GOING AGAINST THE GRAIN: CHEMOTAXIS AND INFECTION IN VIBRIO CHOLERAE

Abstract

Chemotaxis is the process by which motile cells move in a biased manner both towards favourable and away from unfavourable environments. The requirement of this process for infection has been examined in several bacterial pathogens, including Vibrio cholerae. The single polar flagellum of Vibrio species is powered by a sodium-motive force across the inner membrane, and can rotate to produce speeds of up to 60 cell-body lengths (~60µm) per second. Investigating the role of the chemotactic control of rapid flagellar motility during V. cholerae infection has revealed some unexpected and intriguing results.

The Gram-negative bacterium Vibrio cholerae, the causative agent of the severe diarrhoeal disease cholera, is responsible for the deaths of approximately 120,000 people annually¹. Cholera is contracted by ingestion of contaminated water or food and is therefore associated with inadequate sanitation and poverty. As a result, cholera is endemic mainly in the developing world, despite the fact that V. cholerae is present in temperate zones around the planet. Although there are at least 200 known serogroups of V. cholerae, cholera has generally been associated only with the O1 and O139 serogroups². The O1 serogroup is divided into two major serotypes, Inaba and Ogawa, and these can be further divided into two biotypes, classical and El Tor. Throughout history, cholera has been associated with explosive epidemics and, since the nineteenth century, pandemics.

Humans are the only known vertebrate host for V. cholerae and following ingestion, the organism must survive passage through the gastric barrier of the stomach. V. cholerae is not particularly resistant to low pH³, and this is believed to contribute to the relatively high infectious dose that is required to produce infection in healthy human volunteers⁴. However, V. cholerae can adapt to mildly acidic pH, and these acid-adapted bacteria have a huge competitive advantage during experimental infection when compared with unadapted V. cholerae⁵. Following passage through the stomach, the bacteria then enter the small intestine, which is the main site of infection. After reaching the small intestine, chemotaxis could conceivably be required to locate the appropriate intestinal niche for colonization and virulence-factor expression, an idea that will be explored in this review.

Two well characterized virulence factors of V. cholerae are cholera toxin and the toxin co-regulated pilus (TCP). The TCP is required for colonization in both humans and animal models of infection^6,7, and these pili are believed to mediate microcolony formation on the intestinal epithelium⁸. Cholera toxin is a ribosylating enterotoxin⁹ that is responsible for the profuse watery diarrhoea associated with cholera, known as rice-water stool. The genes that encode these virulence factors are tightly regulated, so that expression does not occur inappropriately, such as during extra-intestinal growth in rich media^10,11. By eliciting diarrhoea, V. cholerae cells are shed from the host and back into the environment in the rice-water stool. It is estimated that cholera patients can shed as many as ten trillion V. cholerae cells per day¹², and these shed V. cholerae are highly motile, as observed by microscopy¹³. There are two possible fates for V. cholerae cells that are shed from the host. First, the bacteria are ingested relatively soon after shedding by another human. This might occur within an explosive epidemic scenario, such as an outbreak in a refugee camp. Second, shed bacteria settle into the environmental-reservoir stage of the life cycle, and might or might not be ingested by another human host at a future point in time.

In addition to being an important cause of morbidity and mortality in the developing world, V. cholerae is also a natural inhabitant of freshwater, brackish and coastal-water habitats¹⁴. A model of the pathogenic and environmental aspects of the V. cholerae life cycle is shown in FIG. 1. In the environmental stage, V. cholerae can exist in a free-living, planktonic form or associate with several environmental hosts. V. cholerae can colonize zooplankton, such as copepods^15,16, phytoplankton, such as cyanobacteria^17,18, and the egg masses of Chironomid insects^19,20. Bacterial cells might enter a viable but non-culturable state²¹, although the requirements for this state are not well defined. V. cholerae also associates with abiotic and chitinous surfaces in the form of biofilms^22,23, and motility is required for this process²⁴. Recent studies have shown that V. cholerae that are present in biofilms have improved survival through the stomach and are probably more infectious than planktonic cells²⁵. Virtually all environmental (and clinical) V. cholerae isolates that have been described are motile, and although it has not been formally proven, it is reasonable to assume that chemotaxis is vital to the fitness of V. cholerae in aquatic environments.

Figure 1. The life cycle of pathogenic Vibrio cholerae.

