Drug–lipid interaction evaluation: why a 19th century solution?

Opinion

Drug–lipid interaction evaluation: why a 19th century solution? Marta M.B. Ribeiro, Manuel N. Melo, Isa D. Serrano, Nuno C. Santos and Miguel A.R.B. Castanho Instituto de Medicina Molecular, Faculdade de Medicina da Universidade de Lisboa, Portugal

The affinity of a drug candidate for a biological membrane (its lipophilicity) is closely related to the pharmacologically crucial events of absorption, biodistribution, metabolization and excretion. The evolution of knowledge of biological membranes during the past two decades contrasts with the rudimentary parameter most commonly used to assess lipophilicity: Po/w, the octanol–water partition coefficient. Po/w is especially unrealistic when testing molecules that are polar or partially charged. By contrast, lipid vesicle-based methods determine the extent of the actual partition of a drug to a membrane much more accurately, and have the additional advantage of enabling the choice of the lipid composition considered most suitable to answer a specific biological or pharmaceutical question. In addition, some of these methods are appropriate for high throughput screening, thus shifting determinations of membrane partition to a more preliminary stage of drug development. This streamlines research and development, by saving the time and money that would be spent on unpromising leads. Introduction Drug–lipid interactions are of major concern in drug development strategies, as they are directly implicated in biodistribution, toxicity and membrane receptor–ligand interaction [1]. Absorption, distribution, metabolization and excretion (ADME) of a drug are strongly influenced by its affinity to the lipid environment (its lipophilicity), which usually translates into an affinity for biological membranes (e.g. blood cells and lipoproteins, or barrier epithelia such as the blood–brain barrier or those found in the intestine and skin) or its accumulation in lipid depots. However, lipophilicity does not exclusively affect ADME. The efficacy of drugs that selectively target specific membranes is crucially dependent on lipid affinity. Some antimicrobial drugs [2], for instance, are included in this class. In these cases, there is a delicate balance between solubility and selectivity towards the target membranes [3]. Even if the drug target is not the lipid bilayer itself or a specific lipid within the bilayer, but rather a bilayerinserted target such as a membrane receptor or transporter, the matter of drug–lipid interaction is still of great importance: drugs acquire their correct orientation and/or conformation upon contact with lipid bilayers [4], and diffuse along these bilayers to meet their targets in a manner that is optimized for ligand–receptor docking [1,5]. Corresponding author: Castanho, M.A.R.B. ([email protected]).

Cell membranes were once regarded as ‘inert’ lipid bilayers. The Singer and Nicholson model [6] portrayed cell membranes as nearly bidimensional boundaries serving as a matrix to sustain membrane proteins, and although phospholipids were not static, little biochemical importance was assigned to them, whether in metabolism or cell signaling. The modern view of cell membranes reveals a completely different picture, considering membranes to be highly functional and heterogeneous structures [7,8] (Figure 1). This evolution of knowledge about membrane structure and function has been fuelled by a greater understanding of biological membrane composition [9,10], heterogeneous lipid clustering [11], and the thermodynamics of protein–lipid interactions [12]. Moreover, sophisticated analytical tools are now being used to gain insights into biological membranes at the molecular level [13,14]. Despite the tide of new data and a new vision of biological membranes, little has changed over the past century in the assessment of drug–lipid affinity. The level of insight that has been attained into the study of the structure, function and dynamics of cell membranes contrasts with the ubiquitous but rudimentary parameter used in most pharmaceutical laboratories worldwide, both in academia and industry: Po/w, the octanol–water partition coefficient. Po/w seems to persist as a reference parameter for membrane affinity, regardless of the fact that it is over a century old and there is a tremendous conceptual gap between the homogenous macroscopic solvent, octanol, and the highly ordered microscopic structure of a lipid bilayer. The reasons behind its past success and its current role as a hindrance to drug research and development deserve attention and reflection. Successful studies in which Po/w was replaced by more realistic predictors of drug–membrane interaction should encourage the pursuit of improved more realistic and robust alternative methods. In this paper, we discuss the advantages and limitations of Po/w, and guide the reader through a critical overview of alternative modern approaches to determining lipophilicity, which might be more affordable, simpler and more reliable than might be thought. In fact, they might shape the future of drug research and development. Po/w and its origins The notion of the distribution of a solute between two phases is an ancient one. Solvent extraction has been central to the purification of essences for millennia, and, in recent centuries, a cornerstone of chemical synthesis.

