Journal of Medical Microbiology (2004), 53, 585–593
Review
DOI 10.1099/jmm.0.05491-0
Switches, cross-talk and memory in Escherichia coli adherence Nicola J. Holden and David L. Gally
Correspondence David L. Gally
Zoonotic and Animal Pathogens Research Laboratory, Medical Microbiology, Teviot Place, University of Edinburgh, Edinburgh, EH8 9AG, UK
[email protected]
Escherichia coli is a successful commensal and pathogen. Its pathogenic diversity stems from the acquisition and expression of multiple virulence-associated loci. Many of the key virulence factors are surface structures involved in adherence and motility. These are important antigens and their expression is limited by phase-variable genetic switches that are considered to act randomly. This review considers the possibility that such stochastic expression within a bacterial population belies sequential or co-ordinate control at the level of the individual bacterium. Co-ordinated expression or cross-talk between virulence loci can lead to a programmed set of events within a bacterium analogous to a simple form of electronic memory that is of benefit during infection.
Introduction Bacterial diseases are often a result of multiple ‘stages’ within a host that can be separated spatially as well as temporally. Each niche will have different requirements and favour a different bacterial gene expression profile. Rather than solely responding to the signals present within a niche, it appears that bacteria are primed to meet the demands of the next stage or niche by producing a heterogeneous population. Diversity is a key bacterial weapon as it generates the ‘ideal’ bacterial subpopulation to occupy a niche and also limits host responses, as expression of important bacterial antigens is restricted. Bacteria are continually generating heterogeneity in the expression of surface antigens during infection. As the interaction with each host is to some extent different, the generation of antigenic and regulatory diversity is essential to microbial survival and evolution. For example, a bacterium that can switch off or on ten different surface components is able to generate over a million different phenotypes. This assumes independent regulation of each component. In reality, expression of such components can be interconnected and this will reduce the number of combinations available to those that are most successful for proliferation in that niche. Heterogeneous presentation of surface-associated structures is generated by both antigenic and phase variation. Antigenic variation enables a bacterium to express different forms of a key antigen whereas phase variation allows the antigen to be switched on or off by individual bacteria. Bacterial adhesins are surface-expressed structures required for colonization of the host but in most cases are also antigenic. Therefore, the expression of many adhesins is phase-variable, reflecting a balance between attachment and immune evasion. The sequencing of Gram-negative enteric pathogens such as Salmonella enterica serovar Typhimurium and Escherichia 05491 & 2004 SGM Printed in Great Britain
coli has shown them to contain multiple fimbrial and nonfimbrial adhesin gene clusters. When functional, this repertoire means the bacteria can bind to an assortment of receptors that may be present in different sites within the host, within different hosts or on surfaces outside the host. Different pathotypes have different adhesin collections, often built on top of a set of generic adhesins. In some cases evolution towards a single niche or host has been driven by or resulted in the silencing of certain adhesin gene clusters. Appropriate adhesin regulation is required during an infection to optimize colonization and prevent removal by both the innate and adaptive arms of the immune response. During any infection in which large numbers of bacteria are produced, there is a chance that this regulation will be fine-tuned through natural selection. This has included the evolution of co-ordinated control mechanisms that act between adhesin gene clusters. These prevent co-expression of certain adhesins and allow sequential expression to consolidate colonization. Regulation of surface components is highly complex and dependent on multiple inputs. This review is limited to crossregulation between virulence loci and does not cover the much wider aspect of environmental sensing and response in bacteria. Transcriptional regulation is conventionally thought of as either global, where a regulator affects a large number of genes, or local, involving a regulator that only controls expression of a specific set of genes, often in its own operon. However, to achieve co-ordinate control or sequential regulation, the most useful regulators are those that are expressed by the operons themselves but can also act on other loci. For example, in order to prevent co-expression of two surface factors, the simplest strategy is for a regulator to be produced when operon X is expressed which represses the expression of operon Y. When operon X is switched off, 585
N. J. Holden and D. L. Gally
operon Y can now be expressed, but only in sequence after operon X. In a single cell, regulators acting positively or negatively in this way could achieve a sequential programme of gene expression events that promote persistence within the host. The main concept discussed in this review is that of sequential expression of surface antigens, in particular adhesins, through cross-talk. Particular programmes have been selected that consolidate attachment or favour proliferation/survival in the next phase of the infection. The response is therefore not just adaptive but incorporates information ‘stored’ from previous events. Combinations of genetic switches are one way that inputs can be stored and can lead to subsequent changes in output (gene expression). This is analogous to how simple relays and gates work in electrical circuits. In this way bacteria can be considered to have a basic ‘memory’ as the term applies to electronic memory in computers. This review will consider virulence gene expression at the level of the single bacterium, in particular the regulation between specific fimbrial and flagellal operons. To limit the extent of the review it will use examples mainly from E. coli but when necessary also from other members of the Enterobacteriaceae. Organization of fimbrial gene clusters The majority of fimbrial gene clusters present in E. coli and Salmonella are exported using a chaperone–usher pathway and they share a standard genetic organization. At the start of the cluster are genes involved in their regulation followed by the main structural subunit. Genes encoding the chaperone, usher, adaptor proteins and finally the fimbrial adhesin make up the middle and end of the cluster. The most well studied are type 1, P and S fimbriae, which recognize Æ-D-mannose-, diagalactoside- and sialyl-containing receptors, respectively. Generally, the structural subunits are co-transcribed from a single promoter located either directly upstream of the main structural subunit, as for type 1 fimbriae, or upstream of a regulator, for example for P fimbriae. For further detail on the structure and assembly of fimbrial gene clusters the reader is referred to Blomfield (2001); Hultgren et al. (1996); Krogfelt (1991); Low et al. (1996).