The planktonic V. cholerae that are shed from humans in rice-water stool are central to the life cycle. These highly motile bacteria can be ingested by a new human host after shedding, or can associate with abiotic surfaces (possibly forming biofilms), copepods, algae and Chironomid egg sacs in the environment. These environmental V. cholerae can, presumably, disassociate from these hosts and be ingested by humans or form associations with a new environmental host. Alternatively, or in addition, environmental host-associated V. cholerae can be ingested by humans, causing infection and resulting in shedding of planktonic V. cholerae, therefore completing the life cycle. Chironomid egg mass panel reproduced with permission from REF. 126 © (2003) American Society for Microbiology.

Chemotaxis in V. cholerae and other bacteria

The single polar flagellum of V. cholerae is covered by an extension of the outer membrane to form a sheath²⁶. By using a sodium-motive force across the inner membrane to power the flagellar motor^27,28, rotation of 100,000 revolutions per minute can be achieved in liquid environments ²⁹, generating speeds of up to 60 cell-body lengths (~60 µm) per second³⁰. The rotation of flagella results in a propulsive force that imparts motility on the bacterial cell (FIG. 2). However, several modes of flagellum-independent bacterial motility have also been identified, including twitching motility (mediated by cycles of extension, adherence and retraction of type IV pili)^31,32 and gliding motility (for example, mediated by secretion of slime)³³. These forms of motility are important for moving over surfaces, and chemotaxis homologues have a role in regulating various aspects of these motilities, including direction of movement and gene expression³².

Figure 2. Flagellar-based motility.

There are many schemes for flagellation in bacteria, of which peritrichous flagella and a single polar (monotrichous) flagellum are two types. a | In the case of peritrichous flagella, such as those found in Escherichia coli, counter-clockwise (CCW) flagellar rotation results in the formation of a helical bundle that propels the cell forward in one direction in a smooth-swimming motion (a ‘run’). By contrast, the presence of clockwise (CW) rotation causes unbundling of the helical bundle, allowing the bacterium to randomly reorient its direction (a ‘tumble’). b | In the case of a single polar flagellum, CCW rotation propels the cell forward in a run, whereas CW rotation propels the cell backward with a concomitant random reorientation.

In the case of flagellar motility, chemotaxis is achieved by modulating the direction or speed of flagellar rotation in response to the surrounding environment (BOX 1). Owing to the work of many laboratories, a tremendous amount of structural and functional information is available on the signal-transduction cascade that results in the alteration of flagellar rotation in the peritrichously flagellated enteric bacteria Escherichia coli and Salmonella enterica serovar Typhimurium (S. typhimurium). This makes chemotaxis one of the most studied and well-understood signalling pathways in biology.

Box 1.

Chemotactic signalling in Escherichia coli

This box provides a brief, non-comprehensive overview of chemotaxis in Escherichia coli (see the figure, which shows the main signal transduction events that occur during chemotaxis in E. coli). For more detailed information, the reader is referred to REFS ^95–100. In the figure, all the Che proteins are shown free in the cytoplasm for the sake of clarity, although, with the exception of Che Y-P, all of these proteins are located at the receptor complex. In the table, the chemotaxis paralogues of E. coli and Vibrio cholerae are listed. The large number of paralogues suggests substantial complexity in chemotactic signalling by V. cholerae.

In E. coli, flagellar rotation alternates between the default direction of counter-clockwise (CCW) and clockwise (CW) rotation. In a chemically homogeneous environment, the cells change direction approximately once per second, and there is no bias for net movement in any particular direction¹⁰⁰. However, in the presence of a concentration gradient of chemoattractant or chemorepellent, this frequency is altered, enabling bacteria to swim up concentration gradients of attractants and down concentration gradients of repellents¹⁰¹. Net movement is achieved by lengthening the period of runs as a cell is experiencing an increasing concentration of attractant, and decreasing the period of runs when there is a decreasing concentration of attractant. The process by which bacteria control the frequency of switching between CCW and CW flagellar rotation in response to chemical gradients is called chemotactic signalling.

The first step in this process is signal reception by chemoreceptors, known as methyl-accepting chemotaxis proteins (MCPs)¹⁰². These are generally transmembrane proteins, although cytoplasmic MCPs have also been found, for example in Rhodobacter sphaeroides¹⁰³. An important feature of MCPs is their clustering at the cell poles^104,105, or in the cytoplasm in the case of cytoplasmic MCPs¹⁰³. MCPs bind specific ligands and ligand occupancy is communicated to the flagella through a signal-transduction cascade. Bacteria can respond to changes of only a few molecules of ligand^106,107 and clustering is thought to be important not only for this sensitivity but also for the signal amplification that is required to achieve efficient chemotaxis¹⁰⁸.

