To reach their site of colonization and/or multiplication, some bacteria must cross the physiological barriers. To cross a barrier, a pathogen must interact with it, resist the host defences, and penetrate across – or in between – the cells constituting the barrier. The most frequently encountered barrier is probably the gastro-intestinal barrier, whose colonization is the first step in the pathogenesis of all enteric and of some systemic infections.
We will not discuss what STM has brought to the understanding of the colonization of the gastro-intestinal tissue (see Section 5.3). West and colleagues published very recently a review on this topic . Instead, we will focus here on the two STM screens that have been undertaken to try to decipher the mechanisms underlying the crossing of the blood–brain barrier (BBB).
The BBB constitutes a particularly impermeable barrier for bacteria. It is thus interesting to understand why only certain bacteria, including E. coli, Neisseria meningitidis, M. tuberculosis, Streptococcus agalactiae, S. pneumoniae and Listeria monocytogenes, are able to cross this barrier and for some of them to be able to invade the brain tissue itself (see for a review ). In most cases, invasion of the meninges is a complication of bloodstream invasion, that occurs after primary colonization of the nasopharyngeal tissue in the case of N. meningitidis or the gastrointestinal tissue, in the case of L. monocytogenes (see for reviews [41,42]). Two different structures separate the blood stream from the central nervous system (CNS): the BBB, and the blood–cerebrospinal fluid barrier that is present at the choroid plexuses. The endothelial cells lining the brain capillaries form the BBB. These microvessels are covered by pericytes and outgrowths of astrocytes. Unlike other capillary cells, the endothelial cells of brain capillaries are joined to one another by tight junctions with high electrical resistance (>1000 Ω cm2).
6.1Gram-positive determinants implicated in brain invasion
Listeria monocytogenes is a Gram-positive bacterium widespread in nature and responsible for sporadic severe infections in humans and other animal species (see , and references therein). This organism is a facultative intracellular parasite capable of invading most host cells, including epithelial cells, hepatocytes, fibroblasts, endothelial cells, and macrophages (see , for a review). Each step of the intracellular parasitism by L. monocytogenes is dependent upon the production of virulence factors . The major virulence genes identified to date (hly, plcA, plcB, mpl, actA, inlA, and inlB) are clustered into two distinct loci on the chromosome and are controlled by a single pleiotropic regulatory activator, PrfA, which is required for virulence . However, it is most likely that many other genes, involved in various stages of the infection in vivo remain to be discovered.
Listeria monocytogenes has tropism for the brain as well as for the meninges, and clinical syndromes due to L. monocytogenes in CNS are meningitis, meningoencephalitis and abscess formation. L. monocytogenes can use several different mechanisms to invade the CNS. Infection from the blood stream occurs either by direct invasion of endothelial cells by free blood-born bacteria or by bacteria circulating in infected leukocytes. In addition a neural route of infection has been described in which bacteria reach the CNS via intra-axonal transport from peripheral tissues (see for a review ).
In order to identify genes possibly involved in passage across the blood–brain barrier, we used STM to identify mutants affected in their multiplication in the brains of infected animals . This study represents the only STM screen performed to date in a Gram-positive pathogen that used the brain as a target organ. We tested a bank of 2000 mutants for their capacity to penetrate and/or to multiply in the brain of infected mice . Forty-one pools of 48 mutants were generated, and each pool was injected into mice at a dose of 106 bacteria per animal to ensure a starting inoculum of approximately 2 × 104 bacteria per mutant in each pool. Sixty putative candidates were identified by dot-blot analysis. However, when tested individually, only eighteen of them appeared to be really attenuated. Determination of the transposon insertion site in these 18 attenuated mutants led to the identification of 10 distinct loci: five loci corresponding to putative cell wall components, and five loci involved proteins participating in various cellular processes. The four mutants that showed the highest LD50 values (>106; i.e., 1.5 log higher than the wild-type strain), all corresponding to genes encoding cell wall components (YtgP, GtcA, PbpX, and Lmo2026) were studied in greater detail. Four identical independent transposon insertions were found in ytgP. This orf encodes a putative integral membrane protein possibly involved in polysaccharide biosynthesis. Of interest, immediately downstream of the inactivated gene, a second orf encodes a homologous protein of identical size. In spite of repeated efforts, we were unable to inactivate this second orf, suggesting that the double mutant might not be viable (unpublished data). The gtcA gene is specifically involved in the glycosylation of teichoic acid (TA) domains. Strikingly, five independent transposon insertions were found in this gene (three of which contained an identical insertion). The fact that integrity of this major cell wall component is required for bacterial virulence suggests that TA might influence the capacity of the bacteria to invade and multiply in the brain, either directly or indirectly by influencing other cell-wall-associated factors involved in virulence. One Tn insertion occurred immediately upstream of a gene encoding a putative penicillin-binding protein sharing 47% identity with PbpX of B. subtilis. PbpX of L. monocytogenes might be involved in cell wall synthesis. The fourth Tn insertion occurred immediately upstream of an orf encoding a 626-residue protein (Lmo2026) of unknown function. The ORF, whose predicted product does not have significant similarity with other proteins in the databases, contains five leucine-rich repeats and a membrane-anchor LPXTG motif in its C-terminal part. Such motifs are generally found in proteins of the internalin family . We are currently testing the possible role of ORF626 in the binding and entry of L. monocytogenes into eucaryotic cells.
