection is most common in neonates and infants, although cases of Salmonella meningitis have been reported in adults with underlying immunodeficiencies [4]. Complications of Salmonella central nervous system (CNS) infection include bacterial colonization, brain abscesses, edema, cerebral infarction, pus collection in cerebral cavities, and brain inflammation [5].

Bacterial meningitis occurs when blood-borne pathogens interact with cerebral endothelial cells and cross the blood–brain barrier (BBB); subsequent bacterial replication within the CNS may provoke an overwhelming host inflammatory response. The BBB, responsible for maintaining biochemical homeostasis within the CNS, consists principally of a single layer of specialized brain microvascular endothelial cells (BMECs). Penetration of the BBB by a bacterial pathogen reflects a complex interplay between the host endothelium and microbial surface components. A recent study demonstrated that oral infection of mice with S enterica serovar Typhimurium resulted in meningitis and brain infection [6]; however, the mechanism(s) whereby the bacterium leaves the bloodstream and gains access to the CNS has not been examined.

We hypothesized that direct interaction with human BMECs (hBMECs) is a primary and essential step in the pathogenesis of Salmonella meningitis, whereupon a combination of bacterial transcytosis and inflammatory mechanisms combine to disrupt BBB integrity. Using our in-vitro BBB model, we demonstrate that S Typhimurium efficiently adheres to, invades, and persists within hBMECs. We also assess the acute response of the BBB to Salmonella infection using microarray, real-time reverse-transcription polymerase chain reaction (RT-PCR), and protein analysis. We identify that components of the Salmonella SPI-1 invasion locus are not required for induction of proinflammatory chemokines IL-8, CXCL1, and CXCL2.

METHODS

Bacterial Strains and Mutants

The wild-type (WT) bacterium used in this study was S. enterica serovar Typhimurium 14028S 1/9, and the isogenic mutant strains lacking the invA and sipB genes were invA::aphT clone1 [7] and sipB::aphT clone 1 [8], respectively. Salmonella strains were grown on Luria Broth (LB), supplemented with 50 μg/mL kanamycin for mutant strains at 37^°C. For infection assays, strains were grown without aeration in LB with 0.3 M sodium chloride at room temperature until they reached an OD₆₀₀ of .4 (∼2 × 10⁸ CFU/mL).

Cell Culture and Infection Assays

A well-characterized hBMEC line, immortalized by transfection by SV40 large T antigen, was generously provided by Kwang Sik Kim (Johns Hopkins University). Cell maintenance and assays for hBMEC adherence, invasion, intracellular survival, and bacterial transcytosis were performed as described previously [9–11]. In brief, hBMECs were grown to confluency and infected with Salmonella at indicated multiplicity of infection (MOI). To quantify intracellular bacteria, extracellular bacteria were killed by the addition of 100 μg/mL gentamicin. Following 2 hours, or at indicated time points, intracellular bacteria were liberated by the addition of 0.1 mL 0.25% trypsin/ethylenediaminetetraacetic acid (EDTA) solution followed by 0.4 mL of 0.025% Triton X-100 and quantified by plating serial dilutions on LB agar plates. Total cell-associated bacteria were quantified prior to addition of antibiotics. For inhibition studies, cytochalasin D (Sigma) was added to hBMEC monolayers at the indicated amount 1 h prior to the addition of bacteria, and remained during the experiment. Transmission electron microscopy was performed as described previously [11] following S Typhimurium infection for 1 h.

RNA Isolation, cDNA Preparation, Microarray Analysis, and qPCR

Monolayers of hBMEC were infected with S Typhimurium for 6 h. RNA extraction, complementary DNA (cDNA) transcription and microarray analysis (Sentrix Human-8 Expression BeadChips, Illumina) were performed as described previously [11]. Quantitative PCR (qPCR) conditions and primers for IL-6, IL-8, CXCL1, CXCL2, CCL20, and adrenomedullin (ADM) are described in [11], the following primer set was used to analyze ADM mRNA levels: 5′- CGTCGGAGTTTCGAAAGAAG -3′ and 5′- CCCTGGAGGTTGTTCATGCT -3′. Relative gene expression was normalized to glyceraldehyde 3-phosphate dehydrogenase transcription using the ΔΔ C_T method.

Chemokine Secretion in hBMEC Supernatants

Supernatants were collected after hBMEC infection with S Typhimurium, ΔinvA, and ΔsipB mutants after 6 h. Concentrations of IL-8, CXCL1 (R&D systems), and CXCL2 (BioSupply UK) were measured using enzyme-linked immunosorbent assays (ELISA) according to manufacturer’s instructions.

Statistical Analysis

GraphPad Prism version 4.03 was used for statistical analyses. Differences in bacterial colony forming units (CFU) were analyzed using the Wilcoxon signed rank test. Differences in percentage of adherence/invasion were analyzed using the Friedman test with Dunn multiple comparison correction. Differences in or cytokine production were tested using one-way ANOVA with Tukey multiple comparison correction. Significance was accepted at P < .05.

