Partitioning of Solutes in Different Solvent Systems: The Contribution of Hydrogen-Bonding Capacity and Polarity NABILEL TAYAR',RUEY-SHIUAN TSAI', BERNARD TESTA", PIERRE-ALAIN CARRUPT', AND ALBERTLEO* Received March 5, 1990, from 'School of Pharmacy, Universify of Lausanne, B.E.P., CH- 7075, Lausanne, Swifzerland, and *Sewer Chemistry Accepted for publication July 26, 1990. Laboratory, Pomona Co//ege,Claremont, CA 9177 1. Abstract Published partition coefficient values of 121 solutes in five solvent systems (1-octanol-water, nheptane-water, chloroform-water, diethyl ether-water, and mbutyl acetate-water) were correlated with solute properties, namely intrinsic molecular volume (indicator of cavity formation) and the solvatochromic parameters d (dipolarity/ polaritability). p (H-bond acceptor basicity), and a (H-bond donor acidity).While the cavity term and the H-bond accepting capacity played a comparable role in all solvent systems. the H-bond donor acidity was significant only in the alkane-water and chloroform-water systems. Comparison of the regression coefficients of p, and a demonstrated the important role that water content at saturation in the organic solvents plays in the partitioning of solutes. Analysis of the differences between 1-octanol-water and mheptanewater partition coefficients (Alog PM. hep) and between 1-octanol-water and chloroform-water partition coefficients (Alog P,*h,) showed that these values mainly quantitate the capacity of solute to donate hydrogen bonds. In contrast. the differences between 1-octanol-water and diethyl ether-water or mbutylacetatewater partition coefficients, (Alog Poc14eeand Alog PM.ba,respectively) contain no structural information. TO,
In recent years, a solvatochromic comparison method has been used to measure the dipolarity/polarizability (+), hydrogen-bond donor acidity (a),and hydrogen-bond acceptor basicity ( p )of different kinds of solvents.14 These parameters were proven to be useful in evaluating and identifying the physicochemical properties governing aqueous solubilities and partitioning of nonelectrolyte solutes between biphasic solvent systems. A general equation (eq 1) has been proposed14 to correlate and predict the solubility (S,) of solutes in a given solvent system:
where V, is the intrinsic molecular volume of the solutes, m, s, b, and a are the regression coefficients which reflect the relative contribution of each parameter to the solubility of the solutes in agiven solvent system, and S , is the intercept. This equation has provided a quantitative explanation of the two major factors influencing solubility, namely endoergic cavity formation, determined essentially by the solute molecular volume (V,) or solvent-accessible surface area, and exoergic solute-solvent intermolecular interactions as represented in eq 1by the values of #, a,and p for the solute. In eq 1, V,/lOO is used so that the parameter measuring the cavity term will roughly cover the same numerical range (-0.0. . . .1.0) as the other independent variables 1T*, p and a. This makes evaluation of the relative contributions of the various terms of eq 1 to the property S, more reliable. An important established application4.5 of this approach is the prediction of l-octanolwater partition coefficents (log P,,) of nonhydrogen bonding solutes, hydrogen-bond acceptor solutes, and weak hydrogenbond donor solutes: 590 I Journal of Pharmaceutical Sciences Vol. 80, No. 6, June 1997
3.60(kO.09)p + 0.45(k0.06) (n = 103; r = 0.989; s = 0.16;F
=
1972)
(2)
where n is the number of solutes, r is the correlation coefficient, F is the Fisher-test for significance of the equation, and s is the standard deviation of the regression (95% confidence limits are given in parentheses). Recently, Kamlet et a1.6 have extended this equation to hundreds of solutes including strong hydrogen-bond donors. In the present study, we have extended this approach to include the partition coefficients and the difference between these partition coefficients (Alog P values; see Thoretical Section) of a number of solutes in various organic solvents; namely, 1-octanol-water, n-heptane-water, chloroformwater, diethyl ether-water, and n-butyl acetate-water solvent systems. More than 100 solutes, belonging to different chemical classes such as alkanes, ketones, aliphatic alcohols, phenols, aromatic acids, aliphatic acids, anilines, and amines, were selected, thus covering a large range in lipophilicity and hydrogen-bonding capacity (i.e., nonhydrogen bonding solutes, weak and strong hydrogen-bond acceptors, weak and strong hydrogen-bond donors). On the basis of eq 1, we have established a number of equations in order to evaluate the relative contribution of V,, #, p, and a to solute partition coefficients in different solvent systems. In addition, the solvatochromic parameters were used to analyze and to validate the structural information content of Alog P values.
