Engaging with benzoyllysine through a π-π-π mechanism

Abstract

Epigenetic modifications have been gaining in prominence as fundamental components of the chromatin regulatory machinery. In this review, we summarize the molecular and structural mechanisms of reading, writing, and erasing of lysine benzoylation, a recently discovered posttranslational modification (PTM) in histones. We highlight a unique nature of the conjugated π system of benzoyllysine that may aid in the development of benzoyllysine-specific effectors indifferent to the saturated acyllysine modifications. We also discuss transcriptional and metabolic functions associated with benzoylation of histones and implications of ingesting of sodium benzoate for human health.

Introduction

Covalent modifications of DNA and histone posttranslational modifications (PTMs), collectively referred to as epigenetic marks play a major role in the regulation of chromatin structure and function. Epigenetic marks alter local chromatin environment, affect DNA accessibility, and accelerate or impede DNA-related processes [1–3]. They constitute the human epigenome, which is less stable than the human genome and undergoes changes during normal cellular processes or in response to internal or external stimuli. Incorrect spatiotemporal distribution of epigenetic marks can disturb normal gene expression programs, aberrantly turning genes on and off, and disrupt the cell growth and survival. A wide array of epigenetic marks has been identified, including methylation, phosphorylation and acylation of histones and methylation of DNA.

Acylation of the side chain of lysine residues in histone proteins is one of the major PTMs. This mark alters the electrostatic potential of histones as it eliminates the positive charge of lysine. Histone tails are particularly enriched in lysine residues, acetylation of which reduces interactions of the tails with the negatively charged DNA, leading to a more open chromatin state associated with transcriptional activation. In acylation reaction, acyltransferase enzymes catalyze the transfer of the acyl group from acyl-CoA to the ε-nitrogen of the ammonium moiety of lysine. In addition to the most common and well characterized lysine acylation modification, acetylation, which was discovered 60 years ago [4], a dozen of other acylation modifications has recently been identified in the human epigenome, including lysine benzoylation (Fig. 1a).

Figure 1: The π-π-π stacking mechanism.

(a) Lysine benzoylation pathway in mammalian cells. Chemical formula of sodium benzoate (NaBz), benzoyl-CoA (Bz-CoA) and benzoyllysine are shown. (b, c) Crystal structure of the sirtuin enzyme Hst2 (blue) (that debenzoylates benzoylated lysine) in complex with the substrate peptide, H3K9bz (orange) (PDB ID 7F4E). π-π stacking contacts are shown as yellow dashed lines, and aromatic residues are labeled. (d) A π-π-π stacking mechanism by which the YEATS domain of Taf14 recognizes crotonyllysine (pink). W81 of Taf14 adopts two conformations, rotamer 1 (light gray) and rotamer 2 (green) (PDB ID 5IOK).

Identification of lysine benzoylation sites

Using a mass spectrometry approach, the Zhao group discovered 22 benzoyllysine sites in human cells, in mouse liver, and in Drosophila cells [5], and 27 benzoyllysine sites were identified in yeast by the Chen group [6]. Although benzoyl-CoA (Bz-CoA) is a common intermediate in the anaerobic (without molecular oxygen) metabolic degradation pathway of aromatic compounds in bacteria [7], in mammals, the most likely source of Bz-CoA is consumption of benzoic acid or sodium benzoate (NaBz). In support, treatment of mammalian cells with NaBz increases production of Bz-CoA and subsequently lysine benzoylation in a dose-dependent manner [5]. The lysine benzoylation PTM is primarily identified in histone proteins [5], and additionally, 207 lysine residues in 149 non-histone proteins implicated in ribosome biogenesis and RNA processing were found to be benzoylated in a proteome-wide screening [6].

