Addressing the Possibility of a Histone-Like Code in Bacteria

Abstract

Acetylation was initially discovered as a post-translational modification (PTM) on the unstructured, highly basic N-terminal tails of eukaryotic histones in the 1960s. Histone acetylation constitutes part of the “histone code”, which regulates chromosome compaction and various DNA processes such as gene expression, recombination, and DNA replication. In bacteria, nucleoid-associated proteins (NAPs) are responsible these functions in that they organize and compact the chromosome and regulate some DNA processes. The highly conserved DNABII family of proteins are considered functional homologues of eukaryotic histones despite having no sequence or structural conservation. Within the past decade, a growing interest in N^ε-lysine acetylation led to the discovery that hundreds of bacterial proteins are acetylated with diverse cellular functions, in direct contrast to the original thought that this was a rare phenomenon. Similarly, other previously undiscovered bacterial PTMs, like serine, threonine, and tyrosine phosphorylation, have also been characterized. In this review, the various PTMs that were discovered among DNABII family proteins, specifically histone-like protein (HU) orthologues, from large-scale proteomic studies are discussed. The functional significance of these modifications and the enzymes involved are also addressed. The discovery of novel PTMs on these proteins begs this question: is there a histone-like code in bacteria?

Graphical Abstract

INTRODUCTION

Histones Regulate Chromosome Compaction and Gene Expression in Eukaryotes

In eukaryotes, meters-long linear stretches of DNA must be compacted in order to fit inside the ~6 μm nucleus of the cell. The DNA must be compacted in a way that allows for accurate replication, proper expression of genes, and correct repair, and therefore, it must remain accessible to various protein machines. This is achieved by wrapping the DNA molecule twice around a nucleosome, which is composed of two copies of each histone, H2A, H2B, H3, H4 (Figure 1A), or one of several histone variants.¹ The degree of chromatin compaction is correlated with the level of gene expression and is regulated by numerous factors, including chromatin remodeling proteins and biochemical post-translational modifications (PTMs) on the N-terminal tails of histones.^2–4 The first modifications discovered on histones were lysine acetylation and methylation and then phosphorylation in the 1960s.^5,6 Since then, many more PTMs have been identified, including multiple lysine acylations; arginine methylation and citrullination; lysine sumoylation and ubiquitination; ADP-ribosylation; proline isomerization; glycosylation; biotinylation; and serine, threonine, and tyrosine phosphorylation.^7,8

Figure 1.

Comparison of DNA organization and condensation in histones and HU orthologues. (A) In the nucleus, the DNA is wrapped approximately two times around a core of histone proteins, composed of two copies of each histone (H2A, H2B, H3, and H4) or a histone variant. Histone H1 serves to hold the wrapped DNA in place and stabilize linker DNA between nucleosomes. The wrapping of DNA around histones at regular intervals condenses and further organizes the chromatin. (B) In bacteria, HU family proteins, either hetero- or homodimers, bind nonspecifically along the chromosome and introduce large bends (105° to > 180°). DNA can also wrap around an HU dimer in a right-handed orientation. These two actions help organize and condense the circular bacterial chromosome analogously to histones.

The presence and combination of these epigenetic modifications are regulatory in nature, and it has been hypothesized that these PTMs influence chromatin structure and transcription. This regulatory process is called the “histone code”. These modifications act as signaling hubs and regulate the interaction of the DNA with histones and other accessory proteins and machines.⁹ This concept was originally proposed on the basis of observations that specific histone modifications, like serine 10 phosphorylation of H3 (H3S10), played a role in both chromatin decompaction and the opposite function of condensation.^10,11 One possible explanation of this phenomenon is that although the S10 phosphorylation mark alone is not sufficient to determine the chromatin state, perhaps it becomes important in the context of the overall PTM landscape. In other words, it may be the case that this mark works in combination with other PTMs on one or more histone tails. Possibly the PTMs have a direct effect on chromatin structure by influencing protein–DNA interactions. Additionally, various combinations of histone PTMs lead to the recruitment of chromatin remodeling or transcription factors, called “reader” proteins, which bind modified histones with specific functional consequences.^9,12 For example, the mammalian protein Brd2, which contains two acetyllysine binding domains called bromodomains, binds to acetylated lysine residues (K5, K8, and K12) in histone H4. Binding of the Brd2 reader protein results in the recruitment of the general transcription factor TATA-binding protein (TBP) to the E2F complex, which activates a subset of cell cycle genes.¹³

