Measurement and prediction of hydrophobicity parameters for highly lipophilic compounds: Application of the HPLC column-switching technique to measurement of log P of diarylpyrazines

Article abstract of DOI:10.1021/js990112s

In the preparatory stage of structure-activity relationship (QSAR) studies of anti-platelet aggregant pyrazine derivatives, log P values (P: 1- octanol/water partition coefficient) of diarylpyrazines were measured by a newly developed HPLC column-switching technique. The system consists of two processes: (1) adsorption of the sample at the top end of a short precolumn, and then (2) quantifying the enriched analyte by a conventional analytical column. By using the log P values thus obtained, the correction factor for the steric hindrance caused by the vicinal diphenyl groups was estimated. The log k values (k; retention factor) were also measured with methanol-buffer (pH 7.4) eluents and related to log P. The eluent of 50% methanol content (M50) gave a good linear relationship over a wide range of log P (-0.3< log P < 5.2), indicating that log k(M50) parameter is useful for predicting the log P value.

Full text of DOI:10.1021/js990112s

Measurement and Prediction of Hydrophobicity Parameters for Highly
Lipophilic Compounds: Application of the HPLC Column-Switching
Technique to Measurement of log P of Diarylpyrazines
,†
CHISAKO YAMAGAMI,* KOZUE ARAKI,† KYOKO OHNISHI,† KAORU HANASATO,† HARUKO INABA,‡
‡
MASAHIRO AONO,‡ AND AKIHIRO OHTA
Contribution from Kobe Pharmaceutical University, Motoyamakita-machi, Higashinada, Kobe, 658-8558, Japan, and
Tokyo University of Pharmacy and Life Science, Horinouchi, Hachioji, Tokyo,192-0392, Japan.
Received April 12, 1999. Accepted for publication September 20, 1999.
log P values for multisubstituted compounds might be
expressed as:
Abstract 0 In the preparatory stage of structure−activity relationship
(QSAR) studies of anti-platelet aggregant pyrazine derivatives, log P
values (P: 1-octanol/water partition coefficient) of diarylpyrazines were
measured by a newly developed HPLC column-switching technique.
The system consists of two processes: (1) adsorption of the sample
at the top end of a short precolumn, and then (2) quantifying the
enriched analyte by a conventional analytical column. By using the
log P values thus obtained, the correction factor for the steric hindrance
caused by the vicinal diphenyl groups was estimated. The log k values
(k; retention factor) were also measured with methanol-buffer (pH 7.4)
eluents and related to log P. The eluent of 50% methanol content
(M50) gave a good linear relationship over a wide range of log P
(−0.3< log P < 5.2), indicating that log kM50 parameter is useful for
predicting the log P value.
log P(YRX) ) log P(HRH) + πX + πY
(2)
Although many other empirical ways for log P calcula-
tions have been proposed subsequently,5-8 most of them
essentially assume such additivity. In accord with expecta-
tion, eq 2 fails in solute systems which involve electronic
and/or steric interactions between the substituents, and
various correction methods are proposed so that the
predicted log P values may approach the experimental
results.6,7
In HPLC approaches, the logarithm of the retention
factor, k, is used as a hydrophobicity parameter. Usually
C-18 bonded silica is used as a stationary phase with
methanol-water mixtures as the mobile phase, and a
Collander type relationship as shown by eq 3 is assumed.9
log P ) a log k + b
(3)
Introduction
Since the log k value depends on the mobile phase
composition, extensive studies have been made to find
optimal mobile phase conditions producing a good linear
relationship.10-15 Many investigators have used the log kW
parameter, (log k extrapolated to 0% methanol) to eliminate
organic solvent effects.10,11,16,17 On the other hand, other
reports use isocratic log k values at an appropriate metha-
nol concentration.13-15,18,19 It is very often observed that
hydrogen bonding effects of solutes with the surrounding
molecules break a linearity, making the HPLC approach
less effective for the estimate of log P. In an attempt to
establish a standard HPLC procedure for predicting reli-
able log P values, we have studied the relationship between
log P and log k for various series of fundamental heteroaro-
matic compounds, such as monosubstituted pyridines,
diazines, and thiophenes, under various HPLC condi-
tions.13-15,20 Our results have demonstrated that hydrogen
donating solutes (amphiprotics) usually behave differently
from nonamphiprotics and should be treated separately.13-15
Moreover, the log kW method tends to overestimate log P
for strong hydrogen acceptors. For this reason, most of
pyridines and diazines which have N-atom(s), a strong
