Correspondence: Charles J. Dorman, Department of Microbiology, Moyne Institute of Preventive Medicine, School of Genetics and Microbiology, Trinity College, Dublin 2, Ireland. Tel.: +353 1 896 2013; fax: +353 1 679 9294; e-mail: email@example.com
Horizontal DNA transfer plays a major role in the evolution of bacteria. It allows them to acquire new traits rapidly and these may confer fitness advantages as the bacteria compete with others in the environment. Historically, the mechanisms of horizontal DNA transfer, chiefly conjugation, transformation and transduction, have received a great deal of attention. Less attention has been focused on the regulatory problems that may accompany the acquisition of new genes by lateral routes. How are these genes integrated into the existing regulatory circuits of the cell? Does a process of ‘plug-and-play’ operate, or are the new genes silenced pending the evolution of regulatory mechanisms that make their expression not only safe but also beneficial to both the gene and its new host? Recent research shows that bacterial nucleoid-associated proteins such as H-NS, HU and Fis are important contributors to the processes of regulatory integration that accompany horizontal gene transfer. A key emerging theme is the antagonism that exists between the DNA–protein–DNA bridging activity of the H-NS repressor and the DNA-bending and DNA-wrapping activities of regulatory proteins that oppose H-NS.
Horizontal transfer of genetic material between bacteria has been studied intensively for decades and we now have a deep knowledge at a molecular level of the processes of conjugation, transduction and transformation. In addition, we have a growing appreciation of the importance of these processes for the evolution of bacteria. Lateral genetic transfer assists bacteria to maintain their genomes by providing material for DNA repair via homologous recombination (Arnold et al., 2008). It can import additional copies of resident genes that can then diversify to form new paralogues with novel properties. Newly acquired genes may encode useful traits such as resistance to antimicrobial agents (including antibiotics), virulence factors, symbiosis factors or novel metabolic processes (Dobrindt et al., 2004; Hensel, 2004; Gal-Mor & Finlay, 2006; Qui et al., 2006; Ayub et al., 2007; Becq et al., 2007; Bourhy et al., 2007; Coburn et al., 2007; Hubber et al., 2007; Ubeda et al., 2007). In each of these cases, a balance must be struck between the benefits offered by the new genes and the possibility that they may compromise the competitive fitness of the bacterium. It was discovered recently that the nucleoid-associated protein H-NS plays an important role in the process of silencing horizontally acquired genes in the Gram-negative bacteria Escherichia coli and Salmonella enterica serovar Typhimurium (S. Typhimurium) (Grainger et al., 2006; Lucchini et al., 2006; Navarre et al., 2006, 2007; Oshima et al., 2006; Dorman, 2007a; Wade et al., 2007). The hypothesis that regards H-NS as a repressor of potentially harmful genes that the bacterium acquires by lateral mechanisms is an attractive one (Navarre et al., 2007). It raises important questions about the subsequent expression of the silenced genes and how this is to be regulated in ways that benefit the bacterium. These issues will be considered here, using examples from Gram-negative pathogens, to illustrate the main principles.
The hns genetic locus was identified in the late 1980s (Higgins et al., 1988) and found to be allelic with many other regulatory loci that had been mapped to the equivalent positions on the E. coli and S. Typhimurium genetic maps over several years (Higgins et al., 1988; May et al., 1990). These mapping data and subsequent genetic analysis showed that hns is a highly pleiotropic regulatory locus whose principal influence on gene expression is a negative one. Cloning and DNA sequencing allowed the protein product of the hns gene to be identified as the nucleoid-associated protein H-NS (Hulton et al., 1990), a molecule that had been studied for many years before the identification of the hns gene (Jacquet et al., 1971; Cukier-Kahn et al., 1972). H-NS is a dimeric nucleic acid-binding protein. It has a dimerization domain in its amino terminus that is connected by a flexible linker to a nucleic acid-binding domain that is located in the carboxyl terminus (Bertin et al., 1999; Dorman et al., 1999; Dorman, 2004). The detailed structure of the protein has not been solved for the whole molecule and there is some controversy about the relative orientation of the two monomers within the dimer (Esposito et al., 2002; Bloch et al., 2003; Cerdan et al., 2003; Dorman, 2004; Ono et al., 2005). Nevertheless, it is safe to assume that H-NS dimers possess two domains capable of interacting with DNA and this allows the protein to form DNA–H-NS–DNA bridges (Fig. 1). The existence of these bridges has been confirmed by single-molecule biophysical studies, and bridging is central to the biological role of the H-NS protein (Dame et al., 2005, 2006; Dorman, 2007b; Stoebel et al., 2008).