Theoretical Section The lipophilicity of solutes, as expressed by their partition coefficients, is of significance from both physicochemical and biological viewpoints.7-10 Indeed, this property reflects the combined effects of a number of other properties which result from the interaction of a solute with its environment. In turn, partition properties are meant to mimic biological partitioning processes. Traditionally, partition coefficients (log P)are measured in solvent systems made of water and a n immiscible organic solvent. Large compilations of partition coefficient data in different solvent systems are available, as comprehensively documented in the Pomona College Medicinal Chemistry Project log P database." Organic solvents frequently used in partition measurement are traditionally classified as follows:7,12.13 (1)amphiprotic solvents such as 1-octanol and 1-hexanol; (2)hydrogen-bond acceptor solvents such as di-n-butyl ether and n-butyl acetate; (3) hydrogenbond donor solvents such as chloroform; and ( 4 )apolar aprotic inert solvents such as n-heptane and cyclohexane. Obviously, 0022-3549/97/0600-0590$01 .00/0 0 1997, American Pharmaceutical Association
the classification of organic solvents given above implies that they will interact differently with a given solute, resulting in different partition coefficientvalues. In fad, each solvent has its own physical properties such as surface tension, dielectric constant, viscosity, waterlorganic solvent mutual solubility, and hydrogen-bonding capacity. Collander14J5 showed that log P values from different solvent pairs are linearly correlated:
logP1 = alogP2 + b
Results and Discussion Solvatochromic Analysis of Partition Coefficients in Various Solvents SYstemS-For the l-octanol-water system, and in analogy with eq 2, a good m ~ e l a t i o nhas been obtained between solute partition coefficient and solvatochromic parameters (eq 6) [95% confidence limit (CL) in parentheses]:
log Pact
(3)
where the subscripts 1and 2 refer to solvent systems 1and 2, respectively. This equation, however, is only valid when the organic solvents have similar physical properties, in particular, similar hydrogen-bonding capacity. These restrictions have prompted Seiler16 to define the parameter ZH,which was conceived as a measure of the hydrogen-bonding capacity of a given solute. This parameter is calculated as the difference between the 1-octanol-water partition coefficient (log Po,,) and the cyclohexane-water partition coefficient (log Pcyc):
An almost identical approach was proposed by Fujita et al.,17 who compared other solvent pairs using the indicator HB to account for the hydrogen-bonding capacity of the solutes. In a previous study,la some of us have extended this approach to other solvent systems. Recently, the parameter ZH has led to a new physicochemical model in the design of brain-penetrating H, histamine receptor antagonists.19 In addition, the ZH parameter was proven to play a n essential role in accounting for human skin penetration of various compounds.20 The physical interpretation of this parameter, considered as a measure of the-hydrogen-bonding capacity, has never been evaluated bv comoarison with other related oarameters. The purpose of this paper is to contribute a deeper understanding of the physicochemical factors enfolded in the hydrogen-bonding parameter Alog P.
