Molecular interactions of 1,4-dihydropyridine derivatives with selected organic solvents: A volumetric, spectroscopic and computational study

Available online at www.sciencedirect.com

Journal of Molecular Structure 875 (2008) 354–363 www.elsevier.com/locate/molstruc

Molecular interactions of 1,4-dihydropyridine derivatives with selected organic solvents: A volumetric, spectroscopic and computational study Anamarija Zega a, Stane Srcˇicˇ a, Janez Mavri b, Marija Besˇter-Rogacˇ

c,*

a

c

University of Ljubljana, Faculty of Pharmacy, SI-1000 Ljubljana, Asˇkercˇeva 7, Slovenia b National Institute of Chemistry, SI-1000 Ljubljana, Hajdrihova 19, Slovenia University of Ljubljana, Faculty of Chemistry and Chemical Technology, SI-1000 Ljubljana, Asˇkercˇeva 5, Slovenia Received 1 March 2007; accepted 7 May 2007 Available online 17 May 2007

Abstract Using a combination of volumetric measurements and FTIR spectroscopy, solute–solvent interactions have been investigated for 1,4DHPs in selected organic solvents that mimic the environments of drug production, delivery and the environment from which they reach the site of its activity. Vibrational analysis of 1,4-DHPs and 1,4-DHPs complexes with the solvents has been performed on a medium– high quantum chemical level. Uncharged 1,4-DHPs act in a protic solvent as hydrogen bond acceptors, mainly via the carbonyl group, and, in all other investigated solvents, as a hydrogen bond donor via the hydrogen on the nitrogen. The behaviour of amlodipine besilate differs significantly from that of other compounds. Volumetric measurements proved an effective method for investigating the interactions of uncharged and charged 1,4-DHPs with solvents, and correlated well with FTIR spectroscopy results. This approach can be extended to consider several solvent molecules, perform thermal averaging and to calculate the vibrational spectrum beyond the harmonic approximation. Atomic simulation of antagonistic activity of this class of compounds, by considering the receptor site and the ionic channel, remains a challenge for future decades.  2007 Elsevier B.V. All rights reserved. Keywords: 1,4-Dihydropyridines; Organic solvents; Apparent molar volume; FTIR spectroscopy; Harmonic frequency

1. Introduction 1,4-dihydropyridine calcium channel antagonists (1,4DHPs) have been used for many years in the treatment of angina pectoris, hypertension and other cardiovascular diseases. According to the common view, their mechanism of action is based on inhibition of the smooth muscle Ltype calcium current, thus decreasing intracellular calcium concentration and inducing smooth muscular relaxation [1,2]. In recent years evidence has accumulated that, besides the smooth muscle effects of these agents, their antioxidant protective effects, which are related to the reactivity of these compounds towards radical species, must also be taken into account [3–7]. Recently, the oxidation of 1,4*

Corresponding author. Tel.: +386 1 2419 410; fax: +386 1 2419 437. E-mail address: [email protected] (M. Besˇter-Rogacˇ).

0022-2860/$ - see front matter  2007 Elsevier B.V. All rights reserved. doi:10.1016/j.molstruc.2007.05.011

DHP has attracted much attention because of its reactivity toward endobiotics such as nitric oxide [8–10]. It has been found that the possible radical scavenging properties, in combination with lipophilic character and high affinity for the 1,4-DHP calcium channel receptors, could be the basis for the protective activity in free radical-involved pathologies that appear to accompany many instances of hypertension and also central nervous system disorders [8]. Currently, atomic simulations of allosteric effects are not possible due to lack of the receptor and ionic channel structures, the size of the system and the time scales on which the allosteric effects take place. Therefore, as an initial step in understanding this complex process, we have considered interactions with a few solvents as models for the receptor binding site. Investigation of solvation and associated states of 1,4DHPs can explain their strong complexing interactions in

A. Zega et al. / Journal of Molecular Structure 875 (2008) 354–363 H N

H3C

nifedipine nitredipine nimodipine

R1 CH3 CH3 C2H4OCH3

R2 CH3 C2H5 CH(CH3)2

X 2-NO2 3-NO2 3-NO2

O

CH3 O

O R1

R2

O

O

X

-

O S O

amlodipine besilate

355

H N

+

H3N

H3C

O

CH3 O

O O

CH3

O Cl

Fig. 1. Structures of 1,4-dihydropyridine compounds.

