1.2. Identification of Acetyl-Histone Peptide Interactions Using Peptide Arrays
The availability of affordable synthetic peptide arrays of high density has stimulated several research projects seeking to identify post-translationally modified linear motifs that are recognized by evolutionary conserved effector modules that act as sensors (or “readers”) for these PTMs. Lack of tools that can be used to deposit these modifications in vitro directly onto nucleosomes, has made it impossible to study such motifs in the context of histones. However, peptide arrays have now been successfully used to identify numerous histone-dependent interactions that lead to significant understanding of the underlying biology. The technology itself was shown to be effective for various classes of reader modules, including CHROMO WD-40 and MBT domains [10]. Peptides immobilized on a solid support have also been used to determine potential acetylation-dependent recognition motifs interpreted by bromodomains. Despite the low false negative rate of the method, the false positive rate is much higher requiring the use of orthogonal biophysical methods to confirm binding events. Several studies exploring specific interactions with histone modifications, as well as large scale studies systematically exploring the landscape of histone modifications have been published and will be summarized here, providing a wealth of information suggesting that SPOT techniques can yield robust and reproducible results in identifying acetylation dependent interactions.
Binding of yeast bromodomains to acetylated human histone peptides was tested using peptide arrays. Biotinylated peptides were spotted onto commercial SAM Biotin Capture Membranes and the membranes were incubated with 14 GST-tagged recombinant yeast bromodomains at room temperature. Membranes were immunoblotted with a GST antibody and several acetylated histone peptides were found to bind to these BRD modules, although no orthogonal methods were used to verify these findings [16]. This study established that the technology can be used to rapidly assess binding to several acetyl-lysine modules resulting in numerous potential interactions that can be further validated in order to establish the underlying biological significance of histone-acetyl-lysine recognition. Several BRD-containing proteins have been tested using this technology yielding novel potential interactions.
Binding of acetylated histone sequences to the six BRDs of human polybromo 1 (PB1) was determined by using either cellulose SPOT arrays, or peptide microarrays on silicon slides. Two dimensional (2D) 1H-15N heteronuclear single quantum correlation (HSQC) NMR spectroscopy was then employed to measure the dissociation constant for the interacting peptides. The interaction of the second bromodomain of PB1 with histone H3 acetylated at lysine 14 (H3K14ac) was measured to be 0.5 mM. An NMR structural model was also determined, suggesting insertion of the acetyl-lysine into the cavity of the bromodomain upon binding [17].
The nucleosome-remodeling factor subunit Bromodomain and PHD finger-containing transcription factor (BPTF or Fetal Alzheimer antigen—FALZ) contains BRD/PHD tandem modules which act together to recognize modifications found on histone tails. The selectivity of the BRD module towards H4K16ac or H4K20ac peptides was established using a SPOT array covering all acetylation sites of human histones, printed on a modified cellulose scaffold. A glutathione S-transferase (GST) construct of the bromodomain module of human BPTF was incubated with an array containing duplicates of 96 modified 15-amino-acids long histone peptides and binding was assessed using a GST antibody. Acetylated H4 peptides were further validated employing in solution isothermal titration calorimetry (ITC) in order to determine thermodynamic binding constants. Intriguingly, this study further demonstrated that the BPTF PHD/BRD tandem modules simultaneously engage two heterotypic trans-histone marks, in the context of full nucleosomes, whereby the PHD module engages histone H3K4me3 and the BRD module engages H4K12ac or H4K16ac or H4K20ac resulting in significant selectivity as well as affinity increase [18].