(n = 78; r = 0.960; s = 0.296; F
=
0.09; F
=
13979)
(n = 44; r
(6)
0.951; s = 0.330; F
=
=
75.8)
(7)
logPba = 6.34(+0.80)V1/100+ 0.91(+0.60)# -
(n = 26; r
+
=
0.30(+0.53)a
-
1.52(&1.12)
0.983; s = 0.259; F = 121.6)
(8)
It follows from eqs 7 and 8 that the forces of interactions that determine the partitioning of solutes in both diethyl ether-water and n-butyl acetate-water systems show marked similarities with those in the 1-octanol-water system. Only the ?ic term is statistically nonsignificant in the diethyl ether-water system, and positively correlated with log P (i.e,, it enhances lipophilicity) in the n-butyl acetate-water system (eq 8) compared with the 1-octanol-water system. For the chloroform-water (ch0 solvent system, the following equation was obtained:
logPchf= 6.00(+0.69)V1/IOO- 0.14(+0.40)@ 3.17(+0.49)p
(5)
The A log P values from different solvent systems were calculated in analoev with e a 4 and are reDorted in Table 11.The intrinsic volume of solutes and their solvatochiomic parameters were collected from the literature- (Table 111).Multiple linear regression analysis was performed using SAS22 and QSAR23 programs running on a VAX 8550. No values were omitted in any equation, but only those solutes have been included for which literature values are available.
248.8)
3.83(+0.82)p - 0.20(+0.41)a - 0.34(+0.57)
log Pcpc= 0.99(?0.01) logPSF - 0.04(?0.02) 0,999; s
=
Equation 6, in analogy with eq 2, shows that the regression coefficients of the V,, p, and ?ic terms are statistically significant, while the regression coefficient of the a term is statistically nonsignificant despite the inclusion of strong hydrogen-bond donor solutes. The differences in the goodness of the fit and the regression coefficients of V, and ?r* between eqs 2 and 6, and their larger CL, could be due to the number and classes of compounds used in the analysis. For diethyl ether-water (dee) and n-butyl acetate-water (ba) solvent systems, where the organic solvent is considered a pure hydrogen-bond acceptor, the following equations were obtained:
3.87(+1.82)p
The partition coefficient data set (Table I) was compiled from the Pomona College Medicinal Chemistry Project log P database.11 To verify the reliability of literature logP values, a series of solutes was measured in our laboratory using a centrifugal partition chromatographic (CPC) technique.21 To measure log P values in the heptane-water system, a preparative coil #10 (i.d. 2.6 mm, total capacity 370 mL) was fitted in an Ito Multi-Layer Coil Separator-Extractor (P.C. Inc., Kim Place, Potomac, MD 20854). The rotation speed of the rotor was -1000 rpm. To extend the range of the measured log P values, we have maximized the stationary phase volume (-36:l volume ratio of stationary and mobile phases) and the flow rate was reduced to 0.5-2.0 mL/min. The other experimental details are the same as previously published.21 The measured logP values (log Pcpc)in the heptane-water system was highly correlated with those measured by conventional shake-flask method and reported in the literature (log PF), as seen in eq. 5:
=
5.83(+0.53)V1/100- 0.74(+0.31)# -
3.51(+0.38)/3 - 0.15(*0.23)a - 0.02(+0.34)
Experimental Section
(n = 30;r
=
-
2.99(+0.27)a
-
0.18 (50.43)
(n = 60; r = 0.974, s = 0.300; F = 221.2) H
*'.
