Covalent Modification of Bromodomain Proteins by Peptides Containing a DNA Damage-Induced, Histone Post-Translational Modification

Abstract

An electrophilic 5-methylene-2-pyrrolone modification (K_MP) is produced at lysine residues of histone proteins in nucleosome core particles upon reaction with a commonly formed DNA lesion (C4-AP). The nonenzymatic K_MP modification is also generated in the histones of HeLa cells treated with the antitumor agent, bleomycin that oxidizes DNA and forms C4-AP. This nonenzymatic covalent histone modification has the same charge as the N-acetyllysine (K_Ac) modification but is more electrophilic. In this study we show that K_MP-containing histone peptides are recognized by, and covalently modify bromodomain proteins that are K_Ac readers. Distinct selectivity preferences for covalent bromodomain modification are observed following incubation with K_MP-containing peptides of different sequence. MS/MS analysis of 3 covalently modified bromodomain proteins confirmed that Cys adduction was selective. The modified Cys was not always proximal to the K_Ac binding site, indicating that K_MP-containing peptide interaction with bromodomain protein is distinct from the former. Analysis of protein adduction yields as a function of bromodomain pH at which the protein charge is zero (pI) or cysteine solvent accessible surface area are also consistent with non-promiscuous interaction between the proteins and electrophilic peptides. These data suggest that intracellular formation of K_MP could affect cellular function and viability by modifying proteins that regulate genetic expression.

Graphical Abstract

K_MP-modified histones are a nonenzymatic, post-translational modification produced from the DNA lesion C4-AP. Chemically synthesized K_MP-containing histone peptides selectively modify cysteine-containing bromodomain proteins, the readers of N-acetyllysine. Covalent modification of nuclear proteins, such as bromodomains, by K_MP could affect their function in cells.

Introduction

N-Acetyllysine (K_Ac) formation in histones is dynamic, and has profound effects on chromatin structure, histone deposition, and gene expression (Scheme 1A).^[1] The addition of an acetyl group by a histone acetyltransferase (HAT) to the ε-amine of lysine, which eliminates the positive charge, reduces intra- and inter-nucleosomal interactions and leads to a more open chromatin structure.^[2] Indeed, the two chromatin states, euchromatin and heterochromatin, exhibit distinct acetylation patterns. The less condensed, transcriptionally active euchromatin is enriched in hyperacetylated histones, while hypoacetylated histones correlate with more compacted heterochromatin.^[1-3] Histone deacetylases (HDACs) help regulate K_Ac levels at sites within histones that are recognized by proteins containing bromodomains (BRDs, Scheme 1B).^[4] BRDs are present in nuclear proteins with diverse roles, and are implicated in the development of certain types of cancer.^[5] The 61 BRDs are classified into eight families based on sequence and structural data analysis.^[4] The vast majority of BRDs identified in human proteins (46) contain one or more cysteine residues distributed within the vicinity of the substrate binding site.^[6] These cysteine residues can be targets for electrophilic small molecules.^[7] Recently, a nonenzymatic, DNA damage-induced, lysine histone post-translational modification (K_MP) was identified in HeLa cells treated with the antitumor agent, bleomycin (Scheme 2).^[8] K_MP contains an electrophilic 5-methylene-2-pyrrolone that is generated by the reaction of the ε-amine of lysine with a DNA lesion.^[8] 5-Methylene-2-pyrrolones react readily with thiols (Scheme 2) and the trapping reaction is reversed by high GSH concentration.^[9] We report herein that bromodomain proteins are covalently modified by K_MP-containing histone peptides. The covalent modification yield varies with respect to bromodomain protein and K_MP-containing peptide sequence. Given the importance of modified histone recognition by BRDs to cellular function, covalent modification by K_MP could affect cellular processes and cell viability.

Scheme 1.

Dynamic formation of N-acetyllysine (K_Ac) in a histone (A) and its recognition by bromodomain proteins (B).

Scheme 2.