Bacterial Nucleoid-Associated Proteins Are Functional Equivalents of Histones

While bacteria contain circular chromosomes that are much smaller than their linear eukaryotic counterparts, bacterial chromosomes must still be compacted at least 1000-fold to fit inside the nucleoid region.¹⁴ Negative supercoiling accounts for one level of chromosome compaction, and the supercoils are further organized into loop domains that are topologically isolated from one another.¹⁵ The domains are in part defined by DNA interactions with small, basic, nucleoid-associated proteins (NAPs),¹⁶ which are essential for chromosome compaction and the coordination of DNA transactions. Bacterial NAPs bind either specifically or nonspecifically at sites distributed throughout the chromosome. They condense and organize the nucleoid by introducing bends and bridging chromosomal loci (as reviewed in ref 17).

The most widely conserved NAPs that are present in all eubacteria are the members of the DNABII family, also called the HU family.¹⁸ The DNABII family of proteins consists of orthologues of histone-like protein (HU) and integration host factor (IHF). These proteins are structurally and functionally related, with the distinction that only IHF binds with sequence specificity.¹⁹ This review focuses on the nonspecific DNA binding proteins, the HU orthologues. This family of proteins has been extensively studied in Escherichia coli. E. coli HU consists of two subunits, α and β, and exists mostly as a heterodimeric species.^20–24 It bends and condenses DNA into a fiber (Figure 1B)^25,26 and binds preferentially to single-stranded (ss) forks and 3′ overhangs or to certain double-stranded (ds) DNA structures (e.g., forks, kinks, junctions, or bends) as well as to RNA.^27–32 At least one function of HU is to form and constrain negative supercoils,³³ which leads to wrapping of DNA in a right-handed orientation for compaction.^26,33–35 HU has a similar amino acid content as eukaryotic histone H2B, and despite the lack of sequence or structural homology, this family of proteins are considered to be functional homologues of eukaryotic histones.³⁶ Like histones, HU regulates several DNA transactions such as replication initiation,^37–39 inversion,⁴⁰ bacteriophage Mu transposition,^41,42 and DNA repair.^43,44 Similarly to eukaryotic histones, there is evidence that HU regulates gene expression. Deletion of both subunits of HU affects the transcription of ~1500 genes, including those involved in translation and stress responses, implicating this protein in global transcriptional control.^45–47 However, E. coli HU mutants exhibit pleiotropic defects, leaving open the possibility that some of these transcriptional changes are not due to the absence of HU per se but are secondary consequences of the mutant phenotype, such as slow growth, prolonged lag phase, or defective cell division. Thus, the role of HU in gene expression is not completely understood.

With the constant improvements of mass-spectrometry-based proteomics, many bacterial PTMs, with lysine acetylation the most extensively characterized,^48–52 have now been cataloged on the global scale. Examination of these data sets led to the observation that many of the HU orthologues are post-translationally modified. In this review, the evolutionary conservation of N^ε-lysine acetylation of HU orthologues is explored. Furthermore, the functional significance of these modifications and their possible enzymatic regulation are addressed. Finally, the discovery of other PTMs on HU proteins is discussed, leading to the exciting possibility of the existence of a histone-like code in bacteria.

N^ε-LYSINE ACETYLATION OF HU PROTEINS IS AN EVOLUTIONARILY CONSERVED PHENOMENON

Prior to the year 2000, only a few examples of acetylated bacterial proteins were known, and N^ε-lysine acetylation was thought to be rare in bacteria.⁴⁸ Around 2009, the first large-scale mass-spectrometry-based proteomic characterizations of a bacterial acetylome were performed in E. coli.^50,51With the rapid advances in mass spectrometry technology and improvements in reagents, such as anti-acetyllysine antibodies, the lists of acetylated proteins have expanded considerably. Indeed, the bacterial acetylomes of more than 30 different species are now catalogued.^48,49 It is apparent that lysine acetylation is far more widespread and perhaps more important than previously thought. So then, what is the functional significance of these thousands of acetylation sites on hundreds of proteins? This open-ended question is ever-present in the field, as the significance of the vast majority of these modifications remains unclear.