hydrogen acceptor, in the parent rings have been found to
yield values for log kW much higher than that corresponding
to the true value for log P. Instead, we have found that
isocratic log k obtained with an eluent containing around
50% methanol usually results in a good linear relationship
with log P, permitting us to reliably predict values for log
P over the range, -1 < log P < 3.13-15,20
The logarithm of 1-octanol/water partition coefficient, log
P, has been used as a measure of hydrophobicity of organic
compounds and has played an important role in structure-
activity relationship studies.1,2 A conventional procedure
for measurement of log P is the shake-flask method which
consists of determining the equilibrium concentration of a
substance in the aqueous or in both the aqueous and the
octanol phases. This method, however, is not usually
applicable to highly lipophilic compounds (log P > 3)
because of extremely low concentrations of test compounds
in the aqueous phase. To solve this problem, some proce-
dures, such as re-extraction of the aqueous phase before
the measurement of concentrations and the use of radio-
labeled solutes, have been used to get a satisfactory
response with the detector used.3
In addition to direct measurements, the log P value is
often estimated by calculations or by retention parameters
derived from reversed-phase HPLC, hereafter referred to
as the HPLC method. The most widely used methodology
for calculating log P was first proposed by Fujita et al.,
being based on an additive-constitutive, free energy related
property of log P.4 Thus, by using π defined as the
difference in log P between a derivative with a given X
substituent (RX) and the parent compound (RH),
πX ) log P(RX) - log P(RH)
(1)
* Corresponding author. Telephone: +81-78-441-7547. Fax: +81-
78-435-2080. E-mail: yamagami@kobepharma-u.ac.jp.
† Kobe Pharmaceutical University.
Even when such empirical predictive methods are known
to be reliable, accurate experimental values of log P for
several compounds are required.
‡ Tokyo University of Pharmacy and Life Science.
© 1999, American Chemical Society and
American Pharmaceutical Association
10.1021/js990112s CCC: $18.00
Published on Web 10/22/1999
Journal of Pharmaceutical Sciences / 1299
Vol. 88, No. 12, December 1999

Table 1sLog P Values of (Di)arylpyrazines
In the course of our QSAR studies of bioactive com-
pounds, we were interested in analyzing the structure-
activity relationship of a series of diaryl pyrazines which
has recently been found, by one of the authors, to have
potent anti-platelet aggregation activity.21 Knowing reliable
log P values of these compounds was a prerequisite for this
purpose. We, therefore, undertook measurement and evalu-
ation of log P values for these compounds. The conventional
shake-flask method failed not only because of their high
lipophilicity but also of very low solubility of some com-
pounds even in octanol. Although we were well experienced
in predicting the log P value of pyrazines with log P < 3
by the HPLC method, it seemed dangerous to apply our
relationship to a wider lipophilicity range (log P > 4).
Accordingly, we attempted to develop a method to deter-
mine experimentally log P values for highly lipophilic
compounds by using the HPLC column-switching system.
The column-switching technique is widely used for on-line
sample enrichment and cleanup, and has been increasingly
applied to environmental analyses and the isolation of
clinical samples.22,23 We expected that this method would
be applicable to our measurements of log P by loading large
volumes of the sample solutions onto a precolumn and
succeeded in obtaining log P values ranging from 3.5 to
5.2.
a
b
R
1
R
2
R
3
R
4
log P
log P
add
1
Ph
H
H
H
H
H
Me
OMe
OEt
Cl
i-Pr
H
Me
Me
2-OMe-Bn
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Me
H
Ph
Et
Ph
Ph
2.07
2.24
3.19
3.52
4.21
4.73
4.05
4.71
4.10
3.75
2.52
5.18
5.20
3.42
3.66
4.22
4.30
4.47
5.00
3.70
3.41
4.10
4.28
4.29
4.94
4.66
4.83
c
2
3
4
5
6
4-OMe-Phd
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
4-Me-Ph
4-F−Ph
4-CN−Ph
Ph
4.4
4.9
5.4
5.9
5.4
7e Ph
8
Ph
9e 4-Me-Ph
10
4-F−Ph
11e 4-CN−Ph
f
12
13
Ph
Ph
f
Ph
4-OMe-Ph
3-OMe-Bn
H
14e 4-OMe-Ph
4.7
5.2
5.7
5.7
5.7
6.3
5.0
4.8
5.2
4.9
4.9
6.3
5.4
6.3
15
16
17
18
19
20
21
22
4-OMe-Ph
4-OMe-Ph
4-OMe-Ph
4-OMe-Ph
4-OMe-Ph
4-OMe-Ph
4-OMe-Ph
4-OMe-Ph
4-OMe-Ph Me
4-OMe-Ph Et
4-OMe-Ph Cl
4-OMe-Ph OMe
4-OMe-Ph OEt
4-OMe-Ph CN
4-OMe-Ph COOMe
4-OMe-Ph Me
23g Ph
Me
Me
Et
Ph
H
Ph
H
24
Ph
25h Ph
i
26
Ph
Ph
Et
j
27
Et
Et
a Measured by the column-switching method except for 1 and 2. b Sum of
log P of pyrazine (−0.26) and ∑ pyrazine-π (R , R , R , R ). c Superscript of
1
2
3
4
the compound number identifies the reference for the preparation: Ohta, A.