0165-6147/$ – see front matter ß 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.tips.2010.06.007 Trends in Pharmacological Sciences 31 (2010) 449–454

449

(Figure_1)TD$IG][ Opinion

Trends in Pharmacological Sciences Vol.31 No.10

phases is a constant [15]. Nernst would improve this statement by restricting it to the description of a single partitioning molecular species, therefore precluding solute ionization or association in either phase [16]. The proportionality constant between the concentrations of each phase (the partition coefficient; see Box 1), would soon be applied to pharmacology through the works of Meyer [17] and Overton [18], who demonstrated a correlation between the potency of several anaesthetics and the extent of their oil–gas or oil–water partition. Not only was this a groundbreaking clinical advance, it was also the first time that an organic phase was used to mimic the hydrophobic properties of cell membranes.

Figure 1. Determinant features of lipid bilayers that modulate drug–membrane interaction. (a) Much like the phospholipid molecules, membrane proteins diffuse laterally by brownian motion when free from specific interactions such as those with the underlying cytoskeleton. (b) Instead of being randomly mixed, phospholipid molecules of different types might segregate into domains or be recruited to the vicinity of some proteins. (c) The juxtaposition of hydrophilic and hydrophobic environments in the bilayer, coupled to the anisotropy of the phospholipid acyl chains, constrains the localization and orientation of both membrane proteins and their ligands. This aspect is central to the enzymatic processes that take place in the cell membrane. (d) The role of the cell membrane in signal transduction is not limited to the passive anchoring of receptor proteins, but can also act as a reservoir for precursors of secondary messengers. This is the case with phosphoinositides, which can be cleaved into the messenger molecules inositol-trisphosphate and diacylglycerols.

It was in the second half of the 19th century that this notion was formally introduced. Berthelot and Jungfleisch first stated that, at constant temperature, the ratio of concentrations of a solute in two partially immiscible

Success and ubiquity At a time when physical chemistry and biology were worlds apart, the possibility of using simple thermodynamic relations to understand and model physiological behaviour was of great importance. The later development of in vitro high throughput screens and combinatorial chemistry, which are now routine in the biopharmaceutical and chemical industries, called for rapid and accurate determination of the physicochemical properties of compounds [19]. In accordance with the findings of Meyer and Overton described above, lipophilicity, commonly gauged as the logarithm of Po/w [20], was recognized as an important molecular descriptor in activity prediction algorithms [commonly termed quantitative structure–activity relationship (QSAR) methods]. In fact, the use of Po/w spread mainly because of the simplicity of both the concept and its experimental determination, and its close relation to lipophilicity. The possibility of theoretically predicting

Box 1. The thermodynamic origin of P and analogous parameters. The partition coefficient, P, of a solute s between two immiscible phases, o and w (octanol and water, for instance, denoted as subscripts), is given by the expression: P o;w ¼

Kp ¼

C s;o ; C s;w

(1)

where Cs,o and Cs,w are the local molar concentrations of s in each phase. This simple and widely used expression requires the assumption that both phases are ideally dilute solutions; that is, that the solute molecules are sufficiently diluted to interact only with the solvent molecules. Under these conditions, the chemical potential of s in any phase p will be given by [27,28]: ms; p ¼ mos; p þ RT ln X s; p  mos; p þ RT ln V po C s; p;