frequencies between 0.01 and 1 % of the bacteria per generation (Braaten et al., 1994; Gally et al., 1993; Morschhauser et al., 1994). This, then, has to be balanced against the proportion switching off expression, which can occur at these rates or faster. The fimbriate population is therefore related to the two frequencies: off-to-on and on-tooff. Given that multiple fimbriae can be expressed by many E. coli strains, the expression status at any one time is a product of the different fimbrial transition rates and, importantly, the status of the starting population. Two examples of cross-talk will now be described in detail, one negative and one positive, involving type 1, P and S fimbriae in E. coli. For these to be considered, a brief overview of their regulation will be given, although the reader is referred to an excellent recent review for more details (Blomfield, 2001). Type 1 fimbriae Expression of type 1 fimbriae is governed by the orientation of a 314 bp invertible element (the fim switch) flanked by two 9 bp inverted repeats (Abraham et al., 1985). The switch contains the main promoter for the structural subunits. Two site-specific recombinases, FimE and FimB, are required, in combination with integration host factor (IHF) (Blomfield et al., 1997; Dorman & Higgins, 1987; Eisenstein et al., 1987) and the leucine-responsive regulatory protein (Lrp) (Roesch & Blomfield, 1998), to orientate the switch in either direction, i.e. from off-to-on and from on-to-off (Gally et al., 1994, 1996; Klemm, 1986). By analogy with other systems, IHF and Lrp are required to bend the fim switch DNA so that the inverted repeats are situated in close proximity, allowing recombination to occur. FimE and FimB bind to recognition sites on either side of the inverted repeats where they are able to recombine the DNA when the switch is in the right configuration (Gally et al., 1996). FimE primarily acts to turn the switch from on-to-off whereas FimB can switch expression on or off, but generally at much lower frequencies than FimE. FimB is therefore the regulator that acts to turn on expression of type 1 fimbriae, and the duration of expression at the single cell level is then defined by FimE activity. P fimbriae
Phase variation The majority of adhesins expressed by E. coli exhibit phase variation. This means that in a population of bacteria only a subset express the adhesin at any one time. The proportion of bacteria expressing the adhesin is a result of a stochastic process influenced by numerous other inputs, often environmental. For example, while bacteria expressing type 1 fimbriae can represent 5 % of the population under certain conditions at 37 8C, if the population is shifted to 28 8C, this temperature signal overrides the benefit of generating a heterogeneous population and expression of type 1 fimbriae becomes virtually undetectable (Gally et al., 1993). Rates of phase variation have been calculated for a number of different fimbriae and off-to-on phase transition occurs at 586
Regulation of P fimbriae is quite distinct from that of type 1 fimbriae. The intergenic region between the divergently encoded regulators PapI and PapB contains the promoter for the main structural subunits of P fimbriae, so that papB is also co-transcribed with the genes encoding the structural subunits (Baga et al., 1985). An RNase III cleavage site located downstream of papB results in rapid degradation of the papB mRNA, whereas the mRNA of the structural genes remains relatively stable (Baga et al., 1988). Expression from the divergent promoters is dependent on the methylation status of two GATC sites located between papI and papB (Blyn et al., 1990). In phase-on cells, the specific regulator PapI in combination with Lrp and the catabolite repressor protein (CRP) forms a nucleoprotein complex protecting the distal Journal of Medical Microbiology 53
Switches, cross-talk and memory in E. coli adherence
GATC site from deoxyadenosine methylase (Dam) methylation (Braaten et al., 1992; Goransson et al., 1989). The second proximal GATC site is modified by Dam and transcription of both papI and papBA proceeds. In phase-off cells, Lrp binds to sites that protect the proximal GATC site from methylation, the distal GATC site is methylated and transcription of papBA is inhibited. Phase-variable expression of related fimbrial gene clusters, such as S fimbriae is also dependent on differential methylation of two GATC sites (van der Woude & Low, 1994). Simplistically, the regulators PapB/ SfaB and PapI/SfaC can be viewed as positive regulators, with both being required for expression, with increasing levels stimulating phase transition rates. However, for PapB, phase transition rates fall off once a certain concentration of the regulator is reached (Forsman et al., 1989). Consequently, PapB can act as both a positive and negative regulator of phase transition rates. Cross-talk between adhesin gene clusters
substantially as a result of this control. The situation therefore becomes highly complex if four or five related clusters are present in the genome. One tangible outcome of this would be that clinical isolates containing multiple fimbrial gene clusters would show much higher frequencies of P and S fimbrial gene expression in comparison to clusters that have been cloned and studied in isolation in laboratory strains. This appears to be the case, shown in the fimbrial kinetics studies of neonatal meningitis-associated E. coli (NMEC) and uropathogenic E. coli (UPEC) (Nowicki et al., 1984, 1985). The autoregulatory role of PapB and homologues is likely to be important in tempering this positive cross-talk (Forsman et al., 1989). Our ongoing work is examining the capacity of these highly related regulators to stimulate or repress expression from other fimbrial operons. Preliminary data indicate that very subtle changes in regulators can prevent their ability to cross-talk and that within a pool of homologous regulators only certain ones may have the ability to act away from their own cluster. This raises the possibility of a hierarchy in expression.