et al. J. Heterocycl. Chem. 1982, 19, 1061−1067. d 4-X-Ph refers to 4-X-
substituted phenyl. e Ohta, A. et al. J. Heterocycl. Chem. 1982, 19, 465−473.
f 2-OMe-Bn and 3-OMe-Bn refer to 2-OMe-benzyl and 3-OMe-benzyl,
respectively. g Ohta, A. et al. Heterocycles 1984, 22, 2317−2321. h Ohta, A.
et al. Heterocycles 1986, 24, 785−792. i Ohta, A. et al. Heterocycles 1987,
26, 2449−2454. j Ohta, A. et al. J. Heterocycl. Chem. 1983, 20, 311−320.
This report describes a novel procedure to measure log
P values above 3 by using HPLC with a column-switching
system. The log P values thus obtained were evaluated in
terms of the additivity of π values and also related to log
k values to examine how the log P values might be
predicted by the HPLC approach. For comparison, some
pyrazine N-oxides (PRfO) were also studied.
Table 2sLog P Values of Variously Substituted Pyrazines (PR) and
Pyrazine N-Oxides (PRfO)
a
a
substituents
PR
H
Me
Et
OMe
OEt
OPr
COOMe
COOEt
2,3-diMe
2,6-diMe
2-Me,3-Et
2-Me,3-Pr
log P
substituents
2-Me,3-Bu
2-Me,3-iBu
2,3-diEt
triMe
2,3-diEt,5-Me
tetraMe
log P
Materials and Methods
40
41
42
43
44
45
46b
2.10
1.96
1.51
0.95
1.95
1.28
2.40
28
29
30
31
32
33
34
35
36
37
38
39
−0.26
0.21
0.69
0.73
1.28
1.84
−0.23
0.28
0.54
0.54
1.07
1.57
Ma ter ia lssCompounds used in this study are given in
Tables 1 and 2. Most of the diaryl pyrazines in Table 1 (1,
3, 4, 8, 10, 12, 13, 15-17, 20-22, and 24) and monosub-
stituted pyrazines (28-35) in Table 2 were prepared as
previously described.21,24 Compounds 5, 6, 18, and 19 were
prepared by treatment of the 5-Cl derivatives with an
appropriate sodium alkoxide according to the method for
preparing the corresponding monosubstituted pyrazines.24
Alkylpyrazines (36-45) were purchased from Tokyo Kasei
Organic Chemicals (Tokyo, J apan). The references for
preparation of the other compounds are given in the
footnotes to Tables 1 and 2. HPLC-grade methanol and
water, 1-octanol of the grade for log P measurements, and
a phosphate buffer solution (pH 7.4) were purchased from
Nacalai Tesque.
Mea su r em en t of log P : Con ven tion a l Sh a k e-F la sk
Meth od sThe log P for compounds with medium or low
hydrophobicity (log P < 3) was measured by a conventional
shake-flask method. The concentration of samples in both
of the water and octanol phases was determined by HPLC.
Colu m n -Sw itch in g Meth od sSamples (0.5-30 mg)
were dissolved in 1 mL of octanol and then shaken with
100 mL of distilled water. The phases were allowed to
separate overnight at 25 °C. Persistent emulsions were not
observed to occur with our compounds. After pipetting off
the octanol phase, the aqueous phase was collected into
centrifuge tubes, preventing contamination by octanol, and
2,5-diMe-3,6-diCl
PRfO
2,5-diMe-3Cl
c
47
0.44
0.19
1.08
3.66
3.41
48d
2,5-diMe-6Cl
49e
2,5-diMe-3,6-diCl
2,3-bis(4-Cl−Ph),5-Cl
2,3-bis(4-Br−Ph)
f
50
f
51
a The data for compounds 28−35 and 36−45 were taken from refs 24 and
29, respectively. Those for the others were measured in this work. b Superscript
of the compound number identifies the reference for the preparation: Baxter,
R. A. et al. J. Chem. Soc. 1948, 1859−1862. c Ohta, A. et al. Chem. Pharm.
Bull. 1979, 27, 2027−2041. d Ohta, A. et al. J. Heterocycl. Chem. 1981, 18,
555−558. e Ohta, A. et al. J. Heterocycl. Chem. 1982, 19, 781−784. fOhta,
A. et al. J. Heterocycl. Chem. 1982, 19, 465−473.
then centrifuged twice for 5-10 min. As the solute con-
centration of lipophilic compounds in the aqueous phase
was too low to detect by HPLC, sample enrichment on a
precolumn was performed before conventional quantitative
analysis by using a column-switching technique as follows.