(2)

where ms,p is the solute chemical potential and Xs,p its molar fraction. The standard state chemical potential, mos; p , corresponds to a hypothetical situation at the limit Xs,p ! 1. The contribution of the solute molecules to the total volume is negligible when the solution is ideally diluted; in addition, in such a solution the solvent molecules will interact chiefly with each other and will therefore have the molar volume of the pure solvent (V po ). This allows the approximation of Xs,p as V po  C s; p .At the partition equilibrium of s between the two phases, its chemical potential in both will be the same and therefore: 







ms;o þ RT lnVo C s;o ¼ ms;w þ RT ln Vw C s;w

C s;o V ¼ w  e C s;w Vo

450

  m m i;w i;o RT

¼ P o;w

(4)

nS;L =V L nS;W =V W

(5)

where nS,i are the number of moles of solute present in the aqueous (i = W) and lipid (i = L) phases. From Equation 5, the fraction of membrane-bound solute (XL) can be related to Kp, at a given lipid concentration ([L]): XL ¼

K p g L ½L 1þK p g L ½L;

(6)

where gL is the molar volume of lipid. This equation can be used for data analysis when techniques such as centrifugation, equilibrium dialysis, membrane filtration and chromatography are used. In these cases, XL is directly determined and Kp can be retrieved from a XL versus [L] plot using a simple non-linear regression. Alternatively, a plot of XL vs. (1XL).[L] is linear, with a slope equal to Kp.gL. By following the variation with [L] of an experimental parameter, p, which changes upon membrane insertion and for which the ensemble average is a linear combination of the contribution of the molecules in the aqueous ( pW) or on the lipid ( pL) phase, the value of Kp can be obtained:

(3)

From Equation 3, at a given temperature, the ratio of the concentrations can be seen to be a constant value, the partition coefficient, which is: 

When using lipid bilayers instead of octanol, a partition constant, Kp, is considered in an analogous way:

p¼

p W þ K p g L ½L p L 1 þ K p g L ½L

(7)

This method has been successfully used based on experimental parameters such as absorption spectrophotometry, fluorescence intensity, fluorescence lifetime and, with some limitations, fluorescence anisotropy, as reviewed in [26].

(Figure_2)TD$IG][ Opinion

Trends in Pharmacological Sciences

Vol.31 No.10

Figure 2. Different types of artificial phospholipid membranes (typical diameter magnitudes for each membrane model are indicated in the figure). Unilamellar vesicles have free-standing membranes, with an aqueous environment on both sides. SUVs are straightforward to produce, typically by sonication, but their membrane is subject to a strong nonphysiological bending tension [55]. LUVs, usually produced by extrusion techniques [56], are a commonly used model to determine Kp values and are sufficiently large for their membrane to be considered locally planar at the molecular scale. Giant unilamellar vesicles (GUVs), produced by lengthier methods such as electroformation or gentle hydration [57] are seldom used to measure affinities, but their large size allows visualization under a microscope and spatial discrimination of the binding of labelled drugs. Micelles are simple to prepare, usually by self-assembly of specific lysophospholipids or fatty acids [58]; however, their departure from a membrane topology makes them unsuitable for any binding study involving in-depth interaction. Immobilized artificial membranes are easier to produce than vesicles [33] but they have the disadvantages that they expose only one lipid monolayer to the aqueous environment and free phospholipid movement is restricted. In the end, the best model for a given study will depend on a compromise between accuracy and ease of preparation.

Po/w by quantitative structure–property relationship (QSPR) methods [21,22] further contributed to the ubiquity of this constant; QSAR studies can be fed predicted Po/w values, allowing for a purely theoretical activity prediction of some drugs. Indeed, a large part of recent literature involving Po/w concerns precisely the implementation of QSPR–QSAR methods. Limitations It is unlikely, however, that a single solute descriptor such as the Po/w can fully represent the complex and diverse interactions that drugs might establish with biological membranes. Departure from physiological relevance results from several factors: the inability of the relatively simple octanol phase to mimic the complex and anisotropic hydrophilic/hydrophobic setting of a phospholipid bilayer; the difficulty in adapting the octanol phase to reflect membranes with different compositions; and the impossibility of including in the apolar phase either proteins, transporters or receptors in their native state (which might be a determinant for binding) or any anionic species that model electrostatic interactions established with the often charged bilayer surface. Indeed, studies of drug behaviour both in octanol and phospholipid bilayer phases indicate

