Physicochemical profiling of sartans: A detailed study of ionization constants and distribution coefficients

Article abstract of DOI:10.1002/hlca.200890051

A detailed investigation of the ionization and lipophilicity profiles of selected sartans (valsartan, losartan, irbesartan, candesartan, candesartan cilexetil), a class of antihypertensives commonly used in therapy, is presented. The pK_a macroconstants were determined by integrated potentiometry, capillary electrophoresis, and UV spectrophotometry. The measured pK_a macroconstants were connected with the ionizable centers present in each molecule with the aid of model compounds. Potentiometric titrations with the GLpKa apparatus were performed to determine the distribution profile (log D vs. pH) of valsartan, while the shake-flask procedure was used to characterize the distribution profile of the other compounds. Valsartan showed a lipophilicity profile consistent with the presence of two acidic centers. Losartan and irbesartan, which contain one acidic and one basic center, displayed the classical bell-shaped profile of ordinary ampholytes. By contrast, a more complex situation emerged in the case of candesartan, due to the large number of ionization equilibria involved. The low solubility of candesartan cilexetil, together with the ease of hydrolysis of the ester moiety, prevented a successful investigation of its ionization and lipophilicity profiles.

Full text of DOI:10.1002/hlca.200890051

468
Helvetica Chimica Acta – Vol. 91 (2008)
Physicochemical Profiling of Sartans: A Detailed Study of Ionization
Constants and Distribution Coefficients
by Paolo Toscoa), Barbara Rolandoa), Roberta Fruttero*a), Yveline Henchozb), Sophie Martelb),
Pierre-Alain Carruptb), and Alberto Gascoa)
a
`
) Dipartimento di Scienza e Tecnologia del Farmaco, Universita degli Studi di Torino, Via Pietro Giuria 9,
I-10125 Torino (phone: þ 390116707850; fax: þ 390116707286; e-mail: roberta.fruttero@unito.it)
b
´
´
`
) Unite de Pharmacochimie, Section des sciences pharmaceutiques, Universite de Geneve,
´
`
Universite de Lausanne, 30 Quai Ernest Ansermet, CH-1211 Geneve 4
A detailed investigation of the ionization and lipophilicity profiles of selected sartans (valsartan,
losartan, irbesartan, candesartan, candesartan cilexetil), a class of antihypertensives commonly used in
therapy, is presented. The pKa macroconstants were determined by integrated potentiometry, capillary
electrophoresis, and UV spectrophotometry. The measured pKa macroconstants were connected with the
ionizable centers present in each molecule with the aid of model compounds. Potentiometric titrations
with the GLpKa apparatus were performed to determine the distribution profile (log D vs. pH) of
valsartan, while the shake-flask procedure was used to characterize the distribution profile of the other
compounds. Valsartan showed a lipophilicity profile consistent with the presence of two acidic centers.
Losartan and irbesartan, which contain one acidic and one basic center, displayed the classical bell-
shaped profile of ordinary ampholytes. By contrast, a more complex situation emerged in the case of
candesartan, due to the large number of ionization equilibria involved. The low solubility of candesartan
cilexetil, together with the ease of hydrolysis of the ester moiety, prevented a successful investigation of
its ionization and lipophilicity profiles.
Introduction. – The lipophilicity of an ionizable compound can be measured as the
partition coefficient P (log P) between two essentially immiscible solvents related, by
definition, to a well-defined electrical state. If the compound can exist in different
electrical states, its lipophilicity is expressed by the distribution coefficient D (log D).
This latter molecular descriptor is the weighted mean of the lipophilicity of each
electrical species under which the compound can exist in a pH-dependent proportion
[1]. Since a drug has to cross several biomembranes during its random walk to the
target, the bioavailability of an ionizable drug will depend on the pH of the
environment in which biomembranes are located and on the pKa value(s) of the drug.
Therefore, it is important to know the lipophilicity profile of a drug, namely log D vs.
pH, a key factor for the control of the pharmacokinetic and the pharmacodynamic
phases of drug action [2][3]. These profiles have been studied for many drugs using,
traditionally, the octanol/H2O partition coefficients to model the partitioning in
biomembranes.
Sartans are an important class of drugs acting as selective antagonists of angiotensin
II (Ang II) at the AT1 receptors. This antagonism is commonly used to treat
hypertension, but additional therapeutic indications are emerging for these drugs such
as the management of congestive heart failure, myocardial infarction, and diabetic
ꢀ 2008 Verlag Helvetica Chimica Acta AG, Zürich

Helvetica Chimica Acta – Vol. 91 (2008)
469
nephropathy [4]. Sartans commonly used in therapy are valsartan (1), losartan (2),
irbesartan (3), candesartan (4), and candesartan cilexetil (5) (Fig. 1). The pKa
constants and the lipophilicity profiles of sartans have been studied only in a scattered
and incomplete manner, and frequently with confusing and conflicting results [5]. This
work is dedicated to a close investigation of the ionization constants and lipophilicity
profiles of these drugs by the combined use of potentiometry, capillary electrophoresis
(CE), and UV spectrophotometry. In addition to the sartan derivatives, selected
tetrazoles and benzimidazoles were included in the study as model compounds (Fig. 2).
Fig. 1. Structures of compounds investigated
Results and Discussion. – The sartan drugs studied here, namely valsartan, losartan,
irbesartan, candesartan, and candesartan cilexetil (1 – 5, resp.; Fig. 1), contain different
ionizable fragments such as acidic functions (carboxy and tetrazole groups) or basic
heterocyclic moieties (1H-imidazole, 4,5-dihydro-5-oxo-1H-imidazole, and 1H-benz-
imidazole rings). As a consequence, their complex ionization schemes lead to intricate
lipophilicity profiles. Since sartans contain ionizable groups characterized by a small
difference between equilibrium constants (low DpKa values), they exist in several
different electrical states depending on pH (as exemplified with the ionization
equilibria of candesartan (4) in Scheme 1). To know the exact equilibrium
concentration of each electrical species at a fixed pH, it would be necessary to know
all the microconstants k related to the equilibria between the individual forms.
However, it is not possible to determine all the microscopic ionization constants due to
their proximity. Therefore, all measured pKa values in the stepwise approach used here

