Cell-based multi-substrate assay coupled to UHPLC-ESI-MS/MS for a quick identification of class-specific HDAC inhibitors

Abstract

Histone deacetylases (HDAC) are involved in several diseases including cancer, cardiovascular and neurodegenerative disorders, and the search for inhibitors is a current topic in drug discovery. Four HDAC inhibitors have already been approved by the FDA for cancer therapy and others are under clinical studies. However, the clinical utility of some of them is limited because of unfavorable toxicities associated with their broad range of HDAC inhibitory effects. Toxicity could be decreased by using HDAC inhibitors with improved specificity. To date, the most popular screening assays are based on fluorescence-labeled substrates incubated with an enzymatic source (cells extracts or recombinant isoforms). Here, we describe a high-throughput cell-based UHPLC-ESI-MS/MS assay able to rapidly predict activity against HDAC1 and HDAC6 in a cell environment. This method is predicted to be a useful tool to accelerate the search for class-selective HDAC inhibitors in drug discovery.

• Class I HDAC
• HDAC6
• high-throughput screening
• histone deacetylase
• isoform selectivity

Introduction

Histone acetylation is one of the post-translational modifications that influences the chromatin state and contributes to the regulation of gene expression. Histone acetylation is controlled by the action of two enzyme families: histone deacetylases (HDACs) and histone acetyltransferases (HATs). HDACs are responsible for the deacetylation of the N-terminal lysine of the core histones H2A, H2B, H3 and H4. To date, 18 HDAC isoforms have been identified and classified into four classes (I–IV). Classes I, II and IV are Zn²⁺-dependent and are referred as classical HDACs. Class I HDACs (HDAC1–3 and HDAC8) are essentially expressed within the nucleus where histones constitute their main substrate proteins^1,2. Class II is divided into two subclasses, IIa (HDAC4, HDAC5, HDAC7 and HDAC9) and IIb (HDAC6 and HDAC10)^3,4. They are mainly located into the cytoplasm, but can shuttle between this compartment and the nucleus, depending on various cellular signals^1,2,4. HDAC11 is the only member of class IV. Class III contains the sirtuins (SIRT), which depend upon NAD⁺for their catalytic activity.

Epigenetic dysregulations of the acetylation process have been involved in the development of several diseases including cancer, cardiovascular and neurodegenerative disorders. Thus, HDAC inhibition has emerged as an interesting therapeutic strategy to restore the HDAC/HAT balance. Four HDAC inhibitors have already been approved for cancer therapy by the FDA. All of them are showing a pan-HDAC inhibitory profile^5,6.

The search for isoform specific HDAC inhibitors is a current topic in drug discovery, given that individual isoforms have been linked to distinct pathological processes^7–13. Some isoform-specific inhibitors were shown to be less toxic compared to pan-HDAC inhibitors^14,15. Moreover, specific HDAC6 inhibition exerted neuroprotection either by decreasing neuronal oxidative stress¹⁶ or by increasing the axonal transport¹⁷ in neurodegenerative models.

The most popular screening assays to characterize HDAC inhibitors are based on fluorescence-labeled substrates incubated with an enzymatic source (single isoform, nuclear extracts or cell lysates). The substrate is usually a small peptide sequence containing an acetyl lysine residue and a fluorophore such as a coumarin^18,19. On one hand, these fluorescence-based in vitro assays are very convenient, reproducible and fast for measuring HDAC inhibition in high-throughput screenings²⁰. On the other hand, they have the disadvantage of potential interference from the autofluorescence of cell constituents and test compounds (fluorescence background or quenching) which may result in false positive or false negative results. Furthermore, these assays require the addition of a peptidase, usually trypsin, to release the fluorophore from the deacetylated substrate. This step could also result in false positives because of the possible protease inhibitory potential of some tested compounds. To reduce the potential pitfalls, few strategies have been considered so far. For example, the separation and detection of the substrate and its deacetylated product in reaction mixtures by high-performance liquid chromatography (HPLC) coupled to a fluorescence detector²¹. Another possibility is the use of mass spectrometry (MS) as a detector, which enables to distinguish the acetylated substrate from its deacetylated product in a quantitative manner²².