that the two are generally not comparable, especially where charged molecules are concerned [23,24]. This is especially relevant considering that charged molecules are involved in more than two-thirds of the new drugs being developed by pharmaceutical companies [25]. Thus, the extrapolation of the octanol–water biphasic system to biomembrane–aqueous phase systems in evaluating partition coefficients is frequently an oversimplification [26]. Alternatives to Po/w Biomembrane model systems, such as large or small unilamellar vesicles (LUV, SUV) or multilamellar vesicles (MLV) (Figure 2), can be used to determine the extent of the actual partition of a drug to a membrane [26] (Box 1) much more realistically than with octanol–water systems. These vesicles can be easily produced with different biologically relevant lipid compositions (e.g. presence of charged phospholipids or different cholesterol content), accounting for the contributions of hydrophobic and electrostatic interactions, membrane fluidity and phase separation [2,4,29–31], or can even be prepared directly from lipids extracted from a certain type of cell [32]. The partition constant, Kp, can be determined successfully based on the physical separation of the molecules that 451

Opinion are free in aqueous solution from those that are associated with the lipid membrane. Methods such as centrifugation, equilibrium dialysis, membrane filtration and various types of chromatography can be used for this purpose. Of these, chromatography methods are the most extensively used as an improved method over Po/w for lipophilicity prediction [33]. Monolayers of phospholipids or phospholipid analogues are covalently bonded at their hydrophobic ends to the surface of silica particles and used as the stationary phase in liquid chromatography, a process termed immobilized artificial membrane (IAM) chromatography [34]. The ability of the ordered phospholipids to elicit ionic interactions makes the method more reliable than simple octanol–water partitions; however, the capacity to evaluate charged compounds accurately is still reduced, and the molecular dynamics of the lipid are not fully displayed. In drug discovery and optimization, IAM chromatography provides a rapid and simple tool to quantify drug–membrane interactions separately from other relevant factors such as molecular weight or hydrogen-bonding capacity. In cases in which the fraction of molecules in either of the separated phases can be measured, Kp can be obtained by application of equation 6 in Box 1. However, incomplete separation of the phases or equilibrium perturbation could impair the accuracy of the determination. Physical separation of free and membrane-associated molecules becomes unnecessary if equation 4 in Box 1 and related methods are used. This is particularly simplified if an intrinsically fluorescent drug (e.g. tryptophan- or tyrosine-containing proteins and peptides) is being studied. Nevertheless, for nonfluorescent drugs, the determination can be also achieved without physical separation by UV–visible light absorption [35], calorimetry [36], zeta potential [37], electron paramagnetic resonance [38], or lipid vesicles labeled with a membrane-potential sensitive probe [31]. The application of partition constant determinations at a preliminary stage of drug development definitely requires the establishment of high throughput methods. In principle, some of the optical spectroscopy-based methods can be adapted for batch analysis of several drug candidates. However, to date, the only commercially available techniques for high throughput measurement of lipid bilayer–water partitioning are solid-supported lipid membranes (SSLM; available under the trade name TRANSIL; www.sovicell.com) and liposome electrokinetic chromatography (LEKC). For a review of the comparison between these and other techniques see [39]. SSLM is based on the use of large porous silica particles non-covalently coated with a lipid bilayer of selected phospholipid composition. In a comparison between the partitions obtained by SSLM and those obtained with octanol–water values for 187 pharmaceutical drugs, a correlation between both parameters was observed for most neutral molecules; nonetheless, as might be expected, there was no correlation for ionizable drugs [40]. The same inconsistencies with the data obtained for octanol–water partitioning were reported for LEKC [41], which is a capillary electrophoresis method in which lipid vesicles are incorporated in the buffer solutions and serve as a pseudostationary phase for separation of uncharged and charged molecules. At variance with most of the previously mentioned methods, LEKC enables the establishment of uni452

