A Label-Free LC/MS/MS-Based Enzymatic Activity Assay for the Detection of Genuine Caspase Inhibitors and SAR Development

Abstract

The resurgence of interest in caspases (Csp) as therapeutic targets for the treatment of neurodegenerative diseases prompted us to examine the suitability of published nonpeptidic Csp-3 and Csp-6 inhibitors for our medicinal chemistry programs. To support this effort, fluorescence-based Csp-2, Csp-3, and Csp-6 enzymatic assays were optimized for robustness against apparent enzyme inhibition caused by redox-cycling or aggregating compounds. The data obtained under these improved conditions challenge the validity of previously published data on Csp-3 and Csp-6 inhibitors for all but one series, namely, the isatins. Furthermore, in this series, it was observed that the nature of the rhodamine-labeled substrate, typically used to measure caspase activity, interfered with the pharmacological sensitivity of the Csp-2 assay. As a result, a liquid chromatography/tandem mass spectrometry–based assay that eliminates label-dependent assay interference was developed for Csp-2 and Csp-3. In these label-free assays, the activity values of the Csp-2 and Csp-3 reference inhibitors were in agreement with those obtained with the fluorogenic substrates. However, isatin 10a was 50-fold less potent in the label-free Csp-2 assay compared with the rhodamine-based fluorescence format, thus proving the need for an orthogonal readout to validate inhibitors in this class of targets highly susceptible to artifactual inhibition.

Keywords

caspases nonpeptidic inhibitors LC/MS/MS assay artifactual inhibition

Introduction

Over the past 40 years, protease research has yielded important drugs, most notably for cardiovascular and viral diseases.¹ In the past decade, several proteases have also been implicated in neurodegenerative disorders, and more recently, caspase (Csp)–6 has been proposed as a target for therapeutic intervention in both Alzheimer² and Huntington³ diseases (HD). The clinical successes in the protease field have been possible because the available peptidomimetic protease inhibitors satisfied the practical pharmacological requirements of peripheral drugs. However, the use of peptidomimetics for CNS indications presents a much greater challenge, particularly for cysteine proteases such as the caspase family members. The first-generation caspase inhibitors were tetrapeptides mimicking caspase substrate cleavage sites and containing a reactive moiety at the C-terminus designed to react with the catalytic cysteine. Although efforts in the de-peptidization of such covalent inhibitors produced highly potent, selective, and cell-active Csp-1 and Csp-3 compounds, these new generations of inhibitors still lack suitable physicochemical properties to achieve acceptable peripheral exposure and, a fortiori, adequate CNS coverage. Consequently, nonpeptidic inhibitors have been sought as unique starting points for the development of CNS-active compounds.

After more than a decade of high-throughput screening (HTS) campaigns, there are still very few reports of nonpeptidic caspase inhibitors, and there is an absence of advanced lead series for the caspases implicated in neurological disorders. Due to this paucity of benchmark nonpeptidic caspase inhibitors and the poor historical HTS tractability of this target class, the current lead identification program for HD focused initially on the in vitro characterization of eight previously published nonpeptidic caspase inhibitor series ( Fig. 1 ) to determine their suitability as lead structures.

Figure 1.

Structures of investigated nonpeptidic caspase inhibitors 1 to 9 representing eight different series.

This article describes the optimization of Csp-2, Csp-3, and Csp-6 fluorescence-based enzymatic assays and highlights inconsistencies with previous studies. It was found that seven of the eight investigated chemical series did not confirm in the Csp-3 or Csp-6 assays. The development of novel, label-free liquid chromatography/tandem mass spectrometry (LC/MS/MS) Csp-2 and Csp-3 assays as secondary, orthogonal, and label-free readout assays for confirming potency and selectivity of the only validated series is also described, and assay results are discussed in comparison with those obtained by the fluorometric method. Although there is increasing interest in mass spectrometric readouts of catalytic end-point assays,⁴ there are only few examples describing such readouts for caspases or cysteine-proteases.⁵ The assays described in this publication are designed for compound profiling and could help close that gap.

Materials and Methods

Fluorescence-Based Caspase Assays

These assays are based on the cleavage of a fluorogenic substrate composed of the recognition sequence of the individual caspases and a fluorophore (either rhodamine or coumarin). All assays were performed in a 384-well microplate format (Matrix Screen Mate; Thermo Fisher Scientific, Hudson, NH) using a total assay volume of 20 µL per well. Standard buffer conditions for Csp-2 consisted of 50 mM MES (pH 6.5), 150 mM NaCl, and 1.5% sucrose. Assay buffer consisted of 50 mM HEPES (pH 7.4) and 100 mM NaCl for Csp-3 and 20 mM PIPES (pH 7.2), 100 mM NaCl, 1 mM EDTA, and 10% sucrose for Csp-6. Buffer additives such as the detergents 3[(3-cholamidopropyl)-dimethylammonio]-propanesulfonic acid (CHAPS), Pluronic F-127 (PF127), Triton X-100 (TX100), and the reducing agents dithiothreitol (DTT), glutathione (GSH), and β-mercaptoethanol (β-MCE) were varied as indicated in the Results section ( Table 1 and Fig. 2 ).

Table 1.

In Vitro Inhibitory Potencies of Compounds 1 and 2 for Csp-3 and Csp-6 Recorded in the Presence of Various Buffer Additives and in Different Assay Formats.

	Compound 1, IC50 (µM)		Compound 2, IC50 (µM)
Buffer Additives	Csp-3	Csp-6	Csp-3	Csp-6
Fluorescence intensity format
	—	—	1.40^a	0.15^a
10 mM DTT	0.08^b	—	—	—
10 mM DTT/0.1% CHAPS	>100	0.16^b	—	—
5 mM β-MCE/0.03% PF127	—	—	6.5	47
5 mM GSH/0.03% PF127	>100	0.46^b	7.2	16
5 mM GSH/0.01% TX100	>100	0.38^b	4.4	46
Label-free, LC/MS/MS-based format
5 mM GSH/0.01% PF127	>100	—	>100	—

β-MCE, β-mercaptoethanol; CHAPS, 3[(3-cholamidopropyl)-dimethylammonio]-propanesulfonic acid; PF127, Pluronic F-127 Csp, caspase; DTT, dithiothreitol; GSH, glutathione; LC/MS/MS, liquid chromatography/tandem mass spectrometry. Long dashes indicate data that are not available.