(i) Positive cross-talk
The fact that some fimbrial clusters have regulators that share a high level of sequence similarity implies that complementation may occur between them. Several studies have proved that regulators with highly related amino acid sequences also share functional homology. P- and S-related fimbriae are highly homologous, sharing 85 and 76 % amino acid identity between the PapI-like proteins prfI and sfaC and the PapBlike proteins prfB and sfaB, respectively (Schmoll et al., 1990). The main structural subunit proteins are less similar, but still share 31 % amino acid identity. The uropathogenic E. coli strain 536 contains both P-related and S fimbrial gene clusters, located on pathogenicity islands (PAI II and PAI III, respectively) (Dobrindt et al., 2001; Hacker et al., 1990). Previous analysis of PAI II showed that its deletion resulted in diminished expression of S fimbriae (Knapp et al., 1986). This was subsequently shown to be a consequence of deleting regulators from the P-related fimbrial cluster that were acting positively on S fimbriae expression (Morschhauser et al., 1994). While the differences measured were minor (two- to fourfold), it must be appreciated that these changes are likely to represent changes in the proportion of the population expressing the fimbriae. The repression of S fimbriae observed when PAI II was deleted was overcome by introduction of either prfB or, to a lesser extent, prfI. The effect of mutating prfB and prfI was also examined in vivo. While both regulatory genes were found to effect virulence in mice, it was not possible to determine whether this result was due to the removal of this positive cross-talk. This would have required information on the expression status of both the sfa and prf clusters of the bacterial population during infection. Our own unpublished data support the idea that the PapI family of proteins are able to act positively on related gene clusters. As described above, this may result in relatively subtle changes in total fimbriation levels, as measured by total population techniques, but the likelihood of co- or subsequent expression at the single cell level is increased http://jmm.sgmjournals.org
Positive cross-talk between adhesin gene clusters may be of benefit for several reasons. (i) As stated, multiple P and S (and other) operons may exist in E. coli pathogens to cope with the receptor diversity in and between hosts. If adhesin expression is controlled by homologous regulators from related clusters then, once one adhesin is expressed, the chances of the others being expressed is increased, thereby increasing the probability of bacterial attachment. This cross-talk would also mean that ‘master’ adhesin operons may exist that are controlled appropriately by environmental cues, and these in turn stimulate expression of other adhesin operons lower in the regulatory hierarchy. (ii) Another possibility is that the concentration of regulator required to stimulate self as opposed to related operons is different. In this case the positive cross-talk could act to bring about sequential fimbrial expression. This means that a cell could run through its repertoire of related adhesins, with the successful being ‘selected’ through their capacity to adhere (Fig. 1). Sequential expression would have advantages over combined expression as it would limit physical interference between structures. (iii) Many fimbrial clusters identified from whole genome sequencing projects contain mutations in particular components such as chaperones and ushers leading to the conclusion that the fimbriae will not be expressed. Co-expression of related clusters through such positive cross-talk could allow complementation to occur and lead to expression of an increased adhesin repertoire. (iv) Finally, there is some evidence that E. coli can detect the fact it is attached to a surface, possibly through changes in membrane stresses (Otto & Silhavy, 2002; Otto & Hermansson, 2004; Zhang & Normark, 1996). Evidence to date supports sensing by the CpxRA two-component pathway, as deletion of cpxR results in impaired adhesion of E. coli (Otto & Silhavy, 2002). It is possible that this sort of attachment signal could either interfere or complement adhesin cross-talk and ensure that a ‘successful’ adhesin is kept in an ‘on’ state. 587
N. J. Holden and D. L. Gally
Fig. 1 Schematic diagram showing one possible scenario of sequential virulence factor expression in a single E. coli. Type 1 fimbriae (Fim), blue; Pap fimbriae (Pap), red; S fimbriae (Sfa), green; flagella (Fla), black. Filled lines, good evidence for cross-regulation; broken lines, some evidence for cross-regulation. Arrows, positive regulation; blocked lines, negative regulation.
(ii) Negative cross-talk
The first evidence that expression of one fimbrial type may repress another was provided by work on S and type 1 fimbriae in a clinical isolate from a case of neonatal sepsis (Nowicki et al., 1985). After 35 h incubation, 33 % of cells were S fimbriate, 52 % type 1 fimbriate, 12 % were afimbriate and 3 % had both S and type 1 fimbriae. If the expression of one had no effect on the other, the percentage co-expressing S and type 1 fimbriae should have been 16 % (52 % of the 33 % S fimbriate population). This was shown very clearly by starting from a fully S fimbriate population. After the bacteria were grown statically in LB broth for 29 h, 80 % of the population still expressed S fimbriae and type 1 fimbriate bacteria represented 9.4 % of the total population. Bacteria that expressed both fimbrial types after 29 h represented only 0.5 % of the total population, rather than 8 % (80 % of 9.4 %). It has subsequently been shown that both SfaB and PapB are able to inhibit the activity of the FimB recombinase on the fim switch (Holden et al., 2001; Xia et al., 2000). This means that while P or S fimbriae are in the phase-on state the expression of PapB/SfaB should prevent type 1 fimbriae from being switched on. In addition, PapB was shown to increase the expression of FimE, which turns off the expression of type 1 fimbriae (Xia et al., 2000). Therefore, even if type 1 fimbriae 588
are being expressed in a bacterium in which P fimbrial expression is switched on, it is likely that type 1 fimbrial expression will then be switched off (Fig. 1). At present, it is unclear whether type 1 fimbrial expression can alter P or S fimbrial expression. The importance of the above negative cross-regulation during infection is unknown, although recent work does point to a transition from type 1 fimbrial expression to P fimbrial expression in UPEC during experimental urinary tract infection in mice (Gunther et al., 2001). This transition could result from the PapB inhibition described above. Differences in type 1 fimbrial expression in the mouse model were observed between bacteria associated either with cystitis or pyelonephritis. For cystitis-associated strains, a mean of 81 % of the bacteria attached to bladder epithelium had the fim switch in the on orientation after 24 h compared with 2 % for pyelonephritis isolates at the same time-point. Classically pyelonephritis strains are more likely to express and contain P and S fimbrial gene clusters and therefore down-regulate type 1 fimbriae expression. What is still unknown is how important the cross-talk between P and type 1 fimbriae is to the expression levels during infection. Our work has shown a range of phase-variation frequencies for P and type 1 fimbriae [Leathart & Gally (1998) and unpublished] and the cross-talk between these clusters will Journal of Medical Microbiology 53
Switches, cross-talk and memory in E. coli adherence
have a direct effect on the level of fimbriation. Expression of particular fimbriae is pro-inflammatory (Wullt et al., 2001), so there is likely to be a relationship between the level of expression and the severity of infection. For example, fimbrial expression is more likely to be repressed in asymptomatic infections (Graham et al., 2001; Hull et al., 1998, 1999; Stenqvist et al., 1987). A recent human trial used an E. coli strain isolated from an asymptomatic patient to investigate the expression of P fimbriae during infection. The strain was transformed with P fimbriae and the in vivo expression status of P fimbriae was shown to decrease rapidly following inoculation of human volunteers. Although this asymptomatic strain does possess the chromosomal genes for expression of type 1 fimbriae, P and F1C, these have been shown not to be expressed in vitro or in vivo. This strain apparently uses another unknown mechanism to persist in the bladder and potentially succeeds by failing to stimulate a pro-inflammatory response. Conversely, the proportion of bacteria expressing fimbriae during symptomatic infections may be high, stimulating a strong response, but leading to a short-lived infection. Host factors (receptors and immune response) in combination with bacterial phase-transition frequencies will govern the degree of fimbriation.