HPLC SystemsThe column-switching system (Figure 1)
consisted of two isocratic HPLC pumps (LC 9A, Shimadzu,
Kyoto), an autoinjector (IS-25, Kyoto Chromato, Kyoto,
1300 / Journal of Pharmaceutical Sciences
Vol. 88, No. 12, December 1999

Figure 1sSchematic representation of the HPLC column-switching system. (A) Sample enrichment process. (B) Analytical process.
J apan) fitted with a 10 mL sample loop, a 10 mL syringe
and a Rheodyne 7125 injection valve, and a universal valve-
switching module ACP 1010 with a Rheodyne 7000 six-
port, two-position valve (Anachem, Luton, England). The
following columns were connected to the six-port valve.
Column I (precolumn for concentration): Capcell Pak C18,
UG 80 (Shiseido, Tokyo, J apan), 35 × 4.6 mm i.d. Column
II (analytical column): Cosmosil 5C18-AR-II (Nacalai
Tesque, Kyoto, J apan). 150 × 4.6 mm i.d.
a 200 µL aliquot of the octanol phase was diluted with
methanol to 1.00 mL. The methanolic solution thus ob-
tained was then diluted with pure water by an appropriate
factor as described below. The resulting homogeneous
solution was analyzed quantitatively in the same manner
as in the analysis of the aqueous phase. Injection volume
was the same as that for the aqueous phase. The second
dilution with water was done at two dilution factors so that
the resulting peak areas, Ao (1) and Ao (2), obtained as
described above would be Ao (1) < AW < Ao (2) and also the
ratio of Ao (2)/Ao (1) would be 2-5. With these data, the P
value was calculated as P ) fAo/ AW, where f represents
the overall dilution factor. Agreement of log P from two Ao
values, Ao (1) and Ao (2), was usually obtained to within
0.05 log unit. The mean value was taken as the P value.
For each compound, measurements were carried out at
least in duplicate, and the P values thereby obtained were
reproducible to within (0.1 log unit for most cases.
Column-Switching HPLC AnalysissA 5-9 mL aliquot
of the centrifuged aqueous phase, giving an appropriate
peak area on the final chromatogram, was automatically
injected onto column I by pump A using pure water as the
mobile phase at a flow rate of 5.0 mL/min (at position A,
Figure 1). This process allowed the solute to be adsorbed
at the top end of the column I. After 4 min, the valve was
switched to the position B (Figure 1), and the preconcen-
trated (adsorbed) analyte was eluted to the analytical
column II (back flush mode) with an aqueous solution
containing 70-90% methanol delivered by pump B. The
signals were analyzed by a photodiode array detector SPD-
MV10A (Shimadzu, Kyoto, J apan). Eluent compositions
were chosen so that the analytical time would be less than
10 min for each sample. The average peak area of a given
sample for the aqueous phase, AW, was obtained from
repeated 5-7 cycles. To analyze the corresponding octanol
phase under comparable conditions to those for the aqueous
phase, the following two-step dilution was performed. First,
Mea su r em en t of log ksThe log k value was deter-
mined as previously described with a Capcell Pak C18
column (50 × 4.6 mm i.d., Shiseido, Tokyo, J apan) at 25
°C.13 Methanol-0.01 M phosphate buffer (pH 7.4) mixtures
containing 50, 60, and 70% MeOH (hereafter designated
as M50, M60, and M70, respectively) prepared by volume
were used as the eluent. Log k parameters were determined
by k ) (tR - t0)/t0, where, t0 and tR are retention times of
methanol and a given analyte, respectively.