Formation of a 5-methylene-2-pyrrolone lysine modification (K_MP) due to DNA oxidation, and its reaction with a protein nucleophile.

K_MP forms concomitantly with a single-strand break in nucleosome core particles from a C4′-oxidized abasic site (C4-AP, Scheme 2).^[10] C4-AP is produced via C4′-hydrogen atom abstraction by a variety of DNA damaging agents, including bleomycin.^[11] K_MP results from nucleophilic attack by a histone lysine on the oxidized abasic site. The modification was identified via a chemoproteomics approach at 17 of 57 lysine residues in the core histones H2A, H2B, H3 and H4 when HeLa cells were treated with bleomycin.^[8] Preference for the DNA damaging agent (bleomycin) to react in the DNA entry/exit regions of the nucleosomes resulted in predominant K_MP formation at the proximal lysines.^[12] It is likely that a more promiscuous DNA damaging agent, such as hydroxyl radical would give rise to a broader distribution of K_MP sites throughout the nucleosome.^[11e] The prevalence of these histone adducts motivated us to examine the interaction between K_MP-containing histone peptides with bromodomain containing proteins.

Results and Discussion

K_MP-Containing histone peptide selection

K_MP is produced at almost one-third of the lysines present in the core histone proteins in bleomycin-treated HeLa cells, most of which are also acetylated.^[8] The histone lysines that are modified are limited by the DNA positions at which C4-AP is produced by bleomycin. It is likely that C4-AP is produced more randomly throughout a nucleosome by a diffusible reactive species, such as hydroxyl radical, which is produced by ionizing radiation.^[11e]

As an example of an N-acyllysine modification, we investigated the interaction of K_MP-containing peptides with bromodomain proteins that are known to selectively recognize N-acetyllysine (K_Ac). The peptide sequences (Scheme 3) chosen were based upon known bromodomain recognition sites of N-acetyllysine residues in histones.^[5b] Histone H4K16Ac is one of the amino acid residues in the “basic patch” motif. H4K16Ac hinders the formation of higher order chromatin structure.^[13] Interestingly, loss of H4K16Ac and trimethylated H4K20 is a hallmark of human cancer.^[14] Intracellular H3K9Ac is associated with active promoters.^[15] In particular, H3K9Ac is important for super elongation complex binding to chromatin to promote a switch from initiation to elongation during transcription.^[16] Given the importance of these modifications, the corresponding peptides in which K_Ac of H4K16Ac (1) and H3K9Ac (5) was replaced by K_MP were prepared. In addition, a scrambled version of the peptide containing H4K16MP (4) was synthesized to help address recognition selectivity. Peptides 1 and 5 are the same length but are comprised of a different amino acid composition and overall charge. Together, these molecules were chosen to test whether K_MP-containing peptides selectively covalently modify bromodomain proteins. Dodecameric peptides were synthesized to ensure adequate lengths for protein recognition, based upon previous studies of peptide-bromodomain binding.^[5b,17] All peptides were prepared containing C-terminal amides, instead of a carboxylate to prevent the introduction of a negative charge and to mimic a peptide bond. The peptides were synthesized via solid-phase support by taking advantage of orthogonal protecting groups for the ε-amine of lysine.^[8,18] The N-termini were either acetylated or fluorescently labeled with rhodamine B linked via a triethylene glycol linker to enhance water solubility and distance these portions of the molecules from the peptide. The final peptides were purified by reverse-phase HPLC and characterized by ESI-MS.^[19]

Scheme 3.

Peptides used in this investigation.