470
Helvetica Chimica Acta – Vol. 91 (2008)
Fig. 2. Structures of some tetrazoles and benzimidazoles used as model compounds
to clarify the physicochemical profiles of sartans are macroconstants K, since they refer
to their stoichiometric ionization. Experimental determinations were accomplished
with potentiometric techniques; when the pKa value to be assessed lied outside the
range which can be determined potentiometrically, electrophoretic and spectrophoto-
metric techniques were used, as mentioned below. The constants thus obtained were
also used for the study of lipophilicity profiles. Only a few values for the ionization
constants of the simple tetrazole models 6 – 8 are reported in literature
(pKa(tetrazole) ¼ 4.89 [6], pKa(5-methyltetrazole (6)) ¼ 5.50 [7], pKa(5-phenyl-1H-
tetrazole (7)) ¼ 4.38 [8]). Similarly, only two pKa values of the 1H-benzimidazole
models 9 – 13 studied herein are known (pKa(1H-benzimidazole (9)) ¼ 5.58; pKa (2-
ethoxy-1H-benzimidazole (10)) ¼ 4.39 [9]). Since the methods used for these studies
are not always comparable, we decided to measure the pKa values of the reference
compounds again under the same experimental conditions used for sartans (Table 1).
The results obtained clearly demonstrate that the acidity of the tetrazole moiety
increases with an aromatic 5-substitution, reflecting a larger delocalization of the
negative charge over the aromatic cycle, together with the influence of the electron-
withdrawing substituent in the case of 8. The lipophilicity profiles of the two aromatic
tetrazoles 7 and 8 confirm the charge delocalization with a low difference between the
partition coefficients of neutral and ionized forms (see Table 1) [10][11]. The ionization
constants for drugs 1 – 5 are collected in Table 2; the plots of weight-% of MeOH vs.

Helvetica Chimica Acta – Vol. 91 (2008)
471
Scheme 1. Micro- and Macro-equilibria of Candesartan (4) Ionization
Table 1. Ionization Constants and Lipophilicity Parameters of Model Compounds 6 – 13a)
b
Compound
pKa
)
CLOGPc)
log PN d
À 0.49
)
log PI e
)
diff (log PN–I
)
6
7
8
5.44f)
À 0.83
0.99
1.56
1.57
2.93
4.41
4.42
–
–
4.32f)g)
3.87f)g)
5.48f)
1.50h)
À 0.94
2.44
2.67
–
–
–
1.81h)
À 0.86
9
–
–
–
–
–
–
–
–
10
11
12
4.30f)
3.68f)g)
2.38i)
–
NDf)g)
2.45j)
13
2.12 and 3.67i)
ND and 3.87f)g)
4.40
–
–
–
a) ND: not determinable (see text for more details). b) Potentiometric method, standard deviation
(SD) < 0.02; CE, SD < 0.04; spectrophotometric method, SD < 0.08. c) Calculated by CLOGP algorithm
[12]; the mean value between the 1H- and 2H-tautomers is reported. d) Logarithm of the partition
coefficient of the neutral species (SD < 0.03), obtained by potentiometry. e) Logarithm of the partition
coefficient of the ionized species (SD < 0.05), obtained by shake-flask procedure. f) Obtained by
potentiometry. g) MeOH as co-solvent was used in percentage ranging from 10 to 45; the extrapolation to
zero co-solvent was obtained by the Yasuda – Shedlovsky procedure. h) Validated by shake-flask
i
j
procedure. ) Obtained by capillary electrophoresis (CE). ) Obtained by spectrophotometry.

472
Helvetica Chimica Acta – Vol. 91 (2008)
Table 2. Ionization Constants of Sartans 1 – 5a)
b
b
b
Compound
pKa1
)
Naturec)
pKa2
)
Naturec)
pKa3
)
Naturec)
1
2
3
4
3.60 Æ 0.05d)
2.95 Æ 0.05d)
3.69 Æ 0.09d)
2.12 Æ 0.19e)
NDd)
Acidic
Basic
Basic
Basic
4.70 Æ 0.01d)
4.25 Æ 0.04d)
4.42 Æ 0.05d)
3.34 Æ 0.36e)
3.67 Æ 0.04d)
Acidic
Acidic
Acidic
Acidic
–
–
–
–
–
–
4.50 Æ 0.24e)
Acidic
4.66 Æ 0.04d)
5
NDd)
a) ND: not determinable (see text for more details). Notation adopted in order of increasing pKa value.
b) Potentiometric method, SD < 0.09; CE, CI95% Æ 0.36. c) Electrical nature of the corresponding pKa
as deduced from the slope of the linear regression obtained by plotting %MeOH vs. pKa (see Fig. 3).
d) Obtained by potentiometry. e) Obtained by CE.
psKa, the ionization profiles (% of the species against pH), and the lipophilicity profiles
are shown in Figs. 3, 4, and 5, respectively.
Valsartan (1). Valsartan contains two acidic centers, the COOH group and the
tetrazole ring; as expected, two pKa values were found potentiometrically (Table 2). By
comparison with simpler tetrazoles, the higher value (pKa2 ¼ 4.70) can be attributed to
the prevalent ionization of the tetrazole ring, while the lower (pKa1 ¼ 3.60) is
consequently linked to the prevalent ionization of the COOH group. The lower acidity
Fig. 3. Plots of weight-% of MeOH vs. psKa for losartan (2), irbesartan (3), candesartan (4), and
candesartan cilexetil (5)