Literature value.²²

High Hill coefficient, partial inhibition.

Figure 2.

Concentration-response curves of 1 for Csp-6 (A) and Csp-3 (B) and concentration-response curves of 2 for Csp-6 (C) and Csp-3 (D) in the presence of different buffer conditions. The following buffer additives were used to characterize inhibitory activity: dithiothreitol (DTT)/ CHAPS (), glutathione (GSH)/Triton X-100 (), GSH/Pluronic F-127 (PF127) (), β-mercaptoethanol/PF127 (), or DTT without detergent (). Experimental details are described in the Materials and Methods. Csp, caspase.

The peptide-based inhibitors Ac-VDVAD-CHO (Csp-2), Ac-DEVD-CHO (Csp-3), and Ac-VEID-CHO (Csp-6) were used as reference compounds on each individual assay plate. Reference and test compound stock solutions were prepared in neat DMSO. The 3-fold dilution series with 11 concentrations were prepared in 100% DMSO. Then, 200 nL of the compound dilutions in 100% DMSO were added to each well containing 10 µL of freshly prepared solutions of 20 nM Csp-2, 0.6 nM Csp-3, or 8 nM Csp-6 in the respective assay buffers using a CyBi-Well vario manipulator (CyBio AG, Jena, Germany). Assay plates were incubated at 37 °C for 5 min before the enzymatic reaction was started by adding 10 µL of the substrate solution. Final substrate concentrations were 20 µM for Csp-2 R110- or AMC-based substrates (-VDVAD-) and 10 µM R110- or AMC-based substrates (-DEVD- or -VEID-), for Csp-3 and Csp-6, respectively. These concentrations correspond to the individually determined apparent K_M values (Suppl. Table S1). The final DMSO concentration used throughout all compound tests was 1%. Plates were incubated for 20 min at 37 °C, and reaction was stopped by adding 3 µL of a 5 M aqueous acetic acid solution followed by a short centrifugation step to remove air bubbles.

In Csp-2 substrate competition experiments, IC₅₀ values of selected Csp-2 compounds were recorded in standard Csp-2 buffer conditions with 5 mM GSH and 0.03% PF127 (w/v) in the presence of 20 µM, 100 µM, or 200 µM of Ac-VDVAD-AMC to investigate the competing effect between substrate and inhibitor.

Fluorescence was quantified using a Tecan Safire 2 microplate fluorometer (Tecan Austria GmbH, Grödig, Austria) with the following wavelength settings: excitation at 485 nm and emission at 535 nm for R110 and excitation at 380 nm and emission at 440 nm for coumarin 120. Under all conditions, it was ensured that the experiments performed for determining compound potencies were conducted in the linear range of the reaction. The sources of the other reagents used in this study are given in the supplemental information.

LC/MS/MS-Based Caspase Assays

In these assays, the enzymatic activities of Csp-2 and Csp-3 were determined by measuring the N-terminal cleavage products of the unlabeled peptides (Ac-VDVADGTQAS-NH₂ and Ac-MDLNDGTQAS-NH₂ for Csp-2 and Ac-MDEVDLASCD-NH₂ for Csp-3). The assays were run in 384-well plates (Matrix Screen Mate; Thermo Fisher Scientific) in a final volume of 30 µL per well. The buffer consisted of 50 mM MES (pH 6.5), 150 mM NaCl, 5 mM GSH, and 0.03% PF127 (w/v), and the final Csp-2 concentration was 20 nM for the Csp-2 assay. For the Csp-3 assay, the buffer consisted of 50 mM HEPES (pH 6.5), 150 mM NaCl, 0.03% PF127 (w/v), and 5 mM GSH, and the concentration of Csp-3 was 1 nM. Compound handling was conducted in the same way as for the fluorogenic assays. The reaction was started by adding the substrate solution. The final substrate concentrations of 100 µM and 20 µM corresponded to the K_M values determined for Csp-2 and Csp-3, respectively (Suppl. Table S1). The reaction was stopped after incubation at 37 °C for 20 min by the addition of trichloroacetic acid (TCA) at a final concentration of 5% (w/v). The LC/MS/MS-based analysis of the cleavage reaction is described in detail below.

Substrate and product peptide detection was achieved by multiple-reaction monitoring (MRM) using an Agilent 1100 series chromatography system with a photodiode array detector (Agilent Technologies, Waldbronn, Germany) connected to an API 3000 mass spectrometer (Life Technologies, Darmstadt, Germany). Source parameters were set as follows: polarity ES+, capillary 5.5 kV, desolvation temperature 550 °C, nebulizer gas 8, curtain gas 8, and entrance potential 10; further analyte-specific parameters are shown in Supplemental Table S2.

Analytes were separated using an Acquity UPLC BEH C18 column (1.7 µm; 2.1 × 30 mm, Waters, Eschborn, Germany) at 60 °C column temperature. Mobile phases: A consisted of 0.1% v/v formic acid in LC/MS-grade water, and B consisted of acetonitrile with 0.1% v/v formic acid. The gradient was started at 1.0% B, increased linearly to 17% B in 1.5 min and then to 98% B in 0.5 min, held isocratic at 98% B for 1.0 min, and returned to initial conditions in 0.5 min. The column was equilibrated with initial conditions for 1.0 min before next injection. For accurate injection, the 10-µL sample loop was 5-fold overfilled. Sample storage temperature was set to 8 °C.

