QSAR study on permeability of hydrophobic compounds with artificial membranes

doi:10.1016/j.bmc.2007.03.040

Bioorganic & Medicinal Chemistry

Volume 15, Issue 11, 1 June 2007, Pages 3756-3767

https://doi.org/10.1016/j.bmc.2007.03.040 Get rights and content

Abstract

We previously reported a classical quantitative structure–activity relationship (QSAR) equation for permeability coefficients (P_app-pampa) by parallel artificial membrane permeation assay (PAMPA) of structurally diverse compounds with simple physicochemical parameters, hydrophobicity at a particular pH (log P_oct and |pK_a − pH|), hydrogen-accepting ability (SA_HA), and hydrogen-donating ability (SA_HD); however, desipramine, imipramine, and testosterone, which have high log P_oct values, were excluded from the derived QSAR equation because their measured P_app-pampa values were lower than calculated. In this study, for further investigation of PAMPA permeability of hydrophobic compounds, we experimentally measured the P_app-pampa of more compounds with high hydrophobicity, including several pesticides, and compared the measured P_app-pampa values with those calculated from the QSAR equation. As a result, compounds having a calculated log P_app-pampa > −4.5 showed lower measured log P_app-pampa than calculated because of the barrier of the unstirred water layer and the membrane retention of hydrophobic compounds. The bilinear QSAR model explained the PAMPA permeability of the whole dataset of compounds, whether hydrophilic or hydrophobic, with the same parameters as the equation in the previous study. In addition, PAMPA permeability coefficients correlated well with Caco-2 cell permeability coefficients. Since Caco-2 cell permeability is effective for the evaluation of human oral absorption of compounds, the proposed bilinear model for PAMPA permeability could be useful for not only effective screening for several drug candidates but also the risk assessment of chemicals and agrochemicals absorbed by humans.

Graphical abstract

To investigate PAMPA permeability of hydrophobic compounds, we experimentally measured the P_app-pampa of compounds with high hydrophobicity, including several pesticides, and compared the measured P_app-pampa values with those calculated from the QSAR equation derived in our previous study. The new bilinear QSAR model explained the PAMPA permeability of the whole dataset of compounds, whether they were hydrophilic or hydrophobic, with the same parameters as the equation in the previous study. In addition, the PAMPA permeability coefficients correlated well with Caco-2 cell permeability coefficients.

Introduction

In recent years, in vitro methods for predicting oral drug absorption have progressed. In vitro models of intestinal absorption generally focus on determining membrane permeability using Caco-2 cells, MDCK cells, artificial membranes, and immobilized artificial membrane (IAM) columns.¹ The parallel artificial membrane permeation assay (PAMPA), proposed by Kansy et al. in 1998,² is a high throughput in vitro assay system that evaluates transcellular permeation. In PAMPA, a 96-well microtiter plate completely filled with aqueous buffer solutions is covered with a hydrophobic filter coated with lipids in an organic solvent solution in a sandwich construction.² The PAMPA permeability of several drugs has been evaluated using various lipids dissolved in organic solvents, such as an egg lecithin in n-dodecane2, 3 or 1,9-decadiene,4, 5, 6 the composition of the lipid membrane that mimics the intestinal brush border membrane in 1,7-octadiene (Bio-mimetic PAMPA),7, 8, 9 the porcine polar brain lipid in dodecane,¹⁰ 100% n-hexadecane (HDM-PAMPA),¹¹ diolphosphatidylcholine in dodecane (DOPC-PAMPA)¹², and a phospholipid mixture in dodecane (Double-Sink PAMPA).¹² The PAMPA is used to screen for human intestinal absorption in pharmaceutical research because several papers indicated that PAMPA permeability is correlated with both Caco-2 cell permeability and human intestinal absorption.2, 3, 5, 6, 7, 9, 13, 14, 15, 16 PAMPA is also applicable for predicting passive blood–brain penetration¹⁰ and human skin permeability.¹⁷