With the daunting task ahead to understand the biological relevance of bacterial protein acetylation, specific examples to date have focused largely on proteins involved in important functions like virulence, transcription, translation, and central carbon metabolism.^48,49,53 From acetylome studies in Bacillus subtilis,⁵⁴ Mycobacterium tuberculosis,⁵⁵ and Acinetobacter baumannii,⁵⁶ it was realized that the histone-like HU family proteins are acetylated. This conserved feature of distant species prompted a deeper examination of the previously published acetylome data (Table 1). HU orthologues were found to contain at least one acetylation site in 21 different species. Despite the fact that some published bacterial acetylomes were performed in species where the genome was not completely annotated, HU acetylation nonetheless showed broad distribution in the phyla Proteobacteria, Deinococcales, Cyanobacteria, Firmicutes, and Actinobacteria. These acetylome analyses were performed using different growth conditions, lysis conditions, acetyllysine antibodies for enrichment, proteomics workflows, and mass spectrometry instrumentation. With this lack of uniformity across studies, it is possible that the acetylation of HU family proteins is actually even more widespread and universally conserved. From these 21 organisms, the most well conserved acetylation sites are K3, K18, and K86, which are also the three most highly conserved lysine residues (Figure 2). The high sequence conservation of the HU family, specifically the positively charged residues, implies that these modifications occur in important functional domains, likely those that make contact with the DNA backbone. There are three regions in which many acetylation sites lie. The first is within the N-terminal 20 amino acids, in which almost every species has at least one site (Figure 2). The next hot spot is in the middle of the protein, centered on amino acid 40, which is less conserved. Finally, the C-terminal 25 amino acids are also enriched for acetylation sites. If these acetylation sites occur in functional domains for DNA binding, acetylation of these residues would partially negate the positive charge of the lysine side chain and could drastically alter the HU–DNA binding properties. To date, the available evidence supports this idea, as discussed below.^56–58

Table 1.

Lysine Acetylation Observed in HU Orthologues

bacterial species and protein	identified acetylation site(s)
Acinetobacter baumannii122
HU-β	K3, K13, K23, K64, K72
Acinetobacter baumannii SK1756
HU-β	K13
Aeromonas hydrophila98
HU-α	K91
HU-β	K3, K9, K18, K75, K80
Bacillus amyloliquefaciens123
Hbs	K3, K37, K86
Bacillus subtilis75
HBsu	K18
Bacillus subtilis54
HBsu	K3, K18, K37, K41, K75, K80, K86
Bacillus subtilis76
HBsu	K18, K19, K23, K80
Clostridium acetobutylicum124
HU	K38
Escherichia coli125
HU-α	K13
Escherichia coli126
HU-β	K67 or K75, K83 or K86a
Escherichia coli127
HU-α	K22, K51, K86
HU-β	K9, K18, K86
Escherichia coli128
HU-α	K3, K13, K18, K22
HU-β	K9, K18, K67
Francisella novicida129
HU-β	K37, K53
Geobacillus kaustophilus130
Hbs	K3, K18, K23, K41
Mycobacterium tuberculosis55
HU-β	K13, K86
Pseudomonas aeruginosa131
HU	K3, K18, K67
Pseudomonas aeruginosa132
HU	K80
Saccharopolyspora erythraea133
SACE_6142 (HU)	K51, K85
Spiroplasma eriocheiris134
HupA	K18, K31, K38, K70, K75, K86
Staphylococcus aureus135
HU	K13
Streptomyces roseosporus136
HU-β	K13, K86
Sulfurospirillum halorespirans137
HU-β	one siteb
Synechococcus sp. PCC 7002138
HU	K68, K86
Synechocystis sp. PCC 139
HU	K3, K13, K15, K24
Thermus thermophilus140
HU	K37
Vibrio alginolyticus141
HU-α	K3, K18, K22, K37, K51, K64, K83
HU-β	K3, K18, K64, K75
Vibrio cholerae142
HU-α	K18, K64, K70, K83
HU-β	K18, K70, K75, K80
Vibrio parahemolyticus143
HU-β	K18, K64

^aThe identified peptides contained two lysine residues, and the exact location of the acetylation site was not specified.

^bSpecific site location information was not provided.

Figure 2.