Journal of Pharmaceutical Sciences / 1301
Vol. 88, No. 12, December 1999

Table 3sPyrazine-π (πPR) and Benzene-π(πPhX
)
Table 4sSubstituent Effects of R
3
a
b
X
πPR
πPhX
compound
R ,R
compound
log
obsd
log
pred
log
obsd
log
b
a
F
0.55
0.96
1.19
0.47
0.95
0.99
1.54
0.25
0.03
0.54
0.14
0.71
0.86
0.56
1.02
−0.02
0.38
−0.57
−0.01
0.51
1.96
−
R
3
P
P
R
3
R ,R
P
P
pred
1
2
1
2
Cl
Br
Me
Et
4
5
6
7
Me
Ph 3.52 3.66 15 Me
4-OMe-Ph 3.66 3.89
4-OMe-Ph 4.22 4.37
4-OMe-Ph 4.30 4.38
4-OMe-Ph 4.47 4.41
4-OMe-Ph 5.00 4.96
4-OMe-Ph 3.70 3.67
OMe Ph 4.21 4.18 16 Et
OEt
Cl
Ph 4.73 4.73 17 Cl
Ph 4.05 4.15 18 OMe
19 OEt
OMe
OEt
CN
20 CN
21 COOMe 4-OMe-Ph 3.41 3.45
COOMe
COOEt
Ph
a Sum of log P(3) and pyrazine-π(R ). b Sum of log P(14) and pyrazine-
3
c
2.33
π(R ).
3
c
4-MeO-Ph
2.50
We also examined the substituent effect of R3 on log P
a Taken from ref 24. b Taken from ref 2. c This work.
when R3 is introduced to 2,3-diarylpyrazines, 3 and 14, in
terms of the ortho-diaryl effect discussed above and pyra-
zine-π values. As the R3 substituent hardly suffers from
steric effects, we may expect normal substituent effects in
these pyrazines. The sum of log P of 3 or 14, and pyrazine-π
of R3 was calculated (log Ppred.) and compared with the
experimental log P. As shown in Table 4, they were in fairly
good agreement. It is not unexpected that the deviation
from the experimental value is somewhat larger than the
expected experimental error in some compounds because
some additional minor corrections for the electronic inter-
actions between the substituents and the ring N-atoms as
well as for the steric effect between R2 and R3 should be
taken into account for detailed analysis.24 These results
revealed that the pyrazine-π is very helpful in making a
rough estimate of log P of even highly lipophilic pyrazines,
provided severe steric effects are not involved. It should
be emphasized that the use of benzene-π instead of pyra-
zine-π for the substituent on the pyrazine ring would lead
to erroneous estimated values.
Comparisons of log P for a series of variously substituted
alky-diphenylpyrazines (4 and 23-27) indicates that steric
repulsion by the vicinal alkyl and phenyl substituents also
requires a correction of -0.6 to -0.7 in log P. As the
corresponding reduction in log P in the vicinal dialkyl
substituents is very small (∼-0.15),29 large correction
factor suggests twisting of the benzene ring even in this
case. Nonplanar conformations were also indicated by the
MO calculations. Measurements and more quantitative
analyses of log P for a larger size data set including more
lipophilic compounds are now in progress.
Results and Discussion
log P of Dia r ylp yr a zin essBefore measuring log P
values for our compounds, we attempted to determine the
log P value of curcumin with our column-switching system.
Curcumin has attracted our interest in the previous
structure-activity relationship study,25 and its experimen-
tal log P value is reported to be 4.26.26 We obtained a value
of 4.28, very close to 4.26, indicating that this newly
developed procedure works satisfactorily. Therefore, we
applied this method to determining log P values of diaryl
pyrazines (Table 1). Inspection of Table 1 shows that
introduction of two hydrophobic aryl substituents increases
extensively the hydrophobicity of molecules. We have
previously shown24 that the π value in the pyrazine system
(Table 3), derived from the log P values of monosubstituted
pyrazines and pyrazine itself, often differs from the widely
used benzene-π values and, therefore, that pyrazine-π
values should be used in the analysis of log P values of
pyrazines. By using pyrazine-π values, where available,
estimated log P values, log Padd, were calculated on
assumption of the additivity, that is, the sum of log P of
pyrazine (-0.26) and pyrazine-π values of the substituents
(Table 1).
It is noteworthy that experimental log P values of all
compounds in Table 1 were much lower than log Padd
values. In particular, in 2,3-diarylpyrazines, 3-7 and 14-
22, the difference was over one log unit. Steric hindrance
between the two benzene rings may contribute to reducing
the hydrophobicity of molecules (ortho-diaryl effect). This
argument would also be supported by comparing the log P
values of the three isomeric diphenyl-methylpyrazines (4,
23, and 24). Among them, only the isomer 4 having two
adjacent phenyl groups gave a log P value (3.52) much
smaller than those for 23 (4.28) and 24 (4.29), which have
two phenyl substituents at 2,5- and 2,6-positions, respec-
tively. An X-ray analysis of 2,3-diphenylpyrazine 3 has
demonstrated that the dihedral angles between the two
phenyl substituents and the pyrazine ring are approxi-
mately 50° and 40°.27 Our semiempirical MO calculations
of 2,3-diphenylpyrazine by the AM1 method28 (the MOPAC
93 program package incorporated in an ANCHOR II
modeling system, Fujitsu, J apan) gave a similar result.