In our previous study,⁶ we reported a classical QSAR equation for PAMPA permeability coefficients (P_app-pampa) of structurally diverse compounds with simple physicochemical parameters, log P_oct, where P_oct is the partition coefficient in the 1-octanol/water system, the absolute value of the difference between the pK_a value of the compound and the experimental pH 7.3, |pK_a − pH|, the surface area occupied by the hydrogen-bond acceptor and donor atoms of each modeled conformer, SA_HA, and SA_HD; however, desipramine, imipramine, and testosterone, having high log P_oct, were excluded from the QSAR equation because their measured P_app-pampa values were lower than calculated. Although we attempted the bilinear analysis of PAMPA permeability coefficients including these three compounds, it was difficult to obtain a statistically significant equation because there were few hydrophobic compounds in the dataset. Since many chemicals and agrochemicals used widely are hydrophobic, it is inevitable that humans are exposed to these hydrophobic chemicals as residues in food and water, or in occupational use;¹⁸ therefore, it is desirable to predict exposure using in vitro or in silico methods such as PAMPA.

In the present study, we experimentally measured the P_app-pampa of more compounds which have particularly high apparent hydrophobicity at the experimental pH, including several pesticides. Then, the P_app-pampa of the compounds, including the hydrophobic compounds, was analyzed using the QSAR approach to investigate whether other physiochemical parameters were significant in determining PAMPA permeability of the hydrophobic compounds. We also examined the effect of the unstirred water layer (UWL) and membrane retention on PAMPA permeability. Furthermore, PAMPA permeability of the hydrophobic compounds was compared with their Caco-2 cell permeability to investigate whether PAMPA is applicable for high-throughput evaluation to predict the exposure of humans to hydrophobic chemicals and agrochemicals, as well as drugs.

Section snippets

Permeability with artificial membranes

Table 1 shows the permeability coefficient through an artificial membrane, P_app-pampa values of added 22 chemicals and 15 agrochemicals with the data reported previously.⁶ We used Tris–HCl buffer (pH 7.3) containing 5–30% DMSO for the measurement of PAMPA permeability (5% DMSO: imidacloprid and 21 chemicals except pyrene; 20% DMSO: 13 agrochemicals except biphenyl and imidacloprid; 30% DMSO: biphenyl and pyrene). The log P_app-pampa values of added compounds were in the range of −5.95 (biphenyl)

Permeability with artificial membranes of hydrophobic compounds

It is important to predict the oral absorption of hydrophobic compounds from the perspective of not only effective screening for drug candidates in the early drug discovery stage but also risk assessment for agrochemicals and endocrine-disrupting chemicals in the environment. In the present study, we measured the PAMPA permeability coefficients of hydrophobic compounds which are useful for the prediction of human intestinal absorption.

The measured log P_app-pampa values of added hydrophobic

Conclusions

In this study, we clarified the barrier by the UWL on membrane surfaces and the membrane retention influencing PAMPA permeability of hydrophobic compounds. The bilinear QSAR model explained the PAMPA permeability of diverse compounds, whether hydrophilic or hydrophobic, with the same physicochemical parameters as those in the previous study. It was shown the PAMPA permeability is able to predict Caco-2 cell permeability, including hydrophobic compounds, which is used for the evaluation of human

Materials

Thirty-eight commercial drugs, 28 chemicals, and 11 agrochemicals were purchased from Nacalai Tesque (Kyoto, Japan), Kokusan Chemical Co. Ltd (Tokyo, Japan), Wako Pure Chemical Industries (Osaka, Japan), Bachem AG (Bubendorf, Switzerland), Kanto Chemical (Tokyo, Japan), Sigma–Aldrich Japan (Tokyo, Japan), Tocris Cookson Ltd (Bristol, UK) or BIOMOL Research Laboratories Inc. (PA, USA). Lecithin from egg yolk was purchased from Sigma (MO, USA). Hydrophobic filter plates (MultiScreen-IP, Clear

Acknowledgments

We are grateful to Nippon Kayaku Co., Ltd and Sankyo Agro Co., Ltd for providing a test compound, chromafenozide, and Dr. Yoshiaki Nakagawa (Kyoto University) for providing compounds, tebufenozide, methoxifenozide, and halofenozide. We also thank Emeritus Professor Toshio Fujita (Kyoto University) and Dr. Kiyohiko Sugano (Pfizer Japan Inc.) for helpful advice and comments on QSAR, and advice on the permeability experiment, respectively. The Caco-2 cell permeability coefficients of