Writing and erasing lysine benzoylation

Genetic deletion of seven histone acetyltransferases (HATs) individually in yeast reveals that benzoylation of lysine is mainly catalyzed by Gcn5 HAT, a core subunit of the SAGA (Spt-Ada-Gcn5 acetyltransferase) complex (Table 1) [6]. Although no experimentally determined structure of the Bz-CoA-bound Gcn5 is available, modeling of the succinyl-CoA-bound human GCN5 [8] suggests that the benzoyl group can occupy the same hydrophobic groove of GCN5 that is occupied by the succinyl group [6]. The opposite reaction, the removal of the benzoyl group from benzoylated lysine residues in mammalian cells is catalyzed by the NAD⁺-dependent protein deacetylases SIRT1 and SIRT2 (Table 1) [5,9]. In yeast, debenzoylation is mediated by the sirtuin enzyme Hst2. The crystal structure of Hst2 in complex with H3K9bz (benzoylated lysine 9 of histone H3) peptide offers insight into the mechanism by which the H3K9bz substrate is recognized by this epigenetic eraser [6] (Fig. 1b). In the complex, the benzoylated lysine is bound in an elongated hydrophobic channel with the benzoyl group being sandwiched between three aromatic residues of Hst2. While the benzene ring of the benzoyl moiety is involved in a π-π interaction with F67 of Hst2, the conjugated amide of benzoyllysine is involved in π-π interactions with F184 and H135 of Hst2 [6] (Fig. 1c). The formation of multiple π-π contacts involving the conjugated benzene-amide π system of benzoyllysine represents yet another example of a π-π-π stacking mechanism originally reported for the conjugated C=C double bond-amide π system of crotonyllysine, which is recognized by the YEATS domain of Taf14 [10,11] (Fig. 1d) and the YEATS domain of AF9 [12,13]. Mutation of F184 or H135 in Hst2 to an alanine substantially decreases the binding of benzoyllysine [6], pointing to a critical role of π-π interactions in the formation of this complex.

Table 1.

Proteins involved in the installation, removal, and recognition of lysine benzoylation PTM. References are in parentheses.

writers	Gcn5	[6]
erasers	SIRT1, SIRT2	[5,9]
	Hst2	[6]
readers	YEATS of YEATS2	[16]
	AF9	[16]
	Sas5	[6]
	Taf14	[6]
	BD of Sth1	[6]
	DPF of MOZ	[16]

Reading benzoyllysine

Several domains or PTM-readers [14,15] that can recognize benzoyllysine, such as YEATS domains of human YEATS2 and AF9 [16] and yeast Sas5 and Taf14 [6], bromodomain (BD) of Sth1 [6], and the DPF (double PHD finger) domain of MOZ [16] have been identified and characterized (Table 1). In the crystal structure of the complex of the YEATS2 YEATS domain with H3K27bz (benzoylated lysine 27 of histone H3) peptide, the benzoyl group is positioned between two aromatic residues, Y262 and W282, with Y262 forming the π-π stacking contacts with the benzene ring of benzoyllysine, and W282 forming the π-π stacking contacts with the amide of benzoyllysine [16] (Fig. 2a, b). In contrast to the YEATS2 YEATS:H3K27bz complex, in which benzoyllysine inserts between the tryptophan-tyrosine set of the aromatic residues, the YEATS domain of AF9 uses a tyrosine-phenylalanine (F59 and Y78) set to engage with H3K9bz [16]. The benzene ring of benzoyllysine in the AF9 YEATS:H3K9bz complex is tilted by 30° with respect to the aromatic rings of F59 and Y78, suggesting that either steric constrains are imposed by an adjacent F28 or tryptophan may provide better geometry for the π-π stacking with benzoyllysine. Indeed, the benzene ring adopts a less tilted position in the absence of such steric constrains and the presence of tryptophan (W82) in the complex between the YEATS domain of Sas5 and H3K27bz peptide [6] (Fig. 2c).

Figure 2: Reading benzoyllysine by the YEATS domains.

(a, b) Ribbon diagram of the structure of the H3K27bz (green)-bound YEATS domain from YEATS2 (PDB ID 6LSD). (c) Structure of the Sas5 YEATS domain (yellow) in complex with H3K27bz peptide (magenta) (PDB ID 7F5M). π-π stacking contacts in (a-c) are shown as yellow dashed lines, and aromatic residues are labeled.

Unlike the YEATS domains, BD of Sth1 and DPF of MOZ have predominantly hydrophobic benzoyllysine-binding pockets. Benzoyllysine in H3K14bz (benzoylated lysine 14 of histone H3) inserts deeply in the hydrophobic cavity of DPF of MOZ and no π-π stacking is formed with a single aromatic residue (F211) present in the binding pocket [16]. This binding mode mimics the binding mechanism by which other acyl modifications of H3K14, including crotonylation, are recognized by DPFs [17–20]. Although BDs from many human and yeast proteins tested do not read lysine benzoylation PTMs [6,16], BD of Sth1 does but displays a ~5 fold weaker binding affinity to H3K14bz peptide than to the corresponding acetylated peptide H3K14ac [6]. The crystal structure of the H3K14bz-bound Sth1 BD shows that benzoyllysine occupies the same hydrophobic binding pocket at the top of the four-helix bundle of BD as acetyllysine occupies [21]. Much like in the canonical BD:acetyllysine complex, the benzoyl group in the Sth1 BD:H3K14bz complex is anchored through a hydrogen bond with asparagine (N1333) (Fig. 3c). However, the benzoyl group is longer than the acetyl group and therefore inserts deeper into the cavity between the BD helices. As a result, the invariable water-shell [22] that lines the acyllysine binding pocket of BDs restricting its size is disturbed, which likely accounts for a decrease in binding activity of this BD toward the benzoylation modification.

