Measurement of Lipophilicity

The partition coefficient was first defined in 1872 by Berthelot and Jungfleisch [3], who wrote "On the Laws that Operate for the Partition of a Substance between two Solvents". It was first used to correlate and explain the potencies of biologically active substances at the turn of the century, by both Meyer [4] and Overton [5] in their studies of narcotic compounds. Overton's work stimulated other investigations of the use of partition coefficients for biological correlations, among them a study by Seidell [6] in 1912. Believing that the partition coefficient of thymol might be relevant to a study of the mode of action of thymol against hookworm, Seidell made measurements using a variety of lipid phases, including olive oil, castor oil, peanut oil, and linseed oil. In those days, measurement was particularly tedious: it was necessary to separate thymol from the oil by a steam distillation, and then to estimate thymol in water by treatment with bromine, titrating the resulting hydrobromic acid produced!

With the development of UV spectroscopy, measurement of the partition coefficient for compounds with strong absorption, nonextreme values, and sufficient solubility in the aqueous phase has become routine, using the "shake-flask" method, partitioning between one of a wide variety of lipid phases, and water or an appropriate buffer solution as the aqueous phase. For many ionizable compounds, compounds of low solubility, and compounds with low UV absorbance or extreme values of partition coefficient then special methods of measurement or alternative lipophilicity parameters have had to be devised.

In 1959, Gaudette and Brodie [7] realized both the possibility for using a partition coefficient to model lipophilic character, and the relevance of lipophilicity to pharmacokinetic processes. They found a parallel between the heptane/buffer partition coefficients of certain drugs, and their rate of entry into cerebrospinal fluid. However, generalised use of logP as a lipophilicity parameter did not come about until after 1964, with the Hansch octanol/water system remaining to this day the standard for both experimental and theoretical investigations. In 1971, Leo, Hansch and Elkins [8] published the first comprehensive review of partition coefficients, with a tabulation of nearly 6000 values, including their own measurements on some 800 in the octanol/water system. The review incorporated an account of the shake-flask method of measurement, which was discussed more exhaustively in a 1973 monograph by Purcell, Bass and Clayton [9].

Octanol/water logP has also been measured by high-performance liquid chromatography [10], and by using a filter-probe to sample selectively from the aqueous or lipid phase so there is no need to fully separate the phases [11, 12], For ionizable compounds, Brandstrom in 1963 [13] was first to use a potentiometric titration technique. One aqueous phase titration, with a pH-meter probe, was carried out in the aqueous phase to determine pKa. A second titration was carried out in the presence of octanol, when partition occurred and the pATa shifted. The difference in pK:i was related to logP. In 1974 Seiler [14] modified this technique so as to determine pKn and logP from a single titration. The technique has now been refined to enable not only simultaneous pKa and log/3 determination, but to allow treatment of substances with multiple ionization constants, ion-pair partitioning, and self-association reactions leading to the formation of oligomers [15, 16].

CH2OCOC8H17 CHOCOC8HI7 Me

PGDP

(j;H2OCOCnH2n+1 CHOCOCnH2n+1

CH2OPOCH2CH2NMe3 O

phospholipid

Figure 1. Structural similarities between propylene glycol dipelargonate (PGDP) and a phospholipid molecule.

Lipophilicity has, since 1964, been traditionally measured in the octanol/water system. However, for particular purposes and for particular sets of compounds, other solvent pairs have been used. Octanol/water values have been shown to be generally satisfactory for modeling serum protein binding and for modeling lipophilic interactions with biological membranes consisting largely of protein, but for other types of membrane then a different solvent system might be more appropriate. In 1989, Leahy et al. [17] suggested that membranes (or receptors) could exist with very different hydrogen bonding characteristics from those of octanol. Thus, membranes may contain neither acceptors nor donors (modeled by an alkane); or contain largely amphiprotic groups (as in a protein, modeled by octanol); largely proton donor groups (which may be modeled by chloroform); or largely proton acceptor groups (as in a phospholipid membrane). Leahy argued for the use of propylene glycol dipelargonate (PGDP) as lipid phase to model phospholipid membranes (Fig. 1) and have accordingly measured many partition coefficients in the PGDP/water system [18].

For many compounds, the traditional equilibrium method of partition coefficient measurement may be impossible, impractical, or inappropriate. As a practical alternative to logP, particularly for biological correlations, much use has been made of parameters derived from chromatographic retention. In 1941 Martin and Synge [19] showed that for reversed phase thin-layer chromatography, Eq. (1) relates partition coefficient, P, to the ratio, Rf, of distances moved by the compound spot and the solvent front in a given time, with K being a constant for the system. In 1950 Bate-Smith and Westall [20] defined the parameter Rm as in Eq. (2) from which Eq. (3) follows. In practice, excellent correlations have been found between Rm and logP taking the form of Eq. (4). Kaliszan [21] has reviewed the use of lipophilicity parameters derived from HPLC, TLC, and paper chromatography.

0 0

Post a comment