Evolutionary conservation of lysine acetylation of HU orthologues. HU orthologous sequences were obtained from the NCBI protein database. For some bacterial species, the HU orthologues exist as heterodimers, and the two subunits are denoted as HupA and HupB. The sequence alignment was performed using Geneious Prime, version 2020.1.1; the level of shading indicates the level of conservation, with black the best conserved. The identified acetylation sites from the acetylome studies listed in Table 1 are highlighted in red. The Actinobacteria (Mycobacterium tuberculosis, Saccharopolyspora erythraea, and Streptomyces roseosporus) contain a unique C-terminal extension that is rich in basic amino acids. The entire extension is about 110 amino acids and is not shown. Species included: Acinetobacter baumannii (Abau), Aeromonas hydrophila (Ahyd), Bacillus amyloliquefaciens (Bamy), Bacillus subtilis (Bsub), Clostridium acetobutylicum (Cace), Escherichia coli (Ecol), Francisella novicida (Fnov), Geobacillus kaustophilus (Gkau), M. tuberculosis (Mtub), Pseudomonas aeruginosa (Paer), Staphylococcus aureus (Saur), Spiroplasma eriocheiris (Seri), S. erythraea (Sery), Sulfurospirillum halorespirans (Shal), S. roseosporus (Sros), Synechococcus sp. (Syn), Thermus thermophilus (Tthe), Vibrio alginolyticus (Valg), Vibrio cholerae (Vcho), and Vibrio parahemolyticus (Vpar).

Structural examination of the physical location of these modification sites reveals that many are found in regions predicted to directly contact the DNA. As shown in Figure 3, the top of the HU dimeric structure contains “arms” that create a channel where the DNA molecule rests. This family of proteins introduces large, flexible bends in the DNA molecule of ~105° to >180°, causing the DNA to bend downward and make contacts along the front and back faces of the HU dimer.⁵⁹ For the E. coli (Figure 3A) and Vibrio alginolyticus (Figure 3B) HupA/HupB heterodimers, most of the modified lysine side chains (red) point toward the DNA backbone, which would be at the center of the HU structure. This implies that acetylation of these residues could affect DNA binding. The same is true of B. subtilis (Figure 3D). In each of these structures, some lysines are oriented to point away from where DNA binding likely occurs, so these sites may mediate other important features like protein–protein interactions. Interestingly, the lysine modifications of HupA from the distantly related, spiral shaped, cell-wall-lacking Spiroplasma eriocheiris (Figure 3C) mostly point away from where the DNA molecule lies. This suggests that although acetylation is broadly conserved, the main regulatory mechanisms by which they function probably vary from species to species.

Figure 3.

Structural location of PTMs on HU orthologues. Computational models of (A) E. coli HupA–HupB heterodimer, (B) V. alginolyticus HupA–HupB heterodimer, (C) S. eriocheiris HupA homodimer, and (D) B. subtilis HBsu homodimer are shown. For the heterodimers depicted in (A) and (B), the HupA subunit is colored green, and the HupB subunit is colored tan. Both monomers of the homodimers are depicted in (C) and (D). Acetylated residues are shown as red sticks, and phosphorylated residues are shown as yellow sticks; succinylated residues that do not overlap with an acetylation site are shown as pink sticks. The comparative models were generated using Phyre2.¹²⁰ The Anabaena HU–DNA cocrystal structure (Protein Data Bank code 1P71) was used as the modeling template. Molecular graphics were produced with PyMOL.¹²¹

FUNCTIONAL SIGNIFICANCE OF N^ε-LYSINE ACETYLATION OF HU PROTEINS

Within the past 5 years, several reports about the functional significance of the acetylation of HU orthologues have been published.^56–58,60 Acetylome analysis of A. baumannii revealed that HU was acetylated at a single site, lysine 13.⁵⁶ In contrast to E. coli, A baumannii HU is encoded by hupB and exists as a homodimer.⁶¹ This acetylation site is well-conserved in other bacterial species (Table 1 and Figure 2). HU protein that was homogeneously acetylated at K13 was generated through the use of genetic code expansion using a strain with an evolved tRNA synthetase/tRNA pair for protein purification.⁶² It was found that lysine acetylation does not alter the oligomeric state but does lower the thermal stability of the protein. Similarly, tetraacetylation of histone H4 lowers the thermal stability of the nucleosome particle.⁶³ Therefore, one plausible explanation for this observation is that acetylation destabilizes the HU dimer. Additionally, the kinetics of HU–DNA binding was altered in that K13-acetylated protein had order of magnitude lower association (k_on) and dissociation (k_off) rate constants compared with unacetylated protein. The equilibrium constants (K_D) of the acetylated and unacetylated forms were similar. It was proposed that lysine 13, which is not predicted to directly contact the DNA, may serve as an epigenetic regulator that controls the availability of the pool of free HU in the cell. In this model, unmodified HU associates with the DNA because of faster kinetics compared with the acetylated form. Furthermore, it is possible that acetylation is required to maintain the HU–DNA complexes because of a lower dissociation rate. While these data suggest a functional role for a specific acetylation site that likely does not contact the DNA, the in vivo significance of these observations was not determined.

