Download Drug Permeability: Role of Properties & Membrane Interactions and more Study notes Statistics in PDF only on Docsity!
Physicochemical properties of drugs and
membrane permeability
S.F. Malan , D.J. Chetty and J. du Plessis
Introduction Understanding a drug’s physicochemical properties, histology and the influence of the combination on how the drug perme- ates membranes helps to predict its transport into and through the body. Drugs that are administered for mucosal absorption at a site, whether buccal, sublingual, vaginal, rectal, nasal, or pulmonary, or given perorally for absorption across the intesti- nal mucosa, must pass through the various layers of the absorb- ing membrane. The drug must also interact with a mélange of paracellular and intracellular fluids, biopolymers, lipids and proteins before entering the circulation. Drug molecules that do not have the requisite physicochemical properties necessary for active-site accessibility may have only minimal or even no biological activity. A mechanistic analysis of the correlation between permeation and physicochemical properties will assist in understanding the factors that influence mucosal permeation. 1 The transdermal penetration of chemicals involves their parti- tioning into, and transport through, the cutaneous layers, namely the stratum corneum, the stratum basale and the upper dermis. Uptake into the blood, on the other hand, is rapid and does not appear to limit the process. The stratum corneum is the principal barrier to skin penetration; it consists of dead cells surrounded by an intercellular matrix of lipids and aqueous layers. Three possible mechanisms have been proposed for the trans- port of solute through the stratum corneum (Fig. 1). The first can be described as a shunt route, which provides a parallel pathway through the sweat ducts and hair follicles without hindrance of the stratum corneum. The second and third pathways are the intercellular and intracellular routes. 2 For the intracellular route, drugs pass directly through the cells of the stratum corneum, whereas in the intercellular route, they diffuse around the cells in a tortuous manner. Perhaps the most widespread procedure for attempting to clarify the mechanisms of percutaneous absorption is to estab-
lish relations between representative penetration parameters (such as permeability coefficients) and measurements of lipophilicity or other related physicochemical properties. 3 Strict homeostasis of the central nervous system environment is maintained by the blood–brain barrier (BBB), which is of the utmost importance for optimal brain function.^4 The existence and function of this barrier has been the focus of many studies since the late 19th century, when it was observed that brain tissue did not stain after injection of a vital dye into the systemic blood circulation of rats. 5 Administration of horseradish peroxidase revealed the anatomical localization of the BBB to be on the level of the cerebral endothelial cells. 6 These cells differ substantially from the peripheral endothelial cells in that they possesses tight junctional structures, 7,8^ which hinder para- cellular transport of hydrophilic compounds across the BBB.^9 The low pinocytotic vesicular activity and absence of fenestrae in the cerebral endothelial plasma membrane (Fig. 2) also limit fluid phase uptake and transcellular transport.10,11^ The functional properties of the blood–brain barrier can be summarized as fol- lows: 11 selective and asymmetric permeability to physiologically important ions, implying the expression of site-specific pumps and ion carriers; selective permeability based on molecular weight and oil/water partition coefficient (lipid-soluble
Review Article South African Journal of Science 98 , July/August 2002 385
a (^) School of Pharmacy, Potchefstroom University for C.H.E., Potchefstroom 2520, South Africa.b GlaxoSmithKline, Parsippany, New Jersey, U.S.A. *Author for correspondence. E-mail: fchsfm@puknet.puk.ac.za
Drug penetration across membranes is complex and dependent on many possible physicochemical interactions. The main physicochemical determinants include partition, the molecular weight and size of the drug molecule, its solubility, ionization state, and hydrogen-bonding capacity. These and molecular confor- mational descriptors and their role in quantitative structure perme- ability relationships (QSPR) are discussed in this review. The identification and quantification of these factors and their inclusion in predictive QSPR algorithms, based on an understanding of the physicochemical properties that govern membrane permeation, facilitate the selection of new drugs and formulations.