DNA bridging and gene regulation
H-NS is not the only DNA-binding protein that is able to form bridges. For example, the LacI protein from E. coli that represses transcription of the lac operon also has this property (Wang et al., 2005). Here, the protein is a tetramer and it can form bridges between pairs of cis-acting regulatory DNA elements called operators. The resulting bridges are very stable and can blockade the lac promoter, resulting in transcriptional repression. The force required to break the DNA–LacI–DNA bridge, 500 pN, is well beyond that which is achievable by a moving RNA polymerase (Wang et al., 2005). The bridges formed by H-NS are somewhat more fragile, requiring 5 pN to break (Dame et al., 2006). Nevertheless, they allow H-NS to repress very many promoters. It is not clear whether bridge formation is needed in every case, but where the matter has been examined in detail, the available data are consistent with an important role for this type of architecture in the H-NS–DNA complexes.
Targets of H-NS-mediated repression
H-NS plays an important role in the repression of genes that have been acquired by horizontal transfer, including the major virulence genes of the Gram-negative bacteria Salmonella, Shigella and Yersinia, among others (Stoebel et al., 2008). The target DNA sequences are A+T-rich, something that marks them out from the core genome in these members of the Enterobacteriaceae, and are associated with regions of intrinsic curvature (Lucchini et al., 2006; Navarre et al., 2006). Curved DNA sequences are likely to make good targets for H-NS binding because they lend themselves to bridge formation. Both curvature and A+T richness are features of bacterial promoters (Pedersen et al., 2000); hence, H-NS does not exclusively target DNA of foreign origin that has been acquired by lateral transfer. For example, all seven of the rRNA operons of E. coli are repressed by H-NS (Pul et al., 2005). Recently, a high-affinity target site for H-NS binding to DNA has been identified (Bouffartigues et al., 2007). This sequence is also A+T-rich and may provide a nucleation site for spreading of the H-NS protein along the DNA, leading to more extensive bridging.
Xenogeneic silencing, the downregulation of the transcription of genes of foreign origin, presents the interesting problem of how those genes are ever to be expressed (Stoebel et al., 2008). The H-NS protein is present at a more-or-less constant level throughout the bacterial growth cycle. The cell seems to match H-NS levels to its chromosomal DNA content (Free & Dorman, 1995), but it may not always succeed, especially if horizontally acquired extrachromosomal DNA is involved (Doyle et al., 2007). This transient decline in the H-NS : DNA ratio may provide a window in which otherwise H-NS-repressed genes can be active.
H-NS can form heteromeric complexes with shorter proteins that are equivalent to the H-NS oligomerization domain but lack the nucleic acid-binding activity, such as H-NST, a protein from enteropathogenic E. coli (Williamson & Free, 2005; Baños et al., 2008). These may be expected to undermine bridge formation, and H-NST has the ability to attenuate H-NS-mediated repression (Williamson & Free, 2005). On the other hand, other partial paralogues such as Hha appear to reinforce the repressive activity of H-NS, at least at some promoters (McFeeters et al., 2007; Vivero et al., 2008). The underlying mechanism by which this happens is not clear.
Physical influences on DNA curvature may undermine the bridges that H-NS forms, and this is true of bridge-mediated repression complexes in Shigella that downregulate the promoters of its main virulence genes (Prosseda et al., 2004). This is because changes in temperature or osmolarity alter DNA curvature by displacement of the bend centre and by increasing or decreasing bend angles (Sinden et al., 1998; Badaut et al., 2002; Prosseda et al., 2004).
Antagonism of bridging activity
Both protein-dependent and protein-independent mechanisms for the relief of H-NS-mediated transcription repression can involve disruption of the DNA–H-NS–DNA bridge. How general might this be as an antirepression mechanism? The ability of the LacI repressor protein to form stable bridges between pairs of lac operators was referred to above, and other DNA-binding proteins such as AraC and LeuO are also capable of forming bridges between different parts of a DNA molecule (Harmer & Schleif, 2001; Chen & Wu, 2005; Chen et al., 2005). Alternative types of protein interactions with DNA might be expected to inhibit bridge formation or undermine existing bridges. For example, a DNA-bending protein or a DNA-wrapping protein that competes with the bridge former for the same DNA sequence might make an excellent antagonist. Most DNA-binding proteins introduce bends at their binding sites. However, there is an important subset that imposes very large distortions indeed. For example, the integration host factor (IHF) can bend DNA through angles of up to 180° (Swinger & Rice, 2004; Sugimura & Crothers, 2006). Interestingly, IHF can disrupt LacI-mediated bridges (Zurla et al., 2007). The proteins that have been shown to antagonize H-NS-mediated repression of transcription also have the ability to distort the path of the DNA helix. The binding of the VirB antirepressor to its target promoters in Shigella flexneri seems to involve the wrapping of DNA around VirB resulting in a loss of H-NS binding at the same promoters (Fig. 1) (Turner & Dorman, 2007). In principle, proteins such as HU, Fis and Lrp could also act to oppose the bridging activity of H-NS. Evidence from atomic force microscopy experiments supports the idea that HU can oppose H-NS (Dame & Goosen, 2002) and it does have strong DNA-bending activity (Swinger & Rice, 2007). Similarly, the Lrp leucine-responsive regulatory protein can alter the trajectory of the DNA helix in ways that would be predicted to undermine H-NS-mediated DNA bridging. Lrp can form multimeric structures of up to at least octamers and these are predicted to have a tyre-on-wheel wrapping-like complex with bound DNA (de los Rios & Perona, 2007). Significantly, Lrp targets a number of genes that are also bound by H-NS (O'Gara & Dorman, 2000; Pul et al., 2005; Kelly et al., 2006).