A decrease in attractant binding to MCPs is communicated to the auto-histidine kinase CheA through the protein CheW, resulting in auto-phosphorylation of CheA to CheA-P¹⁰⁹. CheA and CheW form a ternary complex with the MCPs¹¹⁰ at the cell poles^104,105. Two soluble response regulators, CheY and CheB, compete for phosphorylation by CheA-P¹¹¹. As CheY-P binds to the flagellar switch protein FliM¹¹² and causes CW rotation, the intracellular ratio of CheY to CheY-P controls the direction of flagellar rotation¹¹³. Tumbles are kept brief through rapid CheZ-stimulated dephosphorylation of CheY-P, resulting in turnover of this CW-promoting signal¹¹⁴.

The ability of the chemotaxis system to respond to further increases or decreases in attractant binding is crucial for chemotaxis along gradients. Such adaptation is achieved in E. coli by modulating the methylation state of the MCPs using two proteins, a constitutively active methyltransferase CheR¹¹⁵ and a methylesterase CheB¹¹⁶, the activity of which is stimulated after phosphorylation by CheA-P¹¹⁷. Increased methylation of the MCPs dampens the response to ligand binding, whereas decreased methylation sensitizes this response. This allows for adaptation because the decision-making reaction (CheA auto-phosphorylation) can be set to approximately the pre-stimulus level. Null mutations in any of the che genes result in a motile but non-chemotactic (swimming blind) phenotype: loss of cheA, cheY, cheW or cheR causes CCW-biased flagellar rotation, whereas loss of cheZ or cheB results in CW-biased rotation.

che gene	E. coli	V. cholerae
MCPs	4	43
cheA	1	3
cheW	1	3 (+1 putative)
cheY	1	4 (+ putative)
cheZ	1	1
cheR	1	3
cheB	1	3
cheV	0	3 (+1 putative)
cheC	0	1 putative
cheD	0	1 putative

The Gram-positive soil bacterium Bacillus subtilis has several additional chemotaxis components — CheC³⁴, CheD³⁵ and CheV³⁶ — that are also present in V. cholerae³⁷, but which are not reviewed here. However, although E. coli and B. subtilis have a single gene encoding each chemotaxis component, genome sequencing has revealed that this situation might be the exception, instead of the rule³⁸. Many bacteria possess multiple paralogues of each component, often organized into independent systems. V. cholerae is no exception and has an impressive number of chemotaxis paralogues (shown in the table in BOX 1), indicating either that chemotactic signalling in V. cholerae might be complex or that other systems in addition to chemotaxis are regulated by these signalling proteins³⁹.

With the exception of the mcp (methyl-accepting chemotaxis protein) genes, which are scattered throughout the genome, most V. cholerae chemotaxis genes are organized into three operons³⁷. Only one of these, operon 2, is important for chemotaxis in vitro. The cheA-2 (REF. ⁴⁰) and cheY-3 (REF. ⁴¹) genes in operon 2 are required for chemotaxis, whereas cheA-1 and cheA-3, from operons 1 and 3 respectively, are dispensable for chemotaxis⁴⁰. Furthermore, there does not seem to be redundancy between operons 1 and 3 with respect to chemotaxis, as the simultaneous deletion of all the cheY paralogues except the one in operon 2 has no effect on chemotaxis in vitro (S.M.B. and A.C., unpublished data). These data are consistent with the absence of strong homology to the FliM-binding region in these CheY paralogues (I. Kawagishi, personal communication).

As mentioned, multiple chemotaxis systems within the same bacterium are common, and in some cases the function of these additional systems is known. The soil bacterium Myxococcus xanthus, which initiates a developmental programme under starvation conditions that culminates in the formation of a fruiting body, has four chemotaxis systems. This organism has S-motility, which is a form of gliding motility that requires two cell-surface components: type IV pili and extracellular matrix fibrils. One system of chemotaxis homologues, Dif, is involved in fibril biogenesis⁴², and the Frz and Che4 chemotaxis systems control the frequency of reversal of gliding direction^43,44. The Che3 cluster regulates entry into the spore-formation developmental programme by controlling developmental gene expression⁴⁵. The opportunistic pathogen Pseudomonas aeruginosa uses one set of chemotaxis genes to control flagellar motility and another to control twitching motility, which is mediated by type IV pili⁴⁶. As V. cholerae chemotaxis operons 1 and 3 are not required for chemotactic control of flagellar motility, it is possible that genes in these operons could regulate flagellum-independent motility in this organism. One such mode of motility has been observed; however, the requirements for this process remain to be determined⁴⁷.

