Sage Journals: Discover world-class research

Abstract

The throughput level currently reached by automatic liquid handling and assay monitoring techniques is expected to facilitate the discovery of new modulators of enzyme activity. Judicious and dependable ways to interpret vast amounts of information are, however, required to effectively answer this challenge. Here, the 3-point method of kinetic analysis is proposed as a means to significantly increase the hit success rates and decrease the number of falsely identified compounds (false positives). In this post-Michaelis–Menten approach, each screened reaction is probed in three different occasions, none of which necessarily coincide with the initial period of constant velocity. Enzymology principles rather than subjective criteria are applied to identify unwanted outliers such as assay artifacts, and then to accurately distinguish true enzyme modulation effects from false positives. The exclusion and selection criteria are defined based on the 3-point reaction coordinates, whose relative positions along the time-courses may change from well to well or from plate to plate, if necessary. The robustness and efficiency of the new method is illustrated during a small drug repurposing screening of potential modulators of the deubiquinating activity of ataxin-3, a protein implicated in Machado–Joseph disease. Apparently, intractable Z factors are drastically enhanced after (1) eliminating spurious results, (2) improving the normalization method, and (3) increasing the assay resilience to systematic and random variability. Numerical simulations further demonstrate that the 3-point analysis is highly sensitive to specific, catalytic, and slow-onset modulation effects that are particularly difficult to detect by typical endpoint assays.

Keywords

enzyme kinetics high-throughput screening drug repurposing ataxin-3 Machado–Joseph disease

Introduction

The study of enzyme kinetics commonly focuses on the initial reaction phases during which constant rate conditions are verified irrespective of the degree of substrate conversion. During the so-called steady-state period, the measured initial reaction rate (v₀) is related to the initial substrate concentration (S₀) according to the Michaelis–Menten equation.¹ The early stages of reactions are of additional importance for the screening of novel enzyme inhibitors since the percentage inhibition computed in terms of apparent reaction rates is known to decrease over time.^2,3 Moreover, endpoint assays performed within the period of constant v₀ are associated with minimal costs per screened compound when large numbers (>10,000) of chemical compounds are being tested.^2,4

With the advent of robotics and sample miniaturization systems, continuous monitoring of multiple reactions became possible through the use of highly automated workstations.⁵ While covering different phases of the reaction, full progress curve analysis has the potential to increase the screening efficiency through the detection of enzyme modulation effects that are either slow to manifest or masked at high substrate concentrations.^4,6,7 Deconvoluting this useful information from the occurrence of, for example, unaccounted enzyme inactivation, instrumental drift, and compound interference, is, however, difficult even considering separated timescales.^8–10 In a previous attempt to simplify high-throughput screenings (HTSs) of dynamic modulation effects, the theoretical relationship between the half maximal inhibitory concentration (IC₅₀) and the substrate conversion was established assuming first-order kinetic models.¹¹ In another example, endpoint enzymatic assays were optimized in order to measure the reaction conversion at the point of maximum difference in product concentration between control and competitively inhibited reactions.⁷

A minimum of 3-point readouts are required to probe the initial, intermediate, and final phases of screening reactions¹² ( Fig. 1A ). More frequent readings are technologically possible,^13,14 although their implementation in the phase of primary screening is avoided in order to keep the assay cost to a minimum.⁴ When conditions of large substrate concentration cannot be adopted, the throughput level of endpoint assays is greatly limited by the duration of the constant rate period, at the end of which different estimates of v₀ start to be produced^15,16 ( Fig. 1B ). Compound-induced changes of background signal can be automatically normalized by adopting a 2-point strategy to measure the increase of product concentration (ΔP_i = P_i – P₀). This advantage, which is shared by kinetic-mode assays, obviates the design of counterassays for compound interference identification.^17–19 Nevertheless, increasing the reaction monitoring to two or more instants does not solve the problem of time-dependent readouts unless a renewed theoretical framework is adopted.

Figure 1.

Numerical example showing possible configurations of endpoint and 3-point HTS assays. (A) The simulated enzymatic reaction (solid line) is assumed to run simultaneously in a total of n = N wells. The product concentration P in each well can be probed three times (i = 0, 1, 2) over periods of time δt (shaded areas) whose duration is determined by the technological throughput level as well as by the value of N. The slope of the dashed line indicates the true initial rate v₀. (B) Changeable estimates of apparent rates (P_i – P₀)/(t_i – t₀) will be produced unless the δt interval is significantly shorter than the full time span.