Such nonplanar conformation would cause resonance in-
hibition, decreasing the degree of conjugation and conse-
quently reducing hydrophobicity below expectations for a
molecule with full delocalization of π-electrons. The use of
the additivity rule overestimated log P for 3 by 1.2 and
that for 14 by 1.3, suggesting that the steric hindrance
between the two vicinal benzene rings makes a negative
contribution of about -1.2 to log P. This finding conforms
to a well-known phenomenon that an extension of the
resonating system requires a positive correction factor.2
Log k: P r ed iction of log P by th e HP LC Ap p r oa ch s
Our previous studies on several series of monosubstituted
heteroaromatic compounds have revealed that the relation-
ship between log P (-1 < log P < 3) and log k obtained
with various compositions of methanol-water eluents
usually require some correction terms for hydrogen-bond-
ing effects and can be expressed by the following general
equation for nonamphiprotics:30
log k ) a log P + FσI + sSHA + const
(4)
In this equation, the σI parameter represents Charton’s
electronic substituent constant,31 and the SHA parameter
is that we have recently proposed as the hydrogen-
accepting indicator for the substituent.30 In general, con-
tributions of the two correction terms are of minor impor-
tance when the eluent contains 50% MeOH (M50), so that
we could assume the log k-log P linearity. However, these
contributions are found to increase with decreasing metha-
nol content. Strong hydrogen-accepting substituents, such
as COOR and CONMe2, deviate substantially from the line
formed by non-hydrogen bonders and weak hydrogen
acceptors in the plot of log k against log P. If such a
predictive capability of M50 eluent holds for compounds
1302 / Journal of Pharmaceutical Sciences
Vol. 88, No. 12, December 1999

to that for nonamphiprotics in the log k-log P plot.13,15 This
phenomenon is unavoidable because octanol is more hy-
drogen accepting than the stationery phase. Finding an
effective procedure for treating these is still to be investi-
gated.
Conclusion
In this study, we developed a novel procedure for
measuring the log P value of highly lipophilic diaryl
pyrazines by the HPLC column-switching technique. This
method enabled us to measure the extremely low concen-
tration of solutes by direct injection of a large volume (5-9
mL) of aqueous phase onto a precolumn. The log P values
thus obtained ranged from about 3 to 5.2. By using these
data, a large correction factor of about -1.2 was estimated
as the o-diphenyl effect, which was thought to be attribut-
able to the loss of coplanarity. The effects of sterically
unhindered substituents attached to the pyrazine ring on
log P could be well estimated by using the pyrazine-π value
obtained in our earlier work.24 The log k values were also
measured with different compositions of methanol-buffer
(pH 7.4) eluents, and a good linear relationship between
log P and log k was found to hold at 50% MeOH, in
conformity with our previous finding,13,14 over the log P
range from -0.3 to 5.2. To our knowledge, there have been
no previous studies where log k values for a set of
compounds determined under exactly the same HPLC
conditions have been related to experimentally measured
log P values over such a wide range. The present result
seems to afford a nice example demonstrating that the use
of the kM50 parameter provides reliable and practical
method to predict quickly the log P value.
Figure 2sRelationship between log k and log P for variously substituted
pyrazines. Open symbols represent (di)arylpyrazines, and closed symbols
represent the other pyrazines.
with much higher log P values, the log kM50 approach would
be of practical value in prediction of log P. Accordingly, we
measured log k values for the present compounds under
the same conditions and studied the relationship between
log P and log k. To make the results comparable with those
previously studied, we selected from many of the related
compounds studied earlier, typical compounds including
esters and N-oxides which were previously classified as
deviants in water rich eluents.13 Some N-oxides of diaryl-
pyrazines were also added to the data set. Compounds used
for log k measurements in addition to the (di)arylpyrazines
in Table 1 are listed in Table 2 with their log P values. It
should be noted that the compounds studied in this work
are non-hydrogen bonding or hydrogen accepting. Log k
values of all the compounds in Tables 1 and 2 obtained with
the M50 and M70 eluents are plotted against log P in
Figure 2. The M50 eluent again produced a satisfactory
linear relationship of the form shown in eq 5:
Development of procedures for obtaining reliable log P
values for lipophilic compounds has become increasingly
important not only in QSAR studies of bioactive compounds
but also in environmental chemistry and toxicology where
the log P is used as a parameter for evaluating the fish
bioconcentration and toxicity.32-34 Our present study would
be also expected to be helpful for research in such fields.
References and Notes
1. Hansch, C.; Fujita, T. F-σ-π Analysis. A method for the
correlation of biological activity and chemical structure. J .