The role of chemotaxis in virulence

Although the role of motility during infection has been examined in several bacterial pathogens, the importance of chemotaxis in this process has not been as extensively studied⁴⁸. A priori, chemotaxis would be predicted to work hand-in-hand with motility to enable bacteria to swim towards preferred colonization sites. However, in the case of enteric pathogens, the requirement of motility and chemotaxis ranges from being crucial to being dispensable for infection (TABLE 1). For example, Shigella species are non-motile but are highly infectious, with an infectious dose as low as 10 cells. Clearly, the absence of motility in this organism is not an impediment to infection. Among motile bacteria, it seems that invasive enteric pathogens might not require motility for infection⁴⁹. Some Shigella, Listeria, Yersinia and Salmonella species fall into this category, and each of these cross the epithelial barrier by translocation through M cells (reviewed in REF. 50), which are specialized antigen-sampling cells that lack an overlying mucus layer⁵¹. The ability to invade the epithelium in this manner apparently abrogates the need for motility. By contrast, both flagella and flagellar motility are necessary for Yersinia enterocolitica to bind and invade enterocytes in tissue culture⁵².

Table 1.

Outcome of infection for non-chemotactic mutants

Pathogen/animal model	Outcome of infection	Ref.
Campylobacter jejuni
Mouse intestine colonization	cheY mutant absent from stool after 6 days of infection	118
Ferret disease	cheY mutant does not cause diarrhea	118
Chick colonization	cheY mutant present in caecum at levels 10,000-fold less than the wild-type strain	119
Helicobacter pylori
Gnotobiotic-pig stomach Colonization	cheY1 mutant fails to colonize	120
Mouse stomach colonization	cheY1 and cheAY2 mutants fail to colonize	120
Mouse stomach colonization	tlpA and tlpC (putative chemoreceptors) mutants attenuated 50-fold in competition experiments	121
Infant-mouse stomach colonization	cheY mutant attenuated 5-fold in competition experiment	122
Listeria monocytogenes
Mouse systemic infection (oral inoculation)	ΔcheYA mutant showed no defect for recovery from spleen but showed decreased recovery from liver	123
Proteus mirabilis
Mouse ascending urinary-tract infection	cheW mutant is attenuated 106-fold for infection of the urinary tract and bladder	124
Salmonella enterica serovar Typhimurium
Murine ligated-loop tissue invasion	cheB (CW-biased) mutant attenuated 10-fold for invasion of Peyer’s patches	62
Mouse colitis and systemic infection	cheY mutant attenuated 100-fold in caecum-colonization competition experiment; no attenuation observed in spleen or liver	125
Vibrio anguillarum
Rainbow-trout tissue invasion (immersion inoculation)	cheR mutant has a 400-fold higher LD50 than wild-type; no difference between strains when inoculated intraperitoneally	59
Vibrio cholerae
Infant-mouse small intestine	cheY-3 mutant out-competes wild-type strain 70-fold; CW-biased non-chemotactic mutant attenuated 10-fold in competition experiment	55

CW, clockwise.

Alternatively, some non-invasive pathogens, such as Helicobacter pylori and Campylobacter jejuni, require chemotaxis for infection: neither species are thought to attach to the epithelium, but instead use chemotaxis to stay within the mucus layer that lines the stomach⁵³ and caecal crypts⁵⁴, respectively. Like H. pylori and C. jejuni, V. cholerae does not invade the epithelium; however, unlike these species, V. cholerae does attach to epithelial surfaces, so that it is not clear whether motility would be expected to be important for infection.

Chemotaxis and V. cholerae infection

Surprisingly, chemotaxis actually inhibits the ability of V. cholerae to colonize the small intestine of infant mice. Defined non-chemotactic mutants of El Tor biotype strains that lack cheY-3 or cheA-2 were shown to out-compete the wild-type strain 70-fold in vivo⁴¹ (TABLE 1). This out-competition phenotype correlates with an order-of-magnitude increase in infectivity, as defined by a 10-fold reduction in the number of bacteria that are required to cause an infection⁵⁵. No competitive advantage of these mutants is observed during growth in vitro, showing that the advantage is specific to the host small intestine. This unusual phenotype is explored further below.

