Development of an in Vitro Screen for Compound Bioaccumulation in Haemonchus contortus Author(s): Xin Zhou, Jinxia Nancy Deng, Bernard D. Hummel, Debra J. Woods, Wendy T. Collard, Steven X. Hu, Matthew J. Zaya, Christopher S. Knauer, David P. Thompson, Dawn A. Merritt, Julie K. Lorenz, and Alan A. Marchiondo Source: Journal of Parasitology, 100(6):848-855. Published By: American Society of Parasitologists DOI: http://dx.doi.org/10.1645/14-556.1 URL: http://www.bioone.org/doi/full/10.1645/14-556.1

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research.

J. Parasitol., 100(6), 2014, pp. 848–855 Ó American Society of Parasitologists 2014

DEVELOPMENT OF AN IN VITRO SCREEN FOR COMPOUND BIOACCUMULATION IN HAEMONCHUS CONTORTUS Xin Zhou, Jinxia Nancy Deng, Bernard D. Hummel, Debra J. Woods, Wendy T. Collard, Steven X. Hu, Matthew J. Zaya, Christopher S. Knauer, David P. Thompson, Dawn A. Merritt, Julie K. Lorenz, and Alan A. Marchiondo Veterinary Medicine Research and Development, Zoetis LLC, Kalamazoo, Michigan 49007. Correspondence should be sent to: alan.marchiondo@zoetis. com

ABSTRACT:

The objective of the current study was to establish an in vitro screen and a highly sensitive analytical assay to delineate key physicochemical properties that favor compound bioaccumulation in the L3 life stage of a Haemonchus contortus isolate. Timedependent studies revealed that absorption and elimination kinetics during the first 6 hr of exposure were sufficient to achieve maximum bioaccumulation for the majority of compounds tested. In subsequent studies, the larvae were incubated for 6 hr in a medium containing 146 compounds (5 lM initial concentration), including both human and veterinary medicines, characterized by a broad range of physicochemical properties. Bioaccumulation of the compounds by the nematodes was determined, and multiple physicochemical descriptors were selected for correlation. Data analysis using Bayes classification model and partial least-square regression revealed that clogD7.4, rotatable bond, E-state, and hydrogen bond donor each correlated with compound bioaccumulation in H. contortus L3. The finding that lipophilicity was critical for transcuticle compound permeation was consistent with previous studies in other parasitic species and in adult H. contortus. The finding of additional physicochemical properties that contribute to compound conformational flexibility, polarity, and electrotopological state shed light on the mechanisms governing transcuticle permeation. The relatively poor correlation between transcuticle and transmembrane permeation indicated the distinct mechanisms of compound permeation, likely due to the different constituents, and their contributions to overall transport function, of the lipid membranes and the porous collagen barrier of the nematode cuticle. Our study, for the first time, establishes a high-throughput screen for compound bioaccumulation in a parasitic nematode and further elucidates physicochemical factors governing transcuticular permeation of compounds. Application of this methodology will help explain the basis for discrepancies observed in receptor binding and whole organism potency assays and facilitate incorporation of drug delivery principles in the design of candidate anthelmintics.

porous collagen matrix of the cuticle. Lipophilicity, pKa, and molecular size were determined to be critical for the drug permeation (Thompson et al., 1993). The transcuticular pathway was shown to be the major route of absorption in intact adult stages of A. suum and Haemonchus contortus, based on studies on gastrointestinal ligated and non-ligated nematodes in closed perfusion systems (Ho et al., 1992; Sims et al., 1992). With the advancement of bioanalytical techniques and establishment of more sophisticated ex vivo approaches, recent studies for drug absorption in parasitic nematodes have utilized cold compounds instead of radiolabeled compounds, and compound metabolites were assessed in addition to parent compounds. Absorption studies in H. contortus L3, using known anthelmintics, revealed that permeability coefficient correlated reasonably well with lipophilicity, after factoring in size-restriction effects of the collagen cuticle (Ho et al., 1994). Recent studies with benzimidazole analogs confirmed that passive diffusion through the cuticle was the primary site and mechanism for drug permeation; lipophilicity and concentration gradient were key factors governing drug absorption in A. suum (Mottier et al., 2006). Ex vivo albendazole (ABZ) uptake over time was significantly higher compared with its more polar metabolite ABZ sulfoxide (Alvarez et al., 2001) in A. suum. Higher concentrations of ABZ, compared with its sulfoxide metabolite, were measured in H. contortus recovered from infected treated sheep, confirming that higher lipophilicity may contribute to greater penetration (Alvarez et al., 2000). The absence of ABZ in the peripheral plasma and low ABZ recovered in portal blood in catheterized ABZ-treated sheep further support the concept that transcuticular drug absorption in H. contortus is important in vivo (Alvarez et al., 2000). Overall, the studies designed to understand the mechanism of drug permeation have been greatly restricted in chemical variety, testing only a limited set of model permeants and anthelmintics. Therefore, studies to expand the chemical space are warranted to better understand the chemical features that favor transcuticle permeation.