Helvetica Chimica Acta – Vol. 91 (2008)
473
Fig. 4. Ionization profiles (% of the species against pH) of valsartan (1), losartan (2), irbesartan (3), and
candesartan (4)
Fig. 5. Lipophilicity profiles of compounds 1 – 4 obtained by a fitting procedure

474
Helvetica Chimica Acta – Vol. 91 (2008)
with respect to the simple tetrazole 7 can be ascribed to the electrostatic destabilization
due to the concomitant negative charge on the COOÀ group.
The lipophilicity profile of valsartan (1; Fig. 5) displays the pattern typical of a
molecule with two acidic centers with close pKa values. The ionization profile (Fig. 4)
clearly shows that, at pH < 1.60, this drug exists almost exclusively in the undissociated
form; thus log D1.0 ¼ 3.78 coincides with log PN, the logarithm of the partition
coefficient of the neutral form (H2A; Table 3). At pH > 6.70, and thus at physiological
pH (blood, 7.4), the drug exists almost exclusively as dianion. Consequently, log D7.4
¼
À1.36 is comparable to log P2A, i.e., the logarithm of the partition coefficient of the
completely ionized form (AÀÀ). The measured partition coefficient of the neutral form
(log PN) is lower than the CLOGP value [12], suggesting a possible hydrophobic
collapse between alkyl and aromatic moieties [1], while the diff (log PNÀI) parameter
(log PN À log PI ¼ 5.26) is compatible with the presence of two negative charges.
Table 3. Lipophilicity Data: Calculated Values and Experimental Resultsa)
Compound
CLOGPb)
log PN c
)
log PC d
)
log PA e
)
log P2A f
)
log D7.4 g
)
log Dmax h
)
1
2
3
4
4.86
3.85
6.04
5.18
3.78i)
–
–
–
–
–
À 1.48i)
À 1.36i)
3.78
3.09
2.95
3.36
0.85j)
1.25j)
1.38j)
1.12j)
1.24j)
–
–
1.16j)
–
1.28j)
À 1.35j)
À 1.50j)
a) Potentiometric data, SD < 0.05; shake flask data, SD < 0.1. b) Calculated by CLOGP algorithm [12].
c) Logarithm of the partition coefficient of the neutral species (never present alone except for 1).
d) Logarithm of the partition coefficient of the cationic species. e) Logarithm of the partition coefficient
f
of the anionic species (AÀ ) measured at pH 8.5. ) Logarithm of the partition coefficient of the dianionic
species (AÀÀ ) measured at pH 8.5. g) Logarithm of the apparent partition coefficient at physiological
pH. h) Calculated by fitting procedure. i) Experimentally obtained by potentiometry (when < 0
j
validated by shake-flask procedure). ) Experimentally obtained by shake-flask procedure.
Losartan (2) and Irbesartan (3). Both losartan (2) and irbesartan (3) contain one
acidic center, the tetrazole ring, and one basic center, the imidazole and the 4,5-
dihydro-5-oxo-1H-imidazole ring, respectively. Two pKa values were detected potentio-
metrically for each of these products. These values indicate a not well-separated
dissociation of the protonated basic centre and of the acidic centre (DpKa ¼ 1.30 and
0.73 for 2 and 3, resp.). For these two products, the apparent psKa values obtained by
titration in co-solvent mixtures were plotted against the % composition of the mixtures.
Straight lines with a negative slope (known as the typical behavior of basic centers [13];
plots b in Fig. 3) were obtained for both products using the apparent constants related
to the lower pKa values, while lines with a positive slope (known as the typical behavior
of acidic centers [13]; plots a in Fig. 3) were obtained using the apparent constants
related to the higher pKa values. Consequently, for compounds 2 and 3, the lower pKa
values can be reasonably connected with the prevalent ionization of imidazole (pKa1 ¼
2.95) and 4,5-dihydro-5-oxo-1H-imidazole (pKa1 ¼ 3.69) centers, respectively, while the
higher pKa values can be ascribed to the prevalent dissociation of the tetrazole centers
(pKa2 ¼ 4.25 and pKa2 ¼ 4.42, resp.). The pKa assignment for losartan (2) and irbesartan