RESULTS

Invasion of hBMEC by Salmonella

To test the hypothesis that Salmonella directly invades brain endothelium, we optimized our previously established quantitative hBMEC adherence and invasion assays [9] for S Typhimurium. Confluent monolayers of hBMEC were infected with increasing concentrations (multiplicities of infection, MOI) of S Typhimurium (MOI 1 represents ∼1 × 10⁵ CFU). The percentage of adherent and intracellular bacteria ranged from 4.5%–6% for adherence and from 2%–3% for invasion compared with the original inoculum (Fig. 1A). At the standard inoculum used for subsequent comparative assays (MOI = 1), approximately 35% of total hBMEC-associated Salmonella had invaded the intracellular compartment within the 2 h incubation period, suggesting highly efficient endocytic uptake of surface-adherent organisms. Salmonella invasion required rearrangement of actin cytoskeleton, as we observed a dose-dependent decrease in Salmonella invasion in hBMECs in the presence of cytochalasin D, a potent actin microfilament aggregation inhibitor (Fig. 1B).

Figure 1.

S Typhimurium interaction with brain endothelium. (a) S Typhimurium adheres to (total cell-associated, TCA) and invades hBMECs. The number of recovered bacteria is expressed as a percentage of the initial inoculum. MOI of 0.01, 0.1, and 1 represents approximately 10³, 10⁴, and 10⁵ CFU, respectively. (b) Dose-dependent inhibition of S Typhimurium invasion of hBMECs by cytochalasin D. * P < .05. (c) Transmission electron microscopy of intracellular S Typhimurium 1 h after hBMEC infection (MOI = 1). Bacteria present in intracellular compartments are indicated by arrowheads. (d) S Typhimurium survival and persistence inside hBMECs over time (MOI = 1). (e) Transcytosis of S Typhimurium across confluent hBMEC monolayers seeded on transwells after 1 and 2 h (MOI = .1). (f) Invasion and total cell-associated (TCA; surface adherent plus intercellular bacteria) percentage of WT and SPI-1 mutants ΔinvA and ΔsipB (MOI = 1). Pooled data from three independent experiments are shown for Fig 1A, B, D, and E, where bars represent mean and error bars represent standard error of mean (SEM). Pooled data from two independent experiments are shown in Fig. 1F; bars represent mean and error bars represent standard deviation.

To confirm that Salmonella was indeed present inside hBMECs, we performed electron microscopy. Already after 1 h, Salmonella was present inside hBMECs in membrane-bound intracellular vacuoles (Fig. 1C). Similar vacuoles have not been observed in noninfected monolayers [11]. This observation prompted us to test whether Salmonella may persist or replicate intracellularly after invasion of hBMECs. Statistically similar amounts of intracellular organisms were recovered 2, 4, 8, and 24 h post addition of antibiotics, indicating that S Typhimurium persists within hBMECs following invasion (Fig. 1D). Finally, using a transwell assay [11], we observed that Salmonella was able to transcytose across a confluent hBMEC monolayer from the luminal to basolateral side (Fig. 1E).

Requirement of Salmonella SPI-1 Genes invA and sipB for hBMEC Invasion

The ability of Salmonella to invade epithelial cells requires a cluster of genes termed the Salmonella pathogenicity island 1 (SPI-1) locus. This locus encodes a type III secretion system (TTSS) that delivers bacterial effector proteins into the host cell cytoplasm, inducing actin cytoskeletal rearrangements that facilitate uptake of the organism. We hypothesized that the SPI-1 components invA, a needle complex export protein, and sipB, a translocation machinery component, would contribute to Salmonella hBMEC invasion. Using quantitative hBMEC invasion assays, invA and sipB deficient strains exhibited a 4-fold decrease in hBMEC invasion compared with the WT strain (Fig. 1F), while affinity for the hBMEC surface was similar (Fig. 1F). However, this difference was not statistically significant by nonparametric statistical analysis, suggesting that these components are not essential to the invasion process.

Response of Brain Endothelium to Salmonella Infection

As the response of the BBB to bacterial infection may impact bacterial CNS penetration and the progression of Salmonella meningitis, we further examined the initial transcriptional responses of hBMECs to WT Salmonella infection using microarray analysis. By 6 h post infection, ∼40 hBMEC genes exhibited a greater than 2-fold change in transcript abundance (data not shown), including those involved in stress response, signal transduction, and angiogenesis pathways, as well those regulating innate immunity. Expression of some of the most highly induced cytokines and peptides, including IL-8, CXCL-1 and CXCL-2, IL-6, CCL20, and ADM were confirmed in independent experiments by quantitative RT-PCR (Fig. 2A). Overall, the relative abundance of the different transcripts correlated with the fold increases observed by microarray analysis (data not shown). We also analyzed hBMEC supernatants for the presence of IL-8, CXCL-1, and CXCL-2 proteins following infection with the WT strain or ΔinvA and ΔsipB mutants. Chemokine secretion increased upon infection with WT S Typhimurium, compared with the uninfected control (Fig. 2B–D). Protein expression did not require SPI-1 effectors, as infection with the isogenic ΔinvA and ΔsipB mutants induced chemokine activation and subsequent protein secretion similar to Salmonella WT (Fig. 2B–D).

