Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Some Types of Bacterial Phase Variation Are Epigenetic Methylating DNA sites can control phase variation, enabling bacteria to inherit a gene expression state that remains reversible Marjan W. van der Woude n so many bacterial experiments, assuming that all the individual cells of Escherichia coli or other bacterial species in a particular flask are identical in terms of genotype and phenotype is reasonable and usually works well. Indeed, accepting this premise led several generations of microbiologists to study many microbiological properties and to uncover important biological principles, including metabolic pathways, the basics of nucleic acids and genes, and many mechanisms underlying their regulation and DNA replication. However, individual cells can vary phenotypically, even within a clonal population. Indeed, I Summary • Individual bacterial cells can vary phenotypically, even within a clonal population. • Phase variation provides a means for regulating specific genes between an expressed, or “on” state, and a nonexpressed, or “off” state; these two expression states are heritable and reversible. • The mechanisms that lead to phase variation are either genetic or epigenetic; if epigenetic, they entail no changes in DNA sequence. • More generally, phase variation apparently provides a means for generating diversity and thereby increasing the chances of survival amid changing environmental conditions. • Understanding phase variation may allow us to manipulate bacterial populations to provide health benefits, including better diagnostic tests and vaccines. F. W. Andrewes of St. Bartholomew’s Hospital in London, England, observed morphology variation between colonies of a Salmonella strain as early as 1922. Those differences later proved to be due to antigenic variation of the flagellum. The biological significance and mechanisms underlying this and other heterogeneities within particular bacterial populations are important features that microbiologists continue to study. Population Heterogeneity and Phase Variation The phenomenon described by Andrewes consists of variation in the antigenic properties of a particular protein. Although all cells in a population will express the protein, in some of those cells it has a different antigenic form. A second, related phenomenon that also generates heterogeneous phenotypes within a population is known as “phase variation.” This term describes a means for regulating genes that can switch a specific gene between an expressed, or “on” state (blue in Fig 1), and a nonexpressed, or “off” state (white in Fig 1). The expression state is heritable, yet reversible. Thus, cells in the “on” phase give rise to progeny that predominantly express that particular gene with only a minority that are switched “off” and do not express that gene, and vice versa. Therefore, phase variation results in a population of clonal cells that are not identical, yet can revert to each other’s phenotype (Fig. 1). The frequency of this switch can be as high as 1 per 10 cells per generation or as low as 1 per 10,000 cells. The latter frequency is low enough Marjan W. van der Woude is Senior Lecturer in Microbiology in the Department of Biology and the Hull York Medical School, University of York, United Kingdom. Volume 3, Number 1, 2008 / Microbe Y 21 FIGURE 1 that multiple generations are required to see changes in phenotype in the absence of selective pressures. However, this frequency is still more than 100 times higher than spontaneous mutation rates. The higher frequency and, equally importantly, the reversible nature of phase variation distinguish this hereditary change in gene expression from classical mutations. Furthermore, the molecular mechanisms leading to phase variation are distinct from those causing mutations. Molecular Mechanisms of Phase Variation May Be Genetic or Epigenetic Phase variation is a heritable yet reversible form of regulation and results in a heterogenous population. In the absence of selective pressures the two phenotypes will coexist in a population. (A) What can happen after the expression state of one of two daughter cells of an “off “cell has switched, and both grow to a colony. “On” cells are depicted blue and “off” cells white. (B) Colonies from a streak of a single white colony of a strain in which the agn43 phase-varying promoter drives the lacZ reporter fusion. Colonies with predominantly “on” cells are blue, and colonies with predominantly “off” cells are white on this medium containing the substrate Xgal for LacZ activity. (C) Typical result of streaking an “on” (blue) colony and an “off” (white) colony for single colonies. The plates clearly illustrate the bias in progeny phenotype as a result of parental phenotype. The number of generations during which this bias exists depends on the frequency of switching. 