Immobilized Artificial Membrane HPLC Derived Parameters vs PAMPA

Jul 5, 2016 - IAM and those expected for isolipophilic neutral compounds ... IAM at describing the potential of passage through the BBB as compared to...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/molecularpharmaceutics

Immobilized Artificial Membrane HPLC Derived Parameters vs PAMPA-BBB Data in Estimating in Situ Measured Blood−Brain Barrier Permeation of Drugs Lucia Grumetto, Giacomo Russo, and Francesco Barbato*

Mol. Pharmaceutics 2016.13:2808-2816. Downloaded from pubs.acs.org by DURHAM UNIV on 08/07/18. For personal use only.

Dipartimento di Farmacia, Università degli Studi di Napoli Federico II, Via D. Montesano, 49, I-80131 Naples, Italy ABSTRACT: The affinity indexes for phospholipids (log kWIAM) for 42 compounds were measured by high performance liquid chromatography (HPLC) on two different phospholipid-based stationary phases (immobilized artificial membrane, IAM), i.e., IAM.PC.MG and IAM.PC.DD2. The polar/ electrostatic interaction forces between analytes and membrane phospholipids (Δlog kWIAM) were calculated as the differences between the experimental values of log kWIAM and those expected for isolipophilic neutral compounds having polar surface area (PSA) = 0. The values of passage through a porcine brain lipid extract (PBLE) artificial membrane for 36 out of the 42 compounds considered, measured by the so-called PAMPA-BBB technique, were taken from the literature (P0PAMPA‑BBB). The values of blood−brain barrier (BBB) passage measured in situ, P0in situ, for 38 out of the 42 compounds considered, taken from the literature, represented the permeability of the neutral forms on “efflux minimized” rodent models. The present work was aimed at verifying the soundness of Δlog kWIAM at describing the potential of passage through the BBB as compared to data achieved by the PAMPA-BBB technique. In a first instance, the values of log P0PAMPA‑BBB (32 data points) were found significantly related to the n-octanol lipophilicity values of the neutral forms (log PN) (r2 = 0.782) whereas no significant relationship (r2 = 0.246) was found with lipophilicity values of the mixtures of ionized and neutral forms existing at the experimental pH 7.4 (log D7.4) as well as with either log kWIAM or Δlog kWIAM values. log P0PAMPA‑BBB related moderately to log P0in situ values (r2 = 0.604). The latter did not relate with either n-octanol lipophilicity indexes (log PN and log D7.4) or phospholipid affinity indexes (log kWIAM). In contrast, significant inverse linear relationships were observed between log P0in situ (38 data points) and Δlog kWIAM values for all the compounds but ibuprofen and chlorpromazine, which behaved as moderate outliers (r2 = 0.656 and r2 = 0.757 for values achieved on IAM.PC.MG and IAM.PC.DD2, respectively). Since log P0in situ refer to the “intrinsic permeability” of the analytes regardless their ionization degree, no correction for ionization of Δlog kWIAM values was needed. Furthermore, log P0in situ were found roughly linearly related to log BB values (i.e., the logarithm of the ratio brain concentration/blood concentration measured in vivo) for all the analytes but those predominantly present at the experimental pH 7.4 as anions. These results suggest that, at least for the data set considered, Δlog kWIAM parameters are more effective than log P0PAMPA‑BBB at predicting log P0in situ values for all the analytes. Furthermore, ionization appears to affect differently, and much more markedly, BBB passage of acids (yielding anions) than that of the other ionizable compounds. KEYWORDS: immobilized artificial membrane, blood−brain barrier, parallel artificial membrane permeability assay, in situ brain perfusion, quantitative structure−properties relationships Although recently debated,3 one of the most used measures reflecting in vivo BBB penetration capability of drugs is log BB, which is defined as

1. INTRODUCTION The blood−brain barrier (BBB) is one of the most complex and extensively studied biological barriers intended to protect the mammalian brain integrity against possible injurious substances. It is made of endothelial cells, narrowly adherent to one other such as forming tight junctions, which restrict the passage of solutes.1,2 Drug transport is strongly limited by this peculiar biological structure to pure transcellular diffusion, the paracellular route, i.e., the passage of actives through the gaps between the endothelial cells, being completely hindered. Passive transcellular diffusion takes place for the majority of drugs, but active transport may also occasionally occur affecting their BBB passage. © 2016 American Chemical Society

log BB = log

C brain C blood

in which Cbrain is the concentration of the analyte in the brain tissues, and Cblood is the concentration of the analyte in the Received: Revised: Accepted: Published: 2808

May 4, 2016 June 28, 2016 July 5, 2016 July 5, 2016 DOI: 10.1021/acs.molpharmaceut.6b00397 Mol. Pharmaceutics 2016, 13, 2808−2816

Article

Molecular Pharmaceutics

Table 1. pKa Values and Logarithms of Lipophilicity Values in n-Octanol and of Chromatographic Retention Factors on IAM Stationary Phases for the Compounds Considered seriesa 2 1, 1, 1, 1, 1, 2 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 1 1, 1 1, 1, 1, 2 1, 1, 1, 1, 2 1, 1, 1, 1, 1, 1, 2 1 1, 2 1,