Am. Chem. Soc. 1964, 86, 1616-1626.
log kM50 ) 0.554 log P - 0.384
n ) 51, r ) 0.996, s ) 0.085
(5)
2. Hansch, C.; Leo, A. Exploring QSAR-Fundamentals and
Applications in Chemistry and Biology; American Chemical
Society: Washington, DC, 1995; pp 125-168.
The relationship for the M60 eluent was somewhat
inferior to eq 5 (the data not shown). As the data set
contains molecules having strongly hydrogen-accepting
substituents and/or a sterically hindered conjugated sys-
tem, finding a simple linear equation covering a range of
hydrophobicity of 6 log units seems to be rather surprising.
It is of interest to notice that our earlier work on a series
of monosubstituted pyrazines found the value of 0.58 as
the slope of the log P term, being very close to the value of
0.55 shown by eq 5, in the corresponding equation. Al-
though measuring log k values of monosubstituted pyra-
zines in more methanol-rich eluents such as M60 and M70
has been technically difficult, it has now become very clear,
by combination of all the results obtained so far, that an
optimum eluent composition yielding almost linear rela-
tionships would be at 50% MeOH or at methanol concen-
trations near 50%. Since the measurement of log k is easier
and quicker than log P, particularly in very hydrophobic
compounds, the log kM50 parameter can be expected to be
a powerful tool for predicting the log P value. Care should
be taken, however, in treating amphiprotic compounds by
the HPLC method. When alkyl-bonded stationary phases
are used, amphiprotics often give separate lines parallel
3. Camilleri, P.; Watts, S. A.; Boraston, J . A. A surface area
approach to determination of partition coefficients. J . Chem.
Soc., Perkin Trans. 2 1988, 1699-1707.
4. Fujita, T.; Iwasa, J .; Hansch, C. A new substituent constant,
π, derived from partition coefficients. J . Am. Chem. Soc.
1964, 86, 5175-5180.
5. Rekker, R. The Hydrophobic Fragmental Constant; Elsevi-
er: New York, 1977.
6. CLOGP program, Pomona College Medchem Project.
7. Fujita, T. Substituent effects in the partition coefficient of
disubstituted benzenes: bidirectional Hammett-type rela-
tionships. Prog. Phys. Org. Chem. 1983, 14, 75-113.
8. Bodor, N.; Huang, M.-J . An extended version of a novel
method for the estimation of partition coefficients. J . Pharm.
Sci. 1992, 81, 272-281.
9. Collander, R. The partition of organic compounds between
higher alcohols and water. Acta Chem. Scand. 1951, 5, 774-
780.
10. Braumann, T. Determination of hydrophobic parameters by
reversed-phase liquid chromatography: Theory, experimen-
tal techniques, and application in studies on quantitative
structure-activity relationships. J . Chromatogr. 1986, 373,
191-225.
11. Minick, D. J .; Frenz, J . H.; Patrick, M. A.; Brent, D. A. A
comprehensive method for determining hydrophobicity con-
stants by reversed-phase high-performance liquid chroma-
tography. J . Med. Chem. 1988, 31, 1923-1933.
Journal of Pharmaceutical Sciences / 1303
Vol. 88, No. 12, December 1999

12. Lambert, W. J . Modeling oil-water partitioning and mem-
brane permeation using reversed-phase chromatography. J .
Chromatogr. A. 1993, 514, 469-484.
23. LaCourse, W. R.; Dasenbrock, C. O. Column liquid chroma-
tography: Equipment and instrumentation. Anal. Chem.
1998, 70, 37R-52R.
13. Yamagami, C.; Ogura, T.; Takao, N. Relationship between
capacity factors and octanol-water partition coefficients for
monosubstituted pyrazines and the related pyridines. J .
Chromatogr. 1990, 514, 123-136.
24. Yamagami, C.; Takao, N.; Fujita, T. Analysis and prediction
of partition coefficients of monosubstituted diazines. Quant.
Struct.-Act. Relat. 1990, 9, 313-320.
25. Motohashi, N.; Ashihara, Y; Yamagami, C.; Saito, Y. Anti-
mutagenic effects of dehydrozingerone and its analogs on UV-
induced mutagenesis in Escherichia coli. Mutat. Res. 1997,
377, 17-25.
26. Freese, E.; Levin, B. C.; Pearce, R.; Sreevalsan, T.; Kaufman,
J . J .; Koski, W. S.; Semo, N. M. Correlation between the
growth inhibitory effects, partition coefficients and terato-
genic effects of lipophilic acids. Teratology 1979, 20, 413-
439.
