A High-Throughput Assay to Identify Inhibitors of the Apicoplast DNA Polymerase from Plasmodium falciparum

Materials

The 33-mer DNA hairpin substrate (5′-CCTCTCCGTGTC-TTGTACTTCCCGTCAGAGAGG) containing either Cy3 or hexachlorofluorescein (HEX) dyes at the 5′ end, and either Iowa Black® FQ (IBQ) or Black Hole Quencher®-1 (BQ1) quenchers at the 3′ end were purchased from Integrated DNA Technologies. The hairpin contains a 6-nucleotide-long double-stranded stem and 21-nucleotide-long single-stranded loop. A 9-nucleotide primer (5′-GACGGGAAG) that is complementary to a portion of the loop region was obtained from the Iowa State University DNA Facility. Nickel–agarose was bought from the Sigma-Aldrich Chemical Company, deoxyribonucleotides were acquired from Invitrogen or Sigma-Aldrich, chloroquine from Santa Cruz Biotechnology, and aurintricarboxylic acid (ATA) from Acros Organics.

Purification of apPOL^exo−

The exonuclease-deficient apPOL (D1470N and E1472Q ) was expressed and purified as previously described with a few exceptions. ¹¹ Cells were lysed in 20 mM Tris–HCl pH 8.0, 500 mM NaCl, and 5 mM imidazole. Clarified lysate was loaded on ~5 mL of Ni–agarose resin and washed with 100 mL of 20 mM Tris–HCl pH 8.0, 500 mM NaCl, and 5 mM imidazole followed by a high-salt wash of 100 mL of 20 mM Tris–HCl pH 8.0, 1 M NaCl, and 25 mM imidazole. Protein was then eluted in a minimal volume of elution buffer containing 20 mM Tris–HCl pH 8.0, 500 mM NaCl, and 150 mM imidazole. The resulting protein was then further purified through size exclusion chromatography using a 320 mL HiLoad 26/60 Superdex 200 equilibrated in 20 mM Tris–HCl pH 8.0 and 400 mM NaCl.

Kinetic Assays

The enzymatic activity of apPOL^exo− was determined using the DNA hairpin substrate containing the Cy3 and BQ1 dye–quencher pair ( Fig. 1A ). The reactions were carried out at 25 °C in 20 mM Tris–HCl pH 8.0, 10 mM MgAc, 50 mM KAc, 1 mM DTT, 0.1 mg/ml BSA, and 2% DMSO, and all data were collected on a Cary Eclipse Fluorescence Spectrophotometer using the Cary Kinetics software at an excitation wavelength of 545 nm and an emission wavelength of 570 nm.

OPEN IN VIEWER

Figure 1.

Polymerase activity with the high-throughput DNA substrate. (A) A schematic of the DNA substrate and the steps leading to an increase in fluorescence. Additional details can be found in the text. (B) The increase in fluorescence signal is dependent on protein concentration. A polymerase concentration of 3 nM was determined to be optimal for high-throughput screening because approximately half of the DNA substrate would be consumed after 7.5 minutes. The dotted line represents 1 nM apPOL^exo−, the short dashed line 2 nM apPOL^exo−, the large dashed line 4 nM apPOL^exo−, and the solid line 8 nM apPOL^exo−. The inset is a plot of the initial velocities versus enzyme concentration. The slope of the line represents the apparent-k_cat, which is 4.95 min⁻¹. (C) Denaturing urea–polyacrylamide gel analysis of product formation using ³²P-labeled primer. The unextended primer and fully extended product are marked by arrows at the bottom and top of the gel, respectively. (D) Overlay of polymerase activity based on polyacrylamide gel analysis (♦) and fluorescence assay using the hairpin substrate (gray line).

To validate that the Cy3 dye was not affecting the rate of apPOL^exo− polymerization, two separate reactions were carried out with 3 nM apPOL^exo−, 16 µM deoxynucleoside triphosphates (dNTPs) (~5× the K_M), and 50 nM DNA substrate (which provided a strong fluorescent signal). In the first reaction, polymerase activity was determined on a Cary Eclipse, as stated above. For the second reaction, the primer that is annealed to the hairpin DNA was radioactively labeled with ³²P using T4-polynucleotide kinase (New England Biolabs). The reaction was initiated by mixing equal volumes of apPOL^exo− and a mixture containing dNTPs and DNA. Time points were then removed and quenched with 0.1 M ethylenediaminetetraacetic acid (EDTA) and 80% (v/v) formamide. Each time point was analyzed using a 16% denaturing polyacrylamide gel containing 7.5 M urea in a Tris–borate–EDTA buffer. The gel was visualized using a FujiFilm FLA-5100 Fluorescent Image Analyzer and analyzed using ImageJ (National Institutes of Health, Bethesda, MD).