Fig. 1. Three routes of permeation through the stratum corneum. 2
Fig. 2. Comparison of sections of typical blood capillaries in non-neural and neural (brain) tissue. 12
substances with a molecular weight of up to 400–650 generally penetrate the BBB); expression of barrier-specific transporters for essential metabolic substrates and nutrients like D -glucose and charge-bearing L -amino acids; extrusion of otherwise permeable but possibly harmful substances. As a result of this, the entry of drugs targeted at pharmacological sites in the brain parenchyma is severely restricted. There are several mechanisms that convey chemicals across the blood–brain barrier. These include passive diffusion, carrier- mediated transport, receptor- and absorption-mediated transcytosis, and active efflux (by means of P-glycoprotein). 13 Any one or more of these processes can be used for delivery of drugs to the brain. The structural and physicochemical proper- ties of the drug then determines whether effective transport and equilibration across the BBB results. It is also important to note that the blood–brain barrier is under physiological control and that its transport characteristics are influenced by the presence of other substances and thus also by their physicochemical properties. 14 In this review, we discuss the various physicochemical proper- ties that influence membrane permeation and the different approaches that have been used to translate these descriptors into algorithms for the prediction of membrane permeability.
Permeant properties and membrane permeability Passive diffusion is energy-independent (not driven, for exam- ple, by ATP-dependent processes) and a net flow of molecules in a particular direction occurs because of a chemical potential difference or a concentration gradient. For a drug to leave the polar environment of the plasma and cross the membrane — whether mucosal, dermal or BBB — through the non-polar lipid membranes of the endothelial cells by means of diffusion, very specific physicochemical properties are necessary.^15 The proper- ties of the drug molecule that may have a bearing on membrane penetration include molecular weight and size, solubility, parti- tion coefficient, degree of ionization, surface activity, structural isomerism, intermolecular forces, oxidation-reduction poten- tials, interatomic distances between functional groups, and stereochemistry.
Lipophilicity The lipophilicity of a molecule is related to its free energy of partitioning; partition coefficients are directly related to the free energy of transfer of a substance between two immiscible phases. 16 The most common expression of lipophilicity is the logarithm of the n -octanol–water partition coefficient (log P oct). The n -octanol–water two-phase system is a popular model for assessing partitioning at lipid membranes because of the similar- ities of the n -octanol (long hydrophobic chain and polar hydroxyl group) and membrane lipids. The role of hydrophilicity in transbuccal permeation kinetics has been reported for a group of progestational steroids. 17 Progesterone, a lipophilic molecule, and its hydroxy derivatives were used as model penetrants to investigate the effect of a systematic variation in hydrophilicity on permeation across the rabbit’s buccal mucosa. The transbuccal permeation of proges- terone and its mono-, di- and tri-hydroxy derivatives was shown to follow zero-order kinetics under the experimental conditions imposed. The addition of up to three hydroxyl groups on the progesterone molecule substantially affected the rate of transbuccal permeation. In the case of the mono- and di- hydroxy derivatives, the permeability rose as the number of hydroxy groups increased. By contrast, the transbuccal perme- ability of trihydroxyprogesterone decreased in spite of a marked
increase in its solubility. In general, correlations that involve log P oct thus define empirical relationships and provide only limited insight into the nature of the mucosal barrier and its penetration by the specified structures. The lipid/water partition coefficient of a drug is the basic deter- minant of drug permeability through the stratum corneum. A drug with a log P oct value of approximately 2 is considered to be a potential candidate for transdermal delivery. 18 According to Guy, 19 compounds with a log P oct value between 1 and 3, with relatively low molecular weights and relatively low melting points, for example nicotine and nitroglycerin, are likely to display optimum passive skin permeation.^19 The correlation between BBB permeability and log P oct values (representing lipophilicity) is well described in the literature and includes diverse structures, 4,20^ pyrrolopyrimidines, 21 beta-block- ing agents and non-steroidal anti-inflammatory drugs^22. The lipid coating of malto-dextrin nanoparticles increased their permeation characteristics, especially for charged particles. 23 In general, the correlation is described as parabolic with an optimum log P oct value of about 2 (refs 7, 24, 25). The influence of electronic properties, and resonance and steric effects like conformational crowding and molecular folding on lipo- philicity, should also be considered.^26 Lipophilicity can thus be more accurately expressed in mathematical terms other than simply by log P oct values. In addition to the above, hydropho- bicity or lipophilicity estimates from reverse phase HPLC reten- tion times have been correlated with solvation parameters using linear free energy relationships. 27,