~the ~a term ~ is, statistically significant
(9)
with eqs
For the n-heptan*water (hep) solvent system, where the Organic Solvent iS aprotic and apolar, the correlation between solute partition coefficients and their solvatochromic parameters is expressed by the following equation: Journal of Pharmaceutical Sciences I 591 Vol. 80, No. 6, June 7997
Table I-Partitlon Coefficient Values in the Systems Octanol-Water (OCT), mHeptane-Water (HEP), Chloroform-Water (CHF), Dlethyl Ether-Water (DEE), and mButyl Acetatewater (BA). Solute Dichloromethane Chloroform 1,2-Dichloroethane 1,1,1 -Trichloroethane Pentane Acetonitrile Benzonitrile Dimethylsulfoxide Benzaldehyde Acetone 2-Butanone 2-Pentanone 2-Hexanone Acetophenone Methylacetate Ethylacetate Propylacetate Methylbenzoate Methanol Ethanol 1-Propano1 2-Propanol 1-6utanol lsobutanol fert-Butanol 1-Pentanol 1-Hexanol 1-Heptanol Benzyl alcohol Phenol 2-Nitrophenol 3-Nitrophenol 4-Nitrophenol 2-Cyanophenol 4-Cyanophenol 3-Fluorophenol 2-Chlorophenol 3-Chlorophenol 4-Chlorophenol 2-Bromophenol 4-Bromophenol 2-Methylphenol 3-Methylphenol 4-Methylphenol 2-Methoxyphenol 3-Methoxyphenol 4-Methoxyphenol 4-Trifluoromethylphenol 2-lsopropylphenol 2-fert-Butylphenol 4-fert-Butylphenol 2,4-Dimethylphenol 2,5-Dimethylphenol 2,6-Dimethylphenol 3,4-Dimethylphenol 3,5-Dimethylphenol 3,5-Dichlorophenol 2,6-Dichlorophenol 1-Naphthol 2-Naphthol Acetic acid Chloroacetic acid Dichloroacetic acid Trichloroaceticacid 1-Propionic acid 1-6utyric acid 1-Hexanoic acid Benzoic acid 2-Hydroxybenzoicacid
592 i Journal of PharmaceuticalSciences Vol. 80, No. 6, June 7997
log pot? 1.25 1.95 1.48 2.49 3.39 -0.34 1.56 1.35 1.48 -0.24 0.29 0.91 1.38 1.58 0.18 0.73 1.24 2.1 2 -0.77 -0.31 0.25 0.05 -
0.88 0.76 0.35 1.56 2.03 2.72 1.10 1.46 1.79 2.00 1.91 1.61 1.60 1.93 2.15 2.50 2.39 2.35 2.59 1.95 1.96 1.94 1.32 1.58 1.34 2.82 2.88 3.31 3.31 2.30 2.33 2.36 2.23 2.35 3.62 2.75 2.98 2.84 -0.17 0.22 0.92 1.33 0.33 0.79 1.92 1.87 2.26
log Phep 1.62a 1.56’ 2.71 4.18’ 0.90
’
-
1.12’ -0.91 -0.25 0.43 0.87 1.14 -0.26 0.29 0.90 2.13’ -2.80 -2.10 - 1.52 -1 .92’ -0.70 -0.98’ -1.31 -0.40 0.45 1.01 -0.62’ -0.82‘ 1.04” - 1.23‘ -2.15‘ - 1.85’ -2.29’ -0.83’ 0.76‘ -0.07 -0.12’ 1.04 -0.20 0.25 -0.35 -0.19’ 0.36 -0.88‘ - 1.03‘ -0.15’ 0.96’ 2.31 1.17’ 0.34 0.38 0.82 0.28 0.32 0.35’ 1.31a 0.55 0.30 -2.90 -3.14a -2.72a -2.63a -2.14 -0.96 0.24 -0.72 -0.92
’
’
log Pdee
log Pba
-
-
-
-
-
-
log
PChf
-
-0.22
2.71
-
0.24 -
1.74 -0.21 -
-
2.79 1.16 1 .80 2.56 2.17 - 1.26 -0.85 -0.40 -0.35 0.34 0.34
-