Analytes were quantified using MRM methods by comparing the relative intensities (integration of peak areas under the curve, AUC) obtained in incubations with different concentrations. Software used for mass spectrometry data acquisition and analysis was Analyst 1.4.1 (MDS Sciex, Concord, Canada). Transitions monitored are summarized in the supplemental information (Suppl. Table S2).

Data Analysis

To confirm the statistical validity of the assays, Z′-factors were calculated using the median values and standard deviations of the negative (1% DMSO) and positive controls (1 µM peptide-based inhibitors Ac-VDVAD-CHO, Ac-DVED-CHO, and Ac-VEID-CHO for Csp-2, Csp-3, and Csp-6 respectively). All screening runs with a Z′-factor >0.5 were accepted for further analysis (Suppl. Table S3). Normalized percent inhibition for each compound concentration was calculated using the medium values of the negative and positive controls. Concentration-response curves were analyzed with Evotec’s in-house A+-software using a four-parameter Hill equation. Reported IC₅₀ values are the average of at least two independent experiments. Figures were prepared using GraphPad Prism (GraphPad Software, San Diego, CA).

Synthesis of Substrates and Inhibitors

Label-free substrates were synthesized using standard Fmoc solid-phase peptide synthesis protocols (for more details, see supplemental information). Compounds 1 to 9 ( Fig. 1 ; Tables 1 and 2 ) were prepared according to published procedures, whereas novel compounds 10 and 10a,b ( Table 3 ) were synthesized using a minor variation of the literature method reported by Lee et al.,⁶ the details of which are described in the supplemental information. All final compounds were purified by reverse-phase high-performance liquid chromatography (HPLC). The sources of all other reagents used for synthesis are described in the supplemental information.

Table 2.

In Vitro Caspase-3 Inhibitory Potency of Compounds 3 to 9 Recorded in the Presence of Different Detergent, Reducing Agents, and Substrates.

	IC₅₀ (µM)
Detergent	0.03% PF127	0.01% TX100	0.03% PF127	Literature IC₅₀ (µM)
Reducing agent	5 mM GSH	5 mM GSH	5 mM GSH
Assay readout	Fluorogenic	Fluorogenic	Label free
3	<0.002	<0.002	0.0016	0.0084⁶
4	>100	>100	>100	0.023²⁴
5	>100	>100	>100	40²⁵
6	>100	>100	>100	4.3²⁵
7	24	28	37	0.11²⁶
8	>100	>100	>100	66%, 20 µM²⁸
9	>100	>100	>100	14²⁹
Ac-DEVD-CHO	0.0055	0.0075	0.0066	0.0018¹⁸

GSH, glutathione; PF127, Pluronic F-127; TX100, Triton X-100.

Table 3.

In Vitro Csp-2 and Csp-3 Inhibitory Potency for the Isatin Series 10 Using R110, AMC, or Label-Free Substrates.


Compound ID	Csp-2 Rh110, IC₅₀	Csp-2 AMC, IC₅₀	Csp-2 Label Free, IC₅₀	x-Fold Shift Rh110/AMC	x-Fold Shift AMC/Label Free	Csp-3 Rh110, IC₅₀	Csp-3 AMC, IC₅₀	Csp-3 Label Free, IC₅₀
3	0.77	1.6	3.2	2.1	2.0	<0.002	0.005	0.003
(R/S)-10	0.068	1.2	1.6	18	1.3	15	21	—
(S)-10a	0.025	1.3	2.0	52	1.5	11	10	—
(R)-10b	28	56	35	2	0.63	76	—	—
Ac-VDVAD-CHO	0.054	0.046	0.047	0.85	1.0	0.031	0.015	0.031
Ac-DEVD-CHO	>1	>1	15	—	—	0.005	0.007	0.007

IC₅₀ values are reported in µM. Csp, caspase. Long dashes indicate that the data are not available.

Results and Discussion

Optimization of Fluorescence Assays

Enzymatic assays for Csp-2, Csp-3, and Csp-6 inhibitor screening were implemented using a fluorometric assay principle, which measures the increase in fluorescence after protease cleavage of a specific substrate. Sequence-optimized substrates as previously described were used in these assays.⁷ These substrates show micromolar affinities for caspases as confirmed by substrate titration experiments, well suited for the identification of inhibitors during compound screening. Substrate concentrations for each caspase assay were adjusted to match their K_M values (Suppl. Table S1). Enzyme concentrations were also adjusted to achieve turnover rates of approximately 10% after a 20-min incubation. These turnover rates ensured that the enzyme activity is monitored under linear conditions. Finally, assay conditions were validated using the known peptide-based inhibitors Ac-VDVAD-CHO (Csp-2), Ac-DEVD-CHO (Csp-3), and Ac-VEID-CHO (Csp-6).

Selection of Reducing Agent

Caspases are sensitive to oxidation and require a reducing environment to preserve enzymatic activity. To maintain the enzyme in a reduced state, DTT is commonly used. However, in an HTS campaign for Csp-8, DTT resulted in an unusually high false-positive hit rate attributed to the formation of reactive oxygen species (ROS).⁸ DTT has also been shown to cause indirect inactivation of other enzymes susceptible to oxidation,⁹ and two recent publications^10,11 have stressed the need, whenever possible, to employ weaker reducing agents that do not lead to ROS generation. As there is no apparent requirement for a strong reducing agent to stabilize or enhance caspase activity, the aim was to assess compounds under more physiological assay conditions. GSH was therefore evaluated in titration experiments to determine a concentration that is effective in stabilizing enzyme activity (Suppl. Fig. S10). As GSH was shown to be “caspase compatible”, DTT was substituted with this less reactive reducing agent as part of the assay optimization.