The HU orthologues in the phylum Actinobacteria are encoded by a single hupB gene and are composed of two domains. The N-terminal domain is highly homologous to those of other HU orthologues,⁶⁴ as depicted in Figure 2. The HU orthologues in this phylum also contain a unique C-terminal extension that is rich in basic amino acids.^61,64 Large-scale acetylome analysis showed that the M. tuberculosis HU protein was acetylated at two sites, K13 and K86.⁵⁵ Proteomic analysis of HU alone identified eight total sites, with three present in the conserved N-terminal domain.⁶⁰ By the use of protein acetylated in vitro, it was discovered that acetylation reduced the DNA binding affinity 10-fold relative to the unacetylated protein. Moreover, microscopic analysis of HU–DNA complexes by TEM and AFM demonstrated that acetylation leads to less compacted DNA and other architectural changes.^58,65 Again, the biological significance of these observations remains unknown. However, one previously identified site from both acetylome analysis⁵⁵ and the previously discussed study⁵⁸ (K86) has been further evaluated. Lysine 86 is predicted to fall in the region that directly contacts the DNA (Figure 3D). In support of this, in B. subtilis, alanine substitution at K86 resulted in a 70% reduction of the DNA binding activity of the orthologue HBsu.⁶⁶ When mycobacteria are grown in the presence of a suboptimal concentration of a drug, drug-tolerant variants readily arise.⁶⁰ Drug tolerance in bacterial populations is defined as the ability to survive a transient exposure to antibiotics at a lethal concentration.⁶⁷ Clinically, these slower-growing, tolerant variants often lead to treatment failure and relapses. By the use of the model organism Mycobacterium smegmatis, lysine 86 was mutated into an arginine residue, which cannot be modified.⁶⁰ The HU K86R mutant resulted in specific elimination of the isoniazid-tolerant small colony variant (SCV) subpopulation. These SCV populations have altered gene transcription and metabolic potential. Altogether, these data suggest that acetylation of HU proteins may globally alter transcription, and one possible mechanism of action is by alteration of DNA binding activity through PTMs, analogously to eukaryotic histone modifications.

The acetylation of HBsu from the Gram-positive organism B. subtilis has also been studied. Similar to HU, HBsu binds nonspecifically to curved DNA,⁶⁸ associates with the nucleoid region, and condenses DNA.⁶⁹ HBsu is encoded by a single gene and assembles as a homodimer.⁷⁰ Unlike E. coli HU, HBsu is essential for viability.^70,71 To date, HBsu has been implicated in recombination^72–74 and DNA repair.⁷³ HBsu was identified as acetylated in three different acetylome analyses.^54,75,76 The functional significance of seven of those sites was assessed with regard to nucleoid compaction.⁵⁷ In mutants that could not be acetylated (arginine substitutions), the nucleoids were more compacted compared with the wild type. Additionally, the phenotype of acetyl mimic mutations (glutamine substitutions), specifically HBsuK3Q, -K41Q, and -K86Q, had decreased DNA content, suggesting that acetylation of HBsu regulates DNA replication. Interestingly, one of these acetyl site mutants, HBsuK41Q, was shown to have decreased DNA binding affinity. More work is required to determine the specific functions of HBsu in DNA compaction and replication and how they are modulated by acetylation. Taken together, however, these data suggest that acetylation of HBsu at specific lysine residues regulates both DNA compaction and replication, and reduced DNA binding affinity may be one mechanism by which acetylation induces these cellular responses.