1.05 1.69 2.41 0.39 2.35 0.60 0.20
-
-
1.75 0.43 0.93 -1.15 -0.50 -0.02 -0.19 0.61 0.65 0.34 1.20 1.a0 2.40 0.96 1.64 2.18 2.18 2.01 -
-
-
1.36 1.02 1.01 1.64 1.07 1.23 0.89 1.06 1.70 0.94 0.23
2.05 2.10 2.22 -
1.50 1.59
-
-
-
-
1.60
-
1.82 1.74 - 1.52 -1.35 -0.89 -0.69 -0.96 -0.27 1.15 0.56 0.58
-
1.80
-
1.44 2.08 1.47 -
-
2.53
-
-
-
-
- 1.74 -1
.oo
-0.23 0.32 -
-
0.85 1.60 1.90
-
1.69 2.30 2.42 2.32
-
2.24 2.68 2.51 2.46 2.52 2.24 2.13 2.31 1.72 1.83 1.55 -
-
2.51 2.52 2.63 2.45 2.51 -
2.53 1.77 3.11 -0.35 -0.40 0.420 1.46 1.63 0.27 0.09 0.61 0.69 1.97 1.A9 2.53 2.26 (Continued on next Page)
Table Hontlnued
Solute 4-Hydroxybenzoicacid 3-Nitrobenzoic acid 4-Nitrobenzoic acid
2-Methoxybenzoic acid 4-Methoxybenzoicacid 2-Bromobenzoic acid 5,5-Diethylbarbituricacid Thiopental Pentobarbital Hexobarbital Phenobarbital Secobarbital Succinimide
Benzamide Acetanilide Aniline NMethylaniline N,NDimethylaniline 2-Nitroaniline 3-Nitroaniline 4-Nitroaniline 3-Chloroaniline 4-Chloroaniline
2-Toluidine 4-Toluidine Pyridine
lndole Ethylamine 1-Propylamine 1-Butylamine Benzylamine
Benzene Anisole Nitrobenzene Fluorobenzene Chlorobenzene
Bromobenzene lodobenzene Toluene Biphenyl
Sulfathiazole Sulfamethizole
Sulfaethidole Sulfisoxazole Sulfisomidine Amphetamine Methamphetamine
Nicotine Ephedrine
Aminopyrine Atenolol Propranolol a
- 1.82 - 1.22
1.58 1.83 1.89 1.59 1.96 2.20 0.65 2.59 2.07 1.49 1.47 1.97
-
2.08 -0.15
-
-2.15 0.52 - 1.22' -0.57' -2.75' -0.99'
-
-
-2.28 -1.70 0.03' l.llb 2.40' 0.31' -0.46' - 1.09' 0.71 0.57 0.47 0.44 -0.31 " 0.68 - 1.77 - 1.oo -0.62 -0.21 2.29 2.15' 1.55" 2.46' 2.99" 3.12 3.33 2.89' 4.00 -4.60 -3.83 -2.54 -3.57 -3.85 0.73 1.24 0.03 -0.77 -0.68 - 1.28 1.48
0.64 1.16 0.90 1.66 2.31 1.85 1.37 1.39 1.88 1.88 1.32 1.39 0.65 2.14 -0.13 0.48 0.97 1.09 2.04 2.1 1 1.85 2.27 2.89 2.99 3.25 2.73 4.09
0.05 0.54 1.01 1.01 -0.40 1.76 2.07 1.17 0.93 1.oo 0.16 3.37
0.41 0.86 2.53 0.90 0.91 0.45 2.22 1.38 2.18 0.62 1.98 -1.27 0.1 1 0.78 1.26 1.75 2.26 2.13 1.61 1.23
-
-
1.96 1.92 1.32
-
1.42
-
0.78
-
-
2.36
-
-
0.68 0.88 1.28 1.48 1.70 2.48 - 1.42 -0.22 0.65 0.85
-
1.95 1.71 1.48
-
0.08
-
-0.35 0.26 0.56 1.18 2.80 3.12 2.93 2.85 3.46 3.61 3.70 3.41
-1.18 -0.54 0.1 1 0.32
-0.73 -0.03 0.09 0.74 -0.52 2.20 0.85 1.89 1.10 1.86 -0.13 2.97
-0.72
-
1.36
1.97
-
2.46
-
1.91 -1.06
-
1.46 1.08 0.41 -0.20 -2.00 2.46
-
-
2.00 1.80 1.80
-
1.25
-
-
1.66 0.45
-
-
-
-0.28 0.34 0.90 1.35 -0.40
-
0.78 0.56
-
0.07 0.50
mHexanewater log P. Estimated values according to Sei1er16. 'Measured in this study by CPC;'l average SD