Hydrogen-bonding capacity The importance of hydrogen bonding as a determinant of drug permeation across absorbing membranes has been recog- nized in several studies.29,30^ The relationship between the perme- ation of oral mucosae by a series of hydroxylated progestins and a variety of molecular descriptors was correlated with a wide variety of physicochemical parameters and solvatochromic molecular factors (parameters describing dipolarity/polari- zability, hydrogen-bond donor acidity and hydrogen-bond acceptor basicity; arrived at by averaging multiple normalized solvent effects on a variety of properties involving many diverse types of indicators). 1 The principal determinants of both buccal and sublingual permeation, as derived from stepwise regression, were the hydrogen bond acidity and basicity of the penetrant molecules. Quantitative structure–permeability algorithms based on solvatochromic properties demonstrated significant predictive value when applied to sublingual ( r^2 = 0.88) and buccal ( r^2 = 0.90) permeability. The principal advantage of using solvatochromic descriptors in predicting oral mucosal perme- ation of drugs is that they allow molecular structures to be taken into account. Hydrogen-bonding donor (") and acceptor ($) parameters are generally derived from substructure summation and have been successfully used in relation to steroid drugs to provide predictive ( r^2 = 0.96 for this series of steroid drugs) algorithms for trans- dermal permeability.^30 In a study in which the structure–epidermal permeabilities of seven compounds were compared, the solvatochromic group contribution and two-phase models were more successful than the other determinant parameters at predicting the permeability coefficient ( k p ). 31 The hydrogen-bonding capability of the penetrant, measured by " and $, was also described as the most powerful determinant of diffusion across the stratum corneum. 32 Molecular weight was found to be of minor impor- tance, and the role of dipolarity/polarizability was insignificant.
386 South African Journal of Science 98 , July/August 2002 Review Article
bility of a drug determines the concentration presented to the absorption site, and the partition coefficient strongly influences the rate of transport across the absorption barrier.^49 The thermodynamic activity of a drug in a particular vehicle indicates the active substance’s potential to become available for therapeutic purposes. A saturated solution is therefore prefera- ble for a topical drug delivery system as it represents maximum thermodynamic activity (leaving potential). 50 The saturation level is dependent on the solubility of the drug in the delivery formulation. 51 Since the stratum corneum appears to act as a lipophilic membrane, increased drug lipophilicity may be expected to enhance drug penetration into the skin. 52 Drugs with high distribution but low diffusion rates in the skin will, however, accumulate in the skin. The enhanced lipophilicity of derivatives of prednisolone was shown to contribute to their greater distribution into the stratum corneum than the parent compound. Although the permeation constants of the prednisolone derivatives are lower than that of prednisolone itself, drug retention in the skin increases and diffusion rates decrease with an increase in the lipophilicity of these derivatives. 52 The hydrophilicity of progesterone, a lipophilic steroid, was progressively increased by incorporating one or more hydroxy substituents at different positions on the steroidal skeleton.^53 In studies on the effects of these hydrophilic substituents on the permeation of progesterone, the steady-state rate of permeation across the intact skin and stripped skin of the hairless mouse was approximately proportional to the solubility of drugs in the stratum corneum or in the viable skin, respectively. Further- more, the solubility of progesterone and its hydroxy derivatives in the stratum corneum decreased gradually the greater the hydrophilicity of the penetrant. The solubility of these progestins in the viable skin was observed to be dependent not only on the penetrant’s hydrophilicity, however, but also on the position of the –OH group on the penetrant molecule. The diffusivity of progesterone and its hydroxyl derivatives across the stratum corneum was almost independent of the hydro- philicity of the drugs. 53
State of ionization The degree of ionization of a permeant is a function of both its p K a value and pH at the membrane surface. For many weak acids and bases, only the unionized form possesses appreciable lipid solubility. The mucosal absorption of many compounds is maximal at the pH at which they are mostly unionized and decreases as the degree of ionization increases. 54 A change in solution pH alters the state of ionization of a drug and hence its hydrophilicity/lipophilicity as it crosses a membrane. Owing to the complexity of the mucosae, however, it is unlikely that the pH will remain unchanged along the length of the permeation pathway (that is, through the mucosa). It is more likely that altering the pH so as to favour formation of the unionized drug at the absorbing surface simply causes an enhanced partitioning of unionized drug in the surface layers of the mucosa. It is probable also that the high local concentration of the drug then promotes its diffusion across the mucosa. 55 Most drugs are weak acids or bases and, according to pH–partition theory, may exist in an ionized or non-ionized form, depending upon the pH of the vehicle. The drug’s activity coefficient changes significantly as a function of pH, for pH values greater than p K a for acidic compounds, and less than p K w–