22 Y Microbe / Volume 3, Number 1, 2008 The mechanisms that lead to phase variation are classified either as genetic or epigenetic. If genetic, changes in DNA sequence account for differences in expression state between cells in the “on” and “off” phase. Such genetic changes may involve different molecular mechanisms (Table 1), including changes in reading frame of the coding sequence through slipped-strand mispairing or a flip in orientation of a promoter relative to the transcribed gene through site-specific recombination. In each case the sequence change is small and, importantly, reversible. Meanwhile, epigenetic phase variation involves no change in DNA sequence but is also reversible. Accounting for how a change can be heritable but reversible in the absence of DNA sequence changes has occupied several research labs for a number of years. Collectively, those researchers learned that differential methylation of only a few target DNA sequences accounts for this phenomenon. Specifically, deoxyadenosine methyltransferase (Dam) mediates phase variation to control expression of some outer surface structures in Escherichia coli. More generally, Dam is the main DNA methyltransferase of E. coli, and this enzyme modifies the adenine residue of the 4-bp recognition sequence GATC. A few of the GATC sequences in the E. coli genome control phase variation. David Low, now at the University of California at Santa Barbara, uncovered DNA methylation-dependent phase variation while studying the pilus operon (pap) in E. coli. He and his lab members determined that Dam and the global regulator Lrp are key players in regulating pap phase variation, along with several other proteins. Other fimbriae are regulated by homologous systems. A selection of bacterial phase variation mechanisms and some examples of species in which these occur Phase variation classification Molecular Mechanism Genetic Epigenetic Phase Variation: DNA Methylation-Dependent Ag43 Phase Variation Another example of Dam-dependent phase variation is Ag43, an outer-membrane protein that is involved in biofilm formation. Ag43 from E. coli K12 is being further studied to address questions regarding methylationdependent epigenetic regulation, including stochastic events. Borge Diderichsen at CERIA in Belgium and his collaborators recognized phase variation in the Ag43-encoding gene in 1980, naming the locus flu for “fluffing” because changes in Ag43 expression triggered cells to aggregate. They also noted that flu appeared to be “metastable.” Nearly two decades later, Ian Henderson, working with Peter Owen at Trinity College in Dublin, Ireland, and now at the University of Birmingham in England, renamed the gene agn43 and showed that this phase variation requires Dam and the regulatory protein OxyR. A 76-bp stretch in the regulatory region of agn43 contains three sequence elements that are essential for Ag43 phase variation (Fig. 2A). Ag43 phase variation is a result of regulated transcription, and the first essential element is the promoter, which is 70-dependent and has a sequence that is very similar to a consensus promoter region. Second, this 76-bp stretch contains three Dam target sequences, each consisting of 4 bp, GATC. Dam methylates the adenine residue at these GATC sequences (Fig. 2). Third, the 76-bp stretch contains a site to which the protein OxyR binds; this global regulator mainly coordinates expression of genes in response to oxidative stress. At agn43 the OxyR binding site overlaps the three GATC sequences as well as the -10 sequence of the promoter (Fig. 2); when OxyR binds, transcription is repressed. The conservation of these three sequence components in an otherwise variable stretch of DNA at agn43 genes of different E. coli isolates is striking and suggests that phase variation is a conserved feature. Slipped-strand mispairing Bacterial species Haemophilus influenzae, Neisseria meningitidis, Helicobacter pylori Site specific recombination Escherichia coli, Bacteroides fragilis Epigenetic Recombination Streptococcus pneumoniae, Neisseria meningitidis DNA methylation Escherichia coli, Salmonella enterica Phase variation occurs when OxyR binds agn43 or if Dam gains access and methylates its three unbound target GATC sequences within this site (Fig 2B, top). In the absence of OxyR binding, Dam methylates its target sequences, and this methylation blocks subsequent OxyR binding. OxyR can bind to unmethylated sites, however, in turn blocking Dam access. Since OxyR represses transcription of this gene, agn43 expression is off when OxyR is bound. In contrast, when the GATC sequences are methylated, that gene is expressed and such cells produce Ag43. After cells divide, daughter cells are biased to the parental Ag43 expression state, reflecting the inherited methylation state of the agn43 GATC sequences. Because switching appears to be mainly stochastic, determining whether other specific cellular events contribute to this infrequent event remains a challenge. E. coli cells contain no enzyme to remove methyl groups from their DNA. Thus, switching from “on” to “off” depends on DNA synthesis during replication to generate unmethylated GATC sequences (Fig. 2B). Dam acts on hemimethylated as well as nonmethylated DNA, and hemimethylated DNA in cells tends to become fully methylated. However, OxyR can also bind to hemimethylated agn43 GATC sequences, blocking Dam access, and would confer an “off” phenotype on such cells (Fig 2B). Changes in relative concentrations of Dam and OxyR can bias the switch, according to genetic experiments. Specifically, increasing Dam or decreasing OxyR levels biases cells to the “on” phase, and vice versa. Further, mutating the first GATC sequence, rendering this site unmethylated, also biases cells to the “off” Volume 3, Number 1, 2008 / Microbe Y 23 the “on” phase. The OxyR regulon mediates a specific protective response to oxidative stress, changing the oxidation state of OxyR and its DNA binding recognition sequence. This finding seems consistent with Ag43 phase variation helping cells to withstand oxidative stress through inducing Ag43-dependent aggregation. However, both oxidized and reduced OxyR bind to the agn43 region in a DNA methylation-dependent manner, repressing agn43 transcription. Furthermore, applying oxidative stress to a culture does not increase the percentage of Ag43 “on” cells. This absence of environmental signaling from the OxyRdependent mechanism could mean that OxyR was “hijacked” with Dam to generate constant phenotypic diversity. Besides Dam and OxyR, no other transcription factors are required for regulating agn43, while no environmental signaling alters agn43 phase variation. Thus, a random event—phase variation— dictates whether individual cells produce Ag43. FIGURE 2 ON Special Significance of Epigenetic Phase Variation (A) agn43 regulatory region showing the sequence elements that are essential for phase variation and their organization (not to scale). (B) Events during and after DNA replication that contribute to a change or maintenance of the agn43 “on” expression state. Competition between OxyR and Dam for free hemimethylated DNA is a key event. Newly synthesized DNA that is unmethylated initially is depicted in red. The black cross indicates that methylation by Dam or OxyR binding can not occur. Other symbols are explained in the Key box. phase, while mutating the second or third GATC sequence locks cells in the “off” phase. These data support the notion that Dam and OxyR compete for the agn43 regulatory region and that their relative concentrations are important for controlling phase variation. However, no natural conditions seem to change the Dam or OxyR concentration. One way in which E. coli cells cope with environmental stress, including oxidative stress, is to form aggregates. Extensive clumping occurs in stressed cultures whose cells are fully in 24 Y Microbe / Volume 3, Number 1, 2008 To determine whether phase variation regulatory mechanisms confer survival advantages, it would behoove us to know whether phase variation itself is advantageous, an issue that remains problematic. Phase-varying genes mostly encode proteins that directly or indirectly affect bacterial surface structures, which, for pathogens, are exposed to host immune systems. When genes encoding such structures are switched off, the bacteria can more effectively “hide” from adaptive immune responses than when those same genes are on and producing those telltale surface proteins. Hence, some investigators consider phase variation a mechanism for evading host immune responses. Furthermore, “off” settings for phasevariable proteins might also help pathogens to evade innate immune responses, which some of those surface border proteins can trigger. However, this “immune evasion” hypothesis regarding phase variation tends to focus too narrowly on pathogens. Phase variation instead might be a general means for generating diversity and thereby increasing the chances of sur- vival amid changing environmental condiFIGURE 3 tions, according to some investigators such as Richard Moxon at the University of Oxford in England, who considers such genes as belonging to “contingency loci.” Of course, both the “off” and “on” phases might enhance survival, depending on the situation and role played by a particular phase-variable protein. It proves difficult to determine whether phase variation affects virulence. For example, in studying cystitis and pyelonephritis in mice, Harry Mobley and colleagues at the University of Michigan in Ann Arbor concluded that, although type I E. coli fimbriae are a virulence factor, phase variation of those fimbriae is not. However, phase variation allows several Salmonella serovars with The same level of gene expression in a population can underlie different gene expression patterns in individual cells. This illustrates a drawback of methods that multiple fimbrial genes in common to coexist average the expression level from a population including microarrays. Here the in an individual host animal, which those shades of grey indicate the level of expression of a gene: (A) A completely serovars could not do if they all expressed homogenous population as a result of uniform gene expression is depicted (B) a heterogeneous population with a gradient in the level of gene expression, for every fimbrial type, according to Andreas example as a result of stochastic variation in the concentration of a regulatory Baümler and colleagues, now at the Univerprotein; (C) heterogeneity with “on” and “off” cells due to phase variation. sity of California, Davis. Thus, biological context affects how phase variation contributes to virulence or bacteare mutually exclusive. However, predicting rial survival. Nevertheless, one can speculate DNA sites where regulatory proteins bind is still that different environmental conditions gave a challenge, as is predicting whether or what rise to very different mechanisms for regulating residue changes block such binding. phase variation. Although both the epigenetic An additional complication is that the DNA system of pap and site-specific recombination binding of some regulatory proteins is not