In anthelmintic drug discovery, high-throughput screening programs for new drug candidates are typically conducted using 2 distinct types of studies: receptor binding assays and whole organism efficacy screens monitoring the motility or mobility of the parasite (Woods and Williams, 2007). In addition to drug potency in the receptor binding and whole organism screens, drug absorption, distribution, metabolism, and excretion in the host are critical determinants for the success of anthelmintic candidates. Efforts to identify the physicochemical space for successful drug candidates have been usefully applied to drug discovery programs in human medicine. The most prominent system, or ‘‘rule of five,’’ defined the physicochemical space in relation to solubility and permeability for drug absorption (Lipinski et al., 2001). Recent efforts have attempted to design drugs of appropriate physicochemical properties for improved disposition and safety (Meanwell, 2011). A lower number of hydrogen bond donors was identified for commercial herbicides and insecticides compared with human drugs in an investigation of the physicochemical space for agrochemicals (Tice, 2001). However, in anthelmintic drug discovery, little research has focused on physicochemical properties due to the lack of knowledge regarding how parasites accumulate drugs. In addition, differences among life cycle stages and the inaccessibility of adult stages of parasites complicated the interpretation of findings. Early efforts to define the factors that govern drug absorption in parasitic nematodes were conducted using isolated cuticle segments from Ascaris suum, a gastrointestinal nematode large enough to provide cuticle segments that could be isolated and mounted into traditional Ussing chambers (Ho et al., 1990). The rate of penetration for drugs and model permeants across the cuticle of A. suum was shown to be governed by lipid barriers in the hypodermis and by molecular size and charge restriction barriers formed by the Received 29 April 2014; revised 23 July 2014; accepted 4 August 2014. DOI: 10.1645/14-556.1 848

ZHOU ET AL.—COMPOUND BIOACCUMULATION IN H. CONTORTUS

Recently Burns and colleagues developed a structure-based predictive model for drug bioaccumulation in Caenorhabditis elegans (Burns et al., 2010), in which they built a high-throughput screen and used it to identify several key chemical motifs favoring or limiting drug bioaccumulation in free-living nematodes. However, the application for this model may be limited for anthelmintic discovery due to physiological and external environmental differences between free-living and parasitic nematodes. For example, the permeability properties of cuticles in different species and stages of nematode have not been rigorously explored. A recent study demonstrated that the anthelmintic drug derquantel was weakly active in intact L4 young adult stage C. elegans but highly active in the cut worms (RuizLancheros et al., 2011), which suggested that passage of this and perhaps other molecules through the C. elegans cuticle may be more restricted than it is in other adult parasitic nematodes. Alternatively, it could indicate between-species differences in active uptake or extrusion mechanisms. Second, the microenvironment of the cuticles, e.g., pH value, may vary between different species of nematodes and impact drug absorption, especially the acidic and basic drugs characterized by different pKa values. The cuticle microenvironment pH of adult H. contortus is maintained at ~5.0 due to the secretion of organic acids through pores that traverse the cuticle, whereas little is known about the cuticle pH of C. elegans (Sims et al., 1994, 1996). In addition, the predictive model developed for C. elegans may not be applicable to novel structure scaffolds not contained in the training set of the screened libraries. The aim of the current study is to adapt the high-throughput screen from C. elegans (Burns et al., 2010) to the L3 life stage of a parasitic nematode, H. contortus, but apply a distinct modeling approach to identify the physicochemical properties that correlated with compound permeation and expand the understanding of the mechanisms for transcuticle drug bioaccumulation. The screening library included 146 compounds composed of commercially available human medicines and anthelmintics, which represented diverse structural and physicochemical space. After the screen was established, correlation of drug bioaccumulation and physicochemical properties was performed and revealed that clogD7.4, rotatable bond, E-state, and hydrogen bond donor were critical for the transcuticular permeation of compounds in H. contortus L3.

849

and the centrifugation step were repeated 3 times. Finally, the L3 were resuspended in Basal Medium composed of (w/v) 2% pancreatic digest of casein (Bacto-Tryptone), 1% yeast extract, 5.7% d-glucose, 0.08% dipotassium hydrogen orthophosphaten, 0.08% potassium dihydrogen orthophosphate, 2 mM Hepes, 0.0065% penicillin G, and 0.0065% streptomycin as well as 0.025% gentamicin (v/v), to a concentration of approximate 10 L3/ ll. All reagents and antibiotics were purchased from Sigma-Aldrich (St. Louis, Missouri) or Thermo Fisher Scientific (Pittsburgh, Pennsylvania). Here 500 ll of worm suspension was added to a 96-well polypropylene deepwell plate of 1-ml well volume (Thermo Fisher Scientific). Since DMSO may impact drug permeation, the volume of DMSO solution used was kept to a minimum. To test wells, 2 ll of 5 lM test compound dissolved in DMSO were added resulting in a final DMSO concentration of 0.4%. L3 were then incubated at 37 C for 6 hr. After the incubation, the worm suspension in the plate was transferred to an AcroPrep 0.45-lm GHP membrane 96-well filter plate (Pall Corporation, Port Washington, New York) and drained under vacuum. The worm suspension attached to the filter of the plate was then washed twice with 500 ll 0.1% SDS to remove residual compounds adhering to the nematode surface (Burns et al., 2010), followed by an additional wash with 500 ll Basal Medium. After washing and draining, L3 in each well were re-suspended in 100 ll Basal Medium and transferred to a solid-bottom 96-well 1-ml Deepwell plate, and stored frozen at20 C overnight. After freezing overnight, the worm samples were lysed by adding 100 ll of 23 lysis buffer (100 mM KCl, 20 mM Tris, pH 8.3, 0.4% SDS, 120 lg/ml proteinase K, as in Burns et al., 2010) to each well and incubated at 60 C for 1 hr with mixing. To ensure complete disruption of the nematode by the lysis buffer, the samples were homogenized with a polytron (Thermo Fisher Scientific), and no additional amount of compound was recovered after homogenization. A small aliquot of the worm lysates (2 ll) was used to determine protein concentration, using a BCA Protein Assay kit (Thermo Fisher Scientific), and the resulting protein concentration was used to normalize the compound concentration recovered from the L3. The remaining lysate sample extract was precipitated by the addition of 100 ll acetonitrile to each well and centrifuged. Supernatant (10 ll) was used to quantify test compound concentration in the worm lysate using LC/MS. For each compound, quadruplicate-treatment samples were prepared for quantitative analysis. No-worm controls were conducted in duplicate samples to determine the amount of test compound adhering to the filter plate. No-compound controls were performed in duplicate and used for preparing the calibration standards for each compound at a concentration of 5 and 1 lM. In preliminary studies to establish the screening protocol, a time- and dose-dependent incubation was conducted for a subset of compounds. In the time-dependent study, the selected compounds were incubated for 1, 3, 6, 18, and 24 hr. In the dose-dependent study, several compounds were tested at 5, 25, 60, and 120 lM using 6 hr incubations. LC/MS analytical method