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

2 2 2 2 2 2 2 2 2 2 2 2 2 2

2 2

compound aminopyrine amitriptyline antipyrine buspirone caffeine carbamazepine chlorambucil chlorpromazine cimetidine codeine diazepam diltiazem diphenydramine domperidone doxorubicin flurbiprofen fluoxetine fluphenazine fluvastatin haloperidol hydrocortisone hydroxyzine ibuprofen indomethacin lidocaine loratadine methadone metoclopramide morphine naproxen nicotinamide phenytoin progesterone propranolol pyrilamine quinine risperidone temazepam teobromine theophylline thiourea verapamil

pKa 4.50 9.49 0.65 7.57 0.52 4.82/4.62 9.41 6.80 7.99 3.40 8.94 8.76 9.00 8.00/9.93 4.18 9.96 7.84 4.56 8.29 6.62 4.41 3.96 7.90 4.27 9.05 9.08 9.48/9.25 4.15 3.54 8.28 9.50 9.12 8.57 7.81 11.66/1.58 9.90 8.60 8.90

log PN 15

1.00 4.9215 0.5616 2.6315 0.0715 2.1917 3.4116 5.1918 0.4015 1.3919 2.9920 3.4118 3.1816 3.9015 1.9716 4.1621 4.5016 4.3615 4.3022 4.3015 1.6115 3.5516 4.1316 4.2715 2.4823 4.8027 3.9315 2.7216 0.8916 3.2416 0.3715 2.4715 3.8718 3.2824 3.2715 3.4415 3.0425 2.1915 0.7815 0.0215 1.0826 3.7915

log D7.4 0.99 2.8016 0.56 2.3916 −0.07 2.19 0.6116 2.89 0.19 0.2216 2.99 2.02 1.80 2.29 −0.3316 0.91 2.2816 4.33 1.1716 3.57 1.61 3.48 1.4416 0.68 1.53 4.80 2.26 1.03 −0.0728 0.0916 −0.37 2.42 3.87 0.48 1.80 2.19 1.66 2.19 −0.78 −0.04 −1.08 1.88

log kWIAM.MG 29

0.536 2.8817 0.5996 1.742 0.12829 1.0397 1.2887 1.7995 0.6335 0.8557 1.73129 2.12129 2.21929 2.7907 2.223 1.870 3.181 3.5887 2.21030 2.6705 1.550 2.9087 0.9726 2.39021 1.112 3.354 2.646 1.199 0.7675 1.26021 0.351 1.7876 2.7696 1.8215 2.1097 2.313 2.1897 2.190 −0.1566 −0.13029 −0.817 2.89229

log kWIAM.DD2 29

0.573 3.1227 0.39329 1.986 0.11629 1.7177 1.8977 2.2255 1.0485 1.2907 2.1986 2.78029 2.17029 3.2137 1.764 1.950 3.522 3.9577 2.84330 2.78030 1.660 2.9657 1.1706 2.0806 1.650 3.623 2.828 1.902 1.1805 1.33930 −0.179 1.7896 3.3176 2.4805 1.8937 2.810 2.0287 1.69731 −0.0886 0.10029 −1.081 3.08530

chemical characterb

supplier

B B B B B N B/A B B B B B B B B/A A B B A B N B A A B B B B B/A A B A N B B B B B/A A A N B

Sigma Sigma Sigma Sigma Sigma Sigma Sigma Sigma Sigma Sigma Sigma TCI TCI TCI TCI ACROS TCI Sigma Acros TCI TCI Sigma ACROS ACROS ACROS TCI Sigma TCI Sigma ACROS ACROS ACROS ACROS ACROS TCI TCI TCI Sigma ACROS ACROS ACROS ACROS

a

Series 1: compounds for relationships with logP0PAMPA‑BBB values. Series 2: compounds for relationships with log P0in situ values. bA = acid; B = base; N = neutral; B/A = ampholyte. For the ampholytes the two pKa values reported refer to the acidic and basic functions, respectively.

drug candidates at the early stages of pharmaceutical development. In our previous studies we found that BBB permeation, expressed as log BB values, was inversely related, by highly significant inverse linear relationships, to the polar/electrostatic interaction forces occurring between drugs and membrane phospholipids. These interactions were parametrized by Δlog kWIAM and Δ′log kWIAM values, i.e., the differences between experimental affinity values for phospholipids and the expected affinity values based on n-octanol lipophilicity. The experimental affinity values for phospholipids (log kWIAM) were measured by a HPLC method on phospholipid-based stationary phases (IAM-HPLC). The expected phospholipid affinity values were calculated taking into account either the

blood. However, the setting up of alternative in vitro methods aimed at appreciating BBB penetration potential of drugs is more and more desirable, as log BB determination is timeconsuming and may raise ethical issues. Several efforts have been devoted to the development of methods based on the employment of cultured cell lines. However, astrocyte cell cultures4 are often difficult to grow, and this method is difficult to standardize, preventing the comparison of data determined in different laboratories. Therefore, the methods based on physicochemical properties are increasingly attractive. Although they represent a simplification of the more complex phenomena occurring in vivo, they can offer a rationale on the mechanisms driving barrier passage, useful in the structural optimization of new 2809