different media. Volumetric properties of binary solutions have been studied extensively, as they can contribute to clarification of the various intermolecular interactions existing between the different species in solution. In particular, much effort has gone into determining partial molar volumes at infinite dilution (Vo) where only solvent–solvent and solute–solvent interactions are present. These volumes can contribute to an understanding of liquid solutions. Partial molar volume is, in thermodynamic terms, the response of a chemical potential to external pressure. Therefore, it can be considered as a measure of solute–solvent interaction rather than as a metric parameter. Many groups have applied the widely used additivity rule, not only to predict new values of the limiting partial molar volume but also to understand the underlying molecular basis of the volumetric behaviour. Information on specific interactions, effects of conformation and packing efficiencies can be extracted by use of these simple relations [11,12]. Vibrational spectroscopy is a powerful tool for studying molecular interactions. In dilute solutions, the infrared frequency shifts reflect the solvent–solute interactions. The spectroscopic studies of carbonyl groups in mixed aqueous and non-aqueous solvent systems have shown clearly the existence of di-, mono- and non-hydrogen bonded examples of protic solvent (water, methanol) interactions with a carbonyl group [13,14]. The frequency of the IR band depends on the nature and strength of the intermolecular interactions (such as a hydrogen bond). Recently, the solvent induced frequency shift of the carbonyl stretching band in urea and trimethyl urea in 13 pure solvents has been studied [15]. Studies on solvation in these systems have shown that the acidity and basicity of the solvent significantly affect the solvent induced frequency shift. The carbonyl stretching vibration, m(C@O), of the solutes has been correlated to Gutmann’s acceptor number, AN, and the donor number, DN, of the solvent, and explained by the formation of mono- and di-hydrogen bonded species involving the protic solvent and the carbonyl group of the solute. In this work apparent molar volumes, V/, were determined experimentally for nimodipine, nitrendipine nifedipine and amlodipine besilate (Fig. 1) in acetone, ethanol,

dimethyl sulfoxide and dichloromethane at 25.0 C, using a vibrating tube densimeter. Partial molar volumes at infinite dilution and their dependence on solution composition are discussed and interpreted in terms of the various intermolecular interactions formed in solution. The experimental molar volume values are compared with the volumes calculated employing Traube’s additivity principle [11] and those obtained with the help of the ACD/ ChemSketch commercial program. The deviations from the experimental data are attributed mainly to structural features of the solvents and solute–solvent interactions. The measured and calculated infrared band shifts of m(C@O) have allowed us to present a qualitative discussion on the contribution of hydrogen bonding to the partial molar volumes of the investigated 1,4-DHP at infinite dilution. 2. Experimental The selected 1,4-DHP compounds undergo photochemical oxidation to the pyridine analogues and, in the case of nitro- analogues, to nitrosophenylpyridine [16,17]. To avoid such photodegradation, all laboratory procedures were carried out as quickly as possible in order to minimize exposure to light. 2.1. Materials Amlodipine besilate (C26H31ClN2O8S, MW = 567.1 g/mol), nimodipine (C21H26N2O7, MW = 418.4 g/mol), Table 1 Densities of solvents qo, and solvent parameters [18]a Acetone DMSO Dichloromethane Ethanol Water

qo

AN

DN

p*

e

l

0.7844 1.0955 1.3168 0.7850 0.99705

12.5 19.3 20.4 37.9 54.8

17.0 29.8 1 32 33, 22b

0.71 1.00 0.82 0.54 1.09

20.56 47.0 8.93 24.35 78.54

2.69 3.9 1.14 1.66 1.85

p*, solvatochromic parameter. a Units: qo, kg dm3; DN, kcal mol1; l, D. b Two values are reported in Ref. [18].