14. Yamagami, C.; Fujita, T. Hydrophobicity parameter of het-
eroaromatic compounds derived from various partitioning
systems. In Classical and Three-dimensional QSAR in Agro-
chemistry; Hansch, C., Fujita, T., Eds.; ACS Symposium
Series 606, American Chemical Society: Washington, DC,
1995; pp 36-47.
15. Yamagami, C.; Yokota, M.; Takao, N. Hydrogen-bond effects
of ester and amide groups in heteroaromatic compounds on
the relationship between the capacity factor and the octanol-
water partition coefficient. J . Chromatogr. A 1994, 662, 49-
60.
27. Kitano, Y.; Ashida, T.; Ohta, A.; Watanabe, T.; Akita, Y.
Structure of 2,3-diphenylpyrazine, C16H12N2. Acta Crystal-
logr. 1983, C39, 136-139.
16. Garst, J . E.; Accurate, wide-range, automated, high-perfor-
mance liquid chromatographic method for the estimation of
octanol/water partition coefficients II: Equilibrium in parti-
tion coefficient measurements, additivity of substituent
constants, and correlation of biological data. J . Pharm. Sci.
1984, 73, 1623-1629.
28. Dewar, M. J . S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J . J .
P. AM1: A new general purpose quantum mechanical
molecular model. J . Am. Chem. Soc. 1985, 107, 3902-3909.
29. Yamagami, C.; Takao, N.; Fujita, T. Analysis and prediction
of partition coefficients of disubstituted pyrazines. J . Pharm.
Sci. 1991, 80, 772-777.
30. Yamagami, C.; Kawase, K.; Fujita, T. A new hydrogen-
accepting scale of monosubstituted (di)azines for the rela-
tionship between retention factor and octanol-water parti-
tion coefficient. Quant. Struct.-Act. Relat. 1999, 18, 26-34.
31. Charton, M. Electrical effect substituent constants for cor-
relation analysis. Prog. Phys. Org. Chem. 1981, 13, 119-
252.
32. Fox, K.; Zauke, G.-P.; Butte, W. Kinetics of bioconcentration
and clearance of 28 polychlorinated biphenyl congeners in
zebrafish (Brachydanio rerio). Ecotoxicol. Environ. Saf. 1994,
28, 99-109.
33. Leblanc, G. A. Tropic-level differences in the bioconcentration
of chemicals: implications in assessing environmental bio-
magnification. Environ. Sci. Technol. 1995, 29, 154-160.
34. Devillers, J .; Domine, D.; Bintein, S. Fish bioconcentration
modeling with log P. Toxicol. Methods 1998, 8, 1-10.
17. Dross, K. P.; Mannhold, R.; Rekker, R. F. Drug Lipophilicity
in QSAR Practice: II. Aspects of RM-determinations; Critics
of RM-corrections; Interrelations with partition coefficients.
Quant. Struct.-Act. Relat. 1992, 11 , 36-44.
18. Bechalany, A.; Tsantili-Kakoulidou, A.; Tayar, N. El.; Testa,
B. Measurement of lipophilicity indices by reversed-phase
high-performance liquid chromatography: comparison of two
stationary phases and various eluents. J . Chromatogr. 1991,
541, 221-229.
19. Kucharˇ, M.; Rejholec, V.; Kraus, E.; Miller, V.; Ra´bek, V.
Influence of intramolecular interactions on chromatographic
behaviour of arylaliphatic acids. I. Comparison of reversed-
phase thin-layer and high-performance liquid chromatogra-
phy. J . Chromatogr. 1983, 280, 279-288.
20. Yamagami, C.; Masaki, Y. Relationship between capacity
factor and octanol-water partition coefficients of monosub-
stituted thiophenes. Chem. Pharm. Bull. 1995, 43, 2238-
2242
21. Ohta, A.; Takahashi, H; Miyata, N.; Hirono, H.; Nishio, T.;
Uchino, E.; Yamada, K.; Aoyagi, Y.; Suwabe, Y.; Fujitake,
M.; Suzuki, T.; Okamoto, K. Anti-platelet aggregation activity
of some pyrazines. Biol. Pharm. Bull. 1997, 20, 1076-1081.
22. Dorsey, J . G.; Cooper, W. T.; Siles, B. A.; Foley, J . P.; Barth,
H. G. Liquid chromatography: Theory and methodology.
Anal. Chem. 1996, 68, 515R-568R.
Acknowledgments
We thank Professor Toshio Fujita of EMILE project for useful
discussion. This work was partly supported by Kobe Pharmaceuti-
cal University Collaboration Fund.
J S990112S
1304 / Journal of Pharmaceutical Sciences
Vol. 88, No. 12, December 1999