DOI: 10.1021/acs.molpharmaceut.6b00397 Mol. Pharmaceutics 2016, 13, 2808−2816

Article

Molecular Pharmaceutics lipophilicity of the neutral form of the analytes (log PN) or that of the mixture of neutral and ionized forms existing at the physiological pH 7.4 (log D7.4), yielding Δlog kWIAM and Δ′log kWIAM values, respectively.5−7 However, the proposed model was based on a relatively small number of analytes, and the study of possible relationships between “delta” values and data of BBB passage different from log BB would allow a further validation of the proposed model and a deeper understanding of the molecular mechanisms behind BBB uptake. Recently, the relationships between in situ measured BBB passive permeation data and in vitro passage data, achieved by the so-called “PAMPA-BBB” technique, have been reported in the literature.8 In that study the in situ BBB passive permeability was expressed as the values of P0in situ, i.e., in situ brain perfusion permeability values (on rat and mouse) selected from studies which used some sort of carrier-mediated transport inhibition (e.g., GF120918, PSC833, cyclosporin A, self-inhibition at high concentrations, mdr1a(−/−)/mrp1(−/ −)/brcp knockout rodent model), allowing that the in situ data can be assumed as free of efflux effects. It is important to underline that P0in situ values refer to the permeability of the neutral form of the analytes and represent the “intrinsic permeability” regardless of any effect given by ionization. The in vitro data were achieved by the PAMPA-BBB technique and expressed as P0PAMPA‑BBB values. PAMPA (parallel artificial membrane permeability assay) is a method based on the determination of the permeability of substances from a donor compartment, through a lipid-infused artificial membrane, into an acceptor compartment. Although the first described PAMPA-BBB technique employed a PAMPA membrane made from 2% w/v lecithin dissolved in dodecane,9 the data considered in the study were achieved by an improved model consisting of a new PAMPA-BBB formulation based on 10% w/v porcine brain lipid extract (PBLE), using a 5-fold higher lipid concentration in a more viscous alkane solvent than dodecane and with thinner membranes.8 The relationships between log P0in situ and log P0PAMPA‑BBB for 197 compounds became significant only after the classification of the analytes into four predominant-charge groups (positive, negative, neutral, and zwitterionic) and after, for all groups, an Abraham solvation descriptor10 was added as a second term in the equations. In the present study 36 P0PAMPA‑BBB and 38 P0in situ values were selected from the literature.8,11 The aim was to investigate their possible relationships with various physicochemical parameters, including log PN, log D7.4, log kWIAM, Δlog kWIAM, and Δ′log kWIAM. Finally, the effectiveness of the various physicochemical parameters at predicting P0in situ values was compared to that of P0PAMPA‑BBB values. Indeed, the PAMPA method, especially in its variation PAMPA-BBB, is regarded as one of the in vitro techniques mirroring rather closely the BBB passage taking place in vivo.12 However, to the best of our knowledge its validation in terms of log BB predictivity was performed only on limited sets of compounds,12,13 and the effectiveness of IAM derived indexes in surrogating PAMPABBB data has never been investigated.

2.1. Chromatographic System. The system consisted of the following: LC-10AD liquid chromatographic apparatus (Shimadzu Corporation, Kyoto, Japan); SPD-10AV UV detector (Shimadzu), set at λ of maximum absorbance for each compound; 7725 Rheodyne injection valve (fitted with a 20 μL loop). For data processing, Cromatoplus software for personal computer (Shimadzu) was used. The following analytical HPLC columns were used: • IAM.PC.MG (4.6 mm × 150 mm; 12 μm, 300 Å; Regis Chemical Company, Morton Grove, IL) • IAM.PC.DD2 (4.6 mm × 100 mm, 10 μm, 300 Å; Regis Chemical Company, Morton Grove, IL) 2.2. Chromatographic Conditions. The analyses were performed at room temperature with 0.1 M phosphate buffer at pH 7.0 in mixture with acetonitrile at various percentages. The flow rate was selected according to the retention time of each analyte (1.0, 2.0, and 3.0 mL/min). For sample preparation, each analyte was dissolved in the mobile phase or in methanol to ca. 10−4 M concentration. Chromatographic retention data are reported as log k (the logarithm of the retention factor), calculated by the expression log k = log[(tr − t0)/t0] where tr and t0 are the retention times of the drug and a nonretained compound (acetone), respectively. Direct measurements of log k values in fully aqueous mobile phases (log kWIAM) were only possible for the compounds eluting within 20 min, whereas for the solutes requiring the addition of acetonitrile to the eluent, the log kWIAM values were calculated by an extrapolation method:14 log k values were determined at four different acetonitrile percentages (φ) in the mobile phases (from 10 to 30% v/v), and the intercept values of the linear relationships between log k and φ values, found for all compounds in the range of eluent composition examined (r2 ≥ 0.99), were assumed as log kWIAM values. In particular, for basic and neutral analytes having log PN values higher than 2.5, the analyses were carried out employing 30, 25, 20, and 15% v/v acetonitrile percentages (φ). For highly lipophilic (log PN ≥ 2.5) acidic compounds and for all the other analytes, whose log PN values span the range 2.5−1.0 (2.5 > log PN ≥ 1.0) the employed acetonitrile percentages (φ) were 25, 20, 15, and 10% v/v; fully aqueous mobile phase (φ = 0) was used for analyzing the most polar compounds (log PN < 1.0). All reported log k values are the average of at least three measurements; for each log k value the 95% confidence interval associated with each value never exceeded 0.04. To avoid that the experimental measurements were affected by retention changes due to column aging, the retention times of five test compounds (amlodipine, p-nitroaniline, toluene, isradipine, and ketoprofen) were checked weekly. No correction was done to the experimental retention values since no retention value of test compounds changed more than 4% during the study. 2.3. Lipophilic Parameters. log PN values, i.e., partition coefficients n-octanol/aqueous phase of the neutral form of analytes, were either from the literature15−26 or calculated by the program ALOGPS version 2.1.27 The n-octanol/aqueous buffer at pH 7.4 distribution coefficients (log D7.4) were taken from the literature16,28 or calculated according to the following equations:

2. MATERIALS AND METHODS All samples were obtained from Sigma-Aldrich (Milan, Italy), TCI (Milan, Italy), and Acros-Organics (Rodano, Italy) as reported in Table 1. All chemicals were of HPLC grade and used without further purification. 2810

log D7.4 = log P − log(1 + 107.4 − pK a)

(for acids)

log D7.4 = log P − log(1 + 10 pK a − 7.4)

(for bases)

DOI: 10.1021/acs.molpharmaceut.6b00397 Mol. Pharmaceutics 2016, 13, 2808−2816

Article

Molecular Pharmaceutics Scheme 1. Chemical Structures of the Compounds Considered