MATERIALS AND METHODS Reagents and chemicals Compounds from internal and commercial sources were dissolved in DMSO to obtain 30 mM stock solutions. These stocks were serially diluted with DMSO to a concentration of 1.25 mM prior to addition to culture plates. The compounds that could not be dissolved in DMSO at 30 mM were omitted from testing. A total of 146 compounds including 20 anthelmintics including major classes of macrocyclic lactones, benzimidazoles, levamisole, and amino-acetonitrile derivatives, as well as human medicines for nervous system, cancer, infection, inflammation, gastrointestinal, and cardiovascular diseases, were used in the compound permeation assay in L3 H. contortus. L3 H. contortus accumulation assay Viable L3-stage H. contortus (anthelmintic-susceptible isolate Fort Dodge/ Zoetis acquired 2010), stored at 4 C in water, were harvested by centrifugation for 3 min at 150 g. After centrifugation, the supernatant was removed, and the L3-stage larvae (L3) were exsheathed with 2% (w/v) sodium hypochlorite by mixing and shaking, followed by centrifugation for 3 min at 150 g. After centrifugation, the supernatant was removed, and the L3 were washed with Glucose Tyrode’s balanced salt solution composed of w/v) 0.96% glucose Tyrodes, 0.1% sodium bicarbonate, 0.12% penicillin G, and 0.2% streptomycin sulphate as well as 0.001% 5-fluorocytosine. The wash

A generic ion source condition and fragmentation condition were set up using an API 4000 mass spectrometer (AB SCIEX, Framingham, Massachusetts) for positive and negative mode, respectively: collision gas 5 psi; curtain gas, 30 psi; heated nebulizer temperature, 600 C; ion spray voltage, 5,500 or 4,500 V; nebulizer gas, 50 psi; turbo gas, 70 psi; declustering potential, 50 or 50 V; entrance potential, 10 or 10 V; collision energy, 30 or 30 eV; and collision cell exit potential, 10 or 10 V. The product ion spectrum of each compound was acquired, and the transitions for molecular ion to the most abundant 1 or 2 fragment ions were used for Multiple Reaction Monitoring in quantitative analysis. Samples (10 ll) from the compound accumulation assay were injected on an Acquity UPLC system (Waters, Milford, Massachusetts) with a PDA detector (190–500 nm, 1.2 nm steps) connected with the mass spectrometer described above. The compounds were eluted on a BEH C18 (2 3 50 mm) column (Waters), and the temperature for the column was set at 45 C. The generic chromatography method started at mobile phase 5% B (95:5 acetonitrile: water containing 0.1% formic acid [v/v])/95% A (95:5 water: acetonitrile containing 0.1% formic acid [v/v]) and linear gradient to 95% B over 3 min and was maintained at 95% B for 0.5 min and followed by equilibration back to starting conditions from 3.5 to 5 min. All solvents were HPLC grade. Preliminary study established that the detection limit providing a signal to noise ratio 5:1 for a majority of the compounds was approximately 0.01 lM. In the subsequent assays, 0.01 lM was set as the lower limit of quantitation for accumulation in the nematodes.