the absence of oxidative stress, OxyR is almost all, if not all, in the reduced state, since the cytoplasm is a reducing environment. Gel-shift assays showed that a variant of OxyR that is in the oxidative form can bind to agn43 DNA, albeit with approximately fivefold lower affinity than wild-type OxyR (Wallecha et al., 2003). So, while the expression of these adhesins is mutually exclusive, the molecular basis to the regulation is still far from clear.
Another example of apparent mutual exclusivity between adhesins is the expression of type 1 fimbriae and the phasevariable auto-aggregation factor antigen 43 (Ag43). Ag43 interactions are responsible for the auto-aggregation phenotype and the extent of auto-aggregation is dependent on the proportion of Ag43-positive bacteria in a population. Ag43 belongs to a class of bacterial auto-transporters, so has a distinct mechanism of export compared to the chaperone usher family. Dam methylase function is required for Ag43 phase-variable expression (Henderson & Owen, 1999). The Ag43-positive phenotype was originally denoted flu+ and it was observed that flu and type 1 fimbrial expression were mutually exclusive at the colony level (Hasman et al., 1999). Introduction of the fim cluster, on a plasmid, or fimB on a plasmid to stimulate type 1 fimbriae expression abolished the auto-aggregation phenotype of a strain which is locked on for Ag43 expression. Therefore, initial studies suggested that prevention of Ag43 autoaggregation by the presence of long fimbrial structures is a physical phenomenon and not at the level of co-ordinated regulation. However, it has since been demonstrated that the flu mRNA level increased 20-fold in the absence of fim. Moreover flu transcription was decreased when type 1 fimbriae were overexpressed (Schembri & Klemm, 2001). A proposed mechanism for this was the reducing state of the cell, in which the biosynthesis of a large number of disulphide-containing surface structures constitutes a signal that alters the cellular thiol-disulphide status, which drives OxyR to the reduced state and, in turn, downregulates Ag43. OxyR is able to monitor the reducing conditions in the cell and an OxyR recognition sequence covers all three GATC sites within the flu promoter. OxyR binding can effect the access of Dam, as is the case for Lrp in P phase variation, but the importance of the reduced or oxidized form of OxyR for the regulation is not clear. In
Cross-talk between adhesins and flagella expression
http://jmm.sgmjournals.org
A key role for negative cross-talk would be to ensure that different adhesin types are not co-expressed by the same bacterial cell, as this may increase the chances of generating an immune response but also that one may functionally inhibit the other, for example if a fimbrial adhesin is expressed at the same time as a (shorter) non-fimbrial adhesin, such as Ag43 and type 1 fimbriae. Moreover, cross-talk may prevent complications occurring with the export of heterogeneous fimbrial components, which could be detrimental to the survival of the cell. As described, such regulation could also set up sequential expression which, for example, could ensure that an afimbrial adhesin (producing more intimate bacterial–host cell contact) is in position to take advantage of initial binding by a fimbrial adhesin.
Flagella expression has been shown to contribute to virulence in a number of infection models. In addition, both flagella and type 1 fimbriae have been shown to be important in biofilm formation (Danese et al., 2000; Pratt & Kolter, 1998). For many enteric bacteria, flagellae and fimbrial expression is clearly not mutually exclusive, as both can be demonstrated on the surface of an individual bacterium by electron microscopy. However, this does not rule out the occurrence of co-ordinate regulation between the two types of organelle and there is increasing evidence for regulation between the two. Proteus mirabilis contains a fimbrial adhesin MR/P (mannose-resistant, Proteus-like) that is phase-variable and controlled by an invertible genetic element. The mrp operon contains two divergent transcripts, mrpABCDEFGHJ, with .29 % identity to E. coli papAHCDJKEFG, and mrpI, which is 50 % identical to the fimB gene product at the amino acid level (Bahrani & Mobley, 1994; Li et al., 1999). The main fimbrial promoter is contained within a 251 bp invertible element flanked by two 21 bp inverted repeats (Zhao et al., 1997). MrpI is the sole recombinase that regulates phase variation of MR/P fimbriae in P. mirabilis (Li et al., 2002). Another regulator is encoded by the final gene of the cluster, mrpJ (Li et al., 2001). Its amino acid sequence contains a putative helix–turn–helix domain and also shares homology with the Cro/C1 family of regulators. P. mirabilis transformed with a plasmid encoding mrpJ displayed impaired motility in swimming and swarming, compared to vector only or mrp˜. Moreover, an eightfold reduction in flagella synthesis, shown by Western blots with FlaA antisera, occurred when MprJ was overexpressed. Western blot analy589
N. J. Holden and D. L. Gally
sis also showed that a deletion of mrpJ resulted in a decrease in the level of MrpA (main structural subunit) and FlaA, to about the same extent. Mutation of mrpJ results in an attenuated phenotype in the murine model of ascending urinary tract infection, when mice are co-challenged with a wild-type strain and mrpJ mutant strain. Taken together, these studies show that MrpJ is able to repress flagella production when MR/P fimbriae are expressed. Furthermore, the repression occurs at the level of transcription of the flagella master operon, since transcription from a flaDClacZ fusion strain was repressed with elevated levels of MrpJ. As yet, there is no evidence for direct interactions of MrpJ on fla DNA, implying that this activity is indirect. Related MrpJ regulators are present at the end of the P(papX) and S- (sfaX) fimbrial gene clusters in E. coli (Dobrindt et al., 2001; Marklund et al., 1992). Alignment of PapX with MrpJ shows homologous domains with different levels of conservation. Overexpression of PapX in P. mirabilis resulted in decreased motility and FlaA production, compared to the vector control, suggesting that PapX is a functional homologue of MrpJ in P. mirabilis (Li et al., 2001). However, this role has yet to be demonstrated in E. coli. Earlier work where PapX was mutated showed that it did not affect the degree of Prs piliation or the adhesin characteristics of the strain (Marklund et al., 1992). Recently it has been demonstrated that deletion of fliC, which encodes flagellin, results in a decrease in type 1 fimbrial expression in an invasive E. coli strain isolated from a patient with Crohn’s disease (Barnich et al., 2003). Previous work had shown that type 1 pili-mediated adherence plays an essential part in the invasive ability of this strain (Boudeau et al., 2001). Even deletion of motB, which is normally required for flagellar rotation, results in a failure to express