2811

DOI: 10.1021/acs.molpharmaceut.6b00397 Mol. Pharmaceutics 2016, 13, 2808−2816

Article

Molecular Pharmaceutics

Table 2. Values of the Differences between Observed and Expected Logarithms of Retention Factors on IAM.PC.MG and IAM.PC.DD2 Stationary Phases (Δlog kWIAM.MG, Δ′log kWIAM.MG, Δlog kWIAM.DD2, and Δ′log kWIAM.DD2, Respectively) of Logarithms of PAMPA-BBB Permeation Data (log P0PAMPA‑BBB), of in Situ BBB Permeation Data (log P0in situ), and of log BB

a

compound

Δlog kWIAM.MG

Δlog kWIAM.DD2

Δ′log kWIAM.MG

Δ′log kWIAM.DD2

aminopyrine amitriptyline antipyrine buspirone caffeine carbamazepine chlorambucil chlorpromazine cimetidine codeine diazepam diltiazem diphenydramine domperidone doxorubicin flurbiprofen fluoxetine fluphenazine fluvastatin haloperidol hydrocortisone hydroxyzine ibuprofen indomethacin lidocaine loratadine methadone metoclopramide morphine naproxen nicotinamide phenytoin progesterone propranolol pyrilamine quinine risperidone temazepam theobromine theophylline thiourea verapamil

0.658 −0.345 1.097 0.472 1.164 0.145 −0.648 −1.657 1.267 0.644 0.154 0.185 0.479 0.435 1.517 −0.707 0.314 0.841 −0.486 −0.026 1.151 0.852 −1.579 −0.281 −0.030 0.231 0.266 −0.148 0.983 −0.531 1.643 0.654 0.440 −0.004 0.292 0.351 0.569 1.296 1.486 0.863 1.081 0.631

0.845 −0.679 1.122 0.564 1.500 0.753 −0.335 −1.856 1.943 1.157 0.402 0.548 0.177 0.472 1.028 −1.061 0.158 0.738 −0.314 −0.377 1.298 0.588 −1.810 −1.046 0.384 −0.053 0.056 0.387 1.566 −0.716 1.516 0.534 0.607 0.383 −0.194 0.547 0.180 0.733 2.033 1.432 1.352 0.458

0.667 1.466 1.097 0.677 1.164 0.145 1.743 0.307 1.447 1.643 0.154 1.372 1.658 1.810 3.481 2.069 2.210 0.866 2.187 0.597 1.151 0.912 0.718 2.785 0.781 0.231 1.692 1.295 1.803 2.159 1.643 0.696 0.440 2.387 1.548 1.419 1.747 1.296 1.486 0.880 1.081 2.262

0.855 1.524 1.122 0.814 1.500 0.753 2.574 0.533 2.162 2.372 0.402 1.992 1.611 2.145 3.418 2.316 2.464 0.769 2.938 0.382 1.298 0.660 0.985 2.684 1.371 −0.053 1.791 2.142 2.564 2.556 1.516 0.586 0.607 3.292 1.334 1.846 1.614 0.733 2.033 1.453 1.352 2.443

log P0PAMPA‑BBB −1.27 −6.14 −3.85 −5.92 −4.54 −1.46 −6.40 −3.68 −3.83 −3.18 −2.64 −3.36 −4.23 −2.35 −1.39 −2.36 −3.56 −2.06 −5.17 −3.72 −2.64 −2.67 −3.65 −2.18 −1.11 −4.47 −2.63 −4.34 −3.58 −1.93 −2.63 −2.99 −4.00 −8.00 −6.41 −2.03

log P0in situ

log BB

−3.30 −1.48 −3.98 −2.53 −3.85 −3.74 −0.80 −1.33 −5.92 −3.80 −3.35 −2.81 −1.90 −4.45 −5.55 −0.58 −1.11 −3.35

1.30a −0.10 0.40a −0.06 0.00 −1.70 1.06a −1.42 0.55 0.52 0.30a 0.70a −0.80 −0.83 0.30a 0.50 1.51

−5.85

−0.90a

−1.22 −1.06 −3.24 −3.48 −2.02 −2.86 −5.43 −0.77 −4.88 −4.09 −3.74 −1.26 −2.04 −3.45 −2.94 −3.35

−0.18 −1.26 0.10a

−5.09 −5.45 −2.19

−0.29

0.90a −0.70a −0.16 0.10a −0.14 0.20a 0.64a 0.49 0.60a −0.02

−0.70

Value calculated.8

2.4. In Vitro and in Situ Data. P0PAMPA‑BBB and P0in situ values were taken from the literature.8,11 For the selection of in situ perfusion log P0BBB values, the following criteria were applied. If log P0BBB data on different animal models (rat, mouse, dog, cat, and guinea pig) were available, the data achieved on rat were preferred as they were more numerous. For verapamil and diltiazem, for which log P0BBB data were not available on rat, data achieved on mouse were used instead. If data consisting of more than one value achieved on rat were available, an average of the values was performed. For actively transported analytes, measures of BBB passive diffusion were selected by (i) eliminating the values referred to wild-type animal models; (ii) selecting the measures achieved either after the pharmacological inhibition of the active transport

mechanisms obtained by administering cyclosporin A (Diphenhydramine, Pyrilamine and Loratadine) or determined on mdr1a(−/−)/mrp1(−/−)/brcp knockout rat model; (iii) preferring the measures achieved on nonsaturable models as transport saturation is generally referred to some sort of active transport mechanism involvement. Since for indomethacin three log P0BBB values, based on data measured at different pH values (5.5, 6.5, and 7.4), were available, the one extrapolated from measures at pH 7.4, i.e., the closest to the physiological and experimental pH, was selected. 2.5. Statistical Analysis. Linear regression analysis was performed by a commercially available statistical package for personal computer observing the requirements of significant regression analysis. 2812

DOI: 10.1021/acs.molpharmaceut.6b00397 Mol. Pharmaceutics 2016, 13, 2808−2816

Article

Molecular Pharmaceutics

3.1. Relationships between log P0PAMPA‑BBB and Physicochemical Parameters. A significant linear relationship was found between log P0PAMPA‑BBB and log PN values (Figure 1 and eq 3).