Despite all of the efforts in developing HDAC assays for screening purposes, the search for class-selective HDAC inhibitors requires multiple enzymatic assays on individual isoforms that are not always representative of their selectivity profile in the cell environment. HDACs are indeed differently compartmented within the cell, acting either within complexes or alone, and competing with other cell mechanisms^23–25. It is not rare that a compound identified as selective in an enzymatic screening will finally provide a pan-inhibitory behavior in cells. One example is compound ST-3595 showing an enzymatic selectivity for HDAC6 that could not be observed in Western-blot analyzes of treated cell samples²⁶. Unfortunately, antibody recognition assays are usually expensive and time-consuming to be used as preliminary high-throughput screenings.

Here, we describe a high-throughput UHPLC-ESI-MS/MS strategy for the simultaneous assessment of classes I and II HDAC activity directly in cells, with a focus on the identification of HDAC1 and HDAC6 selective inhibitors, respectively. This strategy takes into account the endogenous expression levels and the activity of individual HDAC isoforms within the cell environment by using a mixture of selective and cell-permeable HDAC substrates.

Methods

Cell-based HDAC inhibition assay

HeLa cells (American Type Culture Collection, Rockville, MD) were seeded in 96-well plates at a density of 6000 cells per well in 100 μl Minimum Essential Medium (Life Technologies, Zug, Switzerland), supplemented with 10% fetal calf serum, penicillin G (100 U/ml) and streptomycin (100 μg/ml), and cultured at 37 °C in a humidified atmosphere of 5% CO₂ for 24 h. Medium was then removed and cells were treated with 25 μl of test compounds (0.5% DMSO in culture medium, final concentration) together with 25 μl of a mixture containing the HDAC substrates MAL (Sigma-Aldrich, Steinheim, Germany), MOCPAC (Sigma-Aldrich) and BATCP (Sigma-Aldrich) or of each substrate added to separate wells at a final concentration of 21 μM each in assay buffer (50 mM Tris at pH 8.0 adjusted with HCl, 137 mM NaCl, 2.7 mM KCl, 1 mM MgCl₂) for 8 h. After incubation, 10 μl cold 6× RIPA buffer supplemented with protease inhibitor cocktail SigmaFast (Sigma-Aldrich) were added to the cells, followed by 160 μl of cold acetonitrile (MeCN). The plate was put for 10 min at −80 °C and then centrifuged at 11 500 rpm for 10 min. Supernatants were transferred into UHPLC-compatible 96-well plates to be analyzed on a UHPLC-ESI-MS/MS system consisting of an Acquity UPLC System (Waters, Milford, MA) connected to a Quattro Micro triple quadrupole mass spectrometer equipped with an ESI source operating in positive-ion mode (Waters). Samples were injected (5 μl) into a C18 Kinetex column (2.6 μm, 100 mm × 3 mm i.d.; Phenomenex, Torrance, CA) and eluted (0.8 ml/min, 40 °C) with MeCN and H₂O both containing 0.1% formic acid. A gradient of 5–45% MeCN in 10 min followed by 2 min with 98% MeCN (or 2 and 1 min, respectively, with a single substrate) was used. This was followed by a washing step with 98% MeCN for 2 min. After the washing step, the column was equilibrated with 5% MeCN during 4 min before the next injection. The conditions used for ESI-MS/MS detection were: cone voltage, 30 V; capillary temperature, 350 °C; source voltage, 3.0 kV; nitrogen was used as the sheath gas (800 l/h). MAL was detected by MS/MS at mass transition 446→346 (MAL) and 404→304 (dMAL) at retention times of 7.78 and 5.80 min, respectively (2.52 and 1.92 min with a single substrate). MOCPAC was detected at mass transition 494→449 (MOCPAC) and 439→394 (dMOCPAC) at retention times of 9.06 and 6.46 min, respectively (2.60 and 2.12 min with a single substrate). Finally, BATCP was detected at mass transition 500→400 (BATCP) and 476→376 (dBATCP) at retention times of 9.98 and 6.33 min, respectively (2.63 and 2.06 min with a single substrate). Peak areas were determined using automatic peak area detection of Masslynx 4.1 (Micromass, Manchester, UK). For each substrate, ratio between areas (deacetylated peptide/acetylated peptide) was determined to obtain the amount of deacetylated substrate. HDAC activity was calculated by dividing the amount of deacetylated substrate between control (100% HDAC activity) and test samples. IC₅₀ values was calculated using GraphPad PRISM (GraphPad Software Inc., San Diego, CA).