Trends in Pharmacological Sciences Vol.31 No.10

versal and consistent partition scales for drug–membrane interaction studies [41]. However, it is also considered to be time-consuming and difficult to automate [39]. These might not be drawbacks for SSLM; nevertheless, this method has other limitations, such as the short range of lipophilicities that can be covered. A survey of the latest literature shows that the recognition of the limitations of Po/w and the suitability of these alternative approaches is not yet universal in the scientific community [42–45]. For instance, anaesthesiology studies traditionally focus on Po/w, and most QSPR–QSAR methods still use it as a lipophilicity descriptor. Furthermore, the use of Po/w is backed up by constant updates on the experimental determination of this parameter [46]. However, in fields that have emerged more recently, such as antimicrobial peptides [47], phospholipid membrane models are almost exclusively used. Into the future The majority of current pharmaceutical research and development budgets and time is spent on economic drug failures. Accordingly, and despite increased investment [48], there has been a decline in the number of new chemical formulations submitted to the US FDA, for instance, over the past decade [49]. Failure at late drug development stages, often after more than a decade of intensive research and investment, is a serious financial setback for any pharmaceutical company [50]. Part of the problem lies in the inadequacy of the in vitro assays or animal models employed to mimic physiological processes in humans. To mitigate this problem, innovative, sensitive and rapid technologies, able to analyze multiple drug candidates, should be adopted. Animal experimentation is still acknowledged as the best approximation of human characteristics in a preclinical phase. However, both academia and industry are being enforced by legislation to implement the ‘3Rs’ strategy (replace, reduce, refine) [51]. The European Commission, for instance, launched in 2005 the European Partnership for Alternative Approaches to Animal Testing (for more information visit http://ec.europa.eu/enterprise/ epaa/index_en.htm). Thus, the need for more reliable in vitro screening methods is reinforced by the need to replace and reduce animal experimentation (two of the three Rs [52]) for legal and ethical [53], as well as economic, reasons. Animal experimentation must be kept to a minimum and this is only possible if the choice of drug candidates from in vitro tests is more selective and effective, which in turn is only achievable with more realistic screening methods. In the end, instead of being a legal hurdle, policies such as the 3Rs enforce the creation and adoption of methods that might ultimately improve the cost-effectiveness and reliability of drug discovery and selection, with potential to reduce to 5 years or less the timeline required for drug development and the consequent costs required [50]. It therefore becomes essential to rethink the use of Po/w and move towards partition parameters calculated with lipid bilayers rather than organic solvents. The knowledge accumulated to date about membrane structures and the currently available analytical tools available should ensure a successful outcome for this endeavour.