Figure 2.

Transcriptional response and protein secretion by hBMECs during S Typhimurium infection. (a) mRNA expression levels for IL-8, CXCL1, CXCL2, CCL20, IL-6, and adrenomedullin (ADM) in hBMECs 6 h post infection with S Typhimurium analyzed by quantitative reverse-transcription polymerase chain reaction (RT-PCR). Protein secretion of (b) IL-8, (c) CXCL1, and (d) CXCL2 in hBMEC supernatants 6 h after infection with S Typhimurium wild-type (WT) or the isogenic ΔinvA and ΔsipB mutants. This experiment was repeated 2 times in triplicate; data from a representative experiment are shown. Bars represent mean and error bars represent standard deviations of 3 wells. UI, uninfected hBMEC control. *** P < .001.

DISCUSSION

Salmonella is a common cause of meningitis in neonates and infants in developing countries such as Malawi, Brazil, Thailand, and Taiwan. Moreover, when it occurs, Salmonella meningitis is associated with high mortality and complications, including significant neurological sequelae and a high relapse rate in surviving patients [1]. The presence of Salmonella in the cerebrospinal fluid clearly indicates bloodstream dissemination to the CNS; however, the basic pathogenic mechanisms by which Salmonella species are able to gain access to the CNS and brain have not been studied. We have demonstrated for the first time the ability of Salmonella to invade the single-cell layer of specialized brain microvascular endothelial cells comprising the BBB. Salmonella penetration of brain endothelium requires rearrangement of the actin cytoskeleton, suggesting specific cellular mechanisms underlying the uptake process. We have routinely visualized pathogen–hBMEC interactions by electron microscope (EM) analysis [11], and intracellular Salmonella were similarly observed within membrane-bound vesicles as soon as 1 h after infection. Finally, we demonstrate that Salmonella persist intracellularly and traverse across hBMEC monolayers.

The SPI loci are major virulence determinants of S enterica. SPI-1 encodes a TTSS that injects effector proteins directly into the cytosol of host cells that modulate actin cytoskeleton dynamics, enabling bacterial uptake into intestinal epithelial cells. Our experiments suggest that bacteria were able to invade the BBB in an SPI-1–independent fashion, in contrast to uptake in epithelial cells [12]. This raises the possibility that an additional, as yet undiscovered mechanism mediates Salmonella uptake into the specialized BBB endothelium. Salmonella adherence to and intracellular persistence (data not shown) in hBMECs were independent of the SPI-1 locus. Intracellular survival mechanisms in hBMECs are likely encoded by SPI-2, as genes within SPI-2 promote Salmonella survival in other cell types [13]. However, further studies are required to identify the adhesion molecule(s) that mediates Salmonella association with hBMECs.

Salmonella interaction with the BBB induced an increased expression of chemokines IL-8, CXCL-1, and CXCL-2. The CXC subfamily of chemokines share a high affinity for receptors on neutrophils, and are known to act as strong neutrophil chemoattractants. In addition, another highly induced chemokine, CCL-20, is chemotactic for both lymphocytes and neutrophils. These results correlate with our previous data demonstrating that the BBB plays an active role in initiating neutrophil recruitment in response to bacterial infection [10]. This first line of CNS defense against bacterial infection may be particularly important during Salmonella infection as neutrophils are cytotoxic to Salmonella, in contrast to macrophages in which Salmonella are able to replicate [14]. Chemokine induction did not require an intact TTSS, as production of IL-8, CXCL-1 and CXCL-2 was similarly induced by the WT and ΔinvA and ΔsipB mutant strains. This is in contrast to studies in intestinal epithelial cells where SPI-1 encoded proteins, such as sipB, initiate substantial inflammatory responses in the intestinal epithelium [15].

Penetration of the BBB by a bacterial pathogen reflects a complex interplay between the host endothelium and microbial surface components. Our results build upon the recent development of a mouse model for Salmonella meningitis [6], and demonstrate for the first time that Salmonella penetrates the brain endothelial cells that constitute the BBB. We speculate that the combination of direct bacterial invasion and the induction of proinflammatory signaling pathways contribute to the severity of Salmonella meningitis. Ongoing studies on the elucidation of BBB penetration as well as the modulation of BBB gene expression and innate defense mechanisms by Salmonella will be critical for developing preventative therapies for CNS infection.

Funding

This work was supported by the National Institute of Neurological Disorders and Stroke/National Institutes of Health (grant AI062758 to S.H.B. and R01NS051247 to K.S.D).