Selection of Detergents

The importance of detergent for inhibitor screening and testing has been extensively studied and described by the Shoichet research group¹² and others.^13,14 To summarize, detergents can ameliorate assay performance by stabilizing the enzyme and reducing nonspecific protein adsorption onto the plastic surface. Also, detergent sensitivity has been proposed as a criterion to weed out aggregation-based inhibitors. For caspases, 0.1% (w/v) CHAPS is universally employed for assaying peptidic and nonpeptidic inhibitors because it strongly increases enzyme activity, especially for Csp-6.¹⁵ Nevertheless, zwitterionic detergents such as CHAPS, when used at millimolar concentrations, can negatively affect compound solubility.¹⁶ Furthermore, at such high concentrations, the detergent itself may bind to the target, thereby interfering with the assay readout. Consequently, and since no study has examined the effect of detergents on promiscuous inhibitors for caspases, two nonionic detergents were evaluated during assay development: TX100, which has been reported to be effective in disrupting compound aggregates,¹⁷ and PF127, which is a mild detergent, is compatible with cell cultivation, and has a very high critical micelle concentration. These two nonionic detergents alleviate false-negative inhibition due to compound precipitation because they can be employed at a lower concentration than zwitterionic detergents, which also limits the risk of detergent binding to the protein. As previously reported for CHAPS,¹⁵ a dramatic increase in Csp-2, Csp-3, and Csp-6 catalytic activity was also observed in the presence of TX100 or PF127, reaching a maximum with 0.01% (w/v) and 0.03% (w/v), respectively. The K_M values in substrate titration experiments were not affected by the different detergents or different detergent concentrations. Since the GSH/TX100 and GSH/PF127 combinations are less prone to interference by redox-cycling or aggregate-forming compounds, they were preferred over the commonly used DTT/CHAPS combination.

Evaluation of Csp-6 Inhibitors

There is evidence from an HD mouse model that the genetic mutation of the motif in Huntingtin (HTT), thought to be cleaved by Csp-6, alleviates the pathogenic effects of mutant HTT.³ The replication, in part or fully, of this beneficial effect with a Csp-6 inhibitor would represent a key step toward the validation of this pathogenic mechanism and the development of a therapeutic agent. Therefore, the identification of Csp-6 small-molecule inhibitors was a high priority for the HD program, which started with the verification of the activity of the two published nonpeptidic inhibitors of Csp-6 and their respective selectivity against Csp-3 ( Fig. 1 , compounds 1 and 2).

First, the sub-micromolar Csp-6 activity of compound 1 was reproduced using the combination of detergent and reducing agent (DTT/CHAPS) reported by Scott et al.,¹⁸ but the corresponding concentration-response curve exhibited a Hill coefficient of 3.3 ( Fig. 2A and Table 1 ). Using GSH/TX100 or GSH/PF127, only partial inhibition was observed. Furthermore, an incomplete concentration-response curve characterized by bell-shaped curves along with a high Hill coefficient of 2.7 for the latter combination of reducing agent and detergent has been obtained. Regarding the Csp-3 inhibition of 1, the published 80-nM activity, determined in the presence of DTT and without detergent, was also reproduced but yielded an equally steep concentration-response curve with a maximum inhibition of ~40% ( Fig. 2B ). This nanomolar activity was completely abolished in the presence of GSH/PF127, GSH/TX100, or even DTT/CHAPS. Such a detergent dependence^12–14 and a steep Hill slope¹⁹ have been proposed as markers of aggregation-based nonstoichiometric inhibition. Although there has been some recent debate on the usefulness and accuracy of detergent sensitivity as a criterion to eliminate artifactual inhibitors,²⁰ a Hill coefficient well above 1 in the absence of detergent combined with complete abrogation of activity in the presence of ionic or nonionic detergent strongly suggests that 1 is not a stoichiometric inhibitor of Csp-3. In the case of Csp-6, the detergent effect was not as pronounced as for Csp-3, but a high Hill coefficient coupled with unusual inhibition curves such as partial inhibition indicated suspicious behavior. These observations were corroborated by a report that an analogous compound of 1 was identified as a possible aggregator in a detergent-based assay.²¹ A nonspecific mechanism of action of 1 on Csp-6 therefore appeared highly probable, thus disqualifying it as a potential lead compound.

Compound 2²² is a literature-reported nonpeptidic inhibitor of Csp-6 with sub-micromolar potency. This compound contains an undesirable malononitrile moiety, but it also has an unprecedented 10-fold selectivity against Csp-3 that may be the basis for lead optimization. In the present Csp-6 assay (GSH/PF127 or GSH/TX100), compound 2 showed only high double-digit micromolar activity, whereas in the Csp-3 assay, the low micromolar literature value was reproduced ( Fig. 2C , D and Table 1 ). Prompted by this >100-fold loss in Csp-6 activity, we attempted to reproduce the published experimental conditions. Surprisingly, the reported protocol did not include a reducing agent despite a 2-h incubation at 37 °C. Consequently, these observations raised the question of whether the nanomolar inhibition of Csp-6 activity measured in the absence of reducing agent was not due to stoichiometric inhibition by 2 but rather the result of a decline in Csp-6 activity under the nonreducing conditions. However, it was also considered that the loss of inhibitory potency of 2 compared with the literature value may be artificially caused by the current modified experimental conditions. The series of structurally related isatins such as 3 are well-characterized potent Csp-3 inhibitors that are very weakly active in the presence of DTT because they react with this reducing agent.²³ Similarly, the potential reactivity of GSH toward the malononitrile group was tested by assaying 2 on Csp-6 in the presence of β-MCE, which is often applied to assay isatins.⁶ Despite this substitution of reducing agent, 2 remained very weakly on Csp-6. Eventually, it was found that 2 quickly degraded under experimental conditions and was also not stable over time in the DMSO stock solution (data not shown), which is probably due to the highly reactive malononitrile. Taken together, these results eliminated 2 from further consideration as a starting point for a compound optimization program.