alsystem of type 1 fimbriae are well suited for tered by methylation, even if binding blocks responding to environmental signals, slipped Dam access. For instance, when the repressor of strand mispairing is not. the glucitol utilizing operon, GutR, binds to that segment of DNA, its footprint includes an unHow Common Is Methylationmethylated target sequence. Its binding, howDependent Phase Variation? ever, does not correlate with DNA methylationdependent regulation of any kind. Indeed, most Dam in conjunction with either the global reguof the several dozen unmethylated GATC selator Lrp or OxyR mediates epigenetic phase quences among the total of 18,000 in the E. coli variation in E. coli and Salmonella, as elucidated chromosome are not associated with methylafor pap and other fimbriae and for agn43, retion-dependent regulation. spectively. Meanwhile, genomes of species beFinally, there is no reason for thinking that longing to at least five orders of the ␥-proteobacDNA-methylation-dependent phase variation teria also encode a Dam homolog, and several can be conferred only by Dam, and not by other also encode OxyR and Lrp. DNA methyltransferases. The only known preConceivably, Dam-dependent phase variation requisite is that they may not be part of a funcsystems similar to those controlling pap and tioning restriction modification system because agn43 occur in these species. Furthermore, an unmethylated target sequence would render Dam-dependent phase variation could involve the cell unviable in the presence of the cognate regulatory proteins other than OxyR and Lrp, as restriction enzyme. Thus, the extent of methyllong as binding of these proteins to DNA and ation-dependent phase variation is not known. methylation at the specific phase variation sites Volume 3, Number 1, 2008 / Microbe Y 25 Phase Variation May Affect Research Results as well as Diagnostics and Vaccines Being Developed No matter how it arises, phenotypic heterogeneity can dramatically affect the outcome of microbiological experiments and how they are interpreted. Phase variation is no exception (Fig. 3). If gene expression in a particular cell population is determined, for example, with an enzymatic assay of a reporter gene or a microarray, a population average is measured. Although the data would indicate the same level of expression in all three cases, individual cells within the population would be registering very different gene expression patterns in each case. Thus, single-cell analyses might be needed to identify this sort of heterogeneity. Understanding phase variation may allow us to manipulate bacterial populations to provide human health benefits. Moreover, phase variation should be taken into account when devel- oping diagnostic tests and vaccines. For instance, diagnostic tests that include serotyping may be sensitive to biochemical changes on the bacterial cell surface that reflect phase variation. Meanwhile, phase variation that leads to immune evasion might undermine the efficacy of some candidate vaccines or therapeutic agents. Analyzing the molecular mechanisms underlying phase variation can yield insights and could help us to better understand the role of how stochastic events influence the dynamics of heterogeneity in cell populations. On a larger scale, we are learning to appreciate how bacteria thrive in diverse environments, sometimes relying on minor variants within the greater natural population. Thus, identifying additional, heterogeneously expressed genes and understanding how such heterogeneity is generated and how it can prove beneficial will deepen our understanding of how microbial populations behave. ACKNOWLEDGMENTS I thank my lab members and Dr. Thomas, Chong, and Pohlschroder for critical reading of this article. Work in my lab has been supported by grants from the National Science Foundation, a Marie Curie International Re-integration grant, the Biotechnology and Biological Sciences Research Council, and the Wellcome Trust. SUGGESTED READING Casadesus, J., and D. Low. 2006. Epigenetic gene regulation in the bacterial world. Microbiol. Mol. Biol. Rev. 70:830 – 856. Hernday, A., M. Krabbe, B. Braaten, and D. Low. 2002. Self-perpetuating epigenetic pili switches in bacteria. Proc. Natl. Acad. Sci. USA 99:16470 –16476. Hernday, A. D., B. A. Braaten, G. Broitman-Maduro, P. Engelberts, and D. A. Low. 2004. Regulation of the pap epigenetic switch by CpxAR: phosphorylated CpxR inhibits transition to the phase ON state by competition with Lrp. Mol. Cell.16:537–547. Lim, H. N., and A. van Oudenaarden. 2007. A multistep epigenetic switch enables the stable inheritance of DNA methylation states. Nature Genet. 39:269 –275. Moxon, R., C. Bayliss, and D. Hood. 2006. Bacterial contingency loci: the role of simple sequence DNA repeats in bacterial adaptation. Annu. Rev. Genet. 40:307–333. van der Woude, M. W., and A. Baumler. 2004. Phase variation and antigenic variation in bacteria. Clin. Microbiol. Rev. 17:581– 611. van der Woude, M. W. 2006. Re-examining the role and random nature of phase variation. FEMS Microbiol. Lett 254:190 –197. Wallecha, A., V. Munster, J. Correnti, T. Chan, and M. W. van der Woude. 2002. Dam- and OxyR-dependent phase variation of agn43: essential elements and evidence for a new role of DNA methylation. J. Bacteriol. 184:3338 –3347. 26 Y Microbe / Volume 3, Number 1, 2008