Covalent modification of bromodomain proteins by K_MP-containing histone peptides

A battery of 8 proteins, two from each of the 4 groups (II, V, VII and VIII) of bromodomains were chosen for the study.^[4] The number of cysteine residues in these proteins ranged from 0 to 12 (Figure 1A). Initial experiments were carried out by separately incubating rhodamine-labeled H4K16MP peptide (1 a) in 10-fold excess of each bromodomain protein (Figure 1). Protein adducts were separated by SDS-PAGE and detected via fluorescence. The gels were also silver stained to directly detect the proteins. Multiple fluorescent bands were observed for several of the proteins (Figure 1A). These were attributed to reaction of more than one peptide with some proteins. The presence of multiple products complicated determining protein adduct yields. Estimated yields, based upon band intensities and comparison to a calibration curve constructed using free peptides, ranged from ~11-37% (Figure S1). The yields did not correlate with the number of cysteines in the bromodomain protein. Importantly, reaction of 1 a with BSA, which contains a single cysteine but is not a member of the bromodomain family is covalently modified in the presence of 10-fold excess peptide (6%) to a significantly lower amount than any of the bromodomain proteins. In addition, PB1(2), which does not contain a cysteine, yielded ≤ 4% (Figure S1), suggesting that other nucleophiles can react with K_MP but less efficiently than thiols. Protein adducts were not detected via fluorescence when either the corresponding N-acetyllysine peptide (H4K16Ac, 3) or unmodified peptide (H4K16, 2) were incubated with 7 of the 8 proteins (Figure 1B, Figure S2).

Figure 1.

Covalent modification of bromodomain proteins by a K_MP-containing peptide A) Incubation of excess 1 a (10 μM) with proteins (1 μM) followed by SDS-PAGE and visualization by either in-gel fluorescence (top row) or silver stain (bottom row). B) Validation that K_MP is required for covalent modification.

These data indicate that H4K16MP (1 a) forms covalent adducts with bromodomain proteins, and that product formation is favored by the presence of a cysteine residue in the protein. The lack of correlation between adduct yield and number of cysteine residues within the protein suggests that the reaction depends on selective reactivity of 1 a with the various bromodomains, and is not the result of nonspecific capture of protein nucleophiles by K_MP-containing peptides.

Additional evidence for selective covalent modification of bromodomain proteins by H4K16MP peptide (1 a) was acquired by comparing covalent modification by this peptide with that of a peptide comprised of the same amino acids in a scrambled sequence (4), and K_MP at a comparable position (Figure 2A). In these experiments, bromodomain protein was in 5-fold excess relative to the fluorescently labeled peptide, and products were quantified using a calibration curve constructed using 1 a. These conditions avoided potential complications due to multi-adduct formation when exploring selectivity of the covalent modification. Reactions were complete within 30 min and the yields reported are following a 60 min incubation (Figure S8). The scrambled peptide (4) formed adducts with the suite of bromodomain proteins. However, the yields with 5 of the 7 proteins that contain cysteines were statistically different (*p < 0.05, unannotated p < 0.0001) from those formed from reaction with 1 a. Similarly, the distribution of yields of modified protein obtained by reaction with a peptide containing H3K9MP (5) was also distinct from that of 1 a (Figure 2B). Covalent modification yields by 5 span a broader range and reach a higher maximum (22–82%) compared to the peptide containing H4K16MP (1 a) (8–35%). It is also notable that the covalent modification yields do not correlate with different net charges on 1 a (z = +4) and 5 (z = +3) (Figure 2B), indicating that the reaction preferences are not the result of electrostatic effects. The distinct reaction patterns of the 3 K_MP-containing peptides with the battery of bromodomain proteins (Figure 2) supports the hypothesis that protein adduction is not the result of nonspecific covalent modification.

Figure 2.

Comparison of yields for covalent modification of bromodomain proteins by K_MP-containing peptides in the presence of excess protein.A)Peptide containing H4K16MP (1 a) versus scrambled H4K16MP peptide (4). B)Peptide containing H3K9MP (5) versus 1 a. [Protein] =0.5 μM, [Peptide] - = 0.1 μM. (*p < 0.05; unannotated p < 0.0001).