850

THE JOURNAL OF PARASITOLOGY, VOL. 100, NO. 6, DECEMBER 2014

FIGURE 1. Time course for compound bioaccumulation by Haemonchus contortus L3. All compounds were tested at 5 lM, and incubations maintained for 1, 3, 6, 18, and 24 hr. Concentrations bioaccumulated in the nematodes were determined as described in Materials and Methods. The data were presented as the average of 4 determinations for each compound at each time point in log scale and normalized to protein amount (1.5 mg) recovered from lysed L3. Representative 6 out of 41 compounds tested using this format were used to make the time course graph. Data analysis Data analysis was conducted using Canvas, a cheminformatics package (Schrodinger, Version 1.6, New York) that provides a range of applications ¨ for structural and data analysis, including fingerprints, similarity searching, substructure searching, selection by diversity, clustering, and building regression and classification models (Duan et al., 2010). The molecular descriptors in relevance to molecular weight, lipophilicity (clogP and clogD7.4), and polarity (PSA) of the compounds were obtained from ACD/ Labs (Version 12.0, Toronto, Ontario, Canada), and the rest of more than 100 physicochemical properties were calculated by Canvas. The membrane permeability of compounds was obtained using a low efflux Madin-Darby canine kidney cell line (MDCKII-LE) in a previously described assay method (Di et al., 2011). Highly correlated descriptors were removed from data analysis. Two analyses were carried out for correlation of compound bioaccumulation with remaining descriptors, Bayes classification model, and quantitative model by partial least square (PLS) analysis for the 146 test compounds, including 20 anthelmintics compounds. The compounds were randomly selected and separated into training set and test set. In the Bayes Classification model, the model precision, sensitivity, specificity, and accuracy were determined using equations I–IV (below), in which TP represents true positive, FP represents false positive, TN represents true negative, and FN represents false negative. In the quantitative PLS regression, predicted compound absorption was plotted against the actual compound absorption using the log scale to indicate how effectively the linear model worked based on the physicochemical properties selected.

I. Precision ¼ TP/ (TPþFP) II. Sensitivity ¼ TP/ (TPþFN) III. Specificity ¼ TN/ (TNþFP) IV. Accuracy ¼ (TPþTN)/ (TPþTNþFPþFN) RESULTS Development of a high-throughput screen for compound accumulation in H. contortus L3 Time-dependent compound bioaccumulation studies were conducted to determine the time required for drug concentrations

FIGURE 2. Dose-dependent compound bioaccumulation by Haemonchus contortus L3. Four compounds were tested at 5, 20, 50, and 120 lM during 6 hr incubations. Concentrations bioaccumulated by the nematodes were determined as described in Materials and Methods. Results presented are the average of 4 determinations for each compound at each treatment concentration in log scale.

in H. contortus L3 to plateau. For these studies, L3 were incubated in the same neutral Basal Medium used for whole organism efficacy screening. Forty-one compounds from the 146 compound-library including human medicines and anthelmintics of diverse structures and physicochemical properties were selected. Compounds were tested at a concentration of 5 lM, which was pharmacologically relevant for anthelmintics in vivo, which typically achieve blood levels of 1 to 10 lM. Results for representative compounds with absorption ranked from low to high are depicted in Figure 1. For the majority of the 41 compounds that reached within-worm concentrations above the detection limit of 0.01 lM, the maximal drug concentrations in the parasites occurred by 6 hr. The kinetics was clearly separated into absorption and disposition phases, in which absorption and elimination prevailed, respectively. To further validate the incubation conditions for compound bioaccumulation assay, 4 compounds from the 41 compounds with various structures and physicochemical properties including verapamil, amitriptyline, terfenadine, and tamoxifen (Fig. 2) were selected for dosedependent study at 5, 20, 50, and 120 lM. Incubation time was fixed at 6 hr to obtain approximately maximum bioaccumulation. The results (Fig. 2) demonstrated that for all the compounds examined, the concentration of the compound in L3 increased proportionally with treatment concentration; the absorption plateau was achieved for each compound at concentration 120 lM. Correlation of physicochemical properties with compound accumulation With the high-throughput screen established, bioaccumulation was determined for each of 146 compounds tested. Initial data analysis was conducted to identify the physicochemical space that can differentiate the compounds with bioaccumulation less than and greater than the lower detection limit 0.01 lM. Among

ZHOU ET AL.—COMPOUND BIOACCUMULATION IN H. CONTORTUS

851

compounds were randomly selected for the training set, and the remaining 35 compounds were assigned to the test set (Fig. 4). Within the training set, 6 compounds with bioaccumulation . 0.01 lM were incorrectly predicted to be , 0.01 lM (false negative); 11 compounds with bioaccumulation , 0.01 lM were predicted to be . 0.01 lM (false positive); whereas 28 compounds (true negative) in Class 1 and 66 compounds in Class 2 (true positive) were correctly categorized. In a test set, 4, 6, 6, and 19 compounds were designated to be false negative, false positive, true negative, and true positive, respectively. To further assess the performance of the classification model, precision, sensitivity, specificity, and accuracy were calculated using the equations described in Materials and Methods (Fig. 4). The high values obtained for these parameters (i.e., all showing values . 0.5) indicated the success of the model to predict compound bioaccumulation. Subsequently, PLS regression was performed for all of the compounds with bioaccumulation to worm concentrations . 0.01 lM (Fig. 5). The predicted concentrations, built on the 4 highly correlated physicochemical properties (clogD7.4, HBD, RB, and E-state), were plotted against the measured concentrations (log scale) for compounds with absorption greater than 0.01 lM. The R2 for training set (76 compounds) and test set (19 compounds) was about 0.3 and 0.4, respectively. All 20 anthelmintics of major classes (Table I) tested showed bioaccumulation concentrations . 0.01 lM (data not shown), indicating that the anthelmintics occupy appropriate physicochemical space for transcuticle permeation. Compound accumulation and membrane permeability