Evaluation of Csp-3 Inhibitors as Starting Point for the Design of Csp-2 or Csp-6 Inhibitors

In the absence of validation data for the two Csp-6 inhibitors 1 and 2 and for a published Csp-2 nonpeptidic inhibitor, we attempted to exploit the relatively large number of literature nonpeptidic Csp-3 inhibitors as starting points for the design of selective Csp-2 or Csp-6 inhibitors. Six chemical series were identified, and representative compounds were synthesized ( Fig. 1 , compounds 3–9) and assayed in the presence of the two combinations of reducing agent and detergent GSH/PF127 or GSH/TX100. Only the already well-characterized isatin 3⁶ reproducibly inhibited Csp-3 activity, whereas in the five other series, compounds (4–6 and 8–9) were inactive at up to 100 µM, and compound 7 was only ~25 µM in both conditions ( Table 2 ). As for Csp-6 unconfirmed inhibitors 1 and 2, we sought to understand the discrepancy between the current findings and the data in the literature. First, compound 4,²⁴ reported as a 23-nM Csp-3 inhibitor when tested in the presence of β-MCE/CHAPS, produced no inhibition in the current conditions at up to 100 µM ( Table 2 ). Compound 4 was assayed with β-MCE/PF127 to rule out a reducing agent–related effect, but again 4 was found to be inactive at up to 100 µM. It was decided not to pursue this compound further as the confirmed detergent dependency for the activity of 4 and the noncompetitive nature of the inhibition²⁴ are hallmarks of nonstochiometric inhibition caused by compound aggregation.¹² In the case of the two analogues 5 and 6,²⁵ reported to have micromolar activity in Csp-3, the absence of both reducing agent and detergent in the original study may account for the lack of Csp-3 inhibition at up to 100 µM under current conditions ( Table 2 ) and justified no further investigation of the compounds. The reason for the 230-fold loss of activity of potent literature compound 7²⁶ in the current assay was elucidated in a subsequent publication from the same authors who first disclosed this compound series.²⁷ They elegantly demonstrated that 7 inactivates Csp-3 in the presence of DTT and oxygen, presumably through a redox cycle that would not take place when DTT is substituted by the weaker reducing agent GSH, hence explaining the lack of activity for 7 when GSH is used. Finally, the last two compounds, 8²⁸ and 9,²⁹ were only double-digit micromolar literature inhibitors in the presence of DTT/CHAPS and showed no activity at up to 100 µM under current conditions. This small loss in inhibitory potency of at least ~5-fold was not encouraging but insufficient to classify these compounds as false positives. Given that 8 and 9 exhibit low reported activity, validation with an orthogonal method was necessary to determine whether they are bona fide Csp-3 inhibitors or possibly redox artifacts as the structures, particularly that of 8, would suggest.

Inconsistent SAR in the Csp-2 Inhibitor Series Leads to the Development of an LC/MS/MS Assay

At this juncture, only the isatin 3 satisfied the criteria for a genuine Csp-3 inhibitor, displaying robust and reproducible inhibition of the enzyme ( Table 2 ). Isatins contain a reactive α-ketoamide moiety that limits their cellular activity,⁶ but in the absence of any alternative, we decided to explore the feasibility of deriving potent and selective inhibitors of Csp-2 or Csp-6 in this series originally developed for Csp-3. This medicinal chemistry effort, which will be reported elsewhere, failed to identify any potent Csp-6 inhibitors; however, the effort did produce a series of potent Csp-2 inhibitors that were highly selective against Csp-3. The lead Csp-2 compound 10a displayed an IC₅₀ value of 25 nM in Csp-2 (GSH/PF127, GSH/TX100) and an unprecedented 440-fold selectivity against Csp-3 ( Table 3 ). This potency was stereoselective as the R-isomer 10b was much less active. To progress with this compound, we needed to confirm that its mode of inhibition was still competitive as reported for the initial Csp-3 lead 3.⁶ This assessment could not be made with the rhodamine 110 (R110)–labeled substrate (Suppl. Fig. S1) because such a bivalent fluorophore, initially selected for its lower susceptibility to fluorescence interference,⁷ provides only a limited linear dynamic range in the assay.³⁰ The Csp-2 and Csp-3 R110-based substrates were therefore replaced with the corresponding Coumarin 120 (AMC) substrates, which carry only one cleavage site and are suitable for determining the mode of action of 10a. These experiments with the AMC-based substrates confirmed competitive inhibition of Csp-2 and Csp-3 by 10a; however, a loss of potency in inhibiting Csp-2 was observed compared with the values obtained with the rhodamine substrate. In particular, lead compound 10a displayed a 50-fold lower potency on Csp-2 inhibition, and its selectivity against Csp-3 decreased from 440-fold to 7.7-fold ( Table 3 ), whereas reference isatin 3 showed no significant activity shift from rhodamine to the AMC substrate. This discrepancy was most likely attributable to the substrate labels as all other experimental conditions were kept identical, so it became necessary to understand the origin of this variation. Except for the determination of the mechanism of inhibition described above, R110 had been preferred to AMC because it is much less susceptible to interference from autofluorescence or fluorescence quenching. However, interference of the label in the substrate recognition and/or cleavage steps is another recognized potential source of artifactual activity³¹ that has been given little attention in the caspase field. In this regard, it is important to note that both AMC and R110 dyes flank the cleavable amide bond (Suppl. Fig. S1) and, ipso facto, are upon substrate binding located within the immediate vicinity of the catalytic residues. As a result, these nonphysiological moieties can theoretically perturb substrate affinity and specificity, which could explain the 50-fold difference in Csp-2 inhibition potency (IC₅₀) of compound 10a. In light of these observations, it became evident that an orthogonal method for IC₅₀ determination was necessary to clarify whether the observed increased potency to inhibit Csp-2 and the observed selectivity in this series was genuine.