To determine the optimum enzyme concentration and time point for the stopped-time high-throughput assay, the concentration of apPOL^exo− was varied from 0 to 8 nM. It was found that for a reaction containing a final concentration of 16 µM dNTPs and 50 nM DNA substrate, 3 nM apPOL^exo− led to 50–90% of substrate use over a time course of 7.5 minutes (a convenient time for the HT robotic platform). The steady-state kinetics values (K_M for nucleotide and the DNA substrate) were established using 20 nM apPOL^exo−. To determine the K_M–DNA, the nucleotide concentration was held at 16 µM while the DNA concentration was varied from 0 to 800 nM. To determine the K_M–dNTPs, the DNA concentration was held at 50 nM and the nucleotide concentration was varied from 0 to 16 µM. Initial rates were analyzed using the Matlab R2011b software (Mathworks) using the standard Michaelis–Menten equation.

ATA was identified through high-throughput screening as an inhibitor of apPOL^exo−. Although ATA has previously been classified as an inhibitor by mimicking DNA, recent studies have shown that this may not always be the case. ¹⁶ For this reason, ATA was selected for further kinetic studies. The IC₅₀ of ATA was determined using 16 µM dNTPs, 50 nM DNA substrate, 3 nM apPOL^exo−, and varying concentrations of ATA from 0 to 12 µM. ATA was incubated with apPOL^exo− in the absence of DNA and nucleotide for several minutes prior to the start of the reaction. The reaction was initiated by simultaneous addition of DNA and dNTPs, and time courses lasted approximately 15 minutes. The slope of the linear portion of the time course was used to determine the initial velocities. The resulting initial velocities were fit with the Matlab R2011b software (Mathworks) using the following equation:

V o = V max ( 1 + I I C 50 )

where V_o represents initial velocity, V_max represents maximum velocity, I is the ATA concentration, and IC₅₀ is the concentration of ATA that produces 50% inhibition.

Determination of the ATA inhibition mechanism was carried out by incubating 20 nM apPOL^exo− with varying concentrations of ATA for several minutes. To establish the inhibition mechanism versus nucleotides, the DNA substrate was held constant at 50 nM, and the dNTPs were varied from 0 to 16 µM. To establish the inhibition mechanism versus DNA, the nucleotide was held constant at 16 µM nucleotides, and the DNA substrate was varied from 0 to 800 nM. The resulting data were fit against inhibition models describing competitive, noncompetitive, mixed, and uncompetitive inhibition using the Dynafit ¹² software. The best-fitting model was determined using the model discrimination analysis function in the Dynafit software.

Validation Assay and Pilot Screen

The validation assay was performed at the University of Iowa High Throughput Screening Facility. The reaction was carried out with a final concentration of 16 µM dNTPs, 50 nM DNA substrate, and 3 nM of apPOL^exo− in 20 mM Tris–HCl pH 8.0, 10 mM MgAc, 50 mM KAc, 1 mM DTT, 0.1 mg/mL BSA, and 2% DMSO in the presence and absence of 8 mM chloroquine. Solutions were loaded into a Nunc 384-well plate (Thermo Scientific) using a MicroLab Pipettor (Hamilton). A 2× protein solution was incubated with 16 mM chloroquine (8 mM final) when applicable, and then mixed with an equal volume of 2× DNA solution to initiate the reaction. After 7.5 minutes at room temperature, the reaction was quenched with 100 mM EDTA. Fluorescence was detected at 545 nm using an Envision High Throughput Plate Reader (Perkin Elmer). Resulting data were analyzed and evaluated for robustness through determination of the signal window and Z’.^13,14 Maximum signal values are those resulting from reactions containing no inhibitors, and minimum signal values are background readings with no fluorescent substrate controls.

A 2880-small-molecule pilot screen composed of the Spectrum Collection (composed of 2320 compounds) and University of Iowa Legacy Collection (composed of 560 compounds) was carried out at the University of Iowa High Throughput Screening Facility. The screen took place under the same conditions as the validation assay with a 2× protein solution being incubated with 20 µM (10 µM final) small molecules. Small molecules were added using a Star Robotic Liquid Handler System (Hamilton). Controls containing no protein, 8 mM chloroquine, and no small molecules were performed on each 384-well plate. Fluorescence was detected at 545 nm using an Envision High Throughput Plate Reader. Resulting data were analyzed using Microsoft Excel.

Purification of apPOL^exo−

The expression of apPOL^exo− is extremely robust with yields approaching 50 mg of pure protein per liter of Luria broth. The purified protein is stable at room temperature for up to 8 hours and for at least 1 year frozen at −80 ^oC. The enzyme is also highly active, ¹¹ enabling a very small amount of protein to be used in each assay (3 nM). In the 384-well format, a single preparation of protein from 1 L of media is adequate to screen more than 13,000 plates.