FIGURE 3. Differentiation of compound accumulation with clogD7.4. A total of 146 compounds were divided into 2 classes according to the compound concentration bioaccumulated by Haemonchus contortus L3 during 6 hr incubations. Three physicochemical properties MW, PSA, and clogD7.4 from ACD/Labs were selected to differentiate compounds with accumulation less than and greater than 0.01 lM.

molecular weight (MW), polar surface area (PSA), and clogD7.4, which were obtained from ACD/Labs, clogD7.4 was determined to be a key factor. Compounds with clogD greater than 2 tended to accumulate in the nematodes, whereas PSA and MW had minimal impact on the compound bioaccumulation (Fig. 3). To expand the understanding of contribution of other physicochemical properties, descriptors including alogP, hydrogen bond acceptor (HBA), hydrogen bond donor (HBD), rotatable bond (RB), topological polar surface area (TPSA), E-State, and molecular refraction (MR) were calculated using Canvas. Among those properties, 4 independent properties—clogD7.4, HBD, RB, and E-state—were used for subsequent data analysis. Using the Bayes Classification model, the compounds were divided into 2 classes based on their absorption: Class 1, absorption , 0.01 lM and Class 2, absorption . 0.01 lM. From the 146 compounds collected of human medicines and anthelmintics tested, 111

Data from studies testing compound permeation through standard cell membranes, using the low efflux MDCKII cell line (Di et al., 2011), were compared against data from the accumulation measured in the H. contortus L3 assay. For compounds with accumulation less than 0.01 lM, membrane permeability ranked 0.783 to 44.7 (3 106 cm s1). For compounds with accumulation greater than 0.01 lM, the correlation was poor (Fig. 6). DISCUSSION Physicochemical properties known to affect drug permeation across membranes include molecular weight, lipophilicity (estimated by clogP), hydrogen bond acceptor, and donor. They have been shown to be good predictors of oral drug absorption and are now commonly known as the ‘‘rule of five’’ (Lipinski et al., 2001). A different but related system to Lipinski’s rule-of-five is the ‘‘Oral PhysChem Score’’ (Lobell et al., 2006), which uses a ‘‘traffic light’’ approach and has provided consistent evidence that molecular weight, clogP, number of rotatable bonds, calculated solubility, and TPSA are key physicochemical properties governing the rate of oral drug absorption. Knowledge of the mechanisms governing transcuticle drug absorption, and the physicochemical space favoring it by parasitic nematodes should facilitate the discovery of anthelmintics by frontloading structural features relevant to delivery in the selection and design of drug candidates. The nematode cuticle resembles the cell membrane in terms of the lipid barrier (membranes associated with the hypodermis) but is also distinct from the cell membrane by virtue of the negatively charged, porous collagen

852

THE JOURNAL OF PARASITOLOGY, VOL. 100, NO. 6, DECEMBER 2014

FIGURE 4. Bayes Classification model for correlation of compound bioaccumulation and physicochemical properties. Compounds were divided into 2 classes according to the compound concentration bioaccumulated by H. contortus L3 during 6 hr incubations. Four physicochemical properties of the compounds including clogD7.4, hydrogen bond donor, rotatable bond as well as E-state were selected for correlation with bioaccumulation. The model was established by randomly selecting compounds and separating them into training set and test set. Precision, sensitivity, selectivity, and accuracy of the model were evaluated as described in Materials and Methods.

matrix that imposes size restrictions on permeant transport. In addition, parasitic nematodes were reported to secrete the organic acid through the aqueous pores in the collagen layer of the cuticle (Sims et al., 1994, 1996), creating a microenvironment pH of ~5.0; this could further influence transport of weakly acidic or basic permeants. Furthermore, the composition of lipid layers varies between cuticle and cell membrane. All of these may explain why the relatively low correlation was observed in the present study between transcuticle and transmembrane drug permeation. The in vitro H. contortus L3 assay as a drug discovery screening tool correlates well with the anthelmintic activities of the major classes of nematocides, in spite of the fact that this free-living stage is not the parasitic target found in the parasitized host. Artificial ex-sheathment of the larvae simulates the removal of the cuticle in the ruminant host prior to developing to a parasitic larval stage. While the ex-sheathment process coupled with the differences in microenvironmental pH, temperature and feeding behavior of the free-living L3 versus the parasitic L4 may well have an impact on subsequent transcuticle, transmembrane permeation, and time/intensity of drug bioaccumulation, dose titration curves of known nematocidal anthelmintics such as moxidectin validate the fidelity of the in vitro L3 assay.

The compound library selected for screening covered a diverse set of human medicines in major therapeutic areas, as well as anthelmintics representing each of major classes, including benzimidazoles, levamisole, macrocyclic lactones, and aminoacetonitriles. The library also covered diverse physicochemical space to facilitate correlation of compound bioaccumulation. The physicochemical properties that were identified in the established high-throughput screen included clogD7.4, HBD, RB, and E-state for the total 146 compounds. In the screen used, the incubation medium was neutral pH and identical to the medium used in whole organism efficacy screening. clogD7.4 rather than clogP was selected for correlation by considering pKa and subsequent ionization of the compounds at neutral pH. Lipophilicity was shown to be one of the most critical physicochemical properties governing transcuticle drug permeation, consistent with previous findings in other parasitic nematodes and in studies testing adult H. contortus (Thompson et al., 1993; Ho et al., 1994; Alvarez et al., 2000; Mottier et al., 2006). Number of HBD usually refers to number of nitrogen and oxygen atoms in a molecule. It is known that excessive hydrogen bonding of polar atoms in a compound to water molecules can generate strongly bound hydration shells that impair the ability of the compound to enter the lipophilic phase of cell membranes. The sum of donor and acceptor impacts oral drug bioavailability (Veber et al., 2002); however, only HBD was