ENZYMATIC CONTROL OF HU ACETYLATION

If this possible histone-like code is a regulatory element, there must be a means to add and remove such modifications. In bacteria, acetylation occurs by two mechanisms, nonenzymatic and enzymatic (as reviewed in refs 48, 49 and 53). Non-enzymatic acetylation is prevalent in bacteria and may be responsible for broad, low-level global acetylation.^48,49 The local environment surrounding a lysine residue influences its nucleophilic properties, which provides a mechanism for autocatalysis. The acetyl donor for nonenzymatic acetylation is predominantly the secondary metabolite acetylphosphate. Enzymatic acetylation occurs via the action of lysine acetyltransferases (KATs), whereby acetyl-CoA serves as the acetyl donor. In eukaryotes, there are five classes of KATs, but only Gcn5-N-acetyltransferase (GNAT) is broadly conserved in all domains of life. The evidence to date suggests that HU family proteins are acetylated enzymatically.^57,58 A mass-spectrometry-based interactome analysis performed with M. tuberculosis HU protein identified an interaction with a protein called enhanced intracellular survival (Eis).⁵⁸ Eis is a GNAT that was known to be involved in acetylation of aminoglycoside antibiotics.⁷⁷ By the use of purified proteins, it was established that Eis directly acetylates HU in vitro⁷⁸ and is capable of acetylating 32 sites when present in excess. Moreover, the Michaelis constant (K_m) was ~65 times lower for HU than for kanamycin. Finally, confocal microscopy analysis of strains overproducing Eis showed nucleoid decompaction, consistent with a role for acetylation in the regulation of nucleoid compaction.⁵⁷ These data strongly support a role for Eis as a bona fide KAT for HU.

In B. subtilis, about 50 deletion mutants of putative acetyltransferases were screened for nucleoid compaction phenotypes.⁵⁷ Five KATs were identified whose deletion resulted in a compacted nucleoid, similar to the arginine-substitution phenotype observed for HBsu. Through genetic bypass experiments, it was determined that two KATs acetylate HBsu at specific sites. One enzyme, YfmK, was shown to directly acetylate HBsu in vitro using purified proteins. Moreover, acetylation at K80 was shown to be significantly reduced in a yfmK mutant compared with the wild type using targeted mass spectrometry, which allows selective and specific quantification. Thus, as with Eis in M. tuberculosis, a KAT directly acetylates a B. subtilis HU family protein, and it is likely that this holds true in other bacteria. In eukaryotes, the reversal of lysine acetylation is catalyzed by the NAD⁺-dependent sirtuins and the Zn⁺-dependent lysine deacetylases (KDACs).⁷⁹ The sirtuins are widely distributed in an evolutionary context and are prevalent in bacteria.⁴⁸ KDAC orthologues are known to exist in only a few species.^80,81 Recently, in M. tuberculosis, the sirtuin orthologue Rv1151c was demonstrated to interact with HU in vivo and deacetylate HU in vitro.^65,78 The enzymes responsible for deacetylation of HU in other bacteria have not yet been identified. It is of course entirely possible that other, novel classes of enzymes exist as well. Future studies will be required to address this possibility and determine the complete mechanism of deacetylation of HU family proteins.

OTHER PTMS DISCOVERED ON HU PROTEINS

As noted above, many other modifications beside acetylation have been discovered that constitute part of the code.^4,8 While the biological significance of many of these modifications is unclear, at least seven amino acids have been observed to undergo modification in histones. For example, in addition to acetylation, lysine residues can be modified by methylation (mono-, di-, and tri-), ubiquitination, sumolyation, biotinylation, ADP-ribosylation, and various additional short-chain acylations like succinylation, malonylation, and crotonylation.⁸² Arginine residues are modified by various methylation states, ADP-ribosylation, and citrullination. Serine, threonine, tyrosine, and histidine can be phosphorylated.⁴ These modifications are mostly involved in gene regulation, chromatin formation, and DNA replication and repair.

Over the past decade, numerous PTMs were discovered and characterized in bacteria,⁸³ the most common of which include protein phosphorylation (Arg, Ser, Thr, Tyr), lysine acetylation and succinylation, glycosylation, and lipidation. Arginine phosphorylation is a relatively newly discovered PTM in some Gram-positive bacteria^84,85 that may regulate important components of stress response systems.^86,87 Arginine phosphorylation is believed to target proteins for degradation to the Clp protease, possibly serving as a protein quality control mechanism.⁸⁸ Another PTM that signals for degradation is the prokaryotic ubiquitin-like protein (Pup).⁸⁹ This modification is known to exist mostly in Actinobacteria, where it is a functional homologue of eukaryotic ubiquitin, targeting proteins to the proteasomal degradation machinery.