type 1 fimbriae in the mutant. The molecular basis of this regulation is unknown but does appear to act at the level of the fim switch or on recombinase expression. Interestingly, the fliC mutation appeared to be affecting expression of another unknown determinant required for invasion. This type of data supports the presence of local regulatory networks essential for co-ordinate control of virulence determinants. The molecular basis of gastrointestinal tract infection by enteropathogenic E. coli (EPEC) has been investigated in detail, and current evidence supports a series of steps that lead to epithelial binding (Frankel et al., 1998; Nataro & Kaper, 1998). The initial attachment could be driven by a number of adhesins including flagella (Giron et al., 2002) and bundle-forming pili (Bfp) (Giron et al., 1991; Tobe & Sasakawa, 2001). The role of Bfp is controversial, as these type IV pili are also important for interbacterial interactions and micro-colony formation (Knutton et al., 1999; Tobe & Sasakawa, 2001). Following initial attachment, further interaction with the host cell is mediated by the translocon components of a type III secretion system (McDaniel et al., 1995), in particular EspA filaments. These structures allow the injection of bacterial proteins into the host cell (Knutton 590
et al., 1998). One of the proteins injected is the translocated intimin receptor (Tir), which inserts into the host cell membrane and binds to the bacterial outer-membrane protein intimin (Kenny et al., 1997). This interaction induces cytoskeletal rearrangements in the host cell and the formation of attaching and effacing lesions (Knutton et al., 1989). Variations on this scheme may prevent phagocytosis or Mcell uptake or actually promote invasion in the case of Salmonella serotypes (Celli et al., 2001; Knodler et al., 2002). While such factors are not necessarily controlled by phase-variable switches, the requirement for different surface factors one after the other will have led to an evolved programme of cross-talk between the different virulence loci, resulting in sequential or co-ordinated expression (Elliott et al., 2000; Giron et al., 2002; Knodler et al., 2002; Roe et al., 2003b). Recent research on H6 flagellar expression EPEC has demonstrated co-ordinated expression with type III secretion components and Bfp (Giron et al., 2002). Moreover, this expression appears linked to host-released factors such as epinephrine (Giron et al., 2002; Sperandio et al., 2003). In enterohaemorrhagic E. coli O157 : H7, the expression of the type III EspA translocon filaments, Tir and intimin is heterogeneous and co-ordinated [Roe et al. (2003b) and unpublished]. In the majority of 60 E. coli O157 : H7 strains examined to date the proportions of bacteria expressing these factors in liquid media are very low (1–5 %) (unpublished). While the locus of enterocyte effacement-encoded regulator (Ler) ensures co-expression of most type III apparatus components and effectors (Elliott et al., 2000), it is clear that a further checkpoint in translocon expression and secretion exists that results in the observed heterogeneity. Initial evidence points to this control being post-transcriptional (Roe et al., 2003a). Again, this variable regulation is likely to have evolved both to limit immune exposure of these key antigens and to allow programmed expression of other surface factors during colonization of the primary bovine host. As with EPEC, these factors are likely to be flagella, fimbriae and outer-membrane proteins that are important in the interaction with host cell receptors. Conclusion Bacteria persist in the host as a consequence of appropriate gene expression enabling occupation of specific niches whilst at the same time minimizing exposure to the immune system. If successful infections can be viewed as occupation of sequential niches, then bacteria are likely to have evolved regulatory pathways that reflect this, for example, the processes that would be required to first bind to and penetrate a mucus layer and then adhere to the underlying epithelial cells. The spread of E. coli biofilms requires both type 1 fimbrial and flagellar expression, presumably reflecting the dynamic interplay required between attachment and motility. As described above, there is evidence that expression of these factors is co-ordinated. In a similar way, attachment by a fimbrial adhesin may lead to consolidation by expression of fimbrial or afimbrial adhesins, for example Journal of Medical Microbiology 53
Switches, cross-talk and memory in E. coli adherence
EPEC attachment by fimbriae followed by intimate attachment as a result of intimin expression. All the time, however, the requirement for these different factors has to be weighed against the consequences of their expression. Therefore the expression of many of these factors is phase-variable. A further level of control would be to have such genetic switches dependent on other regulatory events, rather than being random. This would then permit co-ordinated or sequential expression of surface factors. Computing memory is based on simple binary switches, and there is no reason why bacteria using epigenetic methylation and invertible element switches could not execute a programmed series of events that would be dependent on the initial triggering stimulus. As such, bacterial memory, i.e. a programmed response based on previous recent experience, is entirely possible (Fig. 1). At present there are only a few examples of such logical input-related behaviour, but such pathways are very difficult to demonstrate without techniques for imaging gene expression in single cells. As our capacity to do this increases, we predict that more examples of such memory and programming will be uncovered. An important consequence of this type of regulation is that whole-population expression studies can be misleading. Research on microbial pathogens is increasingly making use of DNA microarrays to dissect complex regulatory pathways, but a serious drawback is that heterogeneous expression is virtually impossible to study by this methodology. This limitation will be particularly true for complex in vivo-derived samples. Future research must concentrate on regulation demonstrated at the single cell level in addition to that of the whole population. In summary, within any niche during infection, environmental monitoring by bacteria generates a certain heterogeneous population. Heterogeneity pre-empts the next stage of infection and allows it to be progressed by a successful bacterial subpopulation. The heterogeneity is required to cope with the unpredictability of different hosts and niches. However, within a single bacterium, its changing surface phenotype could reflect a particular sequential regulatory pathway. Understanding the pathways that control cross-talk between surface-expressed organelles will provide further insights into bacterial pathogenesis. References Abraham, J. M., Freitag, C. S., Clements, J. R. & Eisenstein, B. I. (1985).