Helvetica Chimica Acta – Vol. 91 (2008)
475
(3) further confirms the assignment of the pKa2 ¼ 4.70 of valsartan (1) to the prevalent
dissociation of its tetrazole moiety, as proposed above. The ionization profiles deduced
from the macroconstants are shown in Fig. 4. At physiological pH, the drugs exist
principally under the anionic form (AÀ). The neutral form in equilibrium with the
zwitterionic form is present in a pH-dependent concentration over the pH range 1.0 –
6.2 for 2 and 1.7 – 6.4 for 3. As expected, the lipophilicity behavior of losartan (2) and
irbesartan (3) is quite different from that of valsartan (1). These drugs show the
classical bell-shaped lipophilicity profile of ordinary ampholytes [14]. This is in keeping
with the presence, at pH < 0.95 and pH < 1.69, respectively, of the hydrophilic cationic
forms (H2Aþ; 2, log PC ¼ 0.85; 3, log PC ¼ 1.25), and, at pH > 6.25 and pH > 6.42,
respectively, therefore also at physiological pH, of the sole anionic forms (AÀ; 2,
log PA ¼ 1.12; 3, log PA ¼ 1.24). It should be noted that the log Dmax for 2 and 3 (Table 3)
at a pH around 3 reflects the lipophilicity of a complex mixture of cationic, anionic, and
neutral forms (see distribution profile; Fig. 5), while log PN cannot be determined;
therefore, a direct comparison with the calculated partition coefficients of the neutral
forms (CLOGP values) is not feasible.
Candesartan (4). A more complex situation occurs with candesartan (4). This drug
contains two acidic centers (COOH group and tetrazole ring) and one basic center
(substituted 1H-benzimidazole substructure). Three macroconstant pKa values were
detected for this drug by capillary electrophoresis (CE; Table 2). The plot of meff vs. pH
(Fig. 6) shows three partially overlapping inflection points, corresponding to three pKa
values, two acidic and one basic. Indeed, the minimum meff value is twofold larger than
the maximal meff value. The compound exists mainly as a monocationic form (H3Aþ) at
pH lower than 0, as zwitterionic and neutral forms (H2AþÀ, H2A) at pH 2.8, as
monoanionic forms HAþÀÀ and HAÀ at pH 4.1, and as a dianionic form (AÀÀ) at pH
higher than 6.6, thus also at physiological pH. The proximity of the three pKa values
leads to a high value of the 95% confidence intervals for each of them. Only the two
higher constants were detectable by the potentiometric method, probably because the
third is too low to be measured in the presence of high concentrations of MeOH; it is
important to stress that precipitation occurred when less than 45% of MeOH was used.
These two constants are reasonably associated, in analogy to valsartan (1), to the
prevalent dissociation of COOH (pKa2 ¼ 3.67) and tetrazole (pKa3 ¼ 4.66) moieties,
respectively (Table 2). DpKa ¼ 0.99 indicates that the dissociations of these centers are
not well-separated. These assignments were confirmed by the plots of the apparent psKa
values obtained by potentiometric titration in co-solvent mixtures vs. the %
composition of the mixtures (Fig. 3; 4), where straight lines with positive slopes were
obtained in the case of the former two centers. Consequently, the lowest pKa value
detected by CE (pKa1 ¼ 2.12) should be associated to the prevalent ionization of the
1H-benzimidazole ring. This assignment is confirmed by the analysis of pKa values of
the 1H-benzimidazole models we studied alongside. Actually, the introduction of the
EtO substituent at C(2) of 1H-benzimidazole (9) (!10) induces a decrease of ca. 1.2
pK units in the basicity. The further introduction of the PhCH2 group at N(1) (!11)
reduces the basicity of additional 0.6 pK units. Finally, 13, the substructure present in
candesartan (4), shows basic and acidic pKa values in keeping with the corresponding
pKa values found in the parent drug. The analysis of the acid – base properties of
candesartan (4) shows that this drug can exist under several different electrical forms

476
Helvetica Chimica Acta – Vol. 91 (2008)
Fig. 6. Relationship between meff and pH for candesartan (4) in CE experiments. The symbols ꢁþꢂ, ꢁnꢂ, ꢁÀꢂ
and ꢁÀÀꢂ represent the cationic, neutral, monoanionic, and dianionic forms, resp.
(Scheme 1), whose distribution profile, as deduced from the macroconstants, is shown
in Fig. 4.
The lipophilicity profile of candesartan (4) is complex. At pH < 0, log D ¼ 1.38
represents the lipophilicity of the pure cation (H3Aþ), while, at pH > 8, log D ¼ À1.50
represents the lipophilicity of the dianionic form (AÀÀ). The log Dmax occurs at pH 3.36,
at which a complex mixture of different forms is present. Certainly, the highly lipophilic
neutral form (CLOGP ¼ 5.18) has a great influence on the overall partitioning behavior
of the equilibrium mixture. However, the contribution of the other forms cannot be
neglected a priori, in particular that of zwitterionic species in which intramolecular
effects can reduce the hydrophilicity in spite of the presence of two formal charges [14].
To verify this hypothesis, the conformational domain of the two possible zwitterionic
forms was explored (see Exper. Part). The most stable conformers in a 10-kcal range
from the global minimum are shown in Fig. 7. In form a, folded conformations are
largely preferred, with the protonated 1H-benzimidazole center interacting with the
anionic tetrazole moiety by a strong charge-enhanced H-bond, so that the interactions
with the solvent molecules and the overall hydrophilicity are reduced. In form b, the
actual separation of opposite formal charges is lowered following delocalization of the
positive charge on the 1H-benzimidazole nucleus, as evidenced by the distribution of
electrostatic potential-derived atom-centered charges on one of the stable conformers
(Fig. 8). The behavior of the prodrug candesartan cilexetil (5) is particular owing to its
ease of hydrolysis to the free acid. Two pKa values (one basic and one acidic) were
expected for this product. The potentiometric titrations led to the determination of two
pKa values; the plots of the apparent psKa values obtained by potentiometric titration in
cosolvent mixtures vs. the % composition of the mixtures showed that they are
associated to two acidic centers (Fig. 3; 5). HPLC Stability assays showed that, in
aqueous medium, candesartan cilexetil (5) is largely hydrolyzed to the parent drug.