A large scale systematic study of histone modifications was also carried out in order to establish the motifs that are recognized by 33 human bromodomains. Peptides covering single acetylation sites of the four core histones (H2A, H2B, H3 and H4) as well as the linker histone (H1–4) were spotted onto cellulose membranes and were incubated with individual recombinant bromodomains carrying a hexa-histidine affinity tag. After overnight incubation, membranes were washed and probed with an antibody targeting the histidine tag. In addition, arrays covering multiple acetyl-lysine modifications on each peptide were used, screened against the bromo and extra-terminal (BET) subfamily of human bromodomain containing proteins. The effect of adjacent modifications was further examined on histone H3, focusing on a central acetyl-lysine mark flanked by modifications on lysines (acetylation, methylation, dimethylation, trimethylation) as well as serines and threonines (phosphorylation). This systematic study identified 485 acetylation-dependent linear motifs recognized by human bromodomains [3]. Interestingly, the histone acetyl-transferase PCAF as well as BRDs present in nuclear body proteins (such as SP140 and SP140L) exhibited non-specific binding to many peptides found in the arrays studied. Some BRDs did not exhibit any interactions with the tested arrays, suggesting that they may be binding to non-histone proteins. Systematic orthogonal characterization of many interacting peptides using isothermal titration calorimetry established that peptides with dissociation constants of 0.5 mM or lower were successfully identified on the SPOT arrays; however some false negatives were also present. Centering of peptide sequences onto an acetyl-lysine epitope resulted in different affinity to sequences where the epitope was shifted towards the N- or C-terminus, suggesting that steric and positional effects may be present and should be considered when screening large arrays. Some BRDs were also shown to require multiple PTMs in order to interact with a particular peptide, suggesting interplay between signaling pathways. For instance, the BRD of BPTF did not bind to H3K4 or H3K4ac, however it strongly recognized H3pT3K4acK9ac. Similarly, the BRD of WD repeat domain 9 (WDR9) showed strong dependence on pS10 and pT11 in order to recognize H3K18ac. All members of the BET sub-class also exhibited high affinity towards peptides carrying two acetylated lysines. Structural characterization of this interaction established that both acetyl-lysines engage the protein by binding simultaneously within the acetyl-lysine cavity site regardless of the sequences flanking the two acetyl marks. The distance required between two acetyl sites in order to initiate interaction with BET bromodomains was also investigated; this was achieved by using a systematic SPOT array which altered the length as well as the linker type between two acetyl-lysine marks. The results were confirmed employing in-solution thermodynamic measurements, further demonstrating the power and versatility of the peptide array methodology [3].
Combinations of the reader domains found in P300 were tested against a commercial peptide array (Active Motif) covering 384 peptides spanning the human core histones (H2A, H2B, H3 and H4) employing glutathione S-transferase (GST) tagged recombinant domains (BRD/RING/PHD, BRD/PHD or BRD alone). Mainly histone H4 peptides carrying multiple acetylations were found to bind to different constructs containing the BRD module. Interestingly, a systematic pattern for the strongest spots was observed, whereby a two or three-amino-acid spacer separated multiple Kac sequences (Kac(X)2-3Kac) [19].
The tandem PHD/BRD modules of TRIM24 were also found to interact with 20-amino-acid long histone H3 peptides carrying K9mex (x = 1, 2, 3) or K9ac or K14ac modifications. Peptides were biotinylated and printed onto a streptavidine coated ArrayIt slide prepared with the BioRad VersArray Compact Microarrayer. The interactions were determined by incubating glutathione S-transferase fusion tagged TRIM24 PHD-BRD constructs overnight. Orthogonal methods were used to verify the identified interactions, such as in solution isothermal titration calorimetry as well as pull-downs using the same biotinylated peptides [20].
The PHD/BRD/PWWP triple modules of mouse BS69 (Zinc finger MYND domain-containing protein 11, ZMYND11) were found to bind to histone H3 peptides carrying a K36me3 modification using peptide arrays. Probing of a small focused peptide array spanning 84 histone peptides (including peptides found on histones H3, H4, H2A and H2B) with modifications on selected lysine (acetylation/methylation) and arginine (methylation) residues, identified a strong interaction with H3K36me3 recognized by the tandem reader BRD/PWWP modules of ZMYND11. The membrane was incubated with glutathione S-transferase (GST)–ZMYND11 PHD/BRD/PWWP triple domains and staining with an anti-GST antibody revealed strong interaction with H3K36me3 peptides (in triplicate). It is noteworthy that no acetyl-lysine peptides bound to the protein, in agreement with the crystal structure of the BRD/PWWP tandem module where the conserved asparagine found in bromodomains, responsible for histone peptide interactions is replaced with a tyrosine residue (Y192 in mZMYND11), sterically blocking the Kac binding cavity of the module. Intriguingly the protein was further found to preferentially bind to the H3.3 isoform carrying a K36me3 modification, resulting in co-localization with H3 and H3.3 on gene bodies, functioning as an unconventional transcriptional co-repressor by modulating RNA polymerase II during elongation [21].