Need for Csp-2 Label-Free Assay: Choice of Substrate

In looking for a more physiologically relevant readout, we attempted to develop a secondary assay based on Csp-2 cleavage of an oligopeptide derived from the residues around the cleavage site of a Csp-2 protein substrate. Quenched fluorescent substrates described for caspases in the seminal work of Stennicke et al.³² would satisfy the criteria, but the distal fluorophore and quencher could still introduce nonspecific interactions. Instead, we decided to develop a label-free detection method with an LC/MS/MS readout to rule out any reporter interference. In the context of a project whose goal is the identification of tool compounds for HD target validation, the sequence of amino acids around the HTT Csp-2 cleavage site³³ was chosen as the basis for the design of the oligopeptide substrates. In the particular case of Csp-2, it is well documented that a P5 residue is required for efficient cleavage.³⁴ For this reason, the oligopeptide was extended on the N-terminus side up to the P5 position. In the absence of C-terminus residue requirements for efficient cleavage,³⁴ the first five amino acids of the HTT sequence beyond the Csp-2 cleavage site were also included, resulting in a decapeptide Ac-MDLND↓GTQAS-NH₂. It was to be expected that this decapeptide would be a substrate for Csp-2 as it contained the two key P1 and P4 Asp residues present in the optimal recognition sequence VDVAD. However, the other three residues, particularly the critical P5, were different in HTT than in the optimal sequence. Hence, a hybrid peptide, Ac-VDVAD↓GTQAS-NH₂, was also synthesized in which the N-terminus HTT-derived sequence was replaced by the canonical P5–P1 VDVAD motif.

Label-Free Caspase Assay: Development and Validation

Csp-2 activity was determined by quantifying the concentrations of the substrate and the N-terminal cleavage product Ac-VDVAD using LC/MS/MS analysis. Based on the hydrophilic nature of this product, a C18 column was used to separate the N-terminal cleavage product from substrate and the C-terminal side product ( Fig. 3A ). Positive ion mode (ESI+) was selected as a detection method to quantify the amount of the N-terminal cleavage product. Instrument parameters were adjusted so that the LC/MS/MS method allowed monitoring of the substrate (Ac-VDVADGTQAS-NH₂) and the cleavage products over a concentration range from 0.1 to 50 µM, with lower limits of detection for substrate, N-terminal (Ac-VDVAD), and C-terminal (GTQAS-NH₂) cleavage pro-ducts of 0.1 µM, 0.2 µM, and 1.0 µM, respectively ( Fig. 3B ). Concentration-dependent response of the analytes was linear in the relevant concentration range, thus providing ideal readouts for monitoring the enzymatic cleavage reactions. For product quantification, the N-terminal cleavage product was selected because of the >80-fold higher sensitivity of the MRM method for this cleavage product combined with significant retardation on the chromatography column compared with the early eluting C-terminal cleavage product, which reduces potential matrix effects (Suppl. Table S2).

Figure 3.

Development of Csp-2 liquid chromatography/tandem mass spectrometry (LC/MS/MS)–based assay (A): chromatograms of substrate Ac-VDVADGTQAS-CONH₂ (black peak) and N-terminal cleavage product Ac-VDVAD-CO₂H (red peak). (B) Sensitivity of LC/MS/MS instrumentation: linear response of the integrated LC/MS/MS signal expressed as area under the curve (AUC) in dependence of substrate and product concentrations: Ac-VDVADGTQAS-CONH2 (), Ac-VDVAD-CO₂H (), and H-GTQAS-CONH₂ (). (C) Overnight cleavage kinetics of Huntingtin (HTT)–derived peptide: Ac-MDLNDGTQAS-CONH₂ () and Csp-2 optimized substrate Ac-VDVADGTQAS-CONH₂ (). Insert: Overnight cleavage kinetics of HTT-derived peptide Ac-MDLNDGTQAS-CONH₂ in the presence () and absence of Csp-2 (). Solid lines represent a fit to first-order exponential equation. (D) Plot of enzyme activity versus substrate concentration of the catalysis of Csp-2 with a sequence-optimized substrate () and HTT-derived peptide Ac-MDLNDGTQAS-CONH₂ (). Solid lines represent fits to the Michaelis-Menten equation. Inset: Apparent activity for HTT-derived peptide Ac-MDLNDGTQAS-CONH₂.

The two selected substrates described above were analyzed for cleavage by Csp-2 ( Fig. 3C ). Whereas the sequence-optimized Csp-2 substrate Ac-VDVADGTQAS-NH₂ was cleaved nearly completely (~75%) overnight, only a very small fraction (2.5%) of the HTT-derived substrate was proteolysed during the same period ( Fig. 3C , inset). The linear range of product formation was confirmed with 10% turnover within an incubation time of 20 min at a 20-nM enzyme concentration. Subsequently, K_M determinations were performed by quantifying cleavage product formation within the first 60 min. With the optimized substrate Ac-VDVADGTQAS-NH₂, a hyperbolic dependence of concentration against catalytic Csp-2 activity was obtained ( Fig. 3D ), and the apparent K_M value was 5-fold higher than for labeled substrates in the fluorogenic assay. (Suppl. Table S1). For the nonoptimized HTT-derived substrate Ac-MDLNDGTQAS-NH₂, the turnover was too low to determine a K_M value ( Fig. 3D , inset). This result does not necessarily question the finding that HTT is a physiological substrate for Csp-2; rather, it could be evidence that higher order structural features such as tertiary structure may be required for efficient proteolysis of HTT by Csp-2.

In this Csp-2 label-free assay, the activity of the reference isatin 3 and peptidic inhibitor Ac-VDVAD-CHO were in agreement with the data obtained with the fluorogenic substrates ( Table 3 ). In the new Csp-2 series, the LC/MS/MS Csp-2 assay data were in line with the AMC assay results for the compounds tested ( Table 3 ). The lead compound 10a remained a low micromolar inhibitor as observed in the AMC-based assay, resulting in 80-fold loss of activity compared with the value obtained with the R110-labeled substrate (25 nM). A plausible conclusion is that the R110-based assay overestimated the potency of 10a in Csp-2, and consequently its selectivity against Csp-3 was not 440-fold but only ~5-fold. On the other hand, the AMC-based data appear to be reliable within this small subset of compounds as no potency deviation was observed compared with the label-free data. It is particularly intriguing that the R110-related phenomenon manifested itself for one of the stereoisomers, 10a (50-fold shift), and not for the other, 10b (2-fold shift; Table 3 ). This finding is another illustration that apparent SAR can never be interpreted as evidence of bona fide inhibition, and observed results should be treated with caution.³⁵ The mechanism by which R110 enhances the activity of 10 is unknown at present.