After proposing the closed-form mathematical solution of single-substrate, single-active-site enzymatic mechanisms,^20,21 our group developed the linearization method (LM) for the detection of hidden assay interferences.²² Here we show that the LM principles can be applied to the HTS detection of “interferences” caused by candidate enzyme effectors. The advantages of the new method are demonstrated by combining numerical simulations with a practical example of compound screening for modulators of the deubiquinating activity of polyglutamine-expanded ataxin-3 (Atx3 77Q), a protein associated with spinocerebellar ataxia type 3, also known as Machado–Joseph disease.²³

Materials and Methods

Protein Production and Enzymatic Assay

Atx3 77Q was expressed and purified as previously described.^24–26 A modified version of the method described by Burnett et al.²⁷ was employed to monitor the release of fluorescent 7-amino-4-methylcoumarin (AMC) from the fluorogenic substrate ubiquitin–AMC (Ub-AMC; Boston Biochem, Cambridge, MA) by Atx3 77Q. Enzymatic reactions were performed in 50 mM HEPES, pH 7.5 (20 °C), 0.5 mM EDTA, 0.1 mg/mL bovine serum albumin (BSA), 5% (v/v) glycerol, and 10 mM DTT at 37 °C using 384-well microplates (low flange, black, flat bottom, polystyrene; Corning, Kennebunk, ME) and a total reaction volume of 50 or 30 μL/well. A range of Ub-AMC concentrations lower than 1 μM was used, as per recommendation of the substrate manufacturer. Total enzyme concentration ranged from 0.25 to 0.5 μM. Reactions were started by addition of protein, and ﬂuorescence was monitored at 460 nm (390 nm excitation) using a HIDEX CHAMELEON V plate reader (Turku, Finland). To avoid evaporation, the reaction mixture in each well was overlaid with liquid paraffin (30 μL).

Screening Procedures

Potential modulators of Atx3 77Q activity were screened from 1280 Food and Drug Administration (FDA)-approved chemical compounds contained in the Prestwick Chemical Library (Illkirch, France). An automated bulk dispenser (Multidrop Combi; Thermo Fisher Scientific, Waltham, MA) was used to fill 384-well microplates with 15 μL of 0.75 μM Ub-AMC. An automated liquid handler (JANUS Automated Workstation; PerkinElmer, Waltham, MA) equipped with pin tool replicators (V&P Scientific, San Diego, CA) coupled to a Modular Dispense Technology (MDT) head was used to add 0.1 μL of test compounds (from a 1 mM stock) or DMSO (for controls), after which reactions were started by adding 15 μL of 0.70 μM Atx3 77Q to each well (reaction volume: 30 μL). The MDT head of the automatic liquid handler equipped with 96 tips was used to dispense 30 μL of liquid paraffin to overlay the final reaction mixtures. A total of four 384-well microplates were filled, each plate testing 320 chemical compounds (1 compound per well) and running 32 control reactions in the presence of DMSO (DMSO control reactions). The final reaction mixture contained 0.35 μM Atx3 77Q, 0.375 μM Ub-AMC, and 3.33 μM test compound/DMSO. Assay robustness was evaluated separately for each microplate in terms of the median Z factor:^28,29

Z = 1 - \frac{3 σ_{s} + 3 σ_{c}}{| {\tilde{x}}_{s} - {\tilde{x}}_{c} |}

(1)

where $\tilde{x}$ and σ represent median and standard deviation; index s refers to the analyzed sample, corresponding to 320 test reactions + 32 DMSO control reactions; and index c refers to 32 blank reactions corresponding to negative controls. Two different normalization methods were investigated before and after the application of three exclusion criteria described below. Specifically, values of the Z factor were computed based on (1) initial reaction rates determined over a period of ~1 h and (2) apparent reaction rates measured by the 3-point method.

Criteria for Compound Selection

According to the LM, the following linear relationship between the modified reaction coordinates ΔP_i/Δt_i and –ln(1 – ΔP_i/ΔP_∞)/Δt_i is established under typical conditions of large substrate excess over enzyme, and independently of which reaction time frames or initial substrate concentrations are considered:²²

\frac{Δ P_{i}}{Δ t_{i}} = V_{m a x}^{a p p} - K_{m}^{a p p} [- \frac{\ln (1 - \frac{Δ P_{i}}{Δ P_{\infty}})}{Δ t_{i}}]

(2)

where ΔP_∞ corresponds to the product concentration increase measured at the end of the reaction, and $K_{m}^{a p p}$ and $V_{m a x}^{a p p}$ are apparent kinetic constants. Refer to Figure 1B to recall the differences between ΔP_i/Δt_i and v₀. The values of $K_{m}^{a p p}$ and $V_{m a x}^{a p p}$ expected for interference-free control reactions correspond to the Michaelis constant (K_m) and limiting rate (V_max), respectively. Added compounds inhibiting or promoting enzyme activity will change the location of the LM coordinates to a region below (for inhibitors) or above (for activators) the straight line of slope –K_m and vertical axis intercept V_max that is defined by eq 2 ( Suppl. Fig. S1 ).