356

A. Zega et al. / Journal of Molecular Structure 875 (2008) 354–363

nitrendipine (C18H20N2O6, MW = 360.4 g/mol) and nifedipine (C17H18N2O6, 346.3 g/mol) were available as a gift from LEK d.d. (a Sandoz Company). The derivatives are only sparingly soluble in water and therefore acetone (p.a., Merck, Germany), ethanol (p.a., Merck Germany), dimethyl sulfoxide (DMSO, p.a. Riedel-de Hae¨n, Germany) and dichloromethane (CH2Cl2, p.a., Merck, Germany) were used as solvents. All solvents were of analytical grade. Their properties are listed in Table 1. Solutions for density measurements were prepared by weighing on an analytical balance Sartorius A-2005. At least 10 solutions of nifedipine, nimodipine and nitrendipine were prepared in each solvent in the concentration range 0.005 < m (mol kg1) < 0.1. Due to the low solubility of the amlodipine besilate only solutions in ethanol and DMSO in the narrow concentration range 0.005 < m (mol kg1) < 0.05 were prepared. For FTIR measurements, the concentrations of 1,4-DHP in solvents were between 0.03 and 0.05 mol dm3.

2.2. Solvent parameters The role of a solvent is determined by its bulk properties (relative permittivity e, viscosity g and density qo), and its electron pair donor (Lewis donor) and electron pair acceptor (Lewis acceptor) abilities. The main parameter used for assessing the polarity and polarizability of a solvent is the solvatochromic parameter p*. Some empirical parameters are used successfully in applied solution chemistry to described solvent properties [18]. Gutmann’s donor number (DN) is related to the ability of a solvent to interact with electron pair acceptors such as protons, cations or Lewis acids. A solvent’s ability to interact with electron pair donors is commonly characterized by the acceptor number, AN, of Mayer, Gutmann and Gerger. A commonly used broad classification of solvents distinguishes between protic, aprotic and inert solvents. Barthel et al. [18] adopted a classification into three main classes: (a) protic, (b) dipolar aprotic and (c) low polarity and inert solvents. The acid–base properties, polarity and polarizability of non-aqueous solvents are taken into account with the help of subclasses. According to this classification, ethanol is treated as an amphiprotic hydroxylic solvent. This group of solvents (HL), encompassing water (L = OH) and the alcohols (L = OR), show acidic and basic properties. They behave intrinsically in autoprotolysis reactions, HL þ HL ¼ LHþ 2þ L , as ionogenes, producing their own cations LHþ and 2 anions L.. Because of autoassociation they enable chain (primary alcohols) and network (water) formation. They possess high AN values. DMSO and acetone are dipolar aprotic solvents. DMSO is an aprotic protophilic solvent with relatively high DN value and poor ability to solvate anions with localized charge, in contrast to inorganic cations. Acetone is regarded as an aprotic protophobic solvent. It possesses

poor solvating properties for anions with localized charges and inorganic cations. Dichloromethane can be classified as a low polarity solvent of high polarizability, which possesses a small or zero dipole moment and DN, but is distinguished from the inert solvents (where it is generally classified) by large p*. Parameters of solvents used in this work, including the dipole moments, l, are listed in Table 1. For comparison the parameters for water are also given. 2.3. Density measurements All densities were measured at 25 C with a precision of ±5 · 106 g cm3, using a vibrating tube densimeter (DMA-602, Anton Paar, Austria). An ultrathermostat attached to the instrument controlled the temperature at 25.00 ± 0.05 C. 2.4. FTIR spectroscopy Infrared spectra were recorded on Nicolet Nexus (Nicolet Instrument Co., Madison, WI) equipped with DTGS detector and an attenuated total reflection (ATR) sampling accessory (diamond crystal, Smart DuraSamplIR). Data was acquired using OMNIC version 5.2 (Nicolet Instrument Co.). IR spectra were measured over the range 4000–500 cm1 with a resolution of 32 cm1. Each individual spectrum was the average of 32 scans and all measurements were performed four times. The solvents were scanned as background. Infrared spectra were recorded at room temperature (20–25 C). 2.5. Computation of the harmonic frequencies Harmonic frequencies for nifedipine molecule in vacuo and nifedipine complexed with one solvent molecule were calculated on the medium–high B3LYP/6-31G(d) level. Energy minimization was performed first, followed by vibrational analysis in the harmonic approximation. When performing energy minimization of the complexes, several starting points were considered and those reported correspond to the lowest energy. All vibrational frequencies were real, confirming that the stationary points were minima, rather than saddle points. Gaussian-03 suite of programs [19], implemented on a cluster of Athlon processors, was used. Vibrational eigenvectors were visualized by program MOLDEN [20] and carbonyl stretchings were well separated from the other modes. 3. Results 3.1. Densities Experimental density values, q, for various solutions in the molality, m, ranges studied, obey equations of the type q ¼ A þ Bm þ Cm2 :