Do the other chemotaxis systems present in V. cholerae have a role during infection? The answer to this question seems to be no. Strains with single or combined mutations in each of the cheA and cheY paralogues outside of operon 2 retain full virulence in competition assays with the wild-type strain in the infant-mouse model (S.M.B. and A.C., unpublished data). Interestingly, cheA-1 and cheR-1 from operon 1 were identified as being highly expressed during infection in humans⁵⁶. However, as these genes are not required for infection in mice, the significance of these data are unclear. At present, it is unclear what role, if any, these additional V. cholerae chemotaxis systems have in virulence, or for that matter, for life in the environment. The expression of two genes that encode proteins with homology to MCPs, acfB (accessory colonization factor B) and tcpI, is co-regulated with virulence genes^57,58. However the roles of these proteins during infection is unknown.

The strong competitive advantage that is observed in the absence of chemotaxis in V. cholerae is unusual and interesting. Even within the Vibrio genus, non-chemotactic mutants of polarly flagellated Vibrio anguillarum, a pathogen of marine and estuarine fish, were attenuated 400-fold for infection⁵⁹. The squid symbiont Vibrio fischeri requires motility for colonization of its host Euprymna scolopes⁶⁰, and chemotaxis has been predicted to be important in establishing the symbiosis, although this hypothesis still needs to be tested⁶¹. Given that several non-invasive enteric pathogens require chemotaxis for infection, why is chemotaxis not only dispensable, but an impediment, to experimental infection by V. cholerae? Part of the answer to this question is that non-chemotactic V. cholerae mutants show aberrant distribution within the infant-mouse small intestine.

Intestine colonization by V. cholerae

Whereas wild-type V. cholerae predominantly colonize the lower half of the small intestine, corresponding approximately to the lower jejunum and ileum, non-chemotactic mutants colonize equally well throughout the small intestine⁴¹. This relaxation of tissue specificity allows a greater surface area to be colonized.

Why do non-chemotactic V. cholerae mutants colonize the upper small intestine? A breakthrough in answering this question came from the examination of different types of non-chemotactic mutants in vivo. Both counter-clockwise (CCW)-biased and clockwise (CW)-biased flagellar mutants are non-chemotactic. However, the out-competition phenotype is observed only in the presence of CCW-biased flagellar rotation: a CW-biased mutant has the opposite phenotype, being attenuated 10-fold for infection⁵⁵. As the CW-biased mutant changes the direction in which it is swimming extremely frequently, producing a zig-zag pattern of movement, it is defective in its ability to make net progress in any direction. By contrast, the CCW-biased mutant swims in straight runs for relatively long periods of time. Although the direction in which cells swim is random (with respect to chemical gradients), they nonetheless cover a reasonable distance (~40 µm s⁻¹)⁵⁵. As the diameter of the lumen in the infant-mouse small intestine is approximately 400 µm, it is easy to see how the CCW-biased mutant can make contact with the mucus layer and villi, and how the CW-biased mutant would be impaired in this process. It is unknown whether the same outcome would be observed during infection in humans.

Interestingly, CW-biased mutants of S. typhimurium, as well as flagellated but non-motile mutants, show decreased invasion of Peyer’s patches, which are intestinal lymphoid aggregates that contain M cells. Non-flagellated mutants, however, can invade Peyer’s patches at wild-type levels. The attenuation of both the CW-biased mutants and the flagellated but non-motile mutants is not caused by the lack of chemotaxis or motility but by the resulting unbundled peritrichous flagella which prevent interaction of adherence factors on the bacterial cell surface with host cells⁶².

The expanded tissue range of CCW-biased V. cholerae might only account for part of the 70-fold competitive advantage that is seen in vivo. Previously, it was observed that wild-type V. cholerae use chemotaxis to penetrate the mucus layer and move to the intestinal crypts, which are located at the base of the villi (FIG. 3). Non-chemotactic mutants do not accumulate in this site, and instead accumulate in the mucus and on the luminal side of the villi. This was observed in sections dissected from rabbit ileum^63,64, as well as in the infant-mouse intestine⁶⁵. Based on these observations, and to explain the out-competition phenotype of non-chemotactic V. cholerae, Freter et al. proposed that wild-type V. cholerae (but not non-chemotactic strains) move by chemotaxis to the intestinal crypts, and that antimicrobial substances that are present in this location kill a large fraction of the bacteria. Since this initial observation, antimicrobial peptides named defensins (cryptdins) have been identified. They are released from Paneth cells, which are located at the base of the intestinal crypts⁶⁶. However, as defensins are not expressed in mice until 20 days after birth⁶⁷, the identity of the proposed antimicrobial substances in the infant-mouse small-intestine crypts remains unclear.