3. RESULTS AND DISCUSSION log P0PAMPA‑BBB values were taken into account for a set of 36 analytes consisting of 23 bases, 8 acids, 2 ampholytes, and 3 neutral compounds (series 1). log P0in situ values were taken into account for a set of 38 analytes consisting of 24 bases, 6 acids, 4 ampholytes, and 4 neutral compounds (series 2). Table 1 summarizes the values of pKa, log PN, log D7.4, the logarithms of the chromatographic retention coefficients measured on IAM.PC.MG (log kWIAM.MG) and IAM.PC.DD2 (log kWIAM.DD2) stationary phases, and the suppliers of the substances for all the analytes. As can be seen, 32 compounds belong to both groups. The chemical structures of the analytes taken into account are reported in Scheme 1. As it is evident from Table 1, the analytes taken into account span a remarkably wide range of log PN values (0.02−5.19), covering more than five log PN units. The selection of the analytes was performed with the aim of covering and mirroring as faithfully as possible the chemical diversity of the marketed drugs, and both CNS active (e.g., buspirone, chlopromazine, and diazepam) and CNS inactive drugs (e.g., cimetidine and thiourea) were taken into account. It is interesting to note that log kwIAM.DD2 values cover a wider range than log kwIAM.MG values (5.0 vs 4.4 units). This evidence would suggest a higher selectivity of IAM.PC.DD2 in comparison with IAM.PC.MG column. Table 2 summarizes the values of Δlog kWIAM.MG, Δlog kWIAM.DD2, Δ′log kWIAM.MG, and Δ′log kWIAM.DD2 values, as well as the values of log P0PAMPA‑BBB, log P0in situ, and log BB. For the set of compounds taken into account in the present work, the chromatographic retention data achieved on the two different IAM columns, which support the same phosphatidylcholine analogues but differ from each other in the endcapping, were found strongly collinear (r2 = 0.914; F1,40 = 426.71). This, as already previously verified for other sets of compounds,5−7,29,30,32 once again suggests that IAM retention is based on the interactions with phospholipids with secondary interaction mechanisms playing an only minor role. Indeed, using two different phospholipid based stationary phases was aimed at confirming that the experimental retention data were not noticeably affected by any secondary retention mechanism. This can be reasonably assumed if the retention scales obtained on the two columns are collinear. The calculation of Δlog kWIAM and Δ′log kWIAM was based on the evidence that log kWIAM values of structurally nonrelated neutral compounds having polar surface area (PSA) = 0 relate unambiguously with n-octanol lipophilicity values according to the following equations:29 log k W

IAM.MG

log P0 PAMPA − BBB = 0.939(± 0.085) log P N − 6.210(± 0.276) n = 36 r 2 = 0.782 s = 0.765 F1,34 = 121.63 F1,34 α , 0.001 = 12.90

(3)

Figure 1. Relationship between log P0PAMPA‑BBB and log PN values.

No relationship was found with log D7.4 values (r2 = 0.246, F1,34 = 11.14). The relationships with phospholipid affinity data, log kWIAM.MG and log kWIAM.DD2 (eq 4 and eq 5), as well as those with Δlog kWIAM.MG and Δlog kWIAM.DD2 (eq 6 and eq 7), were poorer. Furthermore, no improvement of the relationships was observed by taking into account Δ′log kWIAM.MG and Δ′log kWIAM.DD2 for acids (data not shown). log P0 PAMPA − BBB = 1.193( ±0.210) log k W IAM.MG − 5.654( ±0.425) n = 36 r 2 = 0.487 s = 1.172 F1,34 = 32.29 F1,34 α , 0.001 = 12.90

(4) log P0

− 5.936(± 0.433)

n = 36 r = 0.531 s = 1.120 F1,34 = 38.56 F1,34 α , 0.001 = 12.90

(5) log P0

PAMPA − BBB

= −1.460(± 0.282) Δlog k W

IAM.MG

− 3.045(± 0.223)

2

n = 36 r = 0.441 s = 1.223 F1,34 = 26.86 F1,34 α , 0.001 = 12.90

(6) log P0 PAMPA − BBB = −1.283(± 0.204) Δlog k W IAM.DD2 − 3.048(± 0.200)

= 0.854(± 0.047) log P − 0.976(± 0.156)

n = 36 r 2 = 0.536 s = 1.114 F1,34 = 39.40 F1,34 α , 0.001 = 12.90

n = 17 r = 0.957 s = 0.214 F1,15 = 331.35 F1,15 α , 0.001 = 16.59

(1)

(7)

The above-reported relationships suggest that, at least for the set of compounds considered, the data of passage measured by PAMPA-BBB substantially reflect n-octanol/water partition coefficients of the neutral forms of the analytes, log PN. Since PAMPA barrier consists of porcine brain lipid extract (PBLE), which is better mirrored by phospholipids than n-octanol, it could appear surprising that phospholipophilicity indexes (log kWIAM) are ineffective at describing the passage through them. However, it was already reported in the literature5−7 that the passage through phospholipids is not described by log kWIAM values, i.e., the affinity indexes for them, but by “delta” values accounting for polar/electrostatic interactions. On the other

log k W IAM.DD2 = 1.039(± 0.051) log P − 1.311(± 0.169) n = 17 r 2 = 0.965 s = 0.232 F1,15 = 417.54 F1,15 α , 0.001 = 16.59