The proposed method of enzyme modifier detection relies on the measurement of product concentration P_i (or, alternatively, substrate concentration S_i) in three distinct moments (indices i = 0, 1, 2). When the effects of the screening compounds on the calibration curve are not known beforehand, indirect monitoring of product concentration is performed using signal readout values F_i instead of P_i. Possible changes in background signal are normalized by using ΔF_i differences in the estimation of midpoint-centered (i = 1) LM coordinates:

X_{1} = - \frac{\ln (1 - \frac{Δ F_{1}}{Δ F_{2}})}{Δ t_{1}}

(3a)

Y_{1} = \frac{Δ F_{1}}{Δ t_{1}}

(3b)

Preferably, the third-point location should coincide with the plateau reached at the end of the reaction, thereby admitting the equivalence between ΔF₁/ΔF₂ and ΔP₁/ΔP_∞ ratios. As the third point becomes located further behind the reaction endpoint, the resolution of the method is expected to decrease ( Suppl. Fig. S2 ). The time- and S₀-independent LM criterion for compound selection is based on the difference between obtained and predicted X₁ or Y₁ values. The stronger the effect of the modulator, the larger the differences ΔX₁ or ΔY₁ will be. During the screening of Atx3 77Q effectors, the resolution of the method is computed in terms of the Y₁ difference:

Δ Y_{1} = Y_{1} - (V_{m a x} - K_{m} X_{1}) κ

(4)

where κ is the known proportionality factor converting product concentration units into F_i units in the absence of test compounds; if all test reactions can be probed simultaneously (i.e., if δt << Δt_i in the example of Fig. 1 ), previous knowledge of the control kinetic parameters (K_m and V_max) is not required. In such cases, distribution histograms are sufficient to statistically select higher (for activators) and lower (for inhibitors) values of X₁ or Y₁.

During compound screening for modulators of Atx3 77Q activity, three exclusion criteria are adopted to eliminate test reactions affected by random experimental error and artifacts. The first criterion accounts for evident deviations from the product buildup trend caused, for example, by sudden fluorescence quenching. To pass this criterion, the following condition has to be verified:

0 < Δ F_{1} \leq Δ F_{2}

(5)

The enzymology principles used to define the other two criteria are described in detail in Supplemental Materials Section 2. The second validation criterion is derived taking as reference the limit case of $K_{m}^{a p p}$ ≈ 0 for which product concentration increases linearly with time:

\frac{Δ t_{1}}{Δ t_{2}} \leq \frac{Δ F_{1}}{Δ F_{2}}

(6)

In the other limit case of $K_{m}^{a p p} ≫ S_{0},$ product concentration is expected to follow an asymptotic exponential growth.²¹ With the values of ΔF₁/ΔF₂ and Δt₁, a time constant for exponential P_i increase is defined setting the physical limits for the maximum Δt₁/Δt₂ ratio. While more stringent than eq 6 for high ΔF₁/ΔF₂ ratios, these limits are used to establish the third validation criterion:

\frac{Δ t_{1}}{Δ t_{2}} \leq - \ln (1 - \frac{Δ F_{1}}{Δ F_{2}}) (\frac{1}{Δ F_{1} / Δ F_{2}} - 1)

(7)

Results

The improvements in robustness brought forward by the 3-point LM assay are illustrated with the screening of 1280 FDA-approved chemical compounds for possible modulators of the deubiquinating activity of Atx3 77Q. This is a purposely small chemical library that allows for kinetic-mode monitoring of all tested reactions running in four 384-well microplates ( Fig. 2 ). The occurrence of spurious phenomena eventually suggested by the 3-point positions can thus be evaluated against what is reported by the full progress curves. In all plates, control-reaction conversions of ~0%, ~60%, and 99.9% were chosen for the 3-point positions along the time-courses. The number of excluded compounds changes greatly from plate to plate, possibly as a consequence of the different chemical natures of the tested molecules, but also owing to inherent random and systematic errors arising, for example, from differences in the effective concentration of active sites of Atx3 77Q, which is an aggregation-prone protein.²⁵

Figure 2.