The in vivo kinetics of bacterial multiplication of these four mutants (monitored in the spleen and the liver) was similar to that of the wild-type strain until day 3. Then, bacterial counts fell from day 4 and all of the mice fully recovered from the infection. Thus, in the four mutants, attenuation did not result in a defect in the early stage of multiplication in the spleen or in the liver. Moreover, in no case was invasion of the brain completely abolished. These results suggest a defect of persistence in the brain rather than a defect in crossing of the BBB. Direct invasion of the vascular endothelium by bacteria is not the only means of cellular infection. Indeed, experimental data had already demonstrated dissemination of intracellular L. monocytogenes, indicating that phagocyte-facilitated invasion has a role in central nervous system infection . We found no difference in vitro in the intracellular multiplication of the four mutants in mouse macrophages.
The fact that insertions in gtcA and ytgp were repeatedly found suggested the existence of possible hot spots of insertion of the transposon used for mutagenesis. A second STM screen further confirmed this assumption that we performed more recently with the same transposon . In that second screen, we chose the liver as a target organ to ensure a higher number of recovered bacteria after infection (the number of bacteria per mouse can be up to ca. 1000-fold higher in the liver than in the brain ). Forty different loci were identified (unpublished data). These included 7 out of the 10 loci that had been identified in our previous screen. Strikingly, as in the first screen, insertions into the genes ytgP and gtcA were the most frequent, with 11 mutants at each locus. This result demonstrated that these two genes, initially involved in survival in the brain, were more generally involved in the overall in vivo proliferation of the bacterium.
6.2Use of cellular models to elucidate the crossing of BBB by Gram-negative bacteria
In addition to the animal model, cellular models, primarily endothelial cells cultures, were used in STM studies. E. coli K1 is the principal cause of meningitis for newborns. 3360 mutants of E. coli K1 were analyzed in an infection model of HBMEC cells (human brain microvascular endothelial cell) . Very few mutants (less than 0.2%) were identified as attenuated in virulence. Interestingly, a mutant disrupted in the traJ gene was found. The TraJ protein is normally implicated in positive gene regulation of genes necessary for conjugation. This does not seem to be related to pathogenesis. Nevertheless, the traJ locus was shown to influence the ability of E. coli K1 to invade HBMEC in vitro and to cross the BBB in vivo . In a subsequent paper, the authors began to elucidate the mechanism by which TraJ participates to virulence. They demonstrated that TraJ contributes to the early systemic dissemination of E. coli K1 in the oral infection process via specific TraJ-dependent bacterial interactions with macrophages . Experiments are in progress to understand this phenomenon. It is worthy to notice that this gene was also identified in an IVET screen conducted by the same team .
Many genes were discovered that participate in interactions with endothelial cells, or are required for persistence in the brain. However, no specific gene implicated directly in the crossing of BBB has ever been found, giving clues for a better understanding of this phenomenon. The currently acknowledged hypothesis is that crossing of the BBB is the result of a concerted and synergistic action of many genes: abolition of crossing would probably need inactivation of several of them.