The histone-like nucleoid structuring (H-NS) protein is a major component of the folded chromosome in Escherichia coli and related bacteria. Functions attributed to H-NS include management of genome evolution, DNA condensation, and transcription. The wide-ranging influence of H-NS is remarkable given the simplicity of the protein, a small peptide, possessing rudimentary determinants for self-association, hetero-oligomerisation and DNA binding. In this review, I will discuss our understanding of H-NS with a focus on these structural elements. In particular, I will consider how these interaction surfaces allow H-NS to exert its different effects.
The histone-like nucleoid structuring (H-NS) protein is a 15.5 kDa peptide conserved in Escherichia coli and closely related bacteria . Originally named protein H1, H-NS was identified as an abundant component of cell extracts that could be isolated on the basis of its DNA-binding activity [2–5]. Subsequent work has shown that H-NS binds AT-rich DNA by the initial recognition of a high-affinity nucleation site, distinguished by the presence of a T:A step, followed by oligomerisation across adjacent AT-rich DNA [1,6–10]. In addition, H-NS can target RNA and influence translation initiation [11,12]. However, the rules for binding this nucleic acid have not been defined. The intracellular abundance of H-NS suggests an impact on many transactions involving DNA [13–16]. Indeed, H-NS has been shown to influence transcription, DNA folding, and genome evolution [13–16]. These different activities are intimately related and often difficult to separate. Hence, it is not clear whether H-NS has a single primary function, with other activities being incidental, or whether H-NS has evolved to simultaneously manage different aspects of cell biology. In this review, I will focus on the relationship between the structure and various functions of H-NS.
Oligomerisation and DNA binding by H-NS
Structural properties of H-NS
The E. coli H-NS protein is 137 amino acids in length and consists of two domains . The domains are joined by a flexible linker (Figure 1A). The N-terminal domain (NTD) is 83 amino acids in length and contains four α-helices (α1, α2, α3, and α4). These permit self-association by ‘head-to-head’ and ‘tail-to-tail’ contacts [17–20]. The head-to-head interaction requires helices α1 to α3. The contact is stabilised by the formation of a coiled-coil, involving a sub-section of α3 from each promoter [17–19]. The coiled-coil has been observed in parallel and antiparallel configurations [17–19]. However, the biologically relevant state is unclear [20,21]. The tail-to-tail interaction is mediated by the C-terminal portion of helix α3 and helix α4. These fold to form a helix-turn-helix motif. Two copies of the motif can interlock in an antiparallel orientation. Crucially, the head-to-head and tail-to-tail interactions can occur simultaneously. This allows H-NS to oligomerise and form filaments . The helix-turn-helix is followed by a linker and the C-terminal domain (CTD) that binds DNA. This domain encompasses two β-sheets (β1 and β2), an α-helix (α5), and a 310 helix . The key determinant for DNA binding is a loop located between β2 and the α-helix-containing side-chain R114 (underlined) within the highly conserved sequence ‘TWTXGRXP’. These amino acids form an AT-hook motif that presents R114 to the minor grove of the DNA . This interaction is favoured by AT-rich sequences that exhibit narrowing of the minor groove. Many Gram-negative bacteria encode H-NS-like proteins identifiable by the AT-hook motif described above . Some of these proteins are homologues of E. coli H-NS (e.g. VicH of Vibrio cholerae), while others serve a different purpose (e.g. Ler, an antagonist of H-NS found in enterohaemorrhagic E. coli). Indeed, many Gram-negative bacteria encode several H-NS-like proteins . As discussed in more detail below, cross-talk between these factors often plays a key role. The situation is less well defined in Gram-positive bacteria and mycobacteria; analogues of H-NS exhibit similar DNA-binding properties, but structural similarity is limited beyond the AT-hook motif [22,24].
Oligomerisation and DNA binding by H-NS.
Topology of complexes formed between H-NS and DNA
Oligomerisation of H-NS with DNA results in the formation of nucleoprotein filaments [25,26]. These filaments can be bridged or linear in organisation. Hence, H-NS can compact DNA or unfurl DNA into an extended conformation (Figure 1B). While such structures have been visualised using atomic force microscopy, and can be detected using a variety of other biophysical techniques, their precise organisation remains somewhat speculative [25–27]. Current models are derived from structural characterisation of the H-NS NTD . Here, consecutive head-to-head and tail-to-tail interactions allow H-NS filamentation in the absence of DNA. Consequently, AT-hook motifs, from alternate H-NS protomers, are well positioned to mediate DNA bridging (Figure 1Bi). Presumably, in linear filaments, consecutive AT-hooks bind adjacent sections of the double helix (Figure 1Bii). Conceivably, this could involve reorganisation of the coiled-coil that mediates the head-to-head interaction [17–20]. This model has been mooted because switching between bridged and linear filaments is responsive to the availability of divalent cations and possibly temperature [25–27]. These factors may also determine the nature of the head-to-head interaction [18–21].