ZHOU ET AL.—COMPOUND BIOACCUMULATION IN H. CONTORTUS

FIGURE 5. PLS regression for correlation of compound bioaccumulation and physicochemical properties. Predicted concentrations in the log scale using the equation derived from 4 physicochemical properties (clogD7.4, HBD, RB, and E-state) was plotted against the experimentally measured concentration in log scale for both training set and test set.

determined to be critical to transcuticle permeation. RB reflects the conformational flexibility of a molecule, which in turn influences oral bioavailability (Veber et al., 2002) as well as transcuticle permeation. Low conformational flexibility is beneficial because it requires less energy for a molecule to adapt a specific conformation required to pass the lipid-rich cuticle. Estate is a descriptor for the electrotopological state of atoms in a molecule (Kier and Hall, 2000), which combines the electronic properties and the topological environment for each atom. E-state

853

FIGURE 6. Correlation of cell membrane permeability and transcuticle bioaccumulation. Compound concentrations bioaccumulated in Haemonchus contortus L3 during 6 hr incubations was compared with compound membrane permeability determined in the MDCKII-LE cell line in a log scale. Only compounds with bioaccumulation more than 0.01 lM and with membrane permeability data available were plotted in the graph.

indices for nitrogen, oxygen, and hydrogen atoms correlated highly with drug transport rates for selected human medicines (Norinder and Osterberg, 2001) and was correlated to transcuticle compound absorption in the current study. It should be noted that the kinetics for drug concentrations in H. contortus L3 could be separated into 2 phases, the absorption phase, during which the rate of drug permeation through cuticle overwhelmed drug elimination processes, and the disposition phase, during which the rate of elimination predominated. The ka and ke differed for different compounds, indicating the difference in the rate of absorption and elimination may be due

TABLE I. Anthelmintics tested in the in the accumulation assay in Haemonchus contortus L3.* Compound name

Chemical class/family

MW

ClogD7.4_ACD

PSA_ACD

Levamisole Pyrantel Demiditraz 2-Desoxoparaherquamide Fenbendazole Marcfortine A Thiabendazole Praziquantel Mebendazole Albendazole Flubendazole Spinosad Isoxazoline Niclosamide Selamectin Triclabendazole Ivermectin Monepantel Doramectin Moxidectin

Imidazothiazole Tetrahydropyrimidine Substituted Imidazole Paraherquamide/marcfortine Benzimidazole Alkaloid Benzimidazole Quinoline Benzimidiazole Benzimidiazole Benzimidazole Spinosyn Isoxazoline Salicylanilide Macrocyclic lactone Benzimidazole Macrocyclic lactone Amino-acetonitrile derivative Macrocyclic lactone Macrocyclic lactone

204 206 200 479 299 477 201 312 295 265 313 731 555 327 769 359 875 473 898 637

0.0629 0.818 1.77 2 2.36 2.4 2.47 2.66 2.83 2.99 3.08 3.38 3.49 3.53 4.42 5.27 5.69 5.81 6.05 6.71

40.9 43.84 28.68 91.34 92.31 71.11 69.81 40.62 84.08 93.31 84.08 111.22 79.79 95.15 154.73 63.21 170.06 111.21 181.06 98.97

*The compounds were arranged in the ascending order of their clogD7.4.