To reinvestigate the Csp-3 inhibitors 3 to 9 that were eliminated due to lack of activity or suspicious behavior such as sensitivity to detergent, an LC/MS/MS Csp-3 assay was set up based on the same principle as previously described for Csp-2. In light of the poor affinity of the oligopeptide derived from the Csp-2 HTT cleavage site, only a hybrid peptide was prepared using the canonical Csp-3 sequence DEVD for P1 to P4, whereas nonessential amino acids (P1′–P5′) originated from HTT’s corresponding residues at one of its Csp-3 cleavage sites (position D513). The resulting decapeptide Ac-MDEVD↓LASCD-NH₂ has been characterized by a K_M of 20 µM, only 2-fold higher than for the corresponding AMC and R110 substrates (Suppl. Table S1). In this newly developed Csp-3 LC/MS/MS assay, all compounds except the isatin 3 were inactive at concentrations up to 100 µM (4–6 and 8–9) or showed only very weak inhibition (compound 7; 37 µM). These findings are in complete agreement with the results from the fluorogenic assays under optimized conditions ( Table 2 ), which confirmed the earlier decision not to pursue these compounds.

Various conditions of caspase proteolytic assays in a fluorescence format were optimized and a novel LC/MS/MS-based method to measure the inhibition of Csp-2 and Csp-3 activity was developed. This label-free LC/MS/MS-based assay, used in combination with carefully selected reducing agents and detergents, has provided tools to identify and correct discrepancies in potencies reported for various caspase inhibitors. Moreover, this LC/MS/MS-based method can be used as a secondary screen to confirm hits from an HTS campaign thanks to the 384-well format, which allows for up to 2000 data points per week and thus represents a well-suited alternative to the more artifact-prone fluorescence-based assay. Although currently not of sufficient throughput to serve as a primary screening assay, this protocol could be used as the basis for further developments in the direction of high-throughput mass spectrometry–based screening applicable to large compound collections.

The weak or lack of caspase activity of the compounds presented here illustrates the difficulties seen in conventional caspase screening to identify viable nonpeptidic leads and shows that, due to the challenging nature of caspases as drug targets, assay conditions measuring the inhibitory potency of compounds in vitro need to be carefully selected. Thorough data interrogation is crucial to ensure that accurate structure-activity relationships are developed. Based on this experience, it is recommended that future caspase screening campaigns compare results obtained under different assay conditions and include at least one orthogonal assay. These steps will enable researchers to distinguish genuine enzyme inhibitors from compounds that produce artifactual inhibition due to compound aggregation, redox cycling, or reliance on nonphysiological substrates.

Footnotes

Acknowledgements

We gratefully acknowledge the supply of recombinant human Csp-2, Csp-3, and Csp-6 by Drs. James Wells and Michelle Arkin from the Small Molecule Discovery Center at the University of California, San Francisco. We also thank Drs. Fred Brookfield, Florence Eustache, and Mark Kerry from Evotec for the synthesis of compounds 1 to 10 and 10a–b.

Declaration of Conflicting Interests

Evotec conducted the research described through a fee-for-service agreement for CHDI Foundation. As employees of CHDI Management and advisors to CHDI Foundation, MCM, CD, HP, and IM-S were all intimately involved in the study design, data collection and analysis, decision to publish, and preparation of the manuscript.

Funding

This work was funded and conducted through fee-for-service contract research on behalf of CHDI Foundation, a privately-funded nonprofit biomedical research organization exclusively dedicated to developing therapeutics that slow the progression of Huntington disease.

Supplementary material for this article is available on the Journal of Biomolecular Screening Web site at .

References

Turk

Targeting Proteases: Successes, Failures and Future Prospects. Nat. Rev. Drug Discov. 2006, 9, 785–799.

LeBlanc

A. C.

The Role of Apoptotic Pathways in Alzheimer’s Disease Neurodegeneration and Cell Death. Curr. Alzheimer Res. 2005, 2, 389–402.

Graham

R. K.

Deng

Slow

E. J.

. Cleavage at the Caspase-6 Site Is Required for Neuronal Dysfunction and Degeneration Due to Mutant Huntingtin. Cell 2006, 125, 1179–1191.

Greis

M. P.

Mass Spectrometry for Enzyme Assays and Inhibitor Screening: An Emerging Application in Pharmaceutical Research. Mass Spectrom. Rev. 2007, 26, 324–339.

Rajapaksha

T. W.

Peter

M. E.

. Assays of Endogenous Caspase Activities: A Comparison of Mass Spectrometry and Fluorescence Formats. Anal. Chem. 2006, 78, 4945–4951.

Lee

Long

S. A.

Murray

J. H.

. Potent and Selective Nonpeptide Inhibitors of Caspases 3 and 7. J. Med. Chem. 2001, 44, 2015–2026.

Liu

Bhalgat

Zhang

. Fluorescent Molecular Probes V: A Sensitive Caspase-3 Substrate for Fluorometric Assays. Bioorg. Med. Chem. Lett. 1999, 9, 3231–3236.

Smith

G. K.

Barrett

D. G.

Blackburn

. Expression, Preparation, and High-Throughput Screening of Caspase-8: Discovery of Redox-Based and Steroid Diacid Inhibition. Arch. Biochem. Biophys. 2002, 399, 195–205.

Tjernberg

Hallén

Schultz

. Mechanism of Action of Pyridazine Analogues on Protein Tyrosine Phosphatase 1B (PTP1B). Bioorg. Med. Chem. Lett. 2004, 14, 891–895.

10.

Soares

K. M.

Blackmon

Shun

T. Y.

. Profiling the NIH Small Molecule Repository for Compounds That Generate H₂O₂ by Redox Cycling in Reducing Environments. Assay Drug Dev. Technol. 2010, 8, 152–174.