Validation criteria applied to the screening of drug repurposing compounds for potential modulators of the enzymatic activity of Atx3 77Q. (A–D) Zoomed progress curves showing mean and standard deviation values of control-reaction readouts (solid and dashed blue lines, respectively); excluded (red lines) and validated (gray lines) screening reactions; and the 3-point positions (symbols) along the excluded progress curves. Full-scale graphs are presented in Supplemental Figure S5 . Filters to eliminate spurious results are successively applied based on eqs 5 (solid red lines), 6 (no results excluded), and 7 (dashed red lines). Different panels present the results obtained in microplates (A) 1, (B) 2, (C) 3, and (D) 4. Areas shaded in light gray represent intervals where validated test reactions can be found.

Most excluded curves are identified upon application of the first filter accounting for evident drops in AMC fluorescence caused by quenching interference, inner filter effects, light scattering, and so forth (solid red lines in Fig. 2 ). Examples of supralinear trends failing to comply with eq 6 are not observed, whereas the third exclusion criterion (eq 7) does eliminate further unreliable readouts (dashed red lines in Fig. 2 ). Unsurprisingly, the number of excluded results increases as the second point is positioned nearer to the third point ( Table 1 ). Higher ΔF₁/ΔF₂ ratios increase the risk of false negatives as the result of possible deviations from physically acceptable trends originated by, for example, signal noise (Supplemental Materials Section 2); by adopting ΔF₁/ΔF₂ ratios below 0.5, the safety margin for exclusion increases, yet the false-positive rates in subsequent phases of hit detection are also expected to increase. All the suspicious curves that were eliminated using a midpoint threshold of ~60% control-reaction conversion ( Fig. 2 ) are indeed affected by systematic drops in fluorescence signal during at least 20% of the total time over which the reactions are monitored ( Table 1 , last column).

Table 1.

Impact of the Midpoint Position on the Total Number of Excluded Results.^a

	Number of Excluded Results (out of 1404)				Percentage of Success^c
Control-Reaction Conversion (%)	Criterion 1 (eq 5)	Criterion 2 (eq 6)	Criterion 3 (eq 7)	Total Excluded	Percentage of Success^c
60^b	69	0	29	98	100
70	123	0	175	298	92.6
50	36	0	9	45	100
40	18	0	2	20	100

The reference case of ~60% control-reaction conversion corresponds to the sum of excluded results represented in Figure 2A–D . In all considered cases, the first and third points are located at ~0% and ~99.9% control-reaction conversions.

Chosen midpoint threshold.

Results are confirmed as bad outliers when the fluorescence signal persistently drops over periods longer than 20% of the full time-course.

The AMC fluorophore used in the Atx3 77Q assay has an excitation wavelength in the UV range, which also excites a large number of screening compounds.³⁰ Values of the median Z factor well below the practical limit of 0.5 confirm that the accuracy of this drug repurposing screening would not be acceptable unless robust exclusion criteria were used to identify assay artifacts ( Table 2 ). Major improvements in robustness are achieved through the elimination of bad outliers (compare the last two columns of Table 2 ) but also by adopting better normalization procedures than those conventionally implemented for endpoint assays (compare the v₀- and Y₁/ΔF₂-based Z factors).

Table 2.

Median Z Factors (eq 1) Calculated for Each Microplate before and after the LM-Based Exclusion Criteria Are Applied.^a

	v₀ Readouts		Y₁/ΔF₂ Readouts^b
Plate No.	Before	After	Before	After
1	–4.52	–4.08	–13.23	0.41
2	–0.49	–0.45	0.28	0.45
3	–1.59	–0.25	–0.18	0.50
4	–0.01	0.17	0.33	0.49

The initial rate and ΔF₂-normalized Y₁ values are used as assay readouts.

The ΔF₂ values used for negative controls correspond to the mean ΔF₂ values obtained during the DMSO control reactions.