Figure 3. Architecture of the small intestine.

Vibrio cholerae colonizes the small intestine in infant and adult humans and in experimentally infected infant mammals. Although several animal models of V. cholerae infection have been developed, the most widely used is colonization of the infant-mouse small intestine following oral inoculation⁹². Important factors for human colonization have been identified using this model, lending credence to its use as a viable model of infection. Furthermore, virulence genes that are crucial for infection have been shown to be expressed in infant mice⁹³. A schematic of the small intestine is shown. a | In the scanning electron micrograph (SEM) of an opened section of infant-mouse small intestine, the numerous and closely packed villi are seen. The villi surround a central lumen (not shown), and intervillous spaces are present between villi. At the base of the villi are the crypts of Lieberkühn. b,c | The villi are composed of absorptive epithelial cells (enterocytes) as well as Goblet cells, which produce mucus. This mucus forms a gel covering the villi, which concentrates at the tips. As V. cholerae is known to colonize epithelial surfaces on the villi and crypts, it must therefore penetrate not only the mucus layer present at the tips of the villi, but also any mucus gel covering the intestinal epithelium.

There are several possible explanations as to why V. cholerae might ‘recklessly’ follow a chemotaxis gradient to the crypt epithelia. First, the signals for maximal expression of cholera toxin might be present at this site, as proposed by Lee et al.⁴¹ Second, V. cholerae that penetrate into the intervillous spaces and colonize epithelia might be better protected from the forces of peristalsis. Third, this migration might be crucial for infection in humans, and V. cholerae uses the same strategy during experimental infection of mice and rabbits, which have more potent antimicrobial mechanisms. Arguing against the first explanation is the finding that non-chemotactic mutants that colonize the upper small intestine are competent for transcriptional induction of TCP and cholera-toxin genes⁵⁵. Moreover, no differences were observed in TCP production itself (S.M.B. and A.C., unpublished results). Although these experiments are not conclusive, they might indicate that the signals for virulence gene expression are not confined to the crypt epithelium. Arguing against the second explanation is the fact that non-chemotactic mutants, which are predicted to fail to enter the intervillous spaces, can out-compete the wild-type strain. Future experiments that visualize the locations and gene-expression patterns of wild-type and non-chemotactic V. cholerae in the infant-mouse small intestine might resolve these issues.

The ability of non-chemotactic V. cholerae mutants to colonize the upper small intestine of infant mice indicates that chemotaxis is used by the wild-type bacteria to avoid colonizing this tissue. As the infant mouse small intestine is ~13 cm in length and peristalsis creates a downward flow of luminal contents, V. cholerae cannot respond to distal small-intestine-specific attractants when transiting the upper small intestine. Therefore, there must be an attractant or repellent gradient that is confined to the upper small intestine and directs V. cholerae to the lumen (FIG. 4). This gradient could consist of a repellent that is at a higher concentration near the intestinal wall, or an attractant that is at a higher concentration in the lumen. Presumably, by the time V. cholerae reaches the distal small intestine, this gradient has dissipated and the bacteria can now respond to another gradient that draws them into the intervillous spaces. We have observed that V. cholerae are chemotactically attracted to the epithelial surface of the upper small intestine in primary tissue culture assays, suggesting a lack of chemorepellents emanating from the intestinal surface⁵⁵. These data point to the presence of a chemoattractant gradient in the lumen of the upper small intestine (arrows in FIG. 4).

Figure 4. Model for the effect of chemotaxis in limiting Vibrio cholerae colonization.

A gradient of chemoattractant is present in the lumen of the upper small intestine, oriented away from the villi (shown by arrows, panels a,c). Using chemotaxis, wild-type V. cholerae responds to this gradient by concentrating within the lumen. However, this chemoattractant gradient is absent from the lumen of the lower small intestine (panels b,d), allowing wild-type V. cholerae to respond to a different chemoattractant gradient that directs them into the intervillous spaces and onto epithelial surfaces (see scanning electron micrograph (SEM) panels). By contrast, non-chemotactic counter-clockwise (CCW)-biased V. cholerae are blind to these gradients and therefore colonize both the upper and lower small intestine, presumably on the luminal side of the mucus gel and villi (panels c,d). The V. cholerae that get stuck in mucus can also multiply, as evidenced by the presence of dividing bacteria in the SEM of mucus. SEM lower panel reproduced with permission from REF. 94 © (2002) American Association for the Advancement of Science.