The BRD/PWWP tandem module of human ZMYND11 was also recently profiled against a small histone peptide library printed on a cellulose membrane, covering modifications found on histones H3 and H4. Histone H3 peptides carrying unmodified, methylated (nono-, di- and tri-methylated) or acetylated K36 residues, as well as various combinations of histone H4 peptides with acetylated lysines, were found to bind to the BRD/PWWP modules of ZMYND11. A glutathione S-transferase fusion tag was used on the recombinant proteins for detection carried out employing an anti-GST antibody. Identified interactions were further confirmed employing in solution profiling using a fluorescent polarization assay or western blotting using the full length protein [22].
The Zinc-finger and MYND containing protein 8 (ZMYND8) also contains on its N-terminus three epigenetic reader domains (PHD/BRD/PWWP) which were found to bind to histone marks. A commercial focused peptide array covering 384 peptides (Active Motif) was used to probe binding of a GST-ZMYND8 construct spanning the PHD/BRD modules of the protein. Incubation with the array and detection with anti-GST revealed strong binding to histone H4 marks carrying multiple acetylations (K5ac, K8ac, K12ac, K16ac) or histone H3 peptides carrying K36ac or K36mex (x = 2, 3). Pull-downs with biotin-labeled peptides and the GST-PHD/BRD, as well as endogenous full length protein, confirmed binding to H4 acetylated lysine marks. Interestingly knockdown of the endogenous acetyl-transferase TIP60, responsible for depositing K16ac on histone H4, reduced the protein’s ability to bind to histones, as demonstrated by fluorescent recovery after photobleaching (FRAP) assays in U2OS cells stably expressing GFP-ZMYND8 [23].
Degenerate peptide arrays have been successfully used to define the positional specificity of many interactions (reviewed elsewhere [6]). The positional effect on a central acetyl-lysine residue was systematically probed using cellulose peptide arrays. In these arrays, within each 11 amino acid long peptide bearing a central acetyl-lysine mark, one position contained a fixed amino acid (i.e., as one of the 20 natural amino acids as well as pS, pT, pY, methylated lysine and acetylated lysine) while the remaining positions were degenerate (i.e., containing all possible amino-acids at equimolar amounts). The 14 yeast BRD modules tagged with a GST and a His6 (as GST-BRD-His6), were incubated with this degenerate histone array. Detection was carried out with a GST antibody revealing preferences for acetyl-lysine recognition in a position- and amino-acid dependent context. Tagged yeast BRDs were then successfully used as affinity reagents to purify from cell extract acetylated histone peptides, acting as pan-specific affinity enrichment reagent, suggesting that BRDs can be engineered in several ways in order to increase their selectivity and affinity towards acetylated sequences, making them good tools for affinity purification-based techniques [24].
Acetylation-dependent interactions successfully identified employing SPOT methods highlight the applicability of the technology and its capacity to generate hypothesis driven research in order to understand the wiring underpinning lysine acetylation signaling networks. In an effort to capture the best experimental conditions necessary for scaling peptide membranes to scan large numbers of linear interacting motifs containing acetylated lysine sequences, we explored published protocols and studied the conditions that give good qualitative results which we then compared to in solution biophysical measurements. Here we summarize our control experiments and highlight step-by step protocols that can be used to systematically study acetylation-dependent interactions employing peptide array technologies to identify interactions of biological significance.