Fluorescence Assays

The hairpin DNA substrate used here was previously developed by others but contained a carboxytetramethylrhodamine (TAMRA) fluorophore and a Dabcyl quencher. ¹⁰ The substrate was used in a continuous DNA polymerase assay in a low-throughput fashion ( Fig. 1A ). ¹⁰ Although the assay, as previously reported, is suitable for continuous assays, because of incomplete quenching of the TAMRA dye fluorescence by the Dabcyl quencher, the signal-to-background ratio is relatively low (~fivefold) and is borderline for high-throughput screening. For this reason, we attempted to find a better fluorophore–quencher pair. We initially replaced the pair with hexachlorofluorescein and the Iowa Black® RQ quencher, but saw only modest improvements in the statistical parameters. Using 96-well plates, this dye–quencher pair resulted in a signal window of ~6.6, a Z’ factor of 0.60, and a signal-to-background ratio of 7.1. We then tried the Cy3 fluorophore and BQ1 quencher pair, which resulted in better statistical parameters. Using 96-well plates, the Cy3–BQ1 dye–quencher pair resulted in a signal window of ~8.9, a Z’ factor of 0.73, and a signal-to-background ratio of 9.7. Because some proteins have a strong affinity for dye molecules, we confirmed that the observed rate was linearly dependent on the concentration of apPOL^exo− concentration. As seen in Figure 1B , the increase in fluorescent signal is proportional to the amount of apPOL^exo− in the reaction, with an apparent-k_cat of 4.95 min⁻¹. To directly compare the increase in fluorescence signal to extension of the primer, we ³²P labeled the primer strand of the substrate and analyzed the polymerase time course using denaturing urea–polyacrylamide gel electrophoresis ( Fig. 1C ). Comparison of the fluorescence signal with the quantification of the gel indicates that the increase in fluorescence is directly proportional to the amount of primer extended ( Fig. 1D ). The gel analysis also indicates that there is some degree of pausing when apPOL^exo− reaches the stem portion of the substrate, where it must perform strand-displacement synthesis to separate the fluorophore from the quencher. We then used the assay in continuous mode to determine the steady-state kinetic parameters of apPOL^exo−, which resulted in apparent-K_M values of 3.4 ± 1.0 µM and 305 ± 99 nM for dNTPs and DNA, respectively ( Fig. 2 ). These values are in close agreement with those determined using the standard ³²P-labeled primer-extension assay. ¹¹

OPEN IN VIEWER

Figure 2.

Steady-state kinetics of apPOL^exo−. The fluorescent hairpin substrate was used, and initial rates were determined from continuous time courses. The initial rates (points) were fit to a steady-state mechanism, and the best fit is represented by the line. (A) A K_M value of 3.2 µM was determined for nucleotide. (B) The K_M value of 305 nM was determined for the hairpin substrate.

High-Throughput Assay and Pilot Screen

Following the tests using 96-well plates, the assay was validated for high-throughput screening using 384-well plates. Half of the plate was used as a minimum signal reference (8 mM chloroquine and 3 nM polymerase), and the other half was used as the maximum signal control (3 nM polymerase added). The reactions were initiated by the addition of DNA and nucleotide, and all wells were quenched with EDTA after 7.5 minutes. The plates were analyzed, and resulting values for the signal window, Z’ factor, and signal-to-background ratio were determined to be 9.27, 0.71, and 17.96, respectively.

We next performed a pilot screen using the Spectrum Collection and the University of Iowa Legacy Collection ( Fig. 3 ). The screen consisted of nine 384-well plates with the first two columns of each plate reserved for reference wells (no polymerase addition), the upper half of the last two columns for the fully inhibited reaction (8 mM chloroquine), and the lower half of the last two columns for the uninhibited reaction (polymerase addition only). The average fluorescent signals from the fully active and inhibited reactions were 1638 ± 143 (signal-to-noise of 11) and 39 ± 14 (signal-to-noise of 2.8), respectively. These data were used to calculate values for the signal window, Z’ factor, and signal-to-background ratio of 7.9, 0.71, and 42, respectively. The increase in the signal-to-background ratio compared to the initial characterization is due to an optimization of the plate-reader photomultiplier voltage prior to fluorescence acquisition.

OPEN IN VIEWER

Figure 3.

High-throughput pilot screen. The small-molecule pilot screen of the Spectrum Collection (2320 compounds) and the University of Iowa Legacy Collection (560 compounds) resulted in 62 hits that inhibit polymerase activity by at least 50%. No polymerase controls (gray X’s that occur periodically near the bottom of the graph) were performed on each of the nine 384-well plates. Each symbol type represents the results from an individual 384-well plate.