ð1Þ

A. Zega et al. / Journal of Molecular Structure 875 (2008) 354–363 0.796

420 400 Vφ/cm3 mol-1

0.794 ρ / kg dm-3

357

0.792 0.790

380 360 340 320

0.788

300 280

0.786

260 0.00

0.01

0.02

0.03 m/mol kg -1

0.04

0.05

0.00

0.06

0.01

0.02

0.03 0.04 m/mol kg -1

0.05

0.06

Fig. 2. Densities of amlodipine besilate (s), nimodipine (j), nitrendipine (h) and nifedipine (n) ethanol solutions as functions of molality at 25 C. The line represents the polynomial fit.

Fig. 3. The apparent molar volumes of amlodipine besilate (s), nimodipine (j), nitrendipine (h) and nifedipine (n) in ethanol solutions as a function of molality at 25 C. The line represents the linear fit.

The concentration dependence of the densities of nifedipine, nitrendipine, nimodipine and amlodipine besilate in ethanol solutions at 25 C is shown in Fig. 2. The lines represent the polynomial fit using Eq. (1). The A, B and C coefficients are shown in Table 2 along with their standard deviations (r). The A values compare well with the densities of solvents listed in Table 1. Only for the ethanol solutions did the coefficients C have to be taken into account.

where M is the solute molar mass, qo the density of solvent and q the density of the solution with molality m. V/ decreased linearly with concentration (Fig. 3). Similar behaviour was observed for all the systems, except for nitrendipine and nimodipine in dichloromethane solutions, where values of V/ increased slightly with increasing concentration. A linear equation in molality was found to precisely model the apparent molar volumes of all species investigated:

3.2. Apparent molar volumes

V / ¼ V o þ S m m;

The apparent molar volumes, V/, of all solutes were calculated in the usual way [21];   M 1 1 1  V/ ¼ þ ; ð2Þ q m q qo

where Vo is the partial molar volume of the solute at infinite dilution and Sm is an estimated slope predicted from the fit of Eq. (3) to the apparent molar volumes, V/. Vo and Sm, together with their standard fit deviations, are given in Table 3. Taking into account the sources of error (instru-

ð3Þ

Table 2 Parameters in the equation q = A + Bm + Cm2 for investigated 1,4-DHPs in various organic solvents at 25.0 Ca A ± rA Acetone Nifedipine Nitrendipine Nimodipine

B ± rB

C ± rC

r · 104b

0.7842 ± 2 · 104 0.7843 ± 2 · 104 0.7845 ± 3 · 104

0.122 ± 0.004 0.115 ± 0.003 0.123 ± 0.006

4.5 3.7 6.5

DMSO Nifedipine Nitrendipine Nimodipine Amlodipine besilate

1.09545 ± 2 · 105 1.09543 ± 2 · 105 1.09548 ± 6 · 105 1.09552 ± 1 · 105

0.0565 ± 0.0003 0.0442 ± 0.0004 0.0370 ± 0.0001 0.111 ± 0.002

0.34 0.46 0.12 2.29

CH2Cl2 Nifedipine Nitrendipine Nimodipine

1.31627 ± 1 · 105 1.31624 ± 6 · 105 1.31615 ± 6 · 105

0.0083 ± 0.0002 0.0099 ± 0.0004 0.034 ± 0.001

0.28 0.43 0.13

Ethanol Nifedipine Nitrendipine Nimodipine Amlodipine besilate

0.78550 ± 1 · 105 0.78545 ± 7 · 105 0.78540 ± 7 · 105 0.78551 ± 1 · 105

0.0985 ± 0.001 0.0980 ± 0.006 0.105 ± 0.006 0.183 ± 0.001

a b

Units: A, kg dm3; B, kg2 dm3 mol1; C, kg3 dm3 mol2. Estimated standard deviation of the fit for solution densities.

