Capturing the interaction between histones and their deacetylases with hydroxamic acid-modified microarrays
Histone deacetylases (HDACs) modify the epigenetic code recorded on histones. This landscape of modifications can be altered with drugs, such as HDAC inhibitors, to silence genes involved in cancer progression. Studying the HDAC‒histone interaction may help elucidate these mechanisms.
The cell nucleus is populated by a handful of different histone deacetylases, or HDACs. These enzymes rush from one end to the other of the genome modifying the histone proteins that scaffold the DNA, altering how the information stored in the genes is expressed. This activity, part of the epigenetic machinery, regulates most biological functions in healthy and diseased cells. However, little is known about which part of the histone each HDAC modifies, which would be the key to understand their individual functions. These were the interactions we sought to explore, by combining the expertise of Christian (Olsen lab) within the targeting of HDACs with Hans’ experience (Maric lab) with peptide microarrays, which he was applying at the time as a postdoctoral fellow in the Strømgaard lab. As a just-hired PhD student with zero experience within epigenetics and peptides, I was first given the task to synthesize non-canonical amino acid building blocks with the ability to target the HDACs. This would take me a month of lab work and leave enough spare time to catch up with the literature.
The first thing I learned was that HDACs are Nε-acyllysine hydrolases with a Zn2+ ion buried in the active site. This zinc ion is essential for catalysis, as it coordinates to and activates the carbonyl group of the amide modification and positions the water nucleophile in the right orientation. The field of HDAC inhibition has provided compounds with groups that can chelate or coordinate to this Zn2+ ion, such as hydroxamic acids, ortho-aminoanilides or carboxylic acids (Figure 1)1. Therefore, we decided to prepare Nε-acetyllysine mimics with a Zn2+-binding group instead of the acetamide moiety, so that the HDACs would be trapped upon recognition of specific peptide substrates. This concept was already applied in a bead-based study by Dirk Schwarzer and coworkers and published during my hiring process2, providing optimism that the strategy would work. Since each HDAC isozyme is expected to target different histone sites, we hoped that they would bind to the peptides on the microarray depending on their sequence, and that this could infer the elusive isozyme selectivity.
Figure 1. Binding of substrates and inhibitors to the active HDAC active site. Co-crystal structures of the nuclear enzyme HDAC8 with an Nε-acetyllysine (Kac) substrate (left, PDB 5DIC), a hydroxamic acid inhibitor (center, PDB 3EW8) and an ortho-aminoanilide inhibitor (right, PDB 4LY1); and chemical structures of the fragment of the ligand shown. The Zn2+ ion is represented as a purple sphere.
Of course, it did not just take a month. While adopting the excellent protocol by David P. Fairlie’s laboratory to prepare L-2-aminosuberic acid3, I realized that I had to introduce more changes to the route than expected if we were to end up with gram amounts of the final amino acid. We initially had in mind a collaboration where we would supply at least a gram of building block to a company that would then perform the microarray synthesis and send us µSPOT histone microarray slides to test. In SPOT peptide synthesis, peptides are synthesized using the Fmoc/tBu strategy, with their C-termini covalently attached to a cellulose membrane4. Alternatively, modified cellulose disks can be used that are partially degraded in acid, creating a soluble peptide-cellulose conjugate that can be printed onto coated glass slides (Figure 2)5. With µSPOT, the peptide-cellulose stock solutions can be printed to generate hundreds of microarray slides ready for testing.
Figure 2. Production of µSPOT peptide microarrays. Peptides are synthesized onto modified cellulose disks which, upon acidic treatment, are partially degraded and can be dissolved to obtain peptide-cellulose stocks. These stocks are printed onto coated microscope slides to generate copies of the microarray.
In a fantastic turn of events, the time spent on building block synthesis resulted in the lucky situation that our collaborators in the Strømgaard lab had set up a SPOT synthesizer and µSPOT microarray printer in the meantime. This way, I had much more freedom to optimize conditions and test smaller selections of peptides before going to full arrays. I could include a handful of peptides bearing different Zn2+-binding groups in a first trial synthesis, and we were really excited when the ones with a hydroxamic acid lit up with a strong HDAC-binding signal. It was time to make use of the throughput and screen the entire sequence of the histones.
I quickly generated more data than I could process and, excited with the results, we could think of a ton of applications. But I was brought back to reality by the more experienced scientists around me: screenings are nice, but you have to validate the results, too. And sure enough, they were right! We therefore postponed the most ambitious experiments and focused on evaluating the predictive power of the microarrays. SPOT peptide synthesis is not perfect and, since the assays are performed directly on the product of the reaction, there could be incomplete peptides and other byproducts that could affect the results. We observed that some peptides presented better properties when resynthesized and tested in solution than what was predicted by the microarray (false negatives). In addition, we realized that µSPOT is great for detecting major changes in affinity but may not have the precision to inform on the subtle effects of all peptide sequence changes. After synthesizing, purifying, and testing multiple substrates, inhibitors, probes and controls, we can now say that we understand much better the potential and the limitations of this technique, and that we are at a stage where more concrete studies can be designed.
In this proof-of-concept study, we show that HDAC binding can be studied in high-throughput and that peptides serve the dual purpose of being combinatorial inhibitor scaffolds and mimics of biological substrates. We could show how a specific histone PTM abolishes HDAC binding and also obtain peptide inhibitors with different HDAC selectivity and with activity in cells. For us, the future is to use µSPOT to generate chemoproteomic tools to study each individual HDAC. We also hope that our data will help design future HDAC inhibitors and permit studying the histone PTM cross-talk.
You can read more about the study in Nature Communications
This post was authored by Carlos Moreno-Yruela (Olsen Lab, University of Copenhagen)
- Kristensen, H. M. E., Madsen, A. S. & Olsen, C. A. in Epigenetic Drug Discovery (eds W. Sippl & M. Jung) 155–184 (John Wiley & Sons, 2018).
- Dose, A. et al. Interrogating substrate selectivity and composition of endogenous histone deacetylase complexes with chemical probes. Angew. Chem. Int. Ed. 55, 1192–1195 (2016).
- Kahnberg, P. et al. Design, synthesis, potency, and cytoselectivity of anticancer agents derived by parallel synthesis from α-aminosuberic acid. J. Med. Chem. 49, 7611–7622 (2006).
- Frank, R. Spot-synthesis: an easy technique for the positionally addressable, parallel chemical synthesis on a membrane support. Tetrahedron 48, 9217–9232 (1992).
- Winkler, D. F. H., Hilpert, K., Brandt, O. & Hancock, R. E. W. in Peptide Microarrays Methods in Molecular Biology (eds Marina Cretich & Marcella Chiari) 157–174 (2009).