Opinion Therefore, having better means and tools already available, why stick to a 19th century solution in the 21st century? Although the use of Po/w is still ubiquitous and thus difficult to overcome, we foresee that the use of Kpbased techniques and methods will nucleate and thus overcome the inertia of tradition. A ‘domino effect’ will then occur; as the use of these techniques and method spreads, it will become easier to adapt other techniques and methods to use Kp instead of Po/w. This process will probably begin among academic QSPR–QSAR groups and innovative biotech companies in cosmetics, a field that is more dynamic and less constrained by regulations and standardization than the pharmaceutical industry. Po/w is a crude but settled, simple and affordable parameter. Lack of alternatives in the first half of the 20th century made it a reference standard with a very large amount of accumulated data available to drug researchers and developers (e.g. the LOGKOW databank; http://logkow. cisti.nrc.ca/logkow/index.jsp). The absolute need for standardization in the pharmaceutical industry demands universal parameters that all researchers and technicians can use to: (i) compare between drugs and (ii) assess the potential of a candidate to become an effective drug. Although the power to evolve to a more reliable parameter (Kp) exists, it is yet to build a large and wide-ranging library of reference values of Kp for known drugs that allows comparison of drug candidates with reference drugs, and the development of new methods to screen drug candidates for assessment of their potential. Consequentially, although the experimental use of Kp is becoming widespread, Po/w is still preferred in theoretical models, and these constitute a very important part of the drug discovery process. Yet, a close look at the recent advances in drug–lipid interaction studies in the field of molecular dynamics simulation [54] shows that the knowledge and computational tools are now available to account not only for the properties of the drug, but for the properties of the lipid. The time has come to embrace the knowledge and tools of analytical chemistry to start the post-Po/w era in drug discovery and development. Acknowledgments This work was supported by project grants PTDC/SAU-FCF/69493/2006, PTDC/QUI/69937/2006 and PTDC/QUI-BIQ/104787/2008 from Fundac¸a˜o para a Cieˆncia e Tecnologia – Ministe´rio da Cieˆncia, Tecnologia e Ensino Superior (FCT-MCTES, Portugal), project grant 230654 - PEP2BRAIN (FP7 Marie Curie Programme, IAPP) and Fundac¸a˜o Calouste Gulbenkian (Portugal). MMBR, MNM and IDS also acknowledge FCT-MCTES fellowships SFRH/BD/42158/2007, SFRH/BD/24778/2005 and SFRH/ BPD/37998/2007, respectively.

References 1 Castanho, M.A. and Fernandes, M.X. (2006) Lipid membrane-induced optimization for ligand-receptor docking: recent tools and insights for the ‘‘membrane catalysis’’ model. Eur. Biophys. J. 35, 92–103 2 Melo, M.N. et al. (2009) Antimicrobial peptides: linking partition, activity and high membrane-bound concentrations. Nat. Rev. Microbiol. 7, 245–250 3 Ben-Dror, S. et al. (2006) On the correlation between hydrophobicity, liposome binding and cellular uptake of porphyrin sensitizers. Photochem. Photobiol. 82, 695–701 4 Domingues, M.M. et al. (2009) Fold-unfold transitions in the selectivity and mechanism of action of the N-terminal fragment of the bactericidal/permeability-increasing protein (rBPI(21)). Biophys. J. 96, 987–996