An invertible element of DNA controls phase variation of type 1 fimbriae of Escherichia coli. Proc Natl Acad Sci U S A 82, 5724–5727. Baga, M., Goransson, M., Normark, S. & Uhlin, B. E. (1985).
Transcriptional activation of a pap pilus virulence operon from uropathogenic Escherichia coli. EMBO J 4, 3887–3893. Baga, M., Goransson, M., Normark, S. & Uhlin, B. E. (1988). Processed
mRNA with differential stability in the regulation of E. coli pilin gene expression. Cell 52, 197–206. Bahrani, F. K. & Mobley, H. L. (1994). Proteus mirabilis MR/P fimbrial operon: genetic organization, nucleotide sequence, and conditions for expression. J Bacteriol 176, 3412–3419. Barnich, N., Boudeau, J., Claret, L. & Darfeuille-Michaud, A. (2003).
http://jmm.sgmjournals.org
Regulatory and functional co-operation of flagella and type 1 pili in adhesive and invasive abilities of AIEC strain LF82 isolated from a patient with Crohn’s disease. Mol Microbiol 48, 781–794. Blomfield, I. C. (2001). The regulation of pap and type 1 fimbriation in
Escherichia coli. Adv Microbiol Physiol 45, 1–49. Blomfield, I. C., Kulasekara, D. H. & Eisenstein, B. I. (1997). Integration
host factor stimulates both FimB- and FimE-mediated site-specific DNA inversion that controls phase variation of type 1 fimbriae expression in Escherichia coli. Mol Microbiol 23, 705–717. Blyn, L. B., Braaten, B. A. & Low, D. A. (1990). Regulation of pap pilin phase variation by a mechanism involving differential dam methylation states. EMBO J 9, 4045–4054. Boudeau, J., Barnich, N. & Darfeuille-Michaud, A. (2001). Type 1 pili-
mediated adherence of Escherichia coli strain LF82 isolated from Crohn’s disease is involved in bacterial invasion of intestinal epithelial cells. Mol Microbiol 39, 1272–1284. Braaten, B. A., Platko, J. V., van der Woude, M. W., Simons, B. H., de Graaf, F. K., Calvo, J. M. & Low, D. A. (1992). Leucine-responsive
regulatory protein controls the expression of both the pap and fan pili operons in Escherichia coli. Proc Natl Acad Sci U S A 89, 4250–4254. Braaten, B. A., Nou, X. W., Kaltenbach, L. S. & Low, D. A. (1994).
Methylation patterns in pap regulatory DNA control pyelonephritisassociated pili phase variation in Escherichia coli. Cell 76, 577–588. Celli, J., Olivier, M. & Finlay, B. B. (2001). Enteropathogenic Escherichia
coli mediates antiphagocytosis through the inhibition of PI 3-kinasedependent pathways. EMBO J 20, 1245–1258. Danese, P. N., Pratt, L. A., Dove, S. L. & Kolter, R. (2000). The outer
membrane protein, antigen 43, mediates cell-to-cell interactions within Escherichia coli biofilms. Mol Microbiol 37, 424–432. Dobrindt, U., Blum-Oehler, G., Hartsch, T., Gottschalk, G., Ron, E. Z., Fu¨nfstu¨ck, R. & Hacker, J. (2001). S-fimbria-encoding determinant sfaI
is located on pathogenicity island III536 of uropathogenic Escherichia coli strain 536. Infect Immun 69, 4248–4256. Dorman, C. J. & Higgins, C. F. (1987). Fimbrial phase variation in
Escherichia coli: dependence on integration host factor and homologies with other site-specific recombinases. J Bacteriol 169, 3840–3843. Eisenstein, B. I., Sweet, D. S., Vaughn, V. & Friedman, D. I. (1987).
Integration host factor is required for the DNA inversion that controls phase variation in Escherichia coli. Proc Natl Acad Sci U S A 84, 6506–6510. Elliott, S. J., Sperandio, V., Giro´n, J. A., Shin, S., Mellies, J. L., Wainwright, L., Hutcheson, S. W., McDaniel, T. K. & Kaper, J. B. (2000). The locus of enterocyte effacement (LEE)-encoded regulator
controls expression of both LEE- and non-LEE-encoded virulence factors in enteropathogenic and enterohemorrhagic Escherichia coli. Infect Immun 68, 6115–6126. Forsman, K., Goransson, M. & Uhlin, B. E. (1989). Auto-regulation and
multiple DNA interactions by a transcriptional regulatory protein in E. coli pili biogenesis. EMBO J 8, 1271–1277. Frankel, G., Phillips, A. D., Rosenshine, I., Dougan, G., Kaper, J. B. & Knutton, S. (1998). Enteropathogenic and enterohaemorrhagic Escher-
ichia coli: more subversive elements. Mol Microbiol 30, 911–921. Gally, D. L., Bogan, J. A., Eisenstein, B. I. & Blomfield, I. C. (1993).