Helvetica Chimica Acta – Vol. 91 (2008)
477
Fig. 7. The most stable conformers in a 10-kcal range from the global minimum for the two zwitterionic
forms of candesartan. A charge-enhanced H-bond is represented with a dotted line; non-polar H-atoms
have been omitted for clarity.
Fig. 8. The distribution of electrostatic potential-derived atom-centered charges on one of the stable
conformers of the zwitterionic form b of candesartan. The actual charge separation is lower than the
presence of opposite formal charges would suggest: indeed, the carboxylate group bears an overall
charge of ca. À 0.7, while the positive charge is largely delocalized over the whole benzimidazole ring. To
avoid cluttering the picture, only charges on the benzimidazole nucleus have been reported.
Therefore, the pKa values determined by potentiometric titration are essentially the
pKa values of candesartan (4). The same picture was found when the CE method was
used. By contrast, no hydrolysis was observed with the simple methyl 1H-benzimida-
zole-carboxylate 12, whose basic pKa value was near that of candesartan (4). This
similarity is conceivably due to two reasons: first, the COOH group of 4 is largely
present in the unionized form when the pKa associated to the benzimidazole
substructure is measured; second, the electronic effects of COOH and COOMe
groups are similar [15]. The ease of hydrolysis and low solubility of candesartan
cilexetil (5) prevented the investigation of its lipophilicity profile.

478
Helvetica Chimica Acta – Vol. 91 (2008)
Conclusions. – pH-Dependent partition profiles and pKa values of some important
sartan drugs were determined. The combined use of the potentiometric and CE
methods allowed the complete determination of all pKa values. The analysis of
cosolvent weight-% vs. psKa plots and the study of the ionization of appropriate model
compounds allowed the assignment of all measured pKa values to specific ionizable
moieties of the drugs. The pH-dependent partition profiles should provide good
descriptors for studies of relationships between structure and distribution of the drugs.
At physiological pH, losartan (2) and irbesartan (3) exist principally in the anionic
form, while valsartan (1) and candesartan (4) are present mainly in their dianionic
form. The study of candesartan cilexetil (5) was complicated by both the facility with
which the prodrug undergoes hydrolysis and its low solubility in H2O.
This work was supported by a MIUR grant (COFIN 2005).
Experimental Part
General. Sartans 1 – 5 were kindly supplied by different pharmaceutical companies: valsartan (1) by
Novartis Farma (I-Origgio), losartan (2) by Merck Sharp & Dohme (I-Roma), irbesartan (3) by Sanofi
Synthelabo (I-Limito), candesartan (4) and candesartan cilexetil (5) by Astra Zeneca (I-Milano). Octanol
was of spectrophotometric grade (Sigma-Aldrich), and MeOH was of HPLC-grade (Carlo-Erba).
Compounds 6 [16], 7 [17], and 10 [9] were synthesized according to literature, while 8 and 9 were
purchased from Sigma-Aldrich. Compound 11 was prepared starting from 14, and compounds 12 and 13
were prepared starting from 15 via 16 (Scheme 2). Flash column chromatography (FC): silica gel (Merck,
Kieselgel 60, 230 – 400 mesh ASTM); PE stands for 40 – 708 petroleum ether. The progress of the
reactions was followed by TLC (5 ꢀ 20-cm plates with a layer thickness of 0.2 mm). Anh. MgSO4 was
used as drying agent for the org. phases. Org. solvents were removed under vacuum at r.t. M.p.: cap.
1
apparatus (Büchi 540). H- and 13C-NMR spectra: Bruker Avance-300 at 300 and 75 MHz, resp., with
Me4Si as the internal standard; d in ppm, J in Hz. Low-resolution (LR) MS: Finnigan Mat-TSQ-700; in
m/z. The results of elemental analyses (C, H, N) for the new 1H-benzimidazole derivatives are within Æ
0.4% of the theoretical values (REDOX, I-Monza).
1-Benzyl-2-ethoxy-1H-benzimidazole (11). AcOH (0.25 ml, 4.4 mmol) was added to a soln. of N-
benzylbenzene-1,2-diamine (14 [18]; 0.87 g, 4.4 mmol) in (EtO)4C (4.0 ml, 19 mmol). The mixture was
heated at 608 for 2 h; after cooling, Et2O was added (30 ml), and the org. layer was washed with 10%
NaHCO3 (3 ꢀ 15 ml), then with brine (10 ml), dried, and evaporated. The crude product was purified by
i
FC (PE/AcOEt 9:1), then recrystallized from hot Pr2O/hexane to give 11 (0.93 g, 84%). White solid.
1
M.p. 69.58 (iPr2O/hexane). H-NMR ((D6)DMSO): 7.43 – 7.03 (m, 9 H); 5.22 (s, 2 H); 4.56 (q, J ¼ 7.05,
2 H); 1.39 (t, J ¼ 7.05, 3 H). 13C-NMR ((D6)DMSO): 156.7; 139.8; 136.7; 133.3; 128.6; 127.5; 127.1; 121.0;
120.5; 117.0; 109.0; 65.9; 44.7; 14.4. CI-MS: 253 ([M þ 1]þ). Anal. calc. for C16H16N2O (252.31): C 76.16, H
6.39, N 11.10; found: C 76.00, H 6.32, N 11.06.
Methyl 1-Benzyl-2-ethoxy-1H-benzimidazole-7-carboxylate (12). FeCl3 · 6 H2O (0.013 g, 0.05 mmol)
and activated charcoal (0.13 g, 11 mmol) were added to a soln. of methyl 2-(benzylamino)-3-nitro-
benzoate (15 [19]; 1.0 g, 3.5 mmol) in a mixture of MeOH (15 ml) and freshly distilled THF (15 ml),
according to a known procedure [20]. After refluxing for 30 min, a soln. of NH2NH2 · H2O (1.04 ml,
21 mmol) in MeOH (15 ml) was added dropwise to the mixture over 10 min. The resulting mixture was
refluxed for further 14 h; after cooling, the insoluble material was removed by filtration through a Celite
pad, washing the filter thoroughly with CH2Cl2 (100 ml). The filtrate was washed with 10% NaHCO3
(3 ꢀ 20 ml), then with brine (10 ml), dried, and evaporated. The residue was dissolved in (EtO)4C
(5.0 ml, 24 mmol), then AcOH was added (0.20 ml, 3.5 mmol). The mixture was heated at 608 for 2 h;
after cooling, Et2O was added (30 ml), and the org. layer was washed with 10% NaHCO3 (3 ꢀ 15 ml),
then with brine (10 ml), dried, and evaporated. The crude product was purified by FC (PE 9/AcOEt 1),
i
then recrystallized from hot Pr2O/hexane to give 12 (0.65 g, 60%). White solid. M.p. 1148 (iPr2O/