HDAC inhibition assay by using a mixture of HDAC isoforms

Test compounds were diluted in DMSO and added to the wells (5% DMSO final concentration in each well). MAL, MOCPAC and BATCP were mixed to a final concentration of 10.5 μM each or used separately at the same concentration. The reaction was initiated by the addition of a mixture of HeLa nuclear extract (20 U/well, Enzo Life Sciences, Lausen, Switzerland) and human recombinant HDAC6 (50 U/well, Sigma-Aldrich) diluted in assay buffer (50 mM Tris at pH 8.0 adjusted with HCl, 137 mM NaCl, 2.7 mM KCl, 1 mM MgCl₂) followed by incubation at 37 °C for 4 h. The reaction was stopped by the addition of cold MeCN. Samples were then treated and analyzed by UHPLC-ESI-MS/MS as described for the cell-based assay (multi-substrate or single-substrate accordingly).

Results and discussion

The general HDAC substrate MAL, the HDAC1-selective substrate MOCPAC and the HDAC6 (class II)-selective substrate BATCP were selected for the method development^18,19,21. All of these substrates are known fluorescence-labeled substrates that have been used to determine specific HDAC activity by fluorimetric measurement¹⁸. Given that the three of them possess overlapping fluorophores, they could not be used as a mixture with a fluorescent detection. By using an ESI-MS/MS detection, the hypothesis that a multi-substrate approach is useful for detecting selective HDAC inhibitors in a cell-based assay was tested (Figure 1).

Figure 1. Scheme of the cell-based multi-substrate assay for the identification of class-selective HDAC inhibitors.

HeLa cells were seeded into 96-well plates and incubated at 37 °C for 24 h. Cells were then treated with standard selective and non-selective HDAC inhibitors or vehicle, together with a mixture of substrates and incubated for 8 h followed by the addition of lysis buffer. Samples were then prepared and injected in a UHPLC system coupled to ESI-MS/MS that was used to monitor the three substrates and their deacetylated products in MRM mode. For each substrate, the ratio between peak areas (deacetylated peptide/acetylated peptide) was determined to calculate HDAC activity. To ensure that the deacetylase activity resulted in detectable and quantifiable signals, the number of cells was carefully adjusted so that the chromatographic peaks of each substrate and product showed a signal/noise ratio ≥ 20 (Figure 2). The three substrates were mixed to reach a final concentration of 21.0 μM each¹⁸. This concentration was chosen to fit the linear range of enzymatic rates, i.e. lower than substrate Km values (Km > 70 μM for each substrate, data not shown).

Figure 2. UHPLC-ESI-MS/MS detection of acetylated and deacetylated substrates with the cell-based multi-substrate assay. (A) Chemical structures of MAL, MOCPAC, BATCP; (B) chromatograms of MAL, MOCPAC, BATCP and their deacetylated products (designated as dMAL, dMOCPAC and dBATCP, respectively) in control and TSA-treated cells (31.25 nM) and (C) HDAC inhibitory activity by TSA at various concentrations with the MAL substrate.

Four HDAC inhibitors with known selectivity profiles were tested (Figure 3). Dose-response curves were built and the IC₅₀ values calculated for each substrate (Figure 1). This allowed to evaluate classes I and II HDAC inhibitory activity of each compound (Table 1) with the multi-substrate approach. To check if the mixture of substrates could interfere with the inhibition results and the selectivity profiles, the assay was conducted by incubating the cells with each substrate individually (single-substrate approach, Table 1). Overall, the IC₅₀ values obtained with both approaches were comparable in terms of activity and selectivity profiles. Therefore, by measuring the deacetylation ratio of MAL, the general HDAC inhibitory profile can be detected (General HDAC), whereas the monitoring of selective substrates MOCPAC (HDAC1) and BATCP (HDAC6) allows the determination of selectivity profiles.