The initial rate analysis of control reactions revealed interplate differences in enzymatic activity despite using the same molar concentration of Atx3 77Q across microplates. Analysis of single-point F₁ data by the software HTS-Corrector confirmed the occurrence of plate effects affecting the quality of the χ² contingency test of hit distribution,^31,32 which is passed after the background signal is corrected (final χ² value of 318.0 for a critical χ² value of 380.7, with α = 0.01). This plate variability is likely due to variations in the oligomerization state of this polyglutamine-expanded protein, whose characterization in terms of the kinetic parameters K_m and V_max was performed using a lower enzyme concentration than during compound screening ( Suppl. Fig. S3 ). The resulting LM straight lines (lines in Fig. 3A–D ) underestimate most of the X₁ and Y₁ values obtained for each microplate (symbols in Fig. 3A–D ), even after correcting parameter V_max by the effective enzyme concentration. While systematically affecting the differences between obtained and predicted Y₁ values (eq 4), this variability has no major influence on the 3-point method of compound selection. Illustrating this, interplate variations in Y₁ estimates ( Fig. 4A ) tend to vanish when the results are represented as ΔY₁ distributions ( Fig. 4B ). Since the 3 points are probed at practically simultaneous time instants, within and across microplates, the dispersion of Y₁ (or X₁) values is not justified by significant variations in the reaction periods considered in each well. Should these variations have occurred, their normalization across microplates would also be possible by representing frequency distributions of ΔY₁ values (see numerical examples in Suppl. Figs. S1 and S4 for details).

Figure 3.

Zoomed LM representation of screening reactions. Full-scale graphs are presented in Supplemental Figure S6. (A–D) Symbols: Intermediate LM coordinates computed for validated (open symbols) and excluded (closed symbols in gray) results using eq 3; for graphical representation purposes, an argument of the logarithm in eq 3a is determined as its module. Dashed lines: Theoretical LM straight lines computed from eq 2 using the values of K_m and V_max obtained during Atx3 77Q assay characterization ( Suppl. Fig. S3 ); parameter V_max is corrected by a factor of (A) 0.94, (B) 0.59, (C) 1.02, and (D) 0.86, accounting for the different Atx3 77Q activities measured at 0.375 μM Ub-AMC during assay characterization and during the control reactions in each microplate (see Supplemental Materials Section 3).

Figure 4.

Histogram representation of the frequency distribution of midpoint LM coordinates for each microplate. (A,B) Only validated data are used to compute (A) Y₁ values from eq 3b and (B) ΔY₁ differences from eq 4. The usage of ΔY₁ as compound selection criterion normalizes the variability associated with the effective concentration of active enzyme.

Discussion

The 3-point method of kinetic analysis answers the current demand for new breakthroughs in the discovery of inhibitors and activators for targets of interest in pharmaceutical research.^33–35 Its first principles are those of the LM recently proposed by us as a new tool to detect enzymatic assay interferences,²² with the main difference that LM curves are now used to detect “interferences” caused by candidate enzyme effectors. Using the screening for Atx3 77Q modulators as a practical example, major improvements in robustness are achieved through judicious elimination of outliers, by adopting advanced normalization methods, and by increasing the assay resilience to systematic and random variability. Simple numerical examples illustrate another way by which the 3-point LM assay can improve HTS efficiency, namely, by increasing the detection of true-positive hits ( Fig. 5 ). Mechanisms of specific (or competitive), catalytic (or uncompetitive), and slow-binding inhibition are simulated assuming S₀ = 10K_m. Substrate concentrations this high are beneficial to increase the signal-to-background ratio and to detect uncompetitive inhibitors, yet values of S₀ ≈ K_m are generally preferred in HTS practice as a compromise to warrant good responsiveness to competitive inhibitors as well.^3,34,36 In addition to a marked increase in sensitivity to competitive inhibitors at high S₀ values, the numerical simulations in Figure 5 demonstrate that the 3-point screening method still marginally increases the detection limits for uncompetitive inhibitors. Therefore, in the light of the new methodology, high substrate concentrations can henceforth be adopted without further constraints than the practical limits imposed by substrate solubility and the desired overall duration of the assay. The LM straight lines describing the catalytic and specific mechanisms in Figure 5B are simulated assuming quasi-equilibrium conditions that do not apply to the case of slow-binding inhibition.³⁷ Transient enzyme modifier mechanisms are particularly difficult to detect solely based on initial rate measurements. In the case of slow-binding inhibitors, single-time-point analysis gives rise to apparent potencies much lower than the reality and, therefore, leads to failures in the identification of true hits.³⁸ Conversely, the 3-point method is capable of greatly improving the assay sensitivity to slow-binding inhibitors if sufficiently long Δt₁ periods are adopted ( Fig. 5C ).

Figure 5.