Figure 3: Reading benzoyllysine by DPF and bromodomain.

(a, b) Electrostatic surface potential (a) and ribbon diagram (b) of the structure of DPF from MOZ in complex with H3K14bz peptide (yellow) (PDB ID 6LSB). (c) Crystal structure of the Sth1 bromodomain (light blue) in complex with H3K14bz peptide (salmon) (PDB ID 7F3S). Water-shell molecules and a hydrogen bond are shown as yellow dashed lines and red sphere, respectively. Benzoyllysine is shown in a stick model and labeled.

Histone benzoylation and gene expression

In cellulo studies have shown that like other forms of histone acylation, benzoylation is evolutionarily conserved from yeast to humans. In mammals, all four core histones, i.e., H3, H4, H2A and H2B can be benzoylated, however primarily at their amino-terminal tails [5]. Such preferential localization suggests a unique function for this epigenetic mark, given all other forms of lysine acylation are typically found in both tails and globular domains of histones. Consistent with this idea, transcriptomic analysis of HepG2 cells after induction of histone benzoylation with sodium benzoate reveals the upregulation of a specific subset of metabolic genes associated with glycerophospholipid metabolism, ovarian steroidogenesis and phospholipase D signaling [5]. At the molecular level, histone benzoylation localizes to the transcription start-sites of active genes, and similar localization is observed for histone acetylation. Acetylation and benzoylation appear to compete for the same set of lysine residues in histones, because genes upregulated by the addition of sodium benzoate are associated with an increase in histone benzoylation and a decrease in histone acetylation [5].

The YEATS domain of YEATS2 has been shown to bind the benzoylated H3K27bz peptide ~6-fold tighter than the acetylated H3K27ac peptide [16]. It is intriguing to speculate that histone benzoylation-mediated transcriptional changes are driven by the recruitment of the GCN5-containing ATAC complex that possesses YEAST2 to promoters enriched in histone benzoylation – perhaps via a ‘read-write’ mechanism that enforces the transcription of genes regulated by this histone PTM. Although more work is needed to define how histone benzoylation is deposited and contributes to transcriptional regulation and what role non-histone proteins that undergo benzoylation play, the accumulated to date data point to important transcriptional and metabolic functions of this modification that are likely conserved across eukaryotes.

Benzoylation and health

The cofactor of benzoylation, Bz-CoA, is a common metabolic intermediate produced in plants, bacteria and in mammalian cells. In humans, the gut microbiome plays a major role in the generation of Bz-CoA from benzoic acid, a compound found in fruit and vegetables, and NaBz, an FDA (Food and Drug Administration) approved preservative, widely used as an antimicrobial or flavoring agent to prevent canned food and beverages from molding [23–25]. Although much of NaBz and benzoic acid consumed from the human diet are rapidly cleared from the body [26], a significant amount of Bz-CoA still produced by the gut microbiome promotes histone benzoylation, particularly in the intestinal cells [24,27]. We note that interestingly, the gut lining and microbiome are not killed off by NaBz given its antimicrobial properties; a finding that may be explained by its rapid clearance [26]. NaBz is considered safe and has been labeled GRAS (generally recognized as safe) by the FDA if consumed less than 5 mg/kg per day and is used as a drug to treat hyperammonemia and urea cycle diseases [27]. NaBz has also been shown to have potential in the treatment of neurodegenerative disorders, autism and depression, though some studies report examples of cellular toxicity from NaBz exposure and adverse effects, such as an increase in oxidative damage and genotoxicity [27]. Thus, while very safe, there are concerns with consumption of NaBz in high doses since this compound ultimately stimulates lysine benzoylation that results in a decrease in the level of lysine acetylation and alters epigenetic regulation. Further work is needed to better understand how NaBz exposure impacts gut epigenetics and if this can be detrimental in any way.