H-NS as a modulator of transcription initiation
It is likely that H-NS has an impact on many processes that require access to the DNA template. Of the many possibilities, transcription initiation has been studied most intensely. Briefly, transcription initiation involves binding of RNA polymerase to a DNA sequence known as a promoter . This is followed by DNA unwinding and creation of a nascent RNA chain. Initial binding of RNA polymerase is mediated by determinants that contact sequence elements in the promoter DNA . Thus, the dissociable RNA polymerase σ70 subunit contacts the promoter −35 (5′-TTGACA-3′) and −10 elements (5′-TATAAT-3′). At some promoters, the α subunit CTD (αCTD) contacts AT-rich sequences further upstream . By constraining DNA topology, H-NS can influence these interactions. This was first demonstrated at the proU and bgl loci that are both subject to repression by H-NS [29,30]. Notably, repression often requires interactions between distal H-NS nucleation sequences either side of the promoter [31,32]. However, it is not clear whether this represents a common mode of repression or whether H-NS exerts its effects via a range of mechanisms. The sections below outline the different models for transcription repression by H-NS that have been documented to date.
RNA polymerase trapping
Interactions between distal H-NS binding targets could drive the formation of a repression loop, a nucleoprotein structure whereby RNA polymerase is trapped and unable to transition to the elongation phase of transcription (Figure 2A). This mode of transcriptional regulation has been demonstrated at a few promoters using biophysical approaches. Hence, an ensnared RNA polymerase can be visualised directly at promoters for the transcription of the hdeAB and rrnB operons [33,34]. In some cases, RNA polymerase acts as a cofactor for looping by bending promoter DNA sandwiched between sites of H-NS binding . Hence, H-NS and the stalled RNA polymerase can be considered as co-repressors of transcription.
H-NS as a modulator of transcription initiation.
Repression by H-NS can, sometimes, be circumvented by RNA polymerase, associated with alternative σ factors [35–37]. For example, the single E. coli dps promoter can be shared by the housekeeping (σ70 associated) and starvation phase (σ38 associated) forms of RNA polymerase . Hence, H-NS binds targets overlapping the dps promoter and prevents the binding of the σ70 RNA polymerase derivative . Conversely, σ38-bound RNA polymerase is insensitive to the presence of H-NS, and co-binding is observed . Importantly, this ternary complex is competent for transcription initiation. A different mechanism was reported at the E. coli hdeAB promoter . In this instance, DNA bending by σ70 RNA polymerase facilitates repression loop formation. For σ38-bound RNA polymerase, bending is less pronounced and repression does not occur .
Repression by direct RNA polymerase interaction
The RNA polymerase αCTD is connected to the αNTD by a flexible linker [38,39]. This allows αCTD to bind AT-rich DNA tracts (UP elements) at various positions upstream of the core promoter . Coincidentally, as for H-NS, αCTD binding requires an interaction between an arginine side-chain and the DNA minor groove . The process can be modulated by transcription factors that simultaneously interact with DNA and αCTD [41,42]. Recently, such an interaction involving H-NS was found to repress transcription . In this example, H-NS binds AT-rich DNA > 114 bp (base pairs) upstream of the LEE5 transcription start site. Oligomerisation of H-NS then permits ‘spreading’ to a promoter proximal site and a specific contact with αCTD. This prevents isomerisation of the LEE5 promoter .
RNA polymerase and transcription factor exclusion
All of the mechanisms described above involve co-binding of H-NS and RNA polymerase to the DNA. However, H-NS may also act to exclude RNA polymerase and gene regulatory proteins from certain targets. Most likely, this requires formation of linear nucleoprotein filaments or bridged filaments lacking large sections of H-NS free DNA. There is restricted biochemical evidence for this model since most promoters exhibit co-binding of H-NS and RNA polymerase in vitro [33–35,43]. Instead, the exclusion model is derived from global chromatin immunoprecipitation analyses. Thus, H-NS occludes certain promoters and transcription factor binding sites [44,45].
Interaction with the nascent RNA transcript
Although not mediated by direct transcriptional control, recent work has shown that H-NS can activate the expression of some genes. This requires direct interaction with the nascent RNA transcript and results in enhanced translation . Thus, numerous mRNAs with a suboptimal ribosome-binding site recruit H-NS to their leader region . By binding in this location, H-NS facilitates correct positioning of the ribosome and increases gene expression.
H-NS modulates transcription elongation
RNA polymerase engaged in transcription elongation must also encounter H-NS bound DNA . Whether such nucleoprotein filaments constrain transcription elongation, or are themselves disrupted by passage of the transcriptional apparatus, is poorly understood. Kotlajich et al.  monitored the transit of RNA polymerase across a section of the bgl operon. It was shown that the effects of H-NS on transcription elongation varied according to the organisation of the H-NS:DNA complex. Thus, linear H-NS:DNA filaments allowed transcription elongation to proceed unhindered. Conversely, bridged complexes induced RNA polymerase pausing (Figure 3). Data obtained in vivo suggest that transcription elongation is prone to premature termination when H-NS-bound DNA is encountered [48–50]. This termination is likely mediated by ρ, a factor that binds to the nascent RNA, and disrupts transcription elongation [51,52]. It is thought that Rho acts more efficiently as the rate of transcription elongation is reduced [48–52].