Numerical simulations of different enzyme modifier mechanisms assuming S₀ = 10K_m. Further details on the numerical methods are given in Supplemental Materials Section 3 and Supplemental Table S1 . (A) Lines: S₀-normalized time-courses for control (solid line) and test reactions of catalytic (dashed green line), specific (dashed red line), and slow-binding (dashed blue line) inhibition. Symbols: Possible selection of reaction coordinates by the 3-point method. (B) Lines: LM representations of the time-courses in A. Quasi-equilibrium conditions are assumed for specific and catalytic inhibition. Equation 3 extended to the full time-course is used to compute the LM curve in the case of slow-binding inhibition. Symbols: Linearized midpoint coordinates corresponding to the 3-point selection. Arrows indicate progress in time. (C) Full bars: Resolution of the 3-point method given as ΔY₁ differences (eq 4) normalized by the expected value of Y₁ for control reactions. Open bars: Relative differences between values v₀ determined from the onset of test and control progress curves.

The cases depicted in Figure 5 are necessarily limited in number, as many other mechanisms of inhibition, inactivation, activation, and so forth, could be studied for various combinations of substrate, enzyme, and enzyme modifier concentrations. Independently of which scenario is considered, the LM remains highly sensitive to minor kinetic variations,²² whereas the precursory application of robust validation criteria prevents false hit proliferation. The highly responsive ΔY₁ measurable not only normalizes plate-to-plate differences ( Fig. 4B ) but also helps in detecting row- and column-specific variability when fed as an input variable to existing methods of bias detection.^31,32 Whenever possible, initial substrate concentrations in the order of magnitude of 10K_m should be adopted so as to warrant high efficiency of the 3-point method during the exclusion and selection of HTS results. Although no fixed procedure is imposed for the 3-point selection, narrow ranges of signal readout variation are not recommended due to noise amplification by the logarithm term in the definition of X₁ (eq 3a). Consequently, as a rule of thumb, the first and third points may coincide with the beginning and the conclusion of the assay, while the second point may correspond to instantaneous substrate concentrations of ~K_m (taking the control-reaction progress curves as reference). The suggested midpoint location may change depending on whether enzyme activators or inhibitors are pursued, since test reactions will finish either much sooner or much later when in the presence of potent activators or inhibitors, respectively.

In summary, harnessing the full power of high-throughput resources is now possible by adopting the 3-point kinetics assay as an alternative to traditional endpoint and kinetic-mode assays. A substantial decrease of false-positive and false-negative rates should be attained by improving the robustness and efficiency of the hit selection methodology. The following advantages of the 3-point method are emphasized:

- Simple implementation: Three-point monitoring can be implemented for a large number of screened reactions running simultaneously—none of the 3 points has to be located within the initial period of constant velocity. Moreover, the relative positions of the 3 points on the time-course may change from well to well or from plate to plate.

- Resilience to experimental variability: Not only time-related variability but also experimental variability can be dealt with by the 3-point methodology. Random changes in, for example, substrate concentration, do not considerably affect the method resolution. Variations in concentration and/or activity of the enzyme can also be accounted for during hit selection.

- Interference-proof: The validation criteria provided by eqs 5–7 are confirmed as powerful tools to eliminate artifacts. On the other hand, the proposed normalization method automatically corrects changes in background signal, which obviates the need for counterassays to identify compound interferences.

- Admissible (and recommended) high substrate concentrations: Although desirable for improving the signal-to-background ratio and detecting uncompetitive inhibitors, high values of S₀ are traditionally associated with low sensitivity to competitive inhibitors. With the 3-point method, high substrate concentrations can be adopted without further constraints than the practical limits imposed by substrate solubility and the desired overall duration of the assay.

- Fundamentally based: Both the validation and selection criteria are based on solid enzymology principles and not on subjective judgment.

- High sensitivity: The usage of LM coordinates warrants high sensitivity to minor kinetic variations, while the precursory application of robust validation criteria prevents false hit proliferation. Simple numerical examples show how the 3-point method has evident advantages in the detection of specific-type enzyme modifiers and of time-dependent modulation effects such as slow-binding inhibition.

As a general rule to be followed by HTS assay designers, we recommend that the first and third points are located at the beginning and the conclusion of the assay, while the midpoint location should coincide with instantaneous substrate concentrations slightly above or below ~K_m, depending on whether enzyme activators or inhibitors are looked for.