Membrane permeability

21

1.07

854

THE JOURNAL OF PARASITOLOGY, VOL. 100, NO. 6, DECEMBER 2014

to drug metabolizing enzymes. Xenobiotic metabolizing enzymes including oxidases, reductases, hydrolases, and transferases have been described in several helminth species (Cvilink et al., 2009). Cytochrome P450 oxidase activity was demonstrated in microsomal preparations of H. contortus L1 and L3 larvae, but the activity detected in the adult stage was very low (Kotze, 1997). The P450 inhibitor piperonyl butoxide was shown to increase the toxicity of rotenone to both L3 and adult H. contortus, suggesting the active P450-mediated metabolism in both stages (Kotze et al., 2006). The activity of P450 in the metabolism of moxidectin was implicated in the adult stage of H. contortus due to the production of an undefined metabolite, which was inhibited by carbon monoxide (Alvinerie et al., 2001). P450 activity was compared in macrocyclic lactone-susceptible and -resistant H. contortus, but no difference was detected, likely due to the low activity in the adult stage towards the substrates used in the study (Kotze, 2000). A recent study demonstrated the transcriptional upregulation of 3 cytochrome P450 genes (cyp-35C1, cyp-35A2, and cyp-35A5) in response to ABZ treatment, indicating a role in drug metabolism (Laing et al., 2010). In addition to the Phase I activity, UDPglucosyltransferases was demonstrated to be present in C. elegans due to the detection of a glucoside metabolite (Laing et al., 2010). When anthelmintics are administered to livestock, the pharmacokinetics is determined at 2 levels, the absorption and elimination in the sheep or cattle, as well as the absorption and elimination in the parasitic nematode. In vivo assessment of drug concentrations in parasitic nematodes would provide additional useful insights into the efficacious time window for novel anthelmintic drugs. The absorption data obtained from the current study was assumed to be dependent only on drug permeation, with minimal efflux. Multidrug resistance (MDR) has emerged as a potentially important factor in this process, with increased P-glycoprotein (Pgp) mRNA expression in an ivermectin-resistant strain of H. contortus (Smith and Prichard, 2002). Pgp acts as an efflux transporter to reduce the drug concentration in H. contortus (Xu et al., 1998; Smith and Prichard, 2002). Use of the MDRreversing agents verapamil and CL347099 increased the efficacy of ivermectin and moxidectin against resistant H. contortus strains in jird models (Molento and Prichard, 1999). Overexpression of 5 Pgps (PgpA, -B, -C, -D, and -E) was selected after ivermectin and moxidectin treatment (Prichard and Roulet, 2007) in adult H. contortus. Recently it has been demonstrated that most of the macrocyclic lactones are Pgp substrates and have a stimulatory effect on the transporter activity of the reference Pgp substrate rhodamine 123 in the eggs of the H. contortus–resistant isolate (Kerboeuf and Guegnard, 2011). The high-throughput screen established in the current study, using H. contortus L3, could facilitate the anthelmintic candidate-selection process by focusing resources on compounds that accumulate more rapidly in parasites, through either more rapid permeation into H. contortus or slower metabolism and excretion, or both. ACKNOWLEDGMENTS We are grateful to Dr. John Wendt and Dr. Ashley Fenwick for helpful discussions on medicinal chemistry aspects of the research.

LITERATURE CITED ALVAREZ, L. I., F. A. IMPERIALE, S. F. SA´NCHEZ, G. A. MURNO, AND C. E. LANUSSE. 2000. Uptake of albendazole and albendazole sulphoxide by

Haemonchus contortus and Fasciola hepatica in sheep. Veterinary Parasitology 94: 75–89. ———, M. L. MOTTIER, S. F. SA´NCHEZ, AND C. E. LANUSSE. 2001. Ex vivo diffusion of albendazole and its sulfoxide metabolite into Ascaris suum and Fasciola hepatica. Parasitology Research 87: 929–934. ALVINERIE, M., J. DUPUY, C. EECKHOUTTE, J. F. SUTRA, AND D. KERBOEUF. 2001. In vitro metabolism of moxidectin in Haemonchus contortus adult stages. Parasitology Research 87: 702–704. BURNS, A. R., I. M. WALLACE, J. WILDENHAIN, M. TYERS, G. GIAEVER, G. D. BADER, C. NISLOW, S. R. CUTLER, AND P. J. ROY. 2010. A predictive model for drug bioaccumulation and bioactivity in Caenorhabditis elegans. Nature Chemical Biology 6: 549–557. CVILINK, V., J. LAMKA, AND L. SKA´LOVA´. 2009. Xenobiotic metabolizing enzymes and metabolism of anthelminthics in helminths. Drug Metabolism Review 41: 8–26. DI, L., C. WHITNEY-PICKETT, J. P. UMLAND, H. ZHANG, X. ZHANG, D. F. GEBHARD, Y. LAI, J. J. FEDERICO 3RD, R. E. DAVIDSON, R. SMITH, ET AL. 2011. Development of a new permeability assay using low-efflux MDCKII cells. Journal of Pharmaceutical Sciences 100: 4974–4985. DUAN, J., S. L. DIXON, J. F. LOWRIE, AND W. SHERMAN. 2010. Analysis and comparison of 2D fingerprints: Insights into database screening performance using eight fingerprint methods. Journal of Molecular Graphics and Modeling 29: 157–170. HO, N. F., T. G. GEARY, C. L. BARSUHN, S. M. SIMS, AND D. P. THOMPSON. 1992. Mechanistic studies in the transcuticular delivery of antiparasitic drugs. II: Ex vivo/in vitro correlation of solute transport by Ascaris suum. Molecular and Biochemical Parasitology 52: 1–13. ———, ———, T. J. RAUB, C. L. BARSUHN, AND D. P. THOMPSON. 1990. Biophysical transport properties of the cuticle of Ascaris suum. Molecular and Biochemical Parasitology 41: 153–165. ———, S. M. SIMS, T. J. VIDMAR, J. S. DAY, C. L. BARSUHN, E. M. THOMAS, T. G. GEARY, AND D. P. THOMPSON. 1994. Theoretical perspectives on anthelmintic drug discovery: Interplay of transport kinetics, physicochemical properties, and in vitro activity of anthelmintic drugs. Journal of Pharmaceutical Sciences 7: 1052–1059. KERBOEUF, D., AND F. GUEGNARD. 2011. Anthelmintics are substrates and activators of nematode P glycoprotein. Antimicrobial Agents and Chemotherapy 55: 2224–2232. KIER, L. B., AND L. H. HALL. 2000. The E-state as the basis for molecular structure space definition and structure similarity. Journal of Chemical Information and Computer Sciences 40: 784–791. KOTZE, A. C. 1997. Cytochrome P450 monooxygenase activity in Haemonchus contortus (Nematoda), International Journal for Parasitology 27: 33–40. ———. 2000. Oxidase activities in macrocyclic-resistant and -susceptible Haemonchus contortus. Journal of Parasitology 86: 873–876. ———, R. J. DOBSON, AND D. CHANDLER. 2006. Synergism of rotenone by piperonyl butoxide in Haemonchus contortus and Trichostrongylus colubriformis in vitro: Potential for drug-synergism through inhibition of nematode oxidative detoxification pathways. Veterinary Parasitology 136: 275–282. LAING, S. T., A. IVENS, R. LAING, S. RAVIKUMAR, V. BUTLER, D. J. WOODS, AND J. S. GILLEARD. 2010. Characterization of the xenobiotic response of Caenorhabditis elegans to the anthelmintic drug albendazole and the identification of novel drug glucoside metabolites. Biochemical Journal 432: 505–514. LIPINSKI, C. A., F. LOMBARDO, B. W. DOMINY, AND P. J. FEENEY. 2001. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Advanced Drug Delivery Reviews 46: 3–26. LOBELL, M., M. HENDRIX, B. HINZEN, J. KELDENICH, H. MEIER, C. SCHMECK, R. SCHOHE-LOOP, T. WUNBERG, AND A. HILLISCH. 2006. In silico ADMET traffic lights as a tool for the prioritization of HTS hits. ChemMedChem 1: 1229–1236. MEANWELL, N. A. 2011. Improving drug candidates by design: A focus on physicochemical properties as a means of improving compound disposition and safety. Chemical Research in Toxicology 24: 1420– 1456. MOLENTO, M. B., AND R. K. PRICHARD. 1999. Effects of the multidrug resistance-reversing agents verapamil and CL 347,099 on the efficacy of ivermectin or moxidectin against unselected and drug selected strains of Haemonchus contortus in jirds (Meriones unguiculatus). Parasitology Research 85: 1007–1011.