Figure 3. HDAC inhibitors used for assay development.

Table 1. IC₅₀ values calculated for HDAC selective and pan-inhibitors in a cell-based assay with multi-substrate and single-substrate approaches.

		IC50 (nM)*
Inhibitor	Approach	General HDAC†	HDAC1‡	HDAC6¶	Selectivity index§
TSA	Multi-substrate	21.6 ± 0.7	19.7 ± 0.9	9.9 ± 0.6	2.0
	Single-substrate	19.2 ± 1.1	21.5 ± 2.0	11.6 ± 1.3	1.9
TBA	Multi-substrate	312.9 ± 9.9	>2000	212.2 ± 19.6	>9.4
	Single-substrate	313.4 ± 17.7	>2000	183.0 ± 14.3	>10.9
MS275	Multi-substrate	368.8 ± 7.4	122.2 ± 6.8	1318.0 ± 6.8	0.09
	Single-substrate	427.8 ± 7.9	204.9 ± 11.4	1132.7 ± 45.0	0.2
ST-3595	Multi-substrate	293.7 ± 21.4	263.8 ± 27.0	333.3 ± 19.0	0.8
	Single-substrate	280.4 ± 20.5	268.3 ± 5.7	299.0 ± 17.0	0.9

*Values are shown as the mean ± SD of three independent experiments in triplicate.

† Calculated by measuring the MAL substrate and its deacetylated product.

‡ Calculated by measuring the MOCPAC substrate and its deacetylated product.

¶ Calculated by measuring the BATCP substrate and its deacetylated product.

§ Selectivity index defined as the ratio IC₅₀ HDAC1/IC₅₀ HDAC6.

Multi-substrate: a mixture of MAL, MOCPAC and BATPC; single-substrate: MAL, MOCPAC and BATPC tested separately.

Selective and non-selective HDAC inhibitors were chosen to verify the applicability of the cell-based multi-substrate assay. Trichostatin A (TSA), a pan-inhibitor²⁷, showed a low-selectivity profile in our cell-based method (Table 1), a result that corresponds to the behavior that has already been observed when looking at protein acetylation. TSA has been shown to increase acetylation of both histones (H3, H4) and α-tubulin, which are targeted by class I HDAC isoforms and HDAC6, respectively^28–30. When the potent and selective HDAC6 inhibitor tubastatin A (TBA)³¹ was evaluated, a clear selective profile could be observed toward HDAC6 (Table 1). The selectivity of TBA toward the acetylation of α-tubulin, compared to histones has already been established by several authors^32–34. To see if the selectivity toward HDAC1 could also be detected, the selective HDAC1 inhibitor MS275 was evaluated. MS275 showed a clear selectivity toward HDAC1 (Table 1), also in line with the results previously observed for endogenous protein targets^35,36. Compound ST-3595 has been previously demonstrated to be an HDAC6-selective inhibitor in enzymatic assays, but it failed to show selectivity toward protein targets in cells²⁶. The evaluation of this compound in the method presented here confirmed that ST-3595 behaved as a pan-HDAC inhibitor in cells, inhibiting both HDAC1 and HDAC6 (Table 1). Given that class I HDACs are known to form complexes with others proteins and isoforms in cells^23–25, one possibility is that this compound acts in the presence of those protein complexes rather than interacting with isolated isoforms. Another hypothesis is the subcellular compartmentalization of the compound that would make it more available to the nuclear class I isoforms. The development of selective HDAC inhibitors coupled to a cell imaging-compatible fluorophore would be of great value to address this question.