Supplemental Material

20200911_SupplementalMaterial_SLAS – Supplemental material for Major Improvements in Robustness and Efficiency during the Screening of Novel Enzyme Effectors by the 3-Point Kinetics Assay

Supplemental material, 20200911_SupplementalMaterial_SLAS for Major Improvements in Robustness and Efficiency during the Screening of Novel Enzyme Effectors by the 3-Point Kinetics Assay by Maria Filipa Pinto, Francisco Figueiredo, Alexandra Silva, António R. Pombinho, Pedro José Barbosa Pereira, Sandra Macedo-Ribeiro, Fernando Rocha and Pedro M. Martins in SLAS Discovery

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was financed by (1) Fundo Europeu de Desenvolvimento Regional (FEDER) funds through the COMPETE 2020—Operacional Programme for Competitiveness and Internationalisation (POCI), Portugal 2020, and by Portuguese funds through Fundação para a Ciência e a Tecnologia/Ministério da Ciência, Tecnologia e Ensino Superior (FCT/MCTES) in the framework of projects POCI-01-0145-FEDER-031323 (PTDC/MED-FAR/31323/2017), POCI-01-0145-FEDER-031173 (PTDC/BIA-BFS/31173/2017), and POCI-01-0145-FEDER-007274 (“Institute for Research and Innovation in Health Sciences”); (2) Base Funding—UIDB/00511/2020 of the Laboratory for Process Engineering, Environment, Biotechnology and Energy (LEPABE), funded by national funds through the FCT/MCTES (PIDDAC); and (3) FEDER through Norte Portugal Regional Operational Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement in the framework of Project Norte-01-0145-FEDER-000008.

PhD fellowships SFRH/BD/109324/2015 (M.F.P.) and SFRH/BD/133009/2017 (F.F.) from FCT are acknowledged. Financial support was provided through the Doctoral Program in Biomedical Sciences (ICBAS-UP).

ORCID iD

Maria Filipa Pinto

Supplemental material is available online with this article.

References

Michaelis

Menten

Die Kinetik Der Invertinwirkung. Biochem. Z. 1913, 49, 333–369.

Copeland

R. A.

Evaluation of Enzyme Inhibitors in Drug Discovery: A Guide for Medicinal Chemists and Pharmacologists. John Wiley & Sons: Hoboken, NJ, 2013.

Copeland

R. A.

Mechanistic Considerations in High-Throughput Screening. Anal. Biochem. 2003, 320, 1–12.

Gutiérrez

O. A.

Danielson

U. H.

Detection of Competitive Enzyme Inhibition with End Point Progress Curve Data. Anal. Biochem. 2006, 358, 11–19.

Inglese

Johnson

R. L.

Simeonov

, et al. High-Throughput Screening Assays for the Identification of Chemical Probes. Nat. Chem. Biol. 2007, 3, 466.

Orsi

B. A.

Tipton

K. F.

Kinetic Analysis of Progress Curves. Methods Enzymol. 1979, 63, 159–183.

Tholander

Improved Inhibitor Screening Experiments by Comparative Analysis of Simulated Enzyme Progress Curves. PLoS One 2012, 7, e46764.

Gunawardena

Some Lessons about Models from Michaelis and Menten. Mol. Biol. Cell 2012, 23, 517–519.

Cornish-Bowden

A. J.

Analysis of Progress Curves in Enzyme Kinetics. Biochem. J. 1972, 130, 637–639.

10.

Waley

S. G.

A Quick Method for the Determination of Inhibition Constants. Biochem. J. 1982, 205, 631–633.

11.

Yuan

Hodge

C. N.

Determining Appropriate Substrate Conversion for Enzymatic Assays in High-Throughput Screening. J. Biomol. Screen. 2003, 8, 694–700.

12.

Sárkány

Rocha

Damas

A. M.

, et al. Chemical Kinetic Strategies for High-Throughput Screening of Protein Aggregation Modulators. Chem. Asian J. 2019, 14, 500–508.

13.

Horiuchi

K. Y.

Wang

, et al. Nanoliter Homogenous Ultra-High Throughput Screening Microarray for Lead Discoveries and Ic50 Profiling. Assay Drug Dev. Technol. 2005, 3, 177–187.

14.

Petersen

K. J.

Peterson

K. C.

Muretta

J. M.

, et al. Fluorescence Lifetime Plate Reader: Resolution and Precision Meet High-Throughput. Rev. Sci. Instrum. 2014, 85, 113101.

15.

Bisswanger

Enzyme Assays. Perspect. Sci. 2014, 1, 41–55.

16.

Cornish-Bowden

The Use of the Direct Linear Plot for Determining Initial Velocities. Biochem. J. 1975, 149, 305–312.

17.

Gribbon

Sewing

Fluorescence Readouts in HTS: No Gain without Pain?