Phosphorylation and lysine acylation are the most common and best understood regulatory modifications in bacteria. As with lysine acetylation, large-scale mass-spectrometry-based “omics” studies in various bacteria have recently been published.^90,91 If a “histone-like” code exists, it is probable that other modifications may be present among HU orthologues and constitute part of this code. HBsu in B. subtilis contains a single arginine phosphorylation site (R61).⁸⁴ The significance of this modification is unclear, but two possibilities are likely. First, phosphorylation of R61 may regulate protein stability or turnover and serve in protein quality control. However, R61 does lie within the arms of the structure⁶⁶ and thus may directly contact the DNA. In fact, mutation of this residue to leucine resulted in a 70% loss of DNA binding activity,⁶⁶ suggesting that it is important for DNA binding. The effects of a charge reversal mutant were not investigated, as modified protein may retain even less DNA binding activity because of repulsive electrostatic interactions of the phosphoryl group with the DNA backbone.^92,93

Examination of the phosphoproteomes from 24 bacterial species revealed that this PTM occurs in ~40% of HU orthologues (Table 2).⁹¹ The distribution and evolutionary conservation do not appear to be as widespread as lysine acetylation, but these studies are limited by the same caveats as described before, including lack of uniformity in study design and incomplete genome annotations. Additionally, TiO₂ affinity chromatography, which is widely used for enrichment of phosphopeptides in eukaryotes,⁹⁴ has not been optimized for use with bacterial samples. In bacteria there are far fewer STY phosphorylations than observed in eukaryotes,⁹¹ so this optimization is essential in order to broaden and expand the identification of these modifications. Nonetheless, in the diverse bacteria where STY phosphorylation of HU has been observed, there is typically a single phosphorylation site, and it occurs within the first 20 amino acids of the protein (Table 2). Structural examination of the T4 location in HBsu in B. subtilis (Figure 3D, yellow) shows that this modification likely occurs in a region of the protein that directly contacts the DNA. This suggests that modifications located near this region may be important for relaxation of compacted DNA, as phosphorylation introduces a large structural disturbance. As there is a 20 amino acid region where these modifications tend to occur, some of them may point in the other direction, away from bound DNA. In these cases, phosphorylation may be important for mediating protein–protein interactions and may recruit other transcription factors or modifying enzymes, analogously to histone modifications in eukaryotes. Further functional studies involving protein phosphorylation of HU orthologues must be performed to test these predictions.

Table 2.

Phosphorylation and Succinylation of HU Orthologues

STY Phosphorylation
Acinetobacter baumannii144	S4
Bacillus subtilis145	T4
Listeria monocytogenes146,147	T5, S18
Rhodopseudomonas palustris148	S13
Sinorhizobium meliloti149	S8, S10
Staphylococcus aureus85	T4
Streptomyces coelicolor150	S4, T44
Synechococcus sp.151	T6, T74
Synechocystis sp.152	S21
R Phosphorylation
Bacillus subtilis84	R61
K Succinylation
Aeromonas hydrophila153	K3, K75, K80, K83
Escherichia coli154
HU-α	K3, K13, K18, K22, K67, K70, K86
HU-β	K9, K18, K67
Porphyromonas gingivalis155	K3
Pseudomonas aeruginosa132	K18, K61
Mycobacterium tuberculosis156	K70, K72, K103
Synechococcus sp.157	K3, K68, K86
Vibrio parahemolyticus97	K3, K13, K22, K64, K83

It has become appreciated that lysine acyl modifications are widespread and prevalent in bacteria.⁹⁵ Numerous large-scale studies have recently been published, with lysine succinylation being the most well studied and understood modification after acetylation. Succinylation utilizes the metabolite succinyl-CoA as a donor, and like acetylation, it is a reversible process.⁹⁶ Succinylation is drastically different than acetylation, as it is bulkier (100 Da vs 42 Da for acetylation) and thus has a larger impact on protein structure and function. Acetylation neutralizes the positive charge of the lysine residue, while succinylation results in a charge reversal (+1 to −1) due to the presence of a carboxylate group. Again, this may have large functional consequences for the modified protein. To date, HU succinylation has been observed in seven different bacterial species (Table 2), although the biological significance of these modifications has not been assessed. It is interesting to note that many of the succinylation sites overlap with the acetylation sites (Table 1), which has been previously noted.^95,97,98 In fact, for the E. coli heterodimer (Figure 3A), all but two lysine residues (pink) were sites for both acetylation and succinylation. This interesting observation suggests that there may be crosstalk between these PTMs. It remains to be determined whether different acylations on the same lysine residue have different functional consequences.