References and notes (38)

C. Zhu et al.
Eur. J. Med. Chem.
(2002)
R. Ano et al.
Bioorg. Med. Chem.
(2004)
R. Ano et al.
Bioorg. Med. Chem.
(2004)
M. Fujikawa et al.
Bioorg. Med. Chem.
(2005)
K. Sugano et al.
J. Biomol. Screen
(2001)
K. Sugano et al.
Int. J. Phram.
(2002)
K. Sugano et al.
Int. J. Pharm.
(2003)
L. Di et al.
Eur. J. Med. Chem.
(2003)
M. Bermejo et al.
Eur. J. Pharm. Sci.
(2004)
A. Avdeef et al.
Eur. J. Pharm. Sci.
(2005)

E.H. Kerns et al.

J. Pharm. Sci.

(2004)

G. Corti et al.

Eur. J. Pharm. Sci.

(2006)

J.A. Ruell et al.

Eur. J. Pharm. Sci.

(2003)

C. Hansch et al.

P. Stenberg et al.

J. Med. Chem.

(2001)

P. Griffin et al.

Occup. Environ. Med.

(1999)

J. Lin et al.

Curr. Topics Med. Chem.

(2003)

M. Kansy et al.

J. Med. Chem.

(1998)

F. Wohnsland et al.

J. Med. Chem.

(2001)

Cited by (87)