Drug Discov. Today 2003, 8, 1035–1043.

18.

Shoichet

B. K.

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

19.

Simeonov

Davis

M. I.

Interference with Fluorescence and Absorbance. Eli Lilly & Company and the National Center for Advancing Translational Sciences: Bethesda, MD, 2018.

20.

Pinto

M. F.

Estevinho

B. N.

Crespo

, et al. Enzyme Kinetics: The Whole Picture Reveals Hidden Meanings. FEBS J. 2015, 282, 2309–2316.

21.

Pinto

M. F.

Martins

P. M.

In Search of Lost Time Constants and of Non-Michaelis-Menten Parameters. Perspect. Sci. 2016, 9, 8–16.

22.

Pinto

M. F.

Ripoll-Rozada

Ramos

, et al. A Simple Linearization Method Unveils Hidden Enzymatic Assay Interferences. Biophys. Chem. 2019, 252, 106193.

23.

Silva

de Almeida

A. V.

Macedo-Ribeiro

Polyglutamine Expansion Diseases: More Than Simple Repeats. J. Struct. Biol. 2018, 201, 139–154.

24.

Almeida

Abreu

I. A.

Matos

C. A.

, et al. Sumoylation of the Brain-Predominant Ataxin-3 Isoform Modulates Its Interaction with P97. Biochim. Biophys. Acta Mol. Basis Dis. 2015, 1852, 1950–1959.

25.

Gales

Cortes

Almeida

, et al. Towards a Structural Understanding of the Fibrillization Pathway in Machado-Joseph’s Disease: Trapping Early Oligomers of Non-Expanded Ataxin-3. J. Mol. Biol. 2005, 353, 642–654.

26.

Silva

Almeida

Fraga

J. S.

, et al. Distribution of Amyloid-Like and Oligomeric Species from Protein Aggregation Kinetics. Angew. Chem. Int. Ed. 2017, 56, 14042–14045.

27.

Burnett

Pittman

R. N.

The Polyglutamine Neurodegenerative Protein Ataxin-3 Binds Polyubiquitylated Proteins and Has Ubiquitin Protease Activity. Hum. Mol. Genet. 2003, 12, 3195–3205.

28.

Gribbon

Lyons

Laflin

, et al. Evaluating Real-Life High-Throughput Screening Data. J. Biomol. Screen. 2005, 10, 99–107.

29.

Zhang

J.-H.

Chung

T. D.

Oldenburg

K. R.

A Simple Statistical Parameter for Use in Evaluation and Validation of High Throughput Screening Assays. J. Biomol. Screen. 1999, 4, 67–73.

30.

Tirat

Schilb

Riou

, et al. Synthesis and Characterization of Fluorescent Ubiquitin Derivatives as Highly Sensitive Substrates for the Deubiquitinating Enzymes Uch-L3 and Usp-2. Anal. Biochem. 2005, 343, 244–255.

31.

Makarenkov

Kevorkov

Zentilli

, et al. HTS-Corrector: Software for the Statistical Analysis and Correction of Experimental High-Throughput Screening Data. Bioinformatics 2006, 22, 1408–1409.

32.

Makarenkov

Zentilli

Kevorkov

, et al. An Efficient Method for the Detection and Elimination of Systematic Error in High-Throughput Screening. Bioinformatics 2007, 23, 1648–1657.

33.

Hanley

Q. S.

The Distribution of Standard Deviations Applied to High Throughput Screening. Sci. Rep. 2019, 9, 1268.

34.

Yang

Copeland

R. A.

Lai

Defining Balanced Conditions for Inhibitor Screening Assays That Target Bisubstrate Enzymes. J. Biomol. Screen. 2009, 14, 111–120.

35.

Murie

Barette

Button

, et al. Improving Detection of Rare Biological Events in High-Throughput Screens. J. Biomol. Screen. 2015, 20, 230–241.

36.

Acker

M. G.

Auld

D. S.

Considerations for the Design and Reporting of Enzyme Assays in High-Throughput Screening Applications. Perspect. Sci. 2014, 1, 56–73.

37.

Baici

Kinetics of Enzyme-Modifier Interactions, 1st ed.; Springer: Berlin, 2015; pp 71–73, 388–391.

38.

Holdgate

G. A.

Meek

T. D.

Grimley

R. L.

Mechanistic Enzymology in Drug Discovery: A Fresh Perspective. Nat. Rev. Drug Discov. 2018, 17, 115–132.

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.

1.00 MB

0.00 MB