11.

Johnston

P. A.

Redox Cycling Compounds Generate H₂O₂ in HTS Buffers Containing Strong Reducing Reagents—Real Hits or Promiscuous Artifacts?

Curr. Opin. Chem. Biol. 2011, 15, 174–182.

12.

Shoichet

B. K.

Screening in a Spirit Haunted World. Drug Discov. Today. 2006, 11, 607–615.

13.

Ryan

A. J.

Gray

N. M.

Lowe

P. N.

. Effect of Detergent on “Promiscuous” Inhibitors. J. Med. Chem. 2003, 46, 3448–3451.

14.

Coan

K. E.

Ottl

Klumpp

Non-Stoichiometric Inhibition in Biochemical High-Throughput Screening. Expert Opin. Drug Discov. 2011, 6, 405–417.

15.

Stennicke

H. R.

Salvesen

G. S.

Biochemical Characteristics of Caspases-3, -6, -7, and -8. J. Biol. Chem. 1997, 272, 25719–25723.

16.

Steuer

Heinonen

K. H.

Kattner

. Optimization of Assay Conditions for Dengue Virus Protease: Effect of Various Polyols and Nonionic Detergents. J. Biomol. Screen. 2009, 14, 1102–1108.

17.

Feng

B. Y.

Shoichet

B. K.

A Detergent-Based Assay for the Detection of Promiscuous Inhibitors. Nat. Protoc. 2006, 1, 550–553.

18.

Scott

C. W.

Sobotka-Briner

Wilkins

D. E.

. Novel Small Molecule Inhibitors of Caspase-3 Block Cellular and Biochemical Features of Apoptosis. J. Pharmacol. Exp. Ther. 2003, 304, 433–440.

19.

Shoichet

B. K.

Interpreting Steep Dose-Response Curves in Early Inhibitor Discovery. J. Med. Chem. 2006, 49, 7274–7277.

20.

Habig

Blechschmidt

Dressler

. Efficient Elimination of Nonstoichiometric Enzyme Inhibitors from HTS Hit Lists. J. Biomol. Screen. 2009, 14, 679–689.

21.

Jadhav

Ferreira

R. S.

Klumpp

. Quantitative Analyses of Aggregation, Autofluorescence, and Reactivity Artifacts in a Screen for Inhibitors of a Thiol Protease. J. Med. Chem. 2010, 53, 37–51.

22.

Chu

Rothfuss

Chu

. Synthesis and In Vitro Evaluation of Sulfonamide Isatin Michael Acceptors as Small Molecule Inhibitors of Caspase-6. J. Med. Chem. 2009, 52, 2188–2191.

23.

Brown

D. G.

Urbanek

R. A.

Jacobs

R. T.

. In Chemical Strategies for Identifying False Positives: Applications to Caspase-3 Screening, 226th ACS National Meeting, New York, NY, September 7–11, 2003, MEDI-108.

24.

Kravchenko

D. V.

Kuzovkova

Y. A.

Kysil

V. M.

. Synthesis and Structure-Activity Relationship of 4-Substituted 2-(2-acetyloxyethyl)-8-(morpholine-4-sulfonyl)pyrrolo[3,4-c]quinoline-1,3-diones as Potent Caspase-3 Inhibitors. J. Med. Chem. 2005, 48, 3680–3683.

25.

Fattorusso

Jung

Crowell

K. J.

. Discovery of a Novel Class of Reversible Non-Peptide Caspase Inhibitors via a Structure-Based Approach. J. Med. Chem. 2005, 48, 1649–1656.

26.

Chen

Y. H.

Zhang

Y. H.

Zhang

H. J.

. Design, Synthesis, and Biological Evaluation of Isoquinoline-1,3,4-Trione Derivatives as Potent Caspase-3 Inhibitors. J. Med. Chem. 2006, 49, 1613–1623.

27.

J. Q.

Zhang

H. J.

. Isoquinoline-1,3,4-Trione Derivatives Inactivate Caspase-3 by Generation of Reactive Oxygen Species. J. Biol. Chem. 2008, 283, 30205–30215.

28.

Park

S. Y.

Park

S. H.

Lee

I. S.

. Establishment of a High-Throughput Screening System for Caspase-3 Inhibitors. Arch. Pharm. Res. 2000, 23, 246–251.

29.

Sakai

Yoshimori

Nose

. Structure-Based Discovery of a Novel Non-Peptidic Small Molecular Inhibitor of Caspase-3. Bioorg. Med. Chem. 2008, 16, 4854–4859.

30.

Cai

S. X.

Zhang

H. Z.

Guastella

. Design and Synthesis of Rhodamine 110 Derivative and Caspase-3 Substrate for Enzyme and Cell-Based Fluorescent Assay. Bioorg. Med. Chem. Lett. 2001, 11, 39–42.

31.

Kaeberlein

McDonagh

Heltweg

. Substrate-Specific Activation of Sirtuins by Resveratrol. J. Biol. Chem. 2005, 280, 17038–17045.

32.

Stennicke

H. R.

Renatus

Meldal

. Internally Quenched Fluorescent Peptide Substrates Disclose the Subsite Preferences of Human Caspases 1, 3, 6, 7 and 8. Biochem. J. 2000, 350, 563–568.

33.

Hermel

Gafni

Propp

S. S.

. Specific Caspase Interactions and Amplification Are Involved in Selective Neuronal Vulnerability in Huntington’s Disease. Cell Death Differ. 2004, 4, 424–438.

34.

Talanian

R. V.

Quinlan

Trautz

. Substrate Specificities of Caspase Family Proteases. J. Biol. Chem. 1997, 272, 9677–9682.

35.

Ferreira

R. S.

Bryant

Ang

K. K.

. Divergent Modes of Enzyme Inhibition in a Homologous Structure-Activity Series. J. Med. Chem. 2009, 52, 5005–5008.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.86 MB

0.00 MB