There is an interesting degree of correspondence in the sites that are bound by the Fis (factor for inversion stimulation) protein and H-NS along the chromosome of E. coli, as determined by chromatin immunoprecipitation experiments (Grainger et al., 2006) and Fis is known to oppose the repressor activity of H-NS at a number of promoters (Zacharias et al., 1991; Falconi et al., 1996). Fis dimers can bend DNA by up to 90° (Kostrewa et al., 1991), making Fis a good candidate to oppose the bridge-building activity of H-NS on adjacent DNA.
DNA supercoiling has the potential to influence the activity of a promoter directly either by modulating the formation of a closed transcription complex or by its isomerization to an open complex (Hatfield & Benham, 2002; Dorman, 2006). It is interesting to note that many of the genes that are repressed by the H-NS bridging protein are also sensitive to changes in DNA supercoiling (Higgins et al., 1988; Dorman et al., 1990; Tobe et al., 1995; Hurme & Rhen, 1998; Beltrametti et al., 1999; Bang et al., 2002; Blot et al., 2006; Becker et al., 2007; Bouffartigues et al., 2007). The same is true of other DNA-bridging proteins: it is interesting to note that LacI-mediated bridges are modulated by changes in the supercoiling of the participating DNA (Normanno et al., 2008) and that promoters controlled by the LeuO DNA-bridging protein are supercoiling-sensitive (Chen & Wu, 2005; Chen et al., 2005). While it is possible that the role of DNA supercoiling at these H-NS-repressed promoters may be confined to influencing the interaction of the DNA with RNA polymerase, it is equally likely that the interaction of the DNA with H-NS is also being affected. The plectonemic wrapping of DNA, which is a feature of the writhing that accompanies linking number reduction in negatively supercoiled DNA, lends itself to the facilitation of protein-mediated bridges. Similarly, changes in the superhelical density of the DNA could compromise the integrity of the bridges.
These considerations, together with the known responsiveness of DNA supercoiling to environmental signals such as temperature and osmolarity, suggest a simple mechanism for the fine-tuning of the transcription profile of the bacterium as it traverses a dynamic environment. The mechanism relies on the variability of two key genome features: DNA topology and a population of DNA-bridging proteins. It is so simple that it offers the cell a means of regulating newly acquired genes immediately. Incoming genes that have the necessary DNA sequence motifs to attract H-NS can be repressed; changes in the topology of the DNA that arise from environmentally induced changes in negative supercoiling can relieve this repression, giving the bacterium the opportunity to ‘try out’ the new gene. Similarly, antagonistic action by proteins such as Fis, IHF or HU that have anti-H-NS activity may also play a role in upregulating the newcomer. What of imported genes that do not fit into this regulatory regime? If they are not repressed and their expression is disadvantageous to the bacterium, competition with other, fitter, organisms will eliminate them. Those genes that possess the appropriate profile for successful regulatory integration can have their expression more finely tuned to the needs of the bacterium through the evolution of additional regulatory inputs.
The ‘plug-and-play’ scenario is unlikely in the case of H-NS-repressed genes that have been acquired laterally: a process of regulatory integration is likely to be required. This scenario is more likely to apply in the case of genes of immediate benefit, such as antibiotic resistance genes obtained by a bacterium living in the presence of the cognate antibiotic. Lateral transfer of genetic information is occurring constantly in bacterial populations in the wild. It is impressive that bacteria have developed mechanisms to solve the problems that accompany this everyday fact of their lives, and that they do so by relying on architectural elements of the nucleoid (nucleoid-associated proteins and DNA topology) to provide a basis for the solution.
We are grateful to Science Foundation Ireland for financial support.