0.17 ± 0.02 0.13 ± 0.11 0.17 ± 0.11 0.03 ± 0.02

0.15 0.77 0.87 0.15

358

A. Zega et al. / Journal of Molecular Structure 875 (2008) 354–363

Table 3 Experimental limiting partial molar volumes, Vo, limiting partial molar volumes obtained by Traube’s additivity concept, V oc , and values from ACD/ ChemSketch program, V o3D , of 1,4-DHPs in dichloromethane, acetone, DMSO and ethanol solutions at 25.0 C M

Nifedipine

Nitrendipine

Nimodipine

Amlodipine besilate

346.34

360.37

418.45

567.1

D

D

D

Acetone

Vo Sm

242.5 ± 0.003 37.2 ± 0.05

25

272.1 ± 0.003 39.4 ± 0.05

11

333.8 ± 0.004 51.4 ± 0.07

3

DMSO

Vo Sm

269.0 ± 4 · 104 13.8 ± 0.006

2

292.1 ± 2 · 104 11.7 ± 2 · 104

9

351.1 ± 2 · 104 11.8 ± 0.003

14

CH2Cl2

Vo Sm

258.3 ± 5 · 106 1.63 ± 1 · 104

9

279.5 ± 8 · 106 2.1 ± 1 · 104

4

337.7 ± 1 · 104 8.83 ± 0.002

1

Ethanol

Vo Sm

281.1 ± 0.002 357.4 ± 0.07

V oc V o3D

267 ± 2 272 ± 3

14

300.0 ± 0.002 251.9 ± 0.06

17

362.7 ± 0.004 332.1 ± 0.1

283 ± 2 289 ± 3

26

D

425.2 ± 0.002 42.6 ± 0.04

2.2

424.9 ± 0.003 141.8 ± 0.09

1.9

337 ± 2 345 ± 3

423 ± 2 445 ± 3

The difference between experimental and calculated values, D ¼ V o  V oc . Sm is the coefficient in Eq. (3) and M the solute molar mass. Units: M, g/mol; Vo, V oc ; V o3D , cm3 mol1; Sm, cm3 kg mol2, D, cm3 mol1.

Table 4 Experimental frequencies of the carbonyl stretching band of 1,4-DHPs in acetone, DMSO, dichloromethane and ethanol solutions at 25.0 Ca Nifedipine m(C@O) I Acetone DMSO CH2Cl2 Ethanol a

1712 1697 1698 1686

Nitrendipine m(C@O) II

m(C@O) I

1708

1712 1695 1696 1687

Nimodipine m(C@O) II

m(C@O) I

1707

1712 1693 1693 1681

Amlodipine besilate m(C@O) II

m(C@O) I

m(C@O) II

1703

1712 1691 1691 1680

1691

1

Units: m(C@O), cm .

mental error, error in mass determination, sample impurities, extrapolation errors) Vo values are certain within 0.5%. 3.3. FTIR spectroscopy The wave numbers for the carbonyl stretching vibrational mode, m(C@O), were determined for all the 1,4DHP derivatives in all the solvents (Table 4). Two bands were observed for the m(C@O) of all four compounds in ethanol; in contrast, in the other solvents only one m(C@O) band was observed in the 1715–1685 cm1 region; each of them higher than m(C@O) of solid 1,4-DHP, which was at 1676 cm1, the value for crystalline nifedipine. IR spectroscopy is used extensively to investigate hydrogen bonding because the peak positions of the XAH and YAR stretches are very sensitive to the extent of formation. The non-bonded XAH and YAR stretches gives rise to relatively sharp peaks, whereas on formation of a hydrogen bond, XAH  Y (where Y is the acceptor atom), peaks are red shifted to lower frequencies and can become broader [22]. The red shift is caused by lengthening of the XAH and YAR bonds which results on hydrogen bond formation. Hence a stronger hydrogen bond will lengthen the bond more and produce a shift to a lower frequencies [23,24].

4. Discussion As expected the partial molar volumes at infinite dilution are greater with greater molar weight of the compounds (Table 3). The observed differences between solvents are far outside