- p K b for basic drugs.^56 Membranes are more permeable to the non-ionized forms, because of their greater lipid solubility.^57 The
pH of the vehicle in which the penetrant is administered, in combination with the drug molecule’s ionization constant, p K a , determines the actual concentrations of the ionized and non-ionized species. The non-ionized molecule is believed to travel by the intracellular route through the stratum corneum, whereas the ionized molecule is more likely to pass through the intercellular spaces. 58 The importance of degree of dissociation or ionization as a factor determining permeation through the BBB is also well documented. 15 This is especially so for drugs with p K a values close to the physiological pH of 7.4; the effect is incorporated in the log D x (log P oct at pH = x ) term, which in many cases domi- nates the contributions of other physico-chemical factors.^55
Stereochemistry and steric interactions Studies have been conducted on the effects of the stereo- chemistry of substituent groups on the permeation of 11"- and 11 $-hydroxyprogesterone, 20"-hydroxyprogesterone (20"- dihydroprogesterone), 20$-hydroxyprogesterone, and 17"- estradiol and estradiol (17$-estradiol) across biological mem- branes. 15 The orientation of the hydroxy group played a pivotal role in determining the permeation rate of the isomers. Depending on the orientation, the hydroxy groups may be sterically hindered. The relative steric hindrance of the hydroxy groups then determines aqueous solubility, which in turn affects the permeation rate. For delivery of compounds across the blood–brain barrier, stereochemical considerations are of further importance as specific transporter systems (with stereospecific activity) are described for large neutral, small neutral, basic and acidic L -amino acids. 13,
Molecular conformational and surface properties As the conformation of a molecule changes, physicochemical properties, such as hydrogen-bonding activity, are also affected and may influence membrane permeability. 26 The three- dimensional shape and conformational flexibility of a drug molecule must therefore be considered when evaluating deter- minants of permeation through mucosae. 60 The surface area of drug molecules accessible to a solvent is another potential influence on their permeation. A theoretical method based on the determination of dynamic surface proper- ties of drug molecules has been developed to predict mucosal permeation. 61 The dynamic polar surface area (PSAd ) accounts for the shape and flexibility of the drug molecule and is related to its hydrogen-bonding capacity. The PSA (^) d is calculated as a statis- tical average in which the surface area of each low-energy conformer is weighted by its probability of existence.^62
Quantitative structure–permeability relationships Various physicochemical parameters have been used to predict the passive diffusion of drugs through membranes and studies have employed these descriptors in quantitative struc- ture–permeability relationships (QSPR). 29,30^ Hydrophobicity terms like molecular volume and solvent-accessible surface area and polarity parameters are used in this regard.^63 Quantitative correlations with hydrogen-bond descriptors, polarizability and partial charge have also been derived. 64 The prediction of membrane permeability of structurally unrelated compounds using just a single predictor is seldom possible, given the many mechanistic and physicochemical factors involved in drug transport through membranes. As a result, most predictive models combine several physicochemical descriptors into a single relationship using multiple linear
388 South African Journal of Science 98 , July/August 2002 Review Article
regression or multivariate analysis. 65, Molecular properties such as lipophilicity, molecular weight and hydrogen-bonding potential are much used in QSPR models. The hydrogen-bond forming capacity of drug mole- cules is a good predictor of mucosal permeation. 66,67^ The prediction of permeability was improved further when drug- membrane electrostatic interactions, namely hydrogen-bond donor and acceptor activity or the index of electricity ( E c ) for drugs, were added to the lipophilicity parameter.^65 The predic- tive equation that was obtained in this case was applicable even to the absorption of zwitterionic drugs. Recent predictive algorithms for skin permeability include one which uses the molecular volume and hydrogen bond donor-acceptor activities. 23 This is a further indication of the importance of hydrogen bonding in skin permeation and is a factor that should be considered qualitatively. 68 Solute size, solute dipolarity/polarizability, hydrogen-bond acidity and basicity produced some instructive relationships. 2 That percu- taneous absorption is mediated by the hydrophobicity and molecular size of the penetrant has been re-emphasized in a recent study involving 114 compounds. 69 The use of dynamic polar surface area (PSA (^) d ) to predict intestinal permeability takes into account the three-dimensional shape and flexibility of the drug molecule and it allows the effects of the internal hydrogen bonding and steric hindrance to be considered. 61 Multivariate data analysis has been used to de- rive predictive models that correlated passive intestinal permeability with physicochemical and conformational descriptors. 