The pilot screen identified 62 compounds that resulted in a fluorescence signal of less than 700 (i.e., causing more than 50% inhibition), giving a hit rate of approximately 2.1%. Among these 62 compounds were several nonspecific inhibitors of nucleic acid enzymes, such as DNA intercalators (e.g., ethidium bromide, epirubicin, and acriflavine) and known or suspected DNA mimics (e.g., suramin and ATA). It is expected that nonspecific inhibitors can be removed by using this assay to perform counterscreens against a panel of selected DNA polymerases (e.g., Escherichia coli Pol I and human pol γ). In addition to removing compounds that act as DNA intercalators and mimics, the counterscreens would also identify compounds that may be interfering with the fluorescent signal of the DNA substrate and/or product. Alternatively, an orthogonal assay similar to the one shown in Figure 1C could be used to rule out dye interference.

ATA was selected for determination of its inhibition mechanism because it appears to be capable of multiple inhibition mechanisms depending on the enzyme it is acting on. Although it is generally assumed that the ATA polymerizes and acts as a nonspecific DNA mimic ¹⁵ (i.e., competitive with DNA), recent studies using HCV helicase found that ATA was competitive with adenosine triphosphate and noncompetitive with DNA. ¹⁶ To determine the inhibition of ATA for apPOL^exo−, we first measured the dose–response of ATA under our standard assay conditions and determined the IC₅₀ to be 0.9 ± 0.2 µM ( Fig. 4A ). We then determined the inhibition mechanism of ATA against dNTPs. The concentrations of ATA and dNTPs were varied above and below their IC₅₀ and K_M, respectively, and the resulting data were fit to a variety of inhibition mechanisms using the Dynafit software. ¹² The best-fitting mechanism for ATA versus dNTPs was true noncompetitive with K_i–ATA and K_M–dNTPs values of 0.99 ± 0.08 µM and 3.6 ± 0.5 µM, respectively.

OPEN IN VIEWER

Figure 4.

Mechanism of aurintricarboxylic acid (ATA) inhibition. (A) The IC₅₀ of ATA was determined by the percentage of polymerase activity (♦) at variable concentrations of ATA in mM. Data were fit to the standard IC₅₀ equation (Eq. 1), and an IC₅₀ of 0.9 mM was determined. The inset image depicts the structure of ATA. (B) Plots of initial velocity versus dNTPs concentration at various concentrations of ATA. ATA concentrations are 0 uM (■), 2 uM (•), and 4 uM (♦). The solid lines are theoretical fits to a noncompetitive mechanism using values given in the text. (C) Plots of initial velocity versus DNA concentration at various concentrations of ATA. ATA concentrations are 0 uM (■), 2 uM (•), and 4 uM (♦). The solid lines are theoretical fits to a noncompetitive mechanism using values given in the text.

We also determined the inhibition mechanism of ATA against the DNA substrate. Again, the concentrations of ATA and DNA were varied, and the data were fit to several possible inhibition mechanisms. The best-fitting mechanism was true noncompetitive with a K_i–ATA and K_M–DNA of 1.47 ± 0.09 µM and 261± 15 µM, respectively. Because the inhibition mechanism of ATA is true noncompetitive against both dNTPs and DNA (i.e., the binding of ATA does not affect the apparent-K_M’s for the substrates), the two determined K_i’s and the IC₅₀ values should be the same. We have determined values of 0.90, 0.99, and 1.47 µM for the IC₅₀, K_i (against dNTPs), and K_i (against DNA), respectively, which are all in reasonable agreement with each other. In addition, the apparent-K_M values determined using the data shown in Figure 2 correspond very well to the K_M values determined using the data in Figure 4B and Figure 4C (3.4 and 3.6 µM for K_M–dNTPs and 305 and 261 nM for K_M–DNA, respectively). Based on this analysis, it appears that in the case of apPOL^exo−, ATA is not acting simply as a DNA mimic, but binds to the enzyme at a unique site that does not directly overlap with either the nucleotide- or DNA-binding site.

In conclusion, we have improved on a previously developed DNA polymerase assay by altering the fluorophore and quencher pair, thereby increasing its dynamic range. We found that the assay provides steady-state kinetic parameters that are comparable to the traditional ³²P-based assay and is both highly sensitive and reproducible using 96- and 384-well plates. A pilot screen resulted in excellent statistical parameters (Z’ factor and signal window) and identified a number of potential inhibitors of the apPOL^exo− DNA polymerase. We used the assay to determine the kinetic inhibition mechanism of ATA and somewhat unexpectedly found that it was not behaving as a DNA mimic but was noncompetitive against both dNTPs and DNA. The robustness of the modified assay strongly suggests that it is well suited to the screening of much larger small-molecule libraries in an effort to identify inhibitors of the P. falciparum apicoplast DNA polymerase that can serve as lead compounds in antimalarial drug-discovery efforts.