One potential source of attractants in the lumen of the upper small intestine are sugars, peptides and amino acids. As food is digested and moved by peristalsis, absorption occurs at the epithelial surface, potentially generating a gradient. A second possible source of chemoattractant is bile. After consumption of food, bile is released into the lumen of the upper small intestine from the bile duct. Bile acids are vital for digestion and absorption of fats and fat-soluble vitamins, and are absorbed by the small-intestine epithelium. Although bile is a repellent for H. pylori⁶⁸, it is an attractant for V. cholerae at physiological concentrations⁶⁹. Also, although bile represses flagellar synthesis in S. typhimurium⁷⁰, it increases the motility of V. cholerae and represses virulence gene expression^71,72. In this scenario, chemotactic V. cholerae that are retained in the lumen because of a bile gradient would be predicted to repress virulence gene expression until the local concentration of bile decreases.

Motility and V. cholerae virulence

When considering bacterial pathogens, it is reasonable to assume that the absence of flagella in a flagellated organism would reduce the ability of that organism to initiate an infection^48,73. As mentioned previously, invasive pathogens are known exceptions to this intuitive rule. To determine the role of motility during infection, the requirement of motility per se must be separated from the ability of a bacterium to use flagella as an adherence factor, as occurs in enteropathogenic E. coli infection⁷⁴. This can be achieved by comparing the ability of a non-flagellated (fla⁻) and a flagellated but non-motile (fla⁺ mot⁻) mutant to colonize. The latter category of mutants is conveniently isolated by disrupting genes that encode the motor proteins that are necessary for flagellar rotation. As the V. cholerae flagellum is sheathed, the flagellar subunits themselves are unlikely to function as adhesins, and consistent with this, no differences with respect to infection have been observed for fla⁻ versus fla⁺ mot⁻ mutants⁴¹.

Is motility per se important for V. cholerae infection? Based on early studies, it was proposed that the ability of V. cholerae to swim into the mucus layer might be an important factor in colonization of the intestinal surface^75,76. However, recent studies using the infant-mouse model of infection have produced conflicting data regarding the importance of motility for infection by the classical biotype^77–79. In the case of the El Tor biotype, the situation is more clear; both fla⁻ and fla⁺ mot⁻ mutants are attenuated 10-fold compared with the wild-type strain⁴¹. Therefore, for the current pandemic strains, which are of the El Tor biotype, motility is important for colonizing the wall of the small intestine.

Perhaps the most convincing argument for a bona fide role of motility in infection comes from live-attenuated V. cholerae vaccine trials in human volunteers. Although live-attenuated vaccines are more immunogenic than whole-cell killed vaccines, they are associated with the production of moderate side effects, including nausea, cramps and diarrhoea; this undesired property is referred to as reactogenicity⁸⁰. Mekalanos and colleagues found that non-motile derivatives of live-attenuated vaccine strains are less reactogenic, without compromising immunogenicity^81,82. These non-motile derivatives are more attenuated than the parent live-attenuated strain in infant mice, but nevertheless seem to be able to colonize the small intestine of human volunteers^80,81. The reduced reactogenicity is believed to be due to the inability of the non-motile derivatives to come into close contact with the intestinal epithelium⁸³, consistent with the hypothesized role of motility in penetrating the mucus layer in animal models.

One complicating factor in measuring the role of motility in V. cholerae virulence is the link between motility and virulence gene expression that is observed in many pathogens. These properties are reciprocally regulated in several organisms, so that motility is repressed upon initiation of virulence gene expression. This occurs in S. typhimurium^84,85 and Bordetella bronchiseptica⁸⁶ and was suggested to occur in V. cholerae⁷⁸. Häse and colleagues showed that increasing the viscosity of the medium, as would occur in mucus, as well as adding inhibitors of flagellar motility resulted in increased expression of the major virulence-gene transcriptional activator ToxT in the classical biotype²⁸. It was therefore proposed that the V. cholerae flagellum might respond to increased viscosity to induce virulence gene expression. This type of response has precedence in the Vibrio genus; the polar flagellum of Vibrio parahaemolyticus serves as a mechanosensor of increased viscosity^87,88. In addition, a recent study by Klose and colleagues indicates that the V. cholerae flagellum might have a mechanosensor function that operates during biofilm formation⁸⁹. However, the effects of increased viscosity on virulence gene induction in V. cholerae were later shown to be independent of the flagellum and instead are probably due to changes in the membrane potential⁹⁰. In another study, differences in virulence gene expression in the El Tor biotype were noted during infection in the presence of ΔflaA and ΔmotAB mutations⁴¹. However, no differences were observed using a more quantitative technique (S.M.B. and A.C., unpublished results), and therefore motility does not seem to be linked to virulence gene expression in this biotype.