Environmental-regulation of the fim switch controlling type 1 fimbrial phase variation in Escherichia coli K-12: effects of temperature and media. J Bacteriol 175, 6186–6193. Gally, D. L., Rucker, T. J. & Blomfield, I. C. (1994). The leucine-
responsive regulatory protein binds to the fim switch to control phase variation of type 1 fimbrial expression in Escherichia coli K-12. J Bacteriol 176, 5665–5672. 591
N. J. Holden and D. L. Gally
Gally, D. L., Leathart, J. & Blomfield, I. C. (1996). Interaction of FimB
and FimE with the fim switch that controls the phase variation of type 1 fimbriae in Escherichia coli K-12. Mol Microbiol 21, 725–738. Giron, J. A., Ho, A. S. Y. & Schoolnik, G. K. (1991). An inducible bundle-
forming pilus of enteropathogenic Escherichia coli. Science 254, 710–713.
accumulation at sites of bacterial adhesion to tissue culture cells: basis of a new diagnostic test for enteropathogenic and enterohemorrhagic Escherichia coli. Infect Immun 57, 1290–1298. Knutton, S., Rosenshine, I., Pallen, M. J., Nisan, I., Neves, B. C., Bain, C., Wolff, C., Dougan, G. & Frankel, G. (1998). A novel EspA-associated
Giron, J. A., Torres, A. G., Freer, E. & Kaper, J. B. (2002). The flagella of
surface organelle of enteropathogenic Escherichia coli involved in protein translocation into epithelial cells. EMBO J 17, 2166–2176.
enteropathogenic Escherichia coli mediate adherence to epithelial cells. Mol Microbiol 44, 361–379.
Knutton, S., Shaw, R. K., Anantha, R. P., Donnenberg, M. S. & Zorgani, A. A. (1999). The type IV bundle-forming pilus of enteropathogenic
Goransson, M., Forsman, K., Nilsson, P. & Uhlin, B. E. (1989).
Escherichia coli undergoes dramatic alterations in structure associated with bacterial adherence, aggregation and dispersal. Mol Microbiol 33, 499–509.
Upstream activating sequences that are shared by two divergently transcribed operons mediate cAMP-CRP regulation of pilus-adhesin in Escherichia coli. Mol Microbiol 3, 1557–1565.
Krogfelt, K. A. (1991). Bacterial adhesion: genetics, biogenesis, and role
Graham, J. C., Leathart, J. B. S., Keegan, S. J., Pearson, J., Bint, A. & Gally, D. L. (2001). Analysis of Escherichia coli strains causing bacteriuria
in pathogenesis of fimbrial adhesins of Escherichia coli. Rev Infect Dis 13, 721–735.
during pregnancy: selection for strains that do not express type 1 fimbriae. Infect Immun 69, 794–799.
Leathart, J. B. S. & Gally, D. L. (1998). Regulation of type 1 fimbrial
Gunther, N. W., IV, Lockatell, V., Johnson, D. E. & Mobley, H. L. T. (2001).
In vivo dynamics of type 1 fimbria regulation in uropathogenic Escherichia coli during experimental urinary tract infection. Infect Immun 69, 2838–2846. Hacker, J., Bender, L., Ott, M., Wingender, J., Lund, B., Marre, R. & Goebel, W. (1990). Deletions of chromosomal regions coding for
expression in uropathogenic Escherichia coli: heterogeneity of expression through sequence changes in the fim switch region. Mol Microbiol 28, 371–381. Li, X., Johnson, D. E. & Mobley, H. L. T. (1999). Requirement of MrpH
for mannose-resistant Proteus-like fimbria-mediated hemagglutination by Proteus mirabilis. Infect Immun 67, 2822–2833. Li, X., Rasko, D. A., Lockatell, C. V., Johnson, D. E. & Mobley, H. L. T. (2001). Repression of bacterial motility by a novel fimbrial gene
fimbriae and hemolysins occur in vitro and in vivo in various extraintestinal Escherichia coli isolates. Microb Pathog 8, 213–225.
product. EMBO J 20, 4854–4862.
Hasman, H., Chakraborty, T. & Klemm, P. (1999). Antigen-43-
Li, X., Lockatell, C. V., Johnson, D. E. & Mobley, H. L. T. (2002).
mediated autoaggregation of Escherichia coli is blocked by fimbriation. J Bacteriol 181, 4834–4841. Henderson, I. R. & Owen, P. (1999). The major phase-variable outer
membrane protein of Escherichia coli structurally resembles the immunoglobulin A1 protease class of exported protein and is regulated by a novel mechanism involving Dam and OxyR. J Bacteriol 181, 2132–2141. Holden, N. J., Uhlin, B. E. & Gally, D. L. (2001). PapB paralogues and their
effect on the phase variation of type 1 fimbriae in Escherichia coli. Mol Microbiol 42, 319–330. Hull, R. A., Rudy, D. C., Wieser, I. E. & Donovan, W. H. (1998). Virulence
factors of Escherichia coli isolates from patients with symptomatic and asymptomatic bacteriuria and neuropathic bladders due to spinal cord and brain injuries. J Clin Microbiol 36, 115–117. Hull, R. A., Rudy, D. C., Donovan, W. H., Wieser, I. E., Stewart, C. & Darouiche, R. O. (1999). Virulence properties of Escherichia coli 83972,
a prototype strain associated with asymptomatic bacteriuria. Infect Immun 67, 429–432. Hultgren, S. J., Jones, C. H. & Normark, S. (1996). Bacterial adhesions
and their assembly. In Escherichia coli and Salmonella, pp. 2730–2756. Edited by F. C. Neidhardt. Washington, DC: American Society for Microbiology. Kenny, B., DeVinney, R., Stein, M., Reinscheid, D. J., Frey, E. A. & Finlay, B. B. (1997). Enteropathogenic E. coli (EPEC) transfers its receptor for
intimate adherence into mammalian cells. Cell 91, 511–520. Klemm, P. (1986). Two regulatory fim genes, fimB and fimE, control
the phase variation of type 1 fimbriae in Escherichia coli. EMBO J 5, 1389–1393. Knapp, S., Hacker, J., Jarchau, T. & Goebel, W. (1986). Large, unstable
Identification of MrpI as the sole recombinase that regulates the phase variation of MR/P fimbria, a bladder colonization factor of uropathogenic Proteus mirabilis. Mol Microbiol 45, 865–874. Low, D., Braaten, B. & van der Woude, M. (1996). Fimbriae. In
Escherichia coli and Salmonella, pp. 146–157. Edited by F. C. Neidhardt. Washington, DC: American Society for Microbiology. Marklund, B. I., Tennent, J. M., Garcia, E., Hamers, A., Baga, M., Lindberg, F., Gaastra, W. & Normark, S. (1992). Horizontal gene
transfer of the Escherichia coli pap and prs pili operons as a mechanism for the development of tissue-specific adhesive properties. Mol Microbiol 6, 2225–2242. McDaniel, T. K., Jarvis, K. G., Donnenberg, M. S. & Kaper, J. B. (1995).