67 The best model so derived used the variables HBD (number of hydrogen bond donors), PSA (polar surface area) and either log D 5.5 or log D 6.5 (octanol/water partition coefficient at pH 5.5 or 6.5, respectively). In addition, statistically good models for predicting in vivo P eff (effective permeability) values were also obtained by using only HBD and PSA, or HBD, PSA and log P oct. The importance of the polar surface area of a molecule, in combination with log P oct and hydrogen bond parameters, as an effective predictor of BBB penetration has also become evident. 70– Advances in computing has made it possible to use molecular mechanics and quantum mechanics as virtual screening tools and theoretical methods to predict mucosal permeation based on quantum mechanical calculations with good correlation.61, Descriptors that are related to the valence regions of molecules were generated and then correlated with intestinal permeabil- ity. 74 A correlation between apparent permeability ( P app) of a group of 10 endothelin antagonists and the dynamic polar surface area of these molecules produced an r^2 of 0.83. r^2 increased to 0.87 when the dynamic non-polar surface area was added to the regression equation. Initially, a two-component partial least squares (PLS) model with r^2 of 0.90 was derived from 20 calculated descriptors. When stepwise variable selection was subsequently used, the model resulted in a three-component PLS model with r^2 of 0.98. The final model contained eight of the original 20 descriptors. The excellent correlation obtained using the quantum mechanics approach to developing QSPR demon- strates that using descriptors that are generated in silico can produce highly predictive models of membrane permeability. Good statistical models, with r^2 ranging from 0.73 to 0.95, that permit simple computational prediction of biopharmaceutical properties like absorption and BBB permeability, have also been derived by using theoretically calculated molecular properties (e.g. molar refractivity, molar volume, refractive index, surface tension, density and log P oct). 74
Most models are, however, derived for homologous series of molecules or structurally related drugs and do not have predic- tive capacity when structural diversity is introduced. Quantita- tive molecular descriptors obtained from 3D molecular field maps using the procedure called VolSurf could remedy this problem and has been shown to have a predictive r^2 of 0.90 for membrane partitioning. 75
Active transport systems As the structure and function of the blood–brain barrier seriously hinder passive transport of compounds into the brain, active transport systems play a more important role than passive diffusion in drug delivery. 9,13^ The importance of the physico- chemical properties of the drug for active transport is obvious in the molecular recognition for transport by specific bidirectional systems as well as for unidirectional systems like the P-glyco- protein efflux system.^13 Electrostatic and steric considerations are especially important for specific transporter systems as described for large neutral, small neutral, basic and acidic L -amino acids, monocarboxylic acids and small basic amines.13, Electrostatic interaction is also one of the main determinants of receptor and absorption-mediated transport of peptides. 13 Cationized albumin, for example, interacts electrostatically with anionic groups on the barrier and this interaction triggers absorption-mediated transcytosis through the BBB. 76 By conjugation of a drug to monoclonal antibodies or peptides with the required properties, delivery to the brain can also be achieved through vector- or receptor-mediated trancytosis. 77,
Conclusions It is evident that membrane penetration is dependent on many possible physicochemical interactions. Though the process of drug penetration across membranes is complex, it is possible to identify the main physicochemical determinants involved. The determinants discussed in this review include partition coeffi- cient, molecular weight and size, solubility, ionization state, and hydrogen-bonding capacity. Hydrogen-bonding capacity seems to be an important co-factor with most of the other prop- erties. Molecular conformation also plays an important role in quantitative structure–permeability relationships. In principle, these variables can be modified to optimize drug design and formulation. The study of drug permeation across membranes and assess- ment of the effect of a drug’s physicochemical properties on permeability have increased our understanding of the molecu- lar factors that influence and control drug transport, and how they interact. Quantitative structure–permeability relationships that account for the various physicochemical factors in predic- tive algorithms today facilitate the selection and formulation of new drugs. Recieved 7 August. Accepted 10 December 2001.
- Chetty D.J., Chen L.H., Lin S.S. and Chien Y.W. (1998). Evaluation of physicochemical and solvatochromic descriptors in quantitative struc- ture-buccal permeability relationships. AAPS PharmSci. 1 (1), S93.
- Abraham M.H., Chada H.S. and Mitchell R.C. (1995). The factors that influence skin penetration of solutes. J. Pharm. Pharmacol. 47 , 8–16.
- Diez-Sales O., Watkinson A.C., Herraez-Domininguez M., Javaloyes C. and Hadgraft J. (1996). A mechanistic investigation of the in vitro human skin permeation enhancing effect of Azone ®^ Int. J. Pharm. 129 , 33–40.