Helvetica Chimica Acta – Vol. 91 (2008)
479
Scheme 2. Synthetic Route to Derivatives 11 – 13
hexane). 1H-NMR ((D6)DMSO): 7.71 – 7.69 (m, 1 H); 7.45 – 7.42 (m, 1 H); 7.29 – 7.16 (m, 4 H); 6.93 – 6.90
(m, 2 H); 5.49 (s, 2 H); 4.61 (q, J ¼ 7.05, 2 H); 1.40 (t, J ¼ 7.05, 3 H). 13C-NMR ((D6)DMSO): 166.0;
158.3; 141.6; 136.8; 128.4; 127.1; 126.1; 120.7; 115.4; 66.5; 52.1; 46.4; 40.0; 39.7; 39.4; 39.1; 38.9; 38.6; 14.3.
CI-MS: 311 ([M þ 1]þ). Anal. calc. for C18H18N2O3 (310.35): C 69.66, H 5.85, N 9.03; found: C 69.49, H
5.87, N 9.05.
1-Benzyl-2-ethoxy-1H-benzimidazole-7-carboxylic acid (13). Compound 12 (0.30 g, 0.97 mmol) was
dissolved in 20 ml of ethanolic 1n KOH and stirred for 16 h. The mixture was then diluted with H2O
(20 ml), the solvent was evaporated, and the aq. phase was washed with Et2O (3 ꢀ 15 ml), then pH was
adjusted to 5 with 37% HCl. The precipitate was collected by filtration, washed with H2O, and dried
(P2O5). The crude product was recrystallized from hot THF/iPr2O to give 13 (0.17 g, 60%). White solid.
M.p. 177.58 (THF/iPr2O). 1H-NMR ((D6)DMSO): 7.72 – 7.62 (m, 1 H); 7.55 – 7.47 (m, 1 H); 7.32 – 7.10 (m,
4 H); 7.02 – 6.90 (m, 2 H); 5.62 (s, 2 H); 4.61 (m, 2 H); 1.39 (m, 3 H). 13C-NMR ((D6)DMSO): 167.4;
158.2; 141.6; 137.4; 128.5; 127.1; 126.4; 120.6; 116.7; 66.4; 46.4; 40.0; 39.8; 39.5; 39.2; 38.9; 38.6; 14.3. CI-
MS: 297 ([M þ 1]þ). Anal. calc. for C17H16N2O3 (296.32): C 68.91, H 5.44, N 9.45; found: C 69.15, H 5.52,
N 9.49.
Determination of Ionization Constants. Ionization constants of compounds 1 – 5, as well as of
tetrazole and of benzimidazole model compounds, i.e., 6 – 8 and 9 – 13, resp., were determined by
potentiometry, capillary electrophoresis (CE), and UV spectrophotometry, according to the need (see
Results and Discussion). Potentiometric titrations were carried out with a GLpKa apparatus (Sirius
Analytical Instruments Ltd., Forrest Row, East Sussex, UK). Apparent ionization constants (psKa) were
obtained in co-solvent mixtures because of the low solubility of compounds in H2O. MeOH is the solvent
of choice for this purpose, because its general effects on pKa values have been extensively studied [13]. At
least five different hydro-organic solns. (ionic strength adjusted to 0.15m with KCl) of the compounds