(2)

The values Δ′log Δlog kW , and Δ′log determined according to the procedure described in our previous work.29 They are the differences between the log kWIAM values observed experimentally and those calculated by eqs 1 and 2 taking into account either log PN, to obtain Δlog kWIAM, or log D7.4, to obtain Δ′log kWIAM values. kWIAM.MG,

= 1.192(± 0.192) log k W

IAM.DD2

2

2

of Δlog kWIAM.MG, kWIAM.DD2 were

PAMPA − BBB

IAM.DD2

2813

DOI: 10.1021/acs.molpharmaceut.6b00397 Mol. Pharmaceutics 2016, 13, 2808−2816

Article

Molecular Pharmaceutics hand, the fact that even “delta” values poorly relate with passage data through PBLE may indicate that the latter involve different phenomena with respect to in vivo BBB passage. 3.2. Relationships between log P0in situ, log P0PAMPA‑BBB, and Physicochemical Parameters. P0in situ are “efflux minimized” in situ brain perfusion permeability values of the neutral forms of the analytes. log P0in situ values for 197 analytes were demonstrated to relate to log P0PAMPA‑BBB values8 but only after they were divided into four predominant-charge groups (at pH 7.4), i.e., bases, acids, neutral, and zwitterions. Furthermore, the achievement of significant relationships required the inclusion in the equations of at least one Abraham linear free energy relation (LFER) solvation descriptor,10 i.e., α (H-bond acidity) and/or β (H-bond basicity). The best relation equations between log P0in situ and log PAMPA‑BBB found by the authors8 included the term α, for the P0 bases, and the term (α + β), for acids and neutral compounds. For zwitterions, log P0in situ values related to the sole term (α − β) and no dependence on log P0PAMPA‑BBB values was found so that the authors concluded that “ampholytes seem to be in a class by themselves, minimally dependent on lipophilicity”. In the present work 38 P0in situ values were taken from the literature,11 and for 32 of such analytes the values of P0PAMPA‑BBB were also reported.8 The log P0in situ values for these compounds moderately related linearly to log P0PAMPA‑BBB values (Figure 2 and eq 8).

Figure 3. Relationship between log P0in situ and log PN values.

line generated by the other analytes (Figure 4 and eqs 10 and 11). Ibuprofen and chlorpromazine are the analytes having the

log P0 in situ = 0.809(± 0.120) log P0 PAMPA − BBB − 0.274(± 0.443) n = 32 r 2 = 0.604 s = 0.989 F1,30 = 45.74 F1,30 α , 0.001 = 13.29

(8)

Figure 4. Relationships between log P0in situ and Δlog kWIAM values. Figure 2. Relationship between log P0in situ and log P0PAMPA‑BBB values.

P0PAMPA‑BBB

lowest Δlog kWIAM values; although their behavior may suggest a parabolic dependence of log P0in situ on Δlog kWIAM, this hypothesis should be verified on more than two data points. However, it is interesting to note that chlorpromazine behaved similarly in a previous work reporting the relationships between Δlog kWIAM and log BB values.5

N

Although log relate quite well to log P values, the latter were less effective at describing in situ permeability (Figure 3 and eq 9) whereas the relationships between log P0in situ and either log kWIAM.MG or log kWIAM.DD2 values were not significant (data not shown). log P0 in situ = 0.692(± 0.112) log P N − 4.970(± 0.347)

log P0 in situ = −2.077(± 0.258) Δlog k W IAM.MG − 2.225(± 0.195)

n = 38 r 2 = 0.517 s = 1.089 F1,36 = 38.57 F1,36 α , 0.001 = 12.61

n = 36 r 2 = 0.656 s = 0.907 F1,34 = 53.34 F1,34 α , 0.001 = 12.90

(10)

(9) 7.4

As expected, no relationship was observed with log D values, P0in situ permeability values being referred to the neutral forms of the analytes (data not shown). In contrast, significant linear relationships were found between log P0in situ and both Δlog kWIAM.MG and Δlog kWIAM.DD2, but only after the exclusion of ibuprofen and chlorpromazine, which deviated from the imaginary regression

log P0

in situ

= −1.818(± 0.176) Δlog k W

IAM.DD2

− 2.266(± 0.157)

2

n = 36 r = 0.757 s = 0.762 F1,34 = 105.98 F1,34 α , 0.001 = 12.90

(11)

By replacing Δlog kWIAM values with Δ′log kWIAM values, either for all the analytes or for acids only, the relationships with log P0in situ were not significant (data not shown). 2814

DOI: 10.1021/acs.molpharmaceut.6b00397 Mol. Pharmaceutics 2016, 13, 2808−2816

Article

Molecular Pharmaceutics These results confirm the soundness of Δlog kWIAM to predict BBB passage, in spite of the fact that Δlog kWIAM values, and not Δ′log kWIAM, must be used for also acidic analytes. Indeed, this may appear in contrast with the results of our previous studies, in which log BB, assumed as a measure of BBB passage, related to Δ′log kWIAM of acidic analytes and not with their Δlog kWIAM values.5−7 However, log P0in situ values express the “intrinsic permeability” of the analytes, regardless of their ionization degree, and these values would seem to greatly overestimate the actual capability to cross the BBB of the acids that are extensively ionized at the physiological pH 7.4. To support this hypothesis, we took into account the log BB values of 34 analytes8 among those considered in the present study. It is important to underline that they should be regarded as rough estimates of the actual values, being from different sources and including 14 out of 34 values calculated in silico. Nevertheless, as can be seen in Figure 5, log BB relate linearly