LC–MS/MS analysis of covalent modification of representative bromodomain proteins by a K_MP-containing histone peptide

Bromodomains are comprised of a particular structural fold containing a left-handed four-helix bundle (αZ, αA, αB, and αC), known as the BRD fold (Figure 3A).^[20] A variable hydrophobic pocket formed from the interhelical loops ZA (between αZ and αA) and BC (between αB and αC) lines the K_Ac binding site and determines the binding specificity. K_Ac is anchored within the central deep hydrophobic cavity through its carbonyl oxygen by the side-chain amide nitrogen of a highly conserved asparagine residue (e.g. N156) present in the BC loop (illustrated for BRD2(1) in Figure 3B). The differences in the sequences surrounding the K_Ac binding site affects the protein surface properties and eventually determines the binding specificity.^[4]

Figure 3.

Bromodomain protein structures. A) The overall structure of the BRD2(1) bromodomain (PDB: 2DVQ). The secondary structure elements are labeled. The binding of a histone H4 peptide acetylated at lysine 12 (H4(1-15)K12Ac) on the central cavity is shown. B) The N-acetyllysine-binding pocket when bound to an H4K12Ac peptide. The stick diagram shows key residues (green) and bound water molecules (red spheres) contributing to N-acetyllysine (yellow) recognition. Hydrogen bonding interactions are indicated by dotted lines. C) X-ray crystal structures of (left to right) BRD4(1) (PDB: 3UVW), TAF1L(2) (PDB: 3HMH), and PB1(1) (PDB: 3IU5) showing cysteine residues (green) and those covalently modified by K_MP-containing peptide 1b (yellow). Conserved asparagine residues in BRD4(1) and TAF1L(2) are highlighted. PB1(1) does not contain a conserved asparagine residue and it is replaced by a tyrosine (Y127).

Covalent modification of one member of groups II, VII and VII was examined using mass spectrometry. We did not examine a member of group V due to the large number of cysteines (12). LC–MS/MS analysis of protease digests of selected bromodomain proteins modified with peptide 1 b showed that cysteine residues modified with K_MP were not always proximal to the N-acetyllysine binding site (Figure 3C, S3-5). For example, in BRD4(1) two cysteine residues, C125 and C136, are located on the αB helix. Neither cysteine is solvent accessible. C136 is ~4 Å from conserved asparagine N140. However, only C125, which is outside of the K_Ac binding site, was detected as a K_MP-containing peptide (1b). TAF1L(2) also contains two cysteines. Neither C1619 or C1638 are proximal to the N-acetyl-lysine binding site. The more solvent accessible cysteine, C1638, was modified by K_MP of 1b. PB1(1) has three cysteine residues. C50 and C138 are in the αZ and αC helices, respectively. Modification of these residues was not detected. The third cysteine, C69, which is modified by K_MP, is located in the interhelical ZA-loop, which lines in the K_Ac binding site. In contrast to BRD4(1) and TAF1L(2), PB1(1) does not contain a conserved asparagine in the binding pocket, instead it is replaced by a tyrosine.

Covalent modification of bromodomain proteins by K_MP-containing histone peptides is independent of protein pi or cysteine accessibility

LC–MS/MS analysis reveals that while the K_MP-containing peptide (1 b) does not necessarily modify the cysteine most proximal to the N-acetyllysine binding site, this modification may still affect bromodomain function. Due to the inherent electrophilicity of K_MP, we analyzed whether covalent modification was determined by the 5-methylene-2-pyrrolone’s inherent reactivity. This was explored by analyzing cross-link yields versus properties of the proteins. If the adduct yield is driven by overall electrostatic attraction between the positively charged histone peptides and the bromodomain, one would expect it to vary inversely with respect to pi of the protein. However, no correlation was observed when the yields of protein adducts with peptide containing H4K16MP (1 a) were plotted versus bromodomain pI of those for which such data was available (Figure 4). A similar lack of correlation between protein pI and protein modification was observed for the scrambled peptide (4) and that based upon H3K9MP (5) (Figure S6).

Figure 4.

Yield of protein covalent modification by K_MP-containing peptide 1 a as a function of pI. [Protein] = 0.5 μM, [1 a] = 0.1 μM. Linear regression performed in Origin 6.1 and coefficient of determination shown on plot.