- De Vries H.E., Kuiper J., De Boer A.G., Van Berkel T.J.C. and Breimer D.D. (1997). The blood–brain barrier in neuroinflammatory diseases. Pharmacol. Rev. 49 (2), 143–156.
- Ehrlich P. (1885). Das Sauerstoffbeduerfnis des Organismus: Eine farbenanalytiche Studie, vol. 8, p. 167. Hirschwald, Berlin.
- Reese T.S. and Karnovsky M.J. (1967). Fine structural localisation of a
Review Article South African Journal of Science 98 , July/August 2002 389
- Gaillard P., Carrupt P-A., Testa B. and Boudon A. (1994). Molecular lipophilicity potential, a tool in 3D QSAR: method and applications. J. Comput. Aided Mol. Des. 8 , 83–96.
- Raevsky O.A., Fetisov V.I., Trepalina E.P., McFarland J.W. and Schaper K.J. (2000). Quantitative estimation of drug absorption in humans for passively transported compounds on the basis of their physico-chemical parameters. Quant. Struct.-Act. Relat. 19 (4), 366–374.
- Sugawara M., Takekuma Y., Yamada H., Kobayashi M., Iseki K. and Miyazaki K. (1998). A general approach for the prediction of the intestinal absorption of drugs: Regression analysis using the physicochemical properties and drug-membrane electrostatic interaction. J. Pharm. Sci. 87 (8), 960–966.
- Winiwarter, S., Bonham, N.M., Ax, F., Hallberg, A., Lennerhans, H and Karlen, A. (1998). Correlation of human jejunal permeability ( in vivo ) of drugs with experimentally and theoretically derived parameters. A multivariate data analysis approach. J. Med. Chem. 41 (25), 4939–4949.
- Ren S. and Lien E.J. (2000). Caco-2 cell permeability vs human gastrointestinal absorption: QSPR analysis. Prog. Drug Res. 54 , 1–23.
- Roberts M.S., Anderson R.A. and Swarbrick J. (1977). Permeability of human epidermis to phenolic compounds. J. Pharm.Pharmacol. 29 , 677–683.
- Cronin M.T., Dearden J.C., Moss G.P. and Murray-Dickson G. (1999). Investiga- tion of the mechanism of flux across human skin in vitro by quantitative struc- ture-permeability relationships. Eur. J. Pharm. Sci. 7 (4), 325–330.
- Clark D.E. (1999). Rapid calculation of polar molecular surface area and its
application to the prediction of transport phenomena. 2. Prediction of blood–brain barrier penetration. J. Pharm. Sci. 88 (8), 815–821.
- Ertl P, Rohde B. and Selzer P. (2000). Fast calculation of molecular polar surface area as a sum of fragment-based contributions and its application to the predic- tion of drug transport properties. J. Med. Chem. 43 , 3714–3717.
- Feher M., Sourial E. and Schmidt J.M. (2000). A simple model for the prediction of blood–brain partitioning. Int. J. Pharm. 201 , 239–247.
- Norinder U., Österberg T. and Artusson P. (1997). Theoretical calculation and prediction of Caco-2 cell permeability using Molsurf parametrization and PLS statistics. Pharm. Res. 14 , 1785–1790.
- Österberg T. and Norinder U. (2000). Prediction of drug transport processes using simple parameters and PLS statistics: The use of ACD/logP and ACD/ChemSketch descriptors. Eur. J. Pharm. Sci. 12 ,327–337.
- Cruciani G., Pastor M. and Guba W. (2000). VolSurf: a new tool for the pharmacokinetic optimisation of lead compounds. Eur. J. Pharm. Sci. 11 Suppl. 2, S29–S39.
- Kang Y.S. and Pardridge W.M. (1994). Brain delivery of biotin bound to a conju- gate of neutral avidin and cationised human albumin. Pharm. Res. 11 , 1257–1264.
- Pardridge W.M. (1999). Non-invasive drug delivery to the human brain using endogenous blood-brain barrier transport systems. Pharm. Sci. Technol. Today 2 , 49–59.
- Derossi D, Chassaing G and Prochiantz A. (1998). Trojan peptides: the penetra- tion system for intracellular delivery. Trends Cell. Biol. 8 (2), 84–87.
Research Letters South African Journal of Science 98 , July/August 2002 391