H-NS as a modulator of transcription elongation.
Anti-silencing of H-NS-repressed genes
Anti-silencing by H-NS-like proteins
As noted above, a variety of H-NS-like proteins have been identified in the genomes of Gram-negative bacteria. Some of these are required for maximal expression of H-NS-bound genes. Well-characterised examples include H-NST of enteropathogenic E. coli and Ler of enterohaemorrhagic E. coli [53,54]. Such proteins appear to modulate complexes formed between DNA and H-NS [53,54]. For example, Ler preferentially binds to the same AT-rich DNA sequences as H-NS and in doing so can displace H-NS from the DNA [15,55,56] (Figure 4A). This may occur because of differences in the NTD of Ler that hinder cooperative DNA-binding and direct interactions with H-NS . In the case of H-NST, heterodimers, formed by interaction with H-NS, prevent nucleoprotein filaments being established . This occurs because H-NST makes head-to-head contacts with H-NS, but is unable to participate in tail-to-tail contacts . Consequently, H-NST ‘poisons’ H-NS filamentation (Figure 4B). Note that, in Figure 4, Ler and H-NST are depicted as disrupting bridged H-NS:DNA complexes. It is equally possible that linear filaments of H-NS and DNA are targeted.
Anti-silencing by H-NS-like proteins.
Anti-silencing by canonical transcription factors
Canonical transcription factors recognise specific DNA sequences that are usually located close to a promoter [28,57]. Depending on the juxtaposition of the promoter and transcription factor binding site, this can result in the up- or down-regulation of transcription . Transcription factors that repress usually occlude the promoter while activators binding further upstream do not [28,57]. Transcription factors can also disrupt the binding of H-NS. This activity was first demonstrated by introducing targets for well-characterised repressor proteins upstream of the H-NS silenced bgl operon in E. coli. Thus, by disrupting the binding of H-NS, proteins that usually function as repressors can stimulate transcription [58,59]. Intriguingly, DNA-folding proteins, such as the integration host factor, may also have anti-silencing activity .
Anti-silencing by altered DNA topology
Linear and bridged H-NS filaments suppress and promote plectoneme formation during DNA supercoiling, respectively . Hence, the reverse could be true; the formation of bridged and linear filaments could be controlled by changes in DNA topology. For example, H-NS:DNA interactions are facilitated by optimal minor groove width, and correct relative positioning of H-NS monomers is required for filament formation. These parameters are both influenced by over- and under-winding of the DNA [62,63]. Speculatively, such changes in DNA topology could modulate transcription by altering the characteristics of nucleoprotein filaments containing H-NS.
Horizontally acquired DNA and transcriptional specificity
Over the last 10 years, genome-scale approaches have allowed global studies of transcription modulation by H-NS [6–8,44,45,64,65]. These studies show that H-NS predominantly blocks transcription at horizontally acquired loci, with a high AT-content, and in doing so may facilitate the incorporation of such genes into the chromosome [1,6]. The observation that horizontally acquired genes are transcribed more frequently in cells lacking H-NS was assumed to result from mRNA synthesis. However, it is now clear that the situation is much more complex [66–70]. In particular, AT-rich genes contain many sequences that resemble promoter elements. Consistent with this, recent work has shown that the majority of promoters, repressed by H-NS, are actually located within open reading frames or far from gene starts [68,69]. This may explain why AT-rich DNA is toxic, and why H-NS can counteract this. Hence, these studies revealed that a key function of H-NS is to ensure transcriptional specificity. For example, at the ehxCABD operon from E. coli O157:H7, H-NS blocks transcription from spurious promoters within genes and, in doing so, facilitates correct recognition of a canonical promoter in the upstream gene regulatory region .
The H-NS protein is a small peptide that can self-associate and bind DNA. This simplistic organisation permits remarkably diverse effects on bacterial chromosomes. An ongoing challenge is to understand the precise organisation of the nucleoprotein filament containing H-NS in vivo. It is likely that this will reveal how silencing is overcome, and how specific transcription of H-NS bound genes is achieved while avoiding concomitant production of spurious transcripts from intragenic promoters. For example, it is possible that H-NS:DNA filaments could adopt a conformation that allows transcription elongation but permits initial binding of RNA polymerase to promoters within genes. Importantly, addressing these challenges will require an integrated approach bringing together the field's expertise in genetics, genomics, structural biophysics, and modelling .
Work in my laboratory is currently funded by the BBSRC [grants BB/N014200/1, BB/N005961/1, and BB/J006076/1], the Leverhulme Trust [RPG-2013-147], and the Human Frontiers Science Program [RGP0014/2014].
The Author declares that there are no competing interests associated with this manuscript.
I thank Charles Dorman and Remus Dame for their helpful comments on this manuscript.
David C. Grainger was awarded the Biochemical Society's Colworth Medal in 2016; this review is based on the Award Lecture.