Trends in Pharmacological Sciences

Vol.31 No.10

5 Sargent, D.F. and Schwyzer, R. (1986) Membrane lipid phase as catalyst for peptide-receptor interactions. Proc. Natl. Acad. Sci. U. S. A. 83, 5774–5778 6 Singer, S.J. and Nicolson, G.L. (1972) Fluid mosaic model of structure of cell-membranes. Science 175, 720–731 7 Mouritsen, O.G. and Andersen, O.S. (1998) Do we need a new biomembrane model? Biol. Skr. Dan. Vid. Selsk. 49, 7–12 8 Simons, K. and Ikonen, E. (1997) Functional rafts in cell membranes. Nature 387, 569–572 9 Dennis, E.A. (2009) Lipidomics joins the omics evolution. Proc. Natl. Acad. Sci. U. S. A. 106, 2089–2090 10 Riezman, H. and van Meer, G. (2004) Lipid pickup and delivery. Nat. Cell Biol. 6, 15–16 11 Semrau, S. and Schmidt, T. (2009) Membrane heterogeneity – from lipid domains to curvature effects. Soft Matter 5, 3174–3186 12 Sanchez, S.A. and Gratton, E. (2005) Lipid-protein interactions revealed by two-photon microscopy and fluorescence correlation spectroscopy. Acc. Chem. Res. 38, 469–477 13 Muller, D.J. (2008) AFM: a nanotool in membrane biology. Biochemistry 47, 7986–7998 14 Santos, N.C. and Castanho, M.A.R.B. (2004) An overview of the biophysical applications of atomic force microscopy. Biophys. Chem. 107, 133–149 15 Berthelot, M. and Jungfleisch, E. (1872) Sur les lois qui president ou partage d’un corps entre deux dissolvants (experiences). Ann. Chim. Phys. 26, 396–407 16 Nernst, W. (1891) Verteilung eines Stoffes zwischen zwei Lo¨sungsmitteln und zwischen Lo¨sungsmittel und Dampfraum. Z. Phys. Chem. 1891, 110–139 17 Meyer, H. (1899) Zur Theorie der Alkoholnarkose. Arch. Exptl. Pathol. Pharmakol. 42, 109–118 18 Overton, E. (1901) Studien uber die Narkose, Verlag von Gustav Fischer 19 Ghose, A.K. et al. (1998) Prediction of hydrophobic (lipophilic) properties of small organic molecules using fragmental methods: An analysis of ALOGP and CLOGP methods. J. Phys. Chem. A 102, 3762– 3772 20 Caron, G. et al. (1999) Combined molecular lipophilicity descriptors and their role in understanding intramolecular effects. Pharm. Sci. Technol. Today 2, 327–335 21 Clark, D.E. and Pickett, S.D. (2000) Computational methods for the prediction of ‘drug-likeness’. Drug Discov. Today 5, 49–58 22 Tetko, I.V. et al. (2006) Can we estimate the accuracy of ADME-Tox predictions? Drug Discov. Today 11, 700–707 23 Austin, R.P. et al. (1995) Partitioning of ionizing molecules between aqueous buffers and phospholipid vesicles. J. Pharm. Sci. 84, 1180– 1183 24 Wallace, B.A. et al. (1981) Conformation of gramicidin A in phospholipid vesicles: circular dichroism studies of effects of ion binding, chemical modification, and lipid structure. Biochemistry 20, 5754–5760 25 Wells, J.I. (1988) Pharmaceutical Preformulation: The Physicochemical Properties of Drug Substances, Ellis Horwood Ltd 26 Santos, N.C. et al. (2003) Quantifying molecular partition into model systems of biomembranes: an emphasis on optical spectroscopic methods. Biochim. Biophys. Acta 1612, 123–135 27 Guggenheim, E.A. (1952) Mixtures, Oxford University Press 28 Staverman, A.J. and Van Santen, J.H. (1941) The entropy of liquid mixtures. Rec. Trav. Chim. Pay. B 60, 76–84 29 Lopes, S.C.D.N. et al. (2005) Lipidic membranes are potential ‘‘catalysts’’ in the ligand activity of the multifunctional pentapeptide neokyotorphin. Chem. Bio. Chem. 6, 697–702 30 Marsh, D. (2009) Cholesterol-induced fluid membrane domains: a compendium of lipid-raft ternary phase diagrams. Biochim. Biophys. Acta 1788, 2114–2123 31 Matos, P.M. et al. (2008) Interaction of peptides with biomembranes assessed by potential-sensitive fluorescent probes. J. Pept. Sci. 14, 407– 415 32 Kramer, S.D. and Wunderli-Allenspach, H. (1996) The pH-dependence in the partitioning behaviour of (RS)-[3H]propranolol between MDCK cell lipid vesicles and buffer. Pharm. Res. 13, 1851–1855 33 Taillardat-Bertschinger, A. et al. (2003) Immobilized artificial membrane HPLC in drug research. J. Med. Chem. 46, 655–665