Application of chemometric tools in the QSAR development of VOC removal in plastic waste recycling
2024, Chemosphere
Deodorization and, in a broader sense, the removal of volatile organic compounds (VOCs) from plastic waste have become increasingly important in the field of plastic recycling, and various new decontamination techniques have been developed. Both in research and industrial practice, the selection of VOCs has been random or unsubstantiated, making it difficult to compare studies and assess decontamination processes objectively. Thus, this study proposes the use of Statistical Molecular Design (SMD) and Quantitative Structure - Activity Relationship (QSAR) as chemometric tools for the selection of representative VOCs, based on physicochemical properties. Various algorithms are used for SMD; hence, several frequently used D-Optimal Onion Design (DOOD) and Space-Filling (SF) algorithms were assessed. Hereby, it was validated that DOOD, by dividing the layers based on the equal-distance approach without so-called ‘Adjacent Layer Bias’, results in the most representative selection of VOCs. QSAR models that describe VOC removal by water-based washing of plastic waste as a function of molecular weight, polarizability, dipole moment and Hansen Solubility Parameters Distance were successfully established. An adjusted-R² value of 0.77 ± 0.09 and a mean absolute error of 24.5 ± 4 % was obtained. Consequently, by measuring a representative selection of VOCs compiled using SMD, the removal of other unanalyzed VOCs was predicted on the basis of the QSAR. Another advantage of the proposed chemometric selection procedure is its flexibility. SMD allows to extend or modify the considered dataset according to the available analytical techniques, and to adjust the considered physicochemical properties according to the intended process.
MemCross: Accelerated Weight Histogram method to assess membrane permeability
2023, Biochimica et Biophysica Acta - Biomembranes
Passive permeation events across biological membranes are determining steps in the pharmacokinetics of xenobiotics. To reach an accurate and rapid prediction of membrane permeation coefficients of drugs is a complex challenge, which can efficiently support drug discovery. Such predictions are indeed highly valuable as they may guide the selection of potential leads with optimum bioavailabilities prior to synthesis. Theoretical models exist to predict these coefficients. Many of them are based on molecular dynamics (MD) simulations, which allow calculation of permeation coefficients through the evaluation of both the potential of mean force (PMF) and the diffusivity profiles. However, these simulations still require intensive computational efforts, and novel methodologies should be developed and benchmarked. Free energy perturbation (FEP) method was recently shown to estimate PMF with a significantly reduced computational cost compared to the adaptive biasing force method. This benchmarking was achieved with small molecules, namely short-chain alcohols. Here, we show that to estimate the PMF of bulkier, drug-like xenobiotics, conformational sampling is a critical issue. To reach a sufficient sampling with FEP calculations requires a relatively long time-scale, which can lower the benefits related to the computational gain. In the present work, the Accelerated Weight Histogram (AWH) method was employed for the first time in all-atom membrane models. The AWH-based protocol, named MemCross, appears affordable to estimate PMF profiles of a series of drug-like xenobiotics, compared to other enhanced sampling methods. The continuous exploration of the crossing pathway by MemCross also allows modeling subdiffusion by computing fractional diffusivity profiles. The method is also versatile as its input parameters are largely insensitive to the molecule properties. It also ensures a detailed description of the molecule orientations along the permeation pathway, picturing all intermolecular interactions at an atomic resolution. Here, MemCross was applied on a series of 12 xenobiotics, including four weak acids, and a coherent structure-activity relationship was established.
In silico model-based exploration of the applicability of parallel artificial membrane permeability assay (PAMPA) to screen chemicals of environmental concern
2022, Environment International
Citation Excerpt :
Fig. 3 shows that after 10.5 h of incubation, less than 15 % of the chemical mass was retained by the membrane for hydrophilic chemicals with log KOW or log DOW lower than 3.5. Such low membrane retention is the case for pharmaceuticals, e.g., desipramine with a membrane retention rate previously measured to be 2 % (Fujikawa et al., 2007) vs predicted by our model to be 1 %. By contrast, more than 90 % of the mass was retained by the membrane for chemicals with log KOW or log DOW greater than 5, which is expected to be the case for many chemicals of environmental concern, e.g., BDE209, DEHP, D5, and DDT (Fig. 3).
Parallel Artificial Membrane Permeability Assay (PAMPA) is an in vitro laboratory method for screening the transmembrane permeability of chemicals. Stemming from medicinal chemistry, PAMPA has the potential for use in the cost-effective high-throughput evaluation of chemicals of environmental concern. However, many chemicals of environmental concern differ substantially from pharmaceuticals in hydrophobicity and volatility. Here, we develop an in silico mass balance model to explore the impacts of chemical properties on chemical mass transfer in PAMPA and PAMPA’s applicability to hydrophobic or volatile chemicals of environmental concern. The model’s performance is evaluated by agreement between predicted and measured permeabilities of 1383 chemicals. The model predicts that the PAMPA measured permeability can be highly uncertain for hydrophobic chemicals because of considerable retention by the artificial membrane and for volatile chemicals because of substantial volatilization to the headspace. Notably, the permeabilities of hydrophobic chemicals are remarkably sensitive to specific experimental conditions, for example, the frequency of stirring and incubation time, challenging the comparison between measurements under different conditions. For hydrophobic chemicals, the PAMPA measured permeability may largely indicate the permeability of the unstirred water layer over the membrane, instead of the “intrinsic” permeability of the membrane, and therefore, may not be of interest for environmental exposure and risk assessments. The model also predicts that the time for mass transfer of highly hydrophobic chemicals to reach the steady state likely exceeds the incubation time, which violates the steady-state assumption used in calculating permeability from measured concentrations. Overall, our theoretical analysis underscores the importance to consider chemical properties when applying the current design of PAMPA to chemicals of environmental concern.