480
Helvetica Chimica Acta – Vol. 91 (2008)
(20 ml, ca. 1 mm in 15 – 68 wt-% MeOH) were initially acidified to pH 1.8 with 0.5n HCl. The solns. were
then titrated with standardized 0.5n KOH to pH 10.5. The titrations were performed under Ar at 25.0 Æ
0.18. The initial estimates of the psKa values (the apparent ionization constants in the H2O/MeOH
mixtures) were obtained by Bjerrum plots; these values were finally refined by a weighted nonlinear
least-squares procedure. Aqueous pKa values were obtained by extrapolation using the Yasuda – Shed-
lovsky (Y– S) procedure (Eqn. 1) [13]:
psKa þ log [H2O] ¼ A · 1/e þ B
(1)
where [H2O] is the molar H2O concentration, and e is the dielectric constant of the mixture (55.5 and
78.3, resp., for pure water), A and B are the coefficients of the linear regression. The plot of wt-% of
MeOH vs. psKa(Y– S plot) was drawn for each compound in order to allow the correct assignment of the
electrical nature for the different ionization sites.
Spectrophotometric measurements were carried out at 258 with a UV/VIS spectrophotometer (UV-
2501PC, Shimadzu) according to [21]. Absorbance spectra of compounds (ca. 100 mm) were recorded
from 220 to 360 nm in several buffers at a constant ionic strength of 50 mm (pH range 0.0 to 8.0 with
increments of 0.5 pH units); absorption readings were made at selected wavelengths which maximize the
difference of absorption between the ionized and the unionized form. For each compound, at least six
pKa values at different pHs were measured.
CE Experiments were performed with an HP3DCE system (Agilent Technologies, D-Waldbronn)
equipped with an on-column diode-array detector, an autosampler, and a power supply able to deliver up
to 30 kV. The separation was performed in a fused silica capillary (BGB Analytik, CH-Bçckten) with an
inner diameter of 50 mm and 32.5 cm total length (24 cm to the UV detector). Samples were introduced
by short-end injection: polarity was reversed and samples were injected at the detector side, thus
reducing the effective length from 24 to 8.5 cm. An injection equivalent to 1.5% of the effective length
was performed by applying a pressure of 12 mbar for 4 s. During analysis, a voltage of 2 to 8 kV was
applied in order to avoid the Joule effect. The capillary was thermostated at 258 by a high-velocity air
stream. UV Detection was carried out at 210 nm for sartans and benzimidazole models, and at 260 nm for
acetone. A two-stage dynamic coating procedure using Ceofix (Analis, B-Namur) was applied to
generate a large and constant electroosmotic flow (EOF) at any pH. Samples were freshly prepared
every day. They were set at a concentration of 100 ppm in 5% acetone (EOF neutral marker). Up to 40%
MeCN was added to some sample solns. to enhance compound solubility. Eighteen buffers were prepared
from pH 1.5 to 10.0 with increments of 0.5 pH units; they were set at a constant ionic strength of 50 mm.
Each compound was injected once at every pH value. The analysis time was lower than 5 min. Effective
mobility (meff) was calculated from the migration times of the analyte and neutral marker. Calculated meff
values were reported as a function of pH, giving rise to a sigmoidal curve. Nonlinear regression
(GraphPad Prism 4.02, GraphPad Software, San Diego, CA; USA) was accomplished to determine
apparent pKa values [22][23].
Determination of Lipophilicity Profiles. For valsartan (1), lipophilicity data were obtained by the
potentiometric method. At least four different titrations (ca. 1 mm) were carried out in a pH range from
1.8 to 12.2, using various volumes of octanol (volume ratios of org. solvent/H2O ranging from 0.05 to 1).
The titrations were carried out under Ar at 25.0 Æ 0.18. In the presence of octanol, the pKa value shifts,
giving an apparent constant called p0Ka; this shift is due to the partitioning of the substance into the org.
phase. The shifts in pKa values are used to determine log PN and log PI (the logarithm of the partition
coefficient of the neutral and the ionized form, resp.). The details of the method have been reported in
[24].
Due to the complex ionization profile of losartan (2), irbesartan (3), and candesartan (4), the
apparent partition coefficients (log DpH) were measured by the shake-flask procedure according to [25].
Several 0.05m buffers were chosen according to the required pH (pH 0.0 to 8.5); ionic strength was
adjusted to 0.15m with KCl. Octanol was added to the buffers, and the two phases were mutually
saturated by shaking for 4 h. The compounds were solubilized in the buffered aq. phase at a concentration
of ca. 0.1 mm, and an appropriate amount of octanol was added. The two phases were shaken for ca.
20 min, by which time the partitioning equilibrium of solutes is reached, and then centrifuged