Additional roles for motility and chemotaxis

In addition to its role in the penetration of the mucus layer and for chemotaxis into the intervillous spaces, motility might serve other functions during the infectious process. It was observed that V. cholerae shed in rice-water stools are highly motile¹³. Because V. cholerae are unlikely to use flagellar motility within microcolonies on the intestinal epithelium, motility must be switched back on prior to exit from the host. Although this might simply be in preparation for life outside the host, it might also facilitate movement from the epithelium into the lumen and therefore in subsequent shedding from the host. It is not known whether chemotaxis is involved in this process, although it is hypothesized below that rice-water-stool V. cholerae are phenotypically non-chemotactic. If this hypothesis is correct, then perhaps chemotaxis is repressed to allow for more efficient accumulation of motile V. cholerae in the lumen for subsequent expulsion in stool. In this speculative scenario, bacteria would be blind to any potential chemoattractant gradient that might otherwise draw them back into the intervillous spaces. Looking beyond the human host, it is probable that the motile state of shed V. cholerae promotes survival in aqueous environments. Motility coupled with chemotaxis would be predicted to have a crucial role in dissemination of shed V. cholerae in the environment and in the location of suitable environmental hosts and surfaces (FIG. 1).

The heightened infectivity of non-chemotactic V. cholerae mutants in experimental infection lends support to a recently proposed mode of natural transmission of cholera. A competitive advantage similar in magnitude is observed with rice-water-stool V. cholerae, which out-compete an in vitro-grown wild-type strain 10–100 fold in infant mice¹³. This competitive advantage is retained after 5 hours of incubation in pond water but is completely lost following overnight growth in vitro of the stool V. cholerae, showing that the phenotype is transient. Despite the fact that rice-water-stool V. cholerae are motile, analysis of their transcriptome reveals repression of several of the chemotaxis paralogues¹³, including cheW-1 and cheR-2, that are required for chemotaxis (S.M.B. and A.C., unpublished data). In addition, 18 of the 43 MCPs that are encoded in the genome were also repressed. If any of these MCPs are receptors for chemo-attractants in the lumen, a chemotaxis defect would also be observed. These data indicate that the stool V. cholerae might be in a transiently non-chemotactic CCW-biased state, and if so, this might in part account for the competitive advantage observed in vivo with the stool bacteria. A second microarray study using a different internal reference did not, however, detect this repression of chemotaxis genes⁹¹. Given the difference with respect to internal standard, as well other methodological differences between these microarray experiments, the two studies cannot be directly compared. It is quite possible that other physiological attributes of the stool V. cholerae contribute to the competitive advantage, and therefore it is important to determine whether stool V. cholerae are non-chemotactic at the phenotypic level to validate the above hypothesis. Whether the competitive advantage of stool V. cholerae in infant mice translates to an increase in infectivity in humans has not been tested.

Conclusions

In terms of pathogenesis, chemotaxis is often thought to be required for efficient colonization. This is true of several non-invasive enteric pathogens, but chemotaxis is either absent, or present but dispensable, in several invasive pathogens. The increased infectivity of non-chemotactic V. cholerae might in fact contribute to an important stage of the V. cholerae pathogenic life cycle. For example, if rice-water-stool V. cholerae truly exist in a transient state of non-chemotaxis, then V. cholerae might have evolved to take advantage of this to improve its chances of infecting new human hosts. It is not yet known whether other pathogens modulate chemotaxis to optimize particular stages of infection. In the case of V. cholerae, it is unlikely that such a motile but non-chemotactic state would persist in aqueous environments for an extended period, although even if maintained for only a few hours, such a state might confer a fitness increase by allowing it to take better advantage of an important growth medium and vehicle of dissemination — us.

Glossary

BIOFILM

Microbial biofilms are populations of microorganisms that are concentrated at an interface (usually solid–liquid) and typically surrounded by an extracellular polymeric substance matrix. Aggregates of cells that are not attached to a surface are sometimes termed ‘flocs’ and have many of the characteristics of biofilms.

PLANKTONIC CELLS

Single cells in suspension, instead of in a biofilm.