We also analyzed the covalent modification yields as a function of the sum total of the cysteine solvent accessible surface area (SASA) for those proteins for which structural data were available (Figure 5). The cysteine SASA for each protein was calculated by summing the SASA for all such residues in the protein, although of the 5 proteins analyzed only BRD4(2) contains more than one cysteine with significant SASA. No correlation between cysteine SASA and covalent modification for H4K16MP (1 a) was observed (Figure 5A). In fact, BRD4(1), the bromodomain protein with the least accessible cysteine residue, provides the highest yield of covalent modification. Interestingly, a positive, albeit weak correlation (R² = 0.26) was observed between the cysteine SASA and yields of protein adducts by the peptide with the scrambled sequence (4, Figure S7). In contrast to the reactivity of H4K16MP (1 a), reaction of 5 with BRD4(1) provides only slightly higher covalent modification than the lowest yield, which is produced by TAF1L(2) (Figure 5B). Of the 3 peptides examined, covalent protein modification by the peptide containing H3K9MP (5) showed the greatest, albeit still weak correlation between reactivity and cysteine SASA (Figure 5B). Overall, these data indicate the notion that the sequence of K_MP-modified histone peptides play a role in binding to bromodomain proteins, and that covalent modification of the proteins is not due to nonspecific reaction with solvent exposed, nucleophilic cysteine residues by a promiscuous electrophile.

Figure 5.

Yield of protein covalent modification by K_MP-containing peptides as a function of cysteine solvent accessible surface area. A) Peptide containing H4K16MP (1 a). B) Peptide containing H3K9MP (5). [Protein] = 0.5 μM, [1 a], [5] = 0.1 μM. Linear regression performed in Origin 6.1 and coefficients of determination shown on plots.

Conclusion

Post-translational histone modifications affect cellular processes by altering chromatin structure and by interacting with reader proteins that control genetic expression. A small number of histone post-translational modifications, such as those resulting from crotonylation (K_Cr) or oxopropenylation (K_Pr) introduce electrophilic species (Scheme 4). These PTMs have not yet been found to covalently modify proteins that interact with chromatin.^[21] The examples of nonenzymatic covalent modifications of proteins that are biologically significant are increasing.^[21b,22] The 5-methylene-2-pyrrolone (K_MP) lysine modification that is produced in HeLa cells treated with bleomycin is more electrophilic than K_Cr, which does not react with free thiols.^[8,21a] In this investigation, we demonstrate that K_MP-containing histone peptides covalently modify bromodomain proteins, readers of histone acetylation. The different reactivity of 3 peptides with 7 bromodomain proteins indicate that the modification is not merely attributable to a promiscuous K_MP electrophile. The cysteine residues that are adducted are not always in the region of the proteins that recognize K_Ac. This could reflect the ability of K_MP to modulate peptide recognition and/or the inherent reactivity of the electrophile with nucleophilic cysteines. The relative nucleophilicity of protein cysteine residues is not the sole source of the preference of a particular protein for the K_MP-containing peptides as the 3 peptides showed different reactivity patterns for a given protein. These results suggest that K_MP-modified histones formed in cells may capture histone lysine reader domains. How or even whether such chemical reactions affect biological processes is unknown, but it is possible that they have the potential to massively interfere with normal cellular functions. For instance, a captured transcription factor (e.g. TFIID, which contains the bromodomain TAF1(2)) could remain in a perpetual on-state, resulting in greater levels than desired of transcription.^[23] Another of many possible scenarios is that covalent capture of BRD4 could alter chromatin structure via recruitment of the SWI/SNF remodeling complex, or even lead to cell death in ARID2 deficient cells.^[24] K_MP is generated in HeLa cells treated with bleomycin, a cytotoxic DNA damaging agent. The data presented raise the possibility that K_MP formation may contribute to the cytotoxicity and/or side effects of the anticancer agent bleomycin and other agents that produce the oxidized abasic site, C4-AP.

Scheme 4.

N-Acylated lysine post-translational modifications.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.