PERSPECTIVES

Eukaryotic histones contain long, unstructured N-terminal and C-terminal tails, which are largely absent from bacterial NAPs, with the exception of the C-terminal extensions seen in the Actinobacteria. While the majority of histone PTMs tend to be located among these tails, a substantial number are found within the histone fold domain; these regulate histone–DNA and histone–histone associations and constitute part of the code.^99,100 It seems possible that the widespread conservation of bacterial HU PTMs (Figure 2 and Tables 1 and 2) may represent the evolutionary origins of the eukaryotic histone code. While this is an exciting possibility, significant questions still remain.

The first important question is the following: what are the functions of these identified modifications? HU family proteins are known to play roles in the regulation of various DNA transactions in addition to nucleoid compaction. Determining how these modifications impact those other roles is essential to our understanding of the language of the code. Another important issue concerns the extent to which modifications are conserved. In other words, is the potential code universal, or do different phyla or species have different languages? It is becoming clearer that other PTMs are waiting to be characterized in bacteria. More recently, large-scale proteomics analyses of other PTMs have become available, such as propionylation^101,102 and malonylation.¹⁰³ The Thermus thermophilus HU is propionylated at K23, and the E. coli HupA and HupB are malonylated at K18 and K70 and at K6, respectively. As enrichment strategies and mass spectrometry workflows continue to improve, the landscape of PTMs on HU family proteins will undoubtedly continue to expand. With new modifications continually being discovered, it will be interesting to move away from characterizing single PTMs and determine how combinations of these modifications function together to regulate protein activity. Mass-spectrometry-based techniques are currently available to identify and quantify multiple different PTMs on histones following nuclei isolation and acid extraction.¹⁰⁴ For this reason, the dynamic interplay of crosstalk between these modifications on eukaryotic histones is already being examined.^105,106 As HU proteins are highly abundant, methods to extract native HU proteins without affinity enrichment have been described,^107,108 but mass spectrometry workflows need to be developed and optimized in order to study multiple PTMs.

In eukaryotes, enzymes that add histone PTMs are called “writers”, and those that remove them are called “erasers”. As we learn more about the complete landscape of HU PTMs, the enzymatic mechanisms by which these modifications are regulated will need to be addressed. It is also important to determine which proteins “read” or decipher the code within the bacterial cell. In eukaryotes, there are specific reader proteins that have domains that bind to specific modified amino acids. For example, proteins containing the bromodomain bind to acetylated lysine residues.¹⁰⁹ This binding typically enables recruitment of other important transcription factors or chromatin remodeling factors and thus regulates gene expression.¹¹⁰ To date, no orthologues or functional homologues of the bromodomain have been discovered in bacteria, which remains an important task going forward toward understanding how this histone-like code affects cellular physiology.

The role of protein acetylation in bacterial virulence and drug resistance is beginning to be elucidated.^111–115 With proper study, targeting the enzymes involved in regulation (i.e., KATs, sirtuins, and KDACs) or the acetylated form of a key protein (i.e., virulence factors, essential genes, etc.) may allow for the design of new antimicrobials. Indeed, altering acetylation (acetyltransferase and deacetylase inhibitors) has been a promising therapy for the treatment of some cancers^116,117 and latent viral infections.¹¹⁸ Additionally, it was recently determined that treatment of mycobacteria with the antibiotic fusidic acid followed by treatment with the Eis inhibitor 1a* (a KAT inhibitor) was more efficient at killing the bacteria than with antibiotics alone.¹¹⁹ This suggests that targeting protein acetylation of HU family proteins may be a therapeutic strategy worth further exploration. Our knowledge of this histone-like code is far from complete, and additional PTMs that constitute it are continually being revealed. More PTM writers, erasers, and readers are waiting to be discovered, and similar to the enzymes involved in acetylation, all of them could represent promising new drug targets.

ABBREVIATIONS

AFM

atomic force microscopy

GNAT

Gcn5-N-acetyltransferase

KAT

lysine acetyltransferase

KDAC

lysine deacetylase

NAP

nucleoid-associated protein

PTM

post-translational modification

SCV

small colony variant

TEM

transmission electron microscopy