To assess whether the multi-substrate method could also be applied to a mixture of HDAC isoforms, cells were replaced by a pool of HDAC isoforms (human recombinant HDAC6 added to a HeLa nuclear extract). IC₅₀ values were obtained with the four HDAC inhibitors previously tested (Table 2) and compared to values obtained with single isoforms (Supplementary Material, Table S1). TSA showed a similar inhibitory profile in the cell-based (Table 1) and enzymatic pool assays (Table 2), and the IC₅₀ values were comparable to those obtained with single enzymes (Table S1)²⁷. In the case of TBA, the selectivity profile toward HDAC6 was shown in both cell-based and enzymatic assays (Tables 1 and 2). The IC₅₀ values obtained in the enzymatic assay were consistent with those from the literature³¹. It is noteworthy that the cell-based assay required higher amounts of TBA for HDAC6 inhibition (Table 1). This is also consistent with concentrations used to detect α-tubulin acetylation upon TBA treatment, usually in the low μM range³¹. Interestingly, the HDAC1-selective compound MS275 showed stronger HDAC inhibition in the cell-based assay than in the enzymatic one (Tables 1 and 2). However, the HDAC1 selectivity was maintained in both methods and the enzymatic assay provided IC₅₀ values consistent with those previously obtained in enzymatic assays²⁷. Finally, the enzymatic assay with ST-3595 confirmed that the HDAC6 selectivity was comparable to what is found in the literature (Table 2)²⁶, but this did not reflect the selectivity profile in cells. As previously stated, the lack of selectivity of ST-3595 within the cell environment can be due to rather complex and multifactorial reasons. Most importantly, the cell-based multi-substrate approach was able to predict the selectivity profile previously demonstrated for ST-3595 toward endogenous HDAC targets in living cells²⁶.

Table 2. IC₅₀ values calculated for HDAC selective and pan-inhibitors using a mixture of HDAC isoforms in both multi-substrate and single-substrate approaches.

		IC50 (nM)*
Inhibitor	Approach	General HDAC†	HDAC1‡	HDAC6¶	Selectivity index§
TSA	Multi-substrate	4.1 ± 0.4	4.4 ± 0.4	4.5 ± 0.4	1.0
	Single-substrate	3.9 ± 0.21	4.9 ± 0.14	4.5 ± 0.17	1.1
TBA	Multi-substrate	11.7 ± 0.1	>2000	13.8 ± 1.7	144.9
	Single-substrate	8.8 ± 0.68	>2000	16.7 ± 1.8	119.8
MS275	Multi-substrate	>8000	436.5 ± 16.2	>8000	<0.05
	Single-substrate	>8000	368.6 ± 24.2	>8000	<0.05
ST-3595	Multi-substrate	136.0 ± 11.6	2838.3 ± 276.8	195.5 ± 21.5	14.5
	Single-substrate	144.6 ± 6.2	3177.7 ± 364.4	169.0 ± 25.5	18.8

*Values are shown as the mean ± SD of three independent experiments in triplicate.

† Calculated by measuring the MAL substrate and its deacetylated product.

‡ Calculated by measuring the MOCPAC substrate and its deacetylated product.

¶ Calculated by measuring the BATCP substrate and its deacetylated product.

§ Selectivity index defined as the ratio IC₅₀ HDAC1/IC₅₀ HDAC6.

Multi-substrate: a mixture of MAL, MOCPAC and BATPC; single-substrate: MAL, MOCPAC and BATPC tested separately. All assays were conducted with a mixture of HeLa nuclear extract and human recombinant HDAC6 as the source of enzymes.

Conclusions

A new cell-based multi-substrate UHPLC-ESI-MS/MS assay was developed as a tool for screening selective HDAC inhibitors. This new strategy was able to predict the selectivity profile of HDAC inhibitors in a cell environment, in line with protein acetylation profiles. Compared to the determination of protein acetylation levels, the cell-based multi-substrate approach is faster and less expensive, being suitable for high-throughput determination of HDAC selectivity. Moreover, it provides IC₅₀ values for individual HDAC isoforms in a single step reflecting the inhibition in a cell environment. In addition, the results provided here encourage the development of new isoform-specific cell-permeable HDAC substrates. In this study, it was assumed that MAL, MOCPAC and BATCP were able to keep their selectivity profiles in living cells, an issue that deserves further investigation. Finally, the method presented may be considered as a useful tool to accelerate the search for class-selective HDAC inhibitors in drug discovery.

Declaration of interest

The authors declare no conflict of interest.

Supplementary material available online

Acknowledgements

C.S.P. acknowledges the Pierre Mercier Foundation for funding. We are thankful to Dr Sabrina Dallavalle for providing compound ST-3595 and to Dr Yung-Sing Wong for fruitful discussions.