453

Opinion 34 Ong, S. et al. (1995) Membrane partition coefficients chromatographically measured using immobilized artificial membrane surfaces. Anal. Chem. 67, 755–762 35 Kaminoh, Y. et al. (1988) Membrane-buffer partition coefficients of tetracaine for liquid-crystal and solid-gel membranes estimated by direct ultraviolet spectrophotometry. Biochim. Biophys. Acta 946, 337–344 36 Heerklotz, H. and Seelig, J. (2000) Correlation of membrane/water partition coefficients of detergents with the critical micelle concentration. Biophys. J. 78, 2435–2440 37 Kim, J. et al. (1991) Binding of peptides with basic residues to membranes containing acidic phospholipids. Biophys. J. 60, 135–148 38 Bianconi, M.L. and Schreier, S. (1991) EPR study of membrane partitioning, orientation, and membrane-modulated alkalinehydrolysis of a spin-labeled benzoic-acid ester. J. Phys. Chem. 95, 2483–2486 39 Hartmann, T. and Schmitt, J. (2004) Lipophilicity – beyond octanol/ water: a short comparison of modern technologies. Drug Discov. Today: Technol. 1, 431–439 40 Loidl-Stahlhofen, A. et al. (2001) Multilamellar liposomes and solidsupported lipid membranes (TRANSIL): screening of lipid-water partitioning toward a high-throughput scale. Pharm. Res. 18, 1782–1788 41 Carrozzino, J.M. and Khaledi, M.G. (2004) Interaction of basic drugs with lipid bilayers using liposome electrokinetic chromatography. Pharm. Ress. 21, 2327–2335 42 Belostotsky, I. et al. (in press) Mitochondrial targeting for photochemotherapy. Can selective tumor cell killing be predicted based on n-octanol/water distribution coefficients? Biotech. Histochem., doi:10.3109/10520295.2010.483656 43 Durcekova, T. et al. (2010) Chemometrical study of the anaesthetical activity of alkoxyphenylcarbamic acid esters. Pharmazie 65, 169–174 44 Fernandes, J.P. et al. (2010) QSAR modeling of a set of pyrazinoate esters as antituberculosis prodrugs. Arch. Pharm. (Weinheim) 343, 91–97

454

Trends in Pharmacological Sciences Vol.31 No.10 45 Toropov, A.A. et al. (2010) QSPR modeling of octanol/water partition coefficient of antineoplastic agents by balance of correlations. Eur. J. Med. Chem. 45, 1639–1647 46 Short, J. et al. (in press) Practical methods for the measurement of logP for surfactants. Ecotoxicol. Environ.Saf., doi:10.1016/j.ecoenv. 2010.03.006 47 Yeaman, M.R. and Yount, N.Y. (2003) Mechanisms of antimicrobial peptide action and resistance. Pharmacol. Rev. 55, 27–55 48 Black, D. and Henry, B. (2008) Found in translation. Realizing the potential of new therapeutic targets. The Biochemist. 12, 28–30 49 Parexel. (2006) Pharmaceutical R&D Statistical Sourcebook 2005/ 2006 50 (2005) Drug Discovery Markets in Europe, Frost & Sullivan 51 European Commission. (2006) The Community Action Plan on the protection and welfare of animals 2006–2010. (DG Health and Consumer Protection edn), European Commission 52 Richmond, J. (2000) 3Rs - past, present and future. Scand. J. Lab. Anim. Sci. 27, 84–92 53 Balls, M. (1994) Replacement of Animal Procedures - Alternatives in Research. Education and Testing. Lab. Anim. 28, 193–211 54 Yesylevskyy, S. et al. (2009) Alternative mechanisms for the interaction of the cell-penetrating peptides penetratin and the TAT peptide with lipid bilayers. Biophys. J. 97, 40–49 55 Szoka, F., Jr and Papahadjopoulos, D. (1980) Comparative properties and methods of preparation of lipid vesicles (liposomes). Annu. Rev. Biophys. Bioeng. 9, 467–508 56 Mayer, L.D. et al. (1986) Vesicles of variable sizes produced by a rapid extrusion procedure. Biochim. Biophys. Acta 858, 161–168 57 Rodriguez, N. et al. (2005) Giant vesicles formed by gentle hydration and electroformation: a comparison by fluorescence microscopy. Colloids Surf. B Biointerfaces 42, 125–130 58 Tanford, C. (1991) The hydrophobic effect: formation of micelles and biological membranes, Krieger