Using in vitro ADME data for lead compound selection: An emphasis on PAMPA pH 5 permeability and oral bioavailability
2022, Bioorganic and Medicinal Chemistry
Membrane permeability plays an important role in oral drug absorption. Caco-2 and Madin-Darby Canine Kidney (MDCK) cell culture systems have been widely used for assessing intestinal permeability. Since most drugs are absorbed passively, Parallel Artificial Membrane Permeability Assay (PAMPA) has gained popularity as a low-cost and high-throughput method in early drug discovery when compared to high-cost, labor intensive cell-based assays. At the National Center for Advancing Translational Sciences (NCATS), PAMPA pH 5 is employed as one of the Tier I absorption, distribution, metabolism, and elimination (ADME) assays. In this study, we have developed a quantitative structure activity relationship (QSAR) model using our ∼6500 compound PAMPA pH 5 permeability dataset. Along with ensemble decision tree-based methods such as Random Forest and eXtreme Gradient Boosting, we employed deep neural network and a graph convolutional neural network to model PAMPA pH 5 permeability. The classification models trained on a balanced training set provided accuracies ranging from 71% to 78% on the external set. Of the four classifiers, the graph convolutional neural network that directly operates on molecular graphs offered the best classification performance. Additionally, an ∼85% correlation was obtained between PAMPA pH 5 permeability and in vivo oral bioavailability in mice and rats. These results suggest that data from this assay (experimental or predicted) can be used to rank-order compounds for preclinical in vivo testing with a high degree of confidence, reducing cost and attrition as well as accelerating the drug discovery process. Additionally, experimental data for 486 compounds (PubChem AID: 1645871) and the best models have been made publicly available (https://opendata.ncats.nih.gov/adme/).
Validating ADME QSAR Models Using Marketed Drugs
2021, SLAS Discovery
Problems with drug ADME are responsible for many clinical failures. By understanding the ADME properties of marketed drugs and modeling how chemical structure contributes to these inherent properties, we can help new projects reduce their risk profiles. Kinetic aqueous solubility, the parallel artificial membrane permeability assay (PAMPA), and rat liver microsomal stability constitute the Tier I ADME assays at the National Center for Advancing Translational Sciences (NCATS). Using recent data generated from in-house lead optimization Tier I studies, we update quantitative structure–activity relationship (QSAR) models for these three endpoints and validate in silico performance against a set of marketed drugs (balanced accuracies range between 71% and 85%). Improved models and experimental datasets are of direct relevance to drug discovery projects and, together with the prediction services that have been made publicly available at the ADME@NCATS web portal (https://opendata.ncats.nih.gov/adme/), provide important tools for the drug discovery community. The results are discussed in light of our previously reported ADME models and state-of-the-art models from scientific literature.
Factors associated with partitioning behavior of persistent organic pollutants in a feto-maternal system: A multiple linear regression approach
2021, Chemosphere
Prenatal exposure to persistent organic pollutants (POPs) has been a matter of particular concern because such exposure can severely affect the health of the fetus. The mechanistic understanding of the partitioning behavior of POPs in the feto-maternal system and the associated factors, however, have rarely been studied. Here, we employed a new approach based on multiple linear regression (MLR) analysis to predict the feto-maternal ratio (FM-ratio) of POPs and to assess the factors associated with feto-maternal partitioning behavior. Two preliminary exploratory MLR models were built using physiological conditions of the participants, and molecular descriptors were calculated with a computational model. The FM-ratio was calculated from the concentrations of polychlorinated biphenyls (PCBs), polychlorinated dibenzo-p-dioxins and furans (PCDD/Fs), and polybrominated diphenyl ethers (PBDEs) in 20 pairs of maternal and cord blood. The models showed that the lipids and cholesterols in the maternal and cord blood and the placenta significantly influence the partitioning of POPs. The body mass index (BMI) change during pregnancy was also related to the FM-ratio. The physicochemical properties associated with lipophilicity and molecular size were also related to the FM-ratio. Even though the results should be interpreted with caution, the preliminary MLR models illustrate that feto-maternal partitioning is governed by transplacental transporting mechanisms, toxicokinetics, and the molecular physicochemical properties of POPs. Overall, the new approach used in this study can improve our understanding of the partitioning behavior in the feto-maternal system.

View all citing articles on Scopus

View full text

QSAR study on permeability of hydrophobic compounds with artificial membranes

Abstract

Graphical abstract

Introduction

Section snippets

Permeability with artificial membranes

Permeability with artificial membranes of hydrophobic compounds

Conclusions

Materials

Acknowledgments

Eur. J. Med. Chem.

Bioorg. Med. Chem.

Bioorg. Med. Chem.

Bioorg. Med. Chem.

J. Biomol. Screen

Int. J. Phram.

Int. J. Pharm.

Eur. J. Med. Chem.

Eur. J. Pharm. Sci.

Eur. J. Pharm. Sci.

J. Pharm. Sci.

Eur. J. Pharm. Sci.

Eur. J. Pharm. Sci.

J. Med. Chem.

Occup. Environ. Med.

Curr. Topics Med. Chem.

J. Med. Chem.

J. Med. Chem.