Helvetica Chimica Acta – Vol. 91 (2008)
481
(10000 rpm, 10 min). The concentration of the solutes in the aq. phase was measured by UV
spectrophotometry (UV-2501PC, Shimadzu) at lmax. For each compound, at least seven log D values at
different pHs were measured. The distribution profiles were obtained by a fitting procedure using Eqn. 2:
P
D ¼ fi ꢁ Pi
(2)
where D is the distribution coefficient, f i is the molar fraction of the electrical species i and Pi is the
partition coefficient of i. The molar fractions were calculated using the experimental pKa values.
The potentiometric method described above for valsartan (1) was adopted to assess the log PN values
of reference tetrazoles 6 – 8, while log PI, which could be determined only for the two most lipophilic
aromatic terms 7 and 8, was measured by the shake-flask procedure.
Conformational Study. The molecular models of the two zwitterionic forms of 4 were constructed
using standard bond lengths and angles with the MOE software package [26]. The ground-state
geometries were fully optimized without geometry constraints according to the AM1 semiempirical
method until the largest component of the gradient was less than 1 · 10À4 Hartree BohrÀ1. All quantum-
mechanical calculations were accomplished with the GAMESS-US software package [27]. The
conformational space of the selected models was explored with MOE by means of quenched molecular
dynamics (QMD), since this method has been reported to be the most appropriate to find highly diverse
conformers characterizing the partitioning behavior of org. molecules [28]. 200 5-ps MD simulations
were performed in vacuo at 600 K with the MMFF 94 force field, starting from the AM1 geometries with
a Boltzmann distribution of the atomic velocities (time step 1 fs). At the end of every 5-ps trajectory, the
final conformer was minimized through a truncated Newton – Raphson geometry optimization until the
gradient was less than 0.001 kcal molÀ1, then stored in a database, and used as the starting geometry for
the subsequent QMD cycle. At the end of the run, duplicate conformers were eliminated from the
database according to the criteria proposed in [28]. The whole QMD procedure was repeated five times
for each model in order to ensure thorough sampling of the conformational space; at the end of all runs,
the same results were found. To ensure a reliable estimation of the relative energies, as well as of the
electrostatic potential, all conformers underwent an additional AM1 optimization followed by a single-
point energy evaluation with a DFT method, using the RB3LYP/6-311 þ G(2d,2p) basis set. Atom-
centered charges were fit to the ab initio electrostatic potential according to the RESP procedure as
implemented in AMBER 9 [29]. All calculations were run on a Linux cluster composed of 24 CPUs
(Pentium IV single core, clock frequency ranging from 1.6 to 3.2 GHz).
REFERENCES
[1] P.-A. Carrupt, B. Testa, P. Gaillard, Rev. Comput. Chem. 1997, 11, 241.
[2] R. A. Conradi, P. S. Burton, R. T. Borchardt, in ꢁLipophilicity in Drug Action and Toxicologyꢂ, Eds.
ˇ
V. Pliska, B. Testa, H. van de Waterbeemd, VCH, Weinheim, 1996, p. 233 – 252.
ˇ
ˇ
[3] V. Pliska, in ꢁLipophilicity in Drug Action and Toxicologyꢂ, Eds. V. Pliska, B. Testa, H. van de
Waterbeemd, VCH, Weinheim, 1996, p. 263 – 294.
[4] E. K. Jackson, in ꢁGoodman and Gilmanꢂs. The Pharmacological Basis of Therapeuticsꢂ, Eds. J. G.
Hardman, L. E. Limbird, McGraw-Hill, New York, 2001, p. 8009 – 8042.
´
´
[5] E. Cagigal, L. Gonzalez, R. M. Alonso, R. M. Jimenez, J. Pharm. Biomed. Anal. 2001, 26, 477.
[6] T. Eicher, S. Hauptmann, ꢁThe Chemistry of Heterocyclesꢂ, Thieme, New York, 1995.
[7] V. A. Ostrovskii, G. I. Koldobskii, N. P. Shirokova, V. S. Poplavskii, Chem. Heterocycl. Compd.
(Engl. Transl.) 1981, 17, 412.
[8] L. D. Hansen, E. J. Baca, P. Scheiner, J. Heterocycl. Chem. 1970, 7, 991.
[9] D. J. Brown, R. K. Lynn, J. Chem. Soc., Perkin Trans. 1 1974, 349.
[10] G. Caron, F. Reymond, P.-A. Carrupt, H. H. Girault, B. Testa, Pharm. Sci. Technol. Today 1999, 2,
327.
[11] G. Bouchard, P.-A. Carrupt, B. Testa, V. Gobry, H. H. Girault, Chem. – Eur. J. 2002, 8, 3478.
[12] Bio-Loom for Windows, v. 1.5, BioByte Corporation, Claremont, CA, USA.

482
Helvetica Chimica Acta – Vol. 91 (2008)
[13] A. Avdeef, J. E. A. Comer, S. J. Thomson, Anal. Chem. 1993, 65, 42.
[14] A. Pagliara, P.-A. Carrupt, G. Caron, P. Gaillard, B. Testa, Chem. Rev. 1997, 97, 3385.
[15] O. Exner, in ꢁCorrelation Analysis in Chemistryꢂ, Eds. N. B. Chapman, J. Shorter, Plenum Press, New
York, 1978, p. 439 – 540.
[16] W. Ogilvie, W. Rank, Can. J. Chem. 1987, 65, 166.
[17] W. G. Finnegan, R. A. Henry, R. Lofquist, J. Am. Chem. Soc. 1958, 80, 3908.
[18] P. Chattopadhyay, R. Rai, P. S. Pandey, Synth. Commun. 2006, 36, 1857.
[19] J. W. Hubbard, A. M. Piegols, B. C. G. Sçderberg, Tetrahedron 2007, 63, 7077.
[20] K. Kubo, Y. Kohara, E. Imamiya, Y. Sugiura, Y. Inada, Y. Furukawa, K. Niahikawa, T. Naka, J. Med.
Chem. 1993, 36, 2182.
[21] A. Albert, E. P. Serjeant, ꢁThe determination of ionization constantsꢂ, 3rd edn., Chapman and Hall,
London, 1984.
[22] L. Geiser, Y. Henchoz, A. Galland, P.-A. Carrupt, J.-L. Veuthey, J. Sep. Sci. 2005, 28, 2374.
[23] Y. Henchoz, J. Schappler, L. Geiser, J. Prat, P.-A. Carrupt, J.-L. Veuthey, Anal. Bioanal. Chem. 2007,
389, 1869.
[24] A. Avdeef, Quant. Struct.-Act. Relat. 1992, 11, 510.
[25] J. C. Dearden, G. M. Bresnen, Quant. Struct.-Act. Relat. 1988, 7, 133.
[26] MOE version 2006.08, Chemical Computing Group Inc., Montreal, Quebec, Canada.
[27] M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T. Elbert, M. S. Gordon, J. H. Jensen, S. Koseki, N.
Matsunaga, K. A. Nguyen, S. J. Su, T. L. Windus, M. Dupuis, J. A. Montgomery, J. Comput. Chem.
1993, 14, 1347.
[28] C. Altomare, S. Cellamare, A. Carotti, G. Casini, M. Ferappi, E. Gavuzzo, F. Mazza, P.-A. Carrupt, P.
Gaillard, B. Testa, J. Med. Chem. 1995, 38, 170.
[29] J. Wang, P. Cieplak, P. A. Kollman, J. Comput. Chem. 2000, 21, 1049.
Received November 27, 2007