A genetic locus of enterocyte effacement conserved among diverse enterobacterial pathogens. Proc Natl Acad Sci U S A 92, 1664–1668. Morschhauser, J., Vetter, V., Emody, L. & Hacker, J. (1994). Adhesin
regulatory genes within large, unstable DNA regions of pathogenic Escherichia coli: cross-talk between different adhesin gene clusters. Mol Microbiol 11, 555–566. Nataro, J. P. & Kaper, J. B. (1998). Diarrheagenic Escherichia coli. Clin
Microbiol Rev 11, 142–203. Nowicki, B., Rhen, M., Va¨isa¨nen-Rhen, V., Pere, A. & Korhonen, T. K. (1984). Immunofluorescence study of fimbrial phase variation in
Escherichia coli KS71. J Bacteriol 160, 691–695. Nowicki, B., Rhen, M., Va¨isa¨nen-Rhen, V., Pere, A. & Korhonen, T. K. (1985). Kinetics of phase variation between S and type-1 fimbriae of
Escherichia coli. FEMS Microbiol Lett 28, 237–242. Otto, K. & Hermansson, M. (2004). Inactivation of ompX causes
increased interactions of type 1 fimbriated Escherichia coli with abiotic surfaces. J Bacteriol 186, 226–234.
inserts in the chromosome affect virulence properties of uropathogenic Escherichia coli O6 strain 536. J Bacteriol 168, 22–30.
Otto, K. & Silhavy, T. (2002). Surface sensing and adhesion of Escherichia
Knodler, L. A., Celli, J., Hardt, W. D., Vallance, B. A., Yip, C. & Finlay, B. B. (2002). Salmonella effectors within a single pathogenicity island are
coli controlled by the Cpx-signaling pathway. Proc Natl Acad Sci U S A 99, 2287–2292.
differentially expressed and translocated by separate type III secretion systems. Mol Microbiol 43, 1089–1103.
Pratt, L. A. & Kolter, R. (1998). Genetic analysis of Escherichia coli biofilm formation: roles of flagella, motility, chemotaxis and type I pili. Mol Microbiol 30, 285–293.
Knutton, S., Baldwin, T., Williams, P. H. & McNeish, A. S. (1989). Actin
592
Journal of Medical Microbiology 53
Switches, cross-talk and memory in E. coli adherence
Roe, A. J., Hoey, D. E. E. & Gally, D. L. (2003a). Regulation, secretion and activity of type III-secreted proteins of enterohaemorrhagic Escherichia coli O157. Biochem Soc Trans 31, 98–103. Roe, A. J., Yull, H., Naylor, S. W., Woodward, M. J., Smith, D. G. E. & Gally, D. L. (2003b). Heterogeneous surface expression of EspA translocon
Tobe, T. & Sasakawa, C. (2001). Role of bundle-forming pilus of
enteropathogenic Escherichia coli in host cell adherence and in microcolony development. Cell Microbiol 3, 579–585. van der Woude, M. W. & Low, D. A. (1994). Leucine-responsive
filaments by Escherichia coli O157 : H7 is controlled at the posttranscriptional level. Infect Immun 71, 5900–5909.
regulatory protein and deoxyadenosine methylase control the phase variation and expression of the sfa and daa pili operons in Escherichia coli. Mol Microbiol 11, 605–618.
Roesch, P. L. & Blomfield, I. C. (1998). Leucine alters the interaction of
Wallecha, A., Correnti, J., Munster, V. & van der Woude, M. (2003).
the leucine-responsive regulatory protein (Lrp) with the fim switch to stimulate site-specific recombination in Escherichia coli. Mol Microbiol 27, 751–761. Schembri, M. A. & Klemm, P. (2001). Coordinate gene regulation by
fimbriae-induced signal transduction. EMBO J 20, 3074–3081. Schmoll, T., Morschhauser, J., Ott, M., Ludwig, B., van Die, I. & Hacker, J. (1990). Complete genetic organization and functional-aspects of the
Escherichia coli S fimbrial adhesin determinant: nucleotide sequence of the genes sfa B, C, D, E, F. Microb Pathog 9, 331–343. Sperandio, V., Torres, A. G., Jarvis, B., Nataro, J. P. & Kaper, J. B. (2003).
Phase variation of Ag43 is independent of the oxidation state of OxyR. J Bacteriol 185, 2203–2209. Wullt, B., Bergsten, G., Samuelsson, M., Gebretsadik, N., Hull, R. & Svanborg, C. (2001). The role of P fimbriae for colonization and host
response induction in the human urinary tract. J Infect Dis 183, S43–S46. Xia, Y., Gally, D., Forsman-Semb, K. & Uhlin, B. E. (2000). Regulatory
cross-talk between adhesin operons in Escherichia coli: inhibition of type 1 fimbriae expression by the PapB protein. EMBO J 19, 1450–1457.
Bacteria–host communication: the language of hormones. Proc Natl Acad Sci U S A 100, 8951–8956.
Zhang, J. P. & Normark, S. (1996). Induction of gene-expression in
Stenqvist, K., Sandberg, T., Lidin-Janson, G., Orskov, F., Orskov, I. & Svanborg-Eden, C. (1987). Virulence factors of Escherichia coli in
Zhao, H., Li, X., Johnson, D. E., Blomfield, I. & Mobley, H. L. T. (1997). In
urinary isolates from pregnant women. J Infect Dis 156, 870–877.
http://jmm.sgmjournals.org
Escherichia coli after pilus-mediated adherence. Science 273, 1234–1236. vivo phase variation of MR/P fimbrial gene expression in Proteus mirabilis infecting the urinary tract. Mol Microbiol 23, 1009–1019.
593