ZHOU ET AL.—COMPOUND BIOACCUMULATION IN H. CONTORTUS

MOTTIER, L., L. ALVAREZ, L. CEBALLOS, AND C. LANUSSE. 2006. Drug transport mechanisms in helminth parasites: Passive diffusion of benzimidazole anthelmintics. Experimental Parasitology 113: 49–57. NORINDER, U., AND T. OSTERBERG. 2001 Theoretical calculation and prediction of drug transport processes using simple parameters and partial least squares projections to latent structures (PLS) statistics. The use of electrotopological state indices. Journal of Pharmaceutical Sciences 90: 1076–1085. PRICHARD, R. K., AND A. ROULET. 2007. ABC transporters and beta tubulin in macrocyclic lactone resistance: Prospects for marker development. Parasitology 134: 1123–1132. RUIZ-LANCHEROS, E., C. VIAU, T. N. WALTER, A. FRANCIS, AND T. G. GEARY. 2011. Activity of novel nicotinic anthelmintics in cut preparations of Caenorhabditis elegans. International Journal for Parasitology 41: 455–461. SIMS, S. M., N. F. HO, T. G. GEARY, E. M. THOMAS, J. S. DAY, C. L. BARSUHN, AND D. P. THOMPSON. 1996. Influence of organic acid excretion on cuticle pH and drug absorption by Haemonchus contortus. International Journal for Parasitology 26: 25–35. ———, ———, L. T. MAGAS, T. G. GEARY, C. L. BARSUHN, AND D. P. THOMPSON. 1994. Biophysical model of the transcuticular excretion of organic acids, cuticle pH and buffer capacity in gastrointestinal nematodes. Journal of Drug Target 2: 1–8.

855

———, L. T. MAGAS, C. L. BARSUHN, N. F. HO, T. G. GEARY, AND D. P. THOMPSON. 1992. Mechanisms of microenvironmental pH regulation in the cuticle of Ascaris suum. Molecular and Biochemical Parasitology 53: 135–148. SMITH, J. M., AND R. K. PRICHARD. 2002. Localization of p-glycoprotein mRNA in the tissues of Haemonchus contortus adult worms and its relative abundance in drug-selected and susceptible strains. Journal of Parasitology 88: 612–620. THOMPSON, D. P., N. F. HO, S. M. SIMS, AND T. G. GEARY. 1993. Mechanistic approaches to quantitate anthelmintic absorption by gastrointestinal nematodes. Parasitology Today 9: 31–35. TICE, C. M. 2001. Selecting the right compounds for screening: does Lipinski’s Rule of 5 for pharmaceuticals apply to agrochemicals? Pest Management Science 57: 3–16. VEBER, D. F., S. R. JOHNSON, H. Y. CHENG, B. R. SMITH, K. W. WARD, AND K. D. KOPPLE. 2002. Molecular properties that influence the oral bioavailability of drug candidates. Journal of Medicinal Chemistry 45: 2615–2623. WOODS, D. J., AND T. M. WILLIAMS. 2007. The challenges of developing novel antiparasitic drugs. Invertebrate Neuroscience 7: 245–250. XU, M., M. MOLENTO, W. BLACKHALL, P. RIBEIRO, R. BEECH, AND R. PRICHARD. 1998. Ivermectin resistance in nematodes may be caused by alteration of P-glycoprotein homolog. Molecular and Biochemical Parasitology 91: 327–335.