compounds, too. However, this is not in contrast with the results of our previous studies6,7 in which the BBB passage of acidic compounds was better described by Δ′log kWIAM. Indeed, in those studies BBB passage was expressed by log BB values, accounting for the ionization degree of the acidic analytes, whereas in the present study logP0in situ values were considered, which account for the permeation potential of the neutral forms regardless of the relative abundance of the species existing at the experimental pH. Although our previous studies demonstrated the existence of an inverse linear dependence between membrane passage data and “delta” parameters, this study sheds new light on the different role played by ionization in the BBB passage of drugs. In fact, for bases and neutral compounds, log P0in situ appear as just scaled up log BB values, since a direct linear relationship, albeit weak, can be observed between them. In contrast, for compounds present predominantly as anions at the experimental pH, such relationship is no longer observed, their log BB being related not to log P0in situ but to log Pcin situ, i.e., BBB permeation values corrected for ionization. These results confirm that ionization affects more markedly BBB passage of acids than that of the other analytes.

■

AUTHOR INFORMATION

Corresponding Author

*Tel: +39 081 678627. Fax: +39 081 678107. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) Van Bree, J. B.; De Boer, A. G.; Danhof, M.; Breimer, D. D. Drug Transport across the Blood–Brain Barrier. I. Anatomical and Physiological Aspects. Pharm. Weekbl. 1992, 14 (5), 305−310. (2) Keaney, J.; Campbell, M. The Dynamic Blood-Brain Barrier. FEBS J. 2015, 282 (21), 4067−4079. (3) Bickel, U. How to Measure Drug Transport across the BloodBrain Barrier. NeuroRx 2005, 2 (1), 15−26. (4) Lundquist, S.; Renftel, M.; Brillault, J.; Fenart, L.; Cecchelli, R.; Dehouck, M.-P. Prediction of Drug Transport through the BloodBrain Barrier in Vivo: A Comparison between Two in Vitro Cell Models. Pharm. Res. 2002, 19 (7), 976−981. (5) Grumetto, L.; Carpentiero, C.; Barbato, F. Lipophilic and Electrostatic Forces Encoded in IAM-HPLC Indexes of Basic Drugs: Their Role in Membrane Partition and Their Relationships with BBB Passage Data. Eur. J. Pharm. Sci. 2012, 45 (5), 685−692. (6) Grumetto, L.; Carpentiero, C.; Di Vaio, P.; Frecentese, F.; Barbato, F. Lipophilic and Polar Interaction Forces between Acidic Drugs and Membrane Phospholipids Encoded in IAM-HPLC Indexes: Their Role in Membrane Partition and Relationships with BBB Permeation Data. J. Pharm. Biomed. Anal. 2013, 75, 165−172. (7) Grumetto, L.; Russo, G.; Barbato, F. Indexes of Polar Interactions between Ionizable Drugs and Membrane Phospholipids Measured by IAM-HPLC: Their Relationships with Data of Blood-Brain Barrier Passage. Eur. J. Pharm. Sci. 2014, 65, 139−146. (8) Tsinman, O.; Tsinman, K.; Sun, N.; Avdeef, A. Physicochemical Selectivity of the BBB Microenvironment Governing Passive Diffusion - Matching with a Porcine Brain Lipid Extract Artificial Membrane Permeability Model. Pharm. Res. 2011, 28 (2), 337−363. (9) Di, L.; Kerns, E. H.; Fan, K.; McConnell, O. J.; Carter, G. T. High Throughput Artificial Membrane Permeability Assay for Blood-Brain Barrier. Eur. J. Med. Chem. 2003, 38 (3), 223−232. (10) Abraham, M. H. Scales of Solute Hydrogen-Bonding - Their Construction and Application to Physicochemical and Biochemical Processes. Chem. Soc. Rev. 1993, 22 (2), 73−83.

in situ

Figure 5. Relationship between log BB and logP0 values. For acids both log P0in situ (pink ■) and log Pcin situ (pink ×) values are reported.

with log P0in situ values for all the compounds, including the negligibly ionized acids phenytoin and theophylline, with the exception of the compounds which are present predominantly as anions at pH 7.4 (flurbiprofen, indomethacin, naproxen, ibuprofen, chlorambucil). These compounds strongly deviate from the regression line generated by the other analytes (bases, neutral, ampholytes, and negligibly ionized acids) and show a trend similar to that of the latter only if log P0in situ are corrected for ionization (log Pcin situ) according to the following expression:8 log Pc in situ = log P0 in situ − (1 + log 10(pH − pK a))

Therefore, log P0in situ values reflect the actual BBB permeation potential for all the analytes but the analytes extensively ionized as anions. This suggests that, while ionization affects BBB penetration of both bases and ampholytes in a roughly constant degree, it impacts BBB penetration of acids to a quite different, and more dramatic, extent.

4. CONCLUSION The present study confirms the soundness of Δlog kWIAM parameters to predict BBB passage of drugs. For prediction of logP0in situ values they were demonstrated to be superior not only to the n-octanol lipophilicity parameters (log PN and log D7.4) but also to PAMPA-BBB data. In this study Δlog kWIAM values were able to describe in situ passage data for acidic 2815

DOI: 10.1021/acs.molpharmaceut.6b00397 Mol. Pharmaceutics 2016, 13, 2808−2816

Article

Molecular Pharmaceutics (11) Avdeef, A. Absorption and Drug Development; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2012; pp 625−663. (12) Müller, J.; Esso, K.; Dargó, G.; Könczöl, Á .; Balogh, G. T. Tuning the Predictive Capacity of the PAMPA-BBB Model. Eur. J. Pharm. Sci. 2015, 79, 53−60. (13) Bicker, J.; Alves, G.; Fortuna, A.; Soares-Da-Silva, P.; Falcão, A. A New PAMPA Model Using an in-House Brain Lipid Extract for Screening the Blood-Brain Barrier Permeability of Drug Candidates. Int. J. Pharm. 2016, 501 (1−2), 102−111. (14) Braumann, T.; Weber, G.; Grimme, L. H. Quantitative Structureactivity Relationships for Herbicides. J. Chromatogr. A 1983, 261, 329−343. (15) Wishart, D. S.; Knox, C.; Guo, A. C.; Shrivastava, S.; Hassanali, M.; Stothard, P.; Chang, Z.; Woolsey, J. DrugBank: A Comprehensive Resource for in Silico Drug Discovery and Exploration. Nucleic Acids Res. 2006, 34 (Suppl. 1), D668−D672. (16) Avdeef, A. Absorption and Drug Development; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2012; pp 201−209. (17) Lombardo, F.; Shalaeva, M. Y.; Tupper, K. A.; Gao, F.; Abraham, M. H. ElogP(oct): A Tool for Lipophilicity Determination in Drug Discovery. J. Med. Chem. 2000, 43 (15), 2922−2928. (18) Biobyte Clog P; Claremont, CA. (19) Gambaro, V.; Argo, A.; Cippitelli, M.; Dell’Acqua, L.; Farè, F.; Froldi, R.; Guerrini, K.; Roda, G.; Rusconi, C.; Procaccianti, P. Unexpected Variation of the Codeine/morphine Ratio Following Fatal Heroin Overdose. J. Anal. Toxicol. 2014, 38 (5), 289−294. (20) Drug-Membrane Interactions and Pharmacodynamics. In Methods and Principles in Medicinal Chemistry; Seydel, J. K.; Wiese, M., Eds.; Wiley-VCH Verlag GmbH: Weinheim, Germany, 2002; Vol. 15, 217−289. (21) Barbato, F.; La Rotonda, M. I.; Quaglia, F. Interactions of Nonsteroidal Antiinflammatory Drugs with Phospholipids: Comparison between Octanol/buffer Partition Coefficients and Chromatographic Indexes on Immobilized Artificial Membranes. J. Pharm. Sci. 1997, 86 (2), 225−229. (22) Drug Bioavailability; van de Waterbeemd, H., Testa, B., Eds.; Methods and Principles in Medicinal Chemistry; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, Germany, 2008; Vol. 40. (23) Barbato, F.; La Rotonda, M. I.; Quaglia, F. Chromatographic Indexes on Immobilized Artificial Membranes for Local Anesthetics: Relationships with Activity Data on Closed Sodium Channels. Pharm. Res. 1997, 14 (12), 1699−1705. (24) Barbato, F.; Caliendo, G.; La Rotonda, M. I.; Morrica, P.; Silipo, C.; Vittoria, A. Relationships between Octanol-Water Partition Data, Chromatographic Indices and Their Dependence on pH in a Set of Beta-Adrenoceptor Blocking Agents. Farmaco 1990, 45 (6), 647−663. (25) Aravagiri, M.; Yuwiler, A.; Marder, S. R. Distribution after Repeated Oral Administration of Different Dose Levels of Risperidone and 9-Hydroxy-Risperidone in the Brain and Other Tissues of Rat. Psychopharmacology (Berl) 1998, 139 (4), 356−363. (26) Govers, H.; Ruepert, C.; Stevens, T.; van Leeuwen, C. J. Experimental Determination and Prediction of Partition Coefficients of Thioureas and Their Toxicity to. Chemosphere 1986, 15 (4), 383− 393. (27) Tetko, I. V.; Gasteiger, J.; Todeschini, R.; Mauri, A.; Livingstone, D.; Ertl, P.; Palyulin, V. A.; Radchenko, E. V.; Zefirov, N. S.; Makarenko, A. S.; Tanchuk, V. Y.; Prokopenko, V. V. Virtual Computational Chemistry Laboratory–Design and Description. J. Comput.-Aided Mol. Des. 2005, 19 (6), 453−463. (28) Avdeef, A.; Barrett, D. A.; Shaw, P. N.; Knaggs, R. D.; Davis, S. S. Octanol-, Chloroform-, and Propylene Glycol Dipelargonat-Water Partitioning of Morphine-6-Glucuronide and Other Related Opiates. J. Med. Chem. 1996, 39 (22), 4377−4381. (29) Grumetto, L.; Russo, G.; Barbato, F. Polar Interactions Drug/ phospholipids Estimated by IAM-HPLC vs Cultured Cell Line Passage Data: Their Relationships and Comparison of Their Effectiveness in Predicting Drug Human Intestinal Absorption. Int. J. Pharm. 2016, 500 (1−2), 275−290.

(30) Grumetto, L.; Russo, G.; Barbato, F. Relationships between Human Intestinal Absorption and Polar Interactions Drug/phospholipids Estimated by IAM-HPLC. Int. J. Pharm. 2015, 489 (1−2), 186− 194. (31) Barbato, F.; Grumetto, L.; Carpentiero, C.; Rocco, A.; Fanali, S. Capillary Electrochromatography as a New Tool to Assess Drug Affinity for Membrane Phospholipids. J. Pharm. Biomed. Anal. 2011, 54 (5), 893−899. (32) Tsopelas, F.; Malaki, N.; Vallianatou, T.; Chrysanthakopoulos, M.; Vrakas, D.; Ochsenkuhn-Petropoulou, M.; Tsantili-Kakoulidou, A. Insight into the Retention Mechanism on Immobilized Artificial Membrane Chromatography Using Two Stationary Phases. J. Chromatogr. A 2015, 1396, 25−33.

2816

DOI: 10.1021/acs.molpharmaceut.6b00397 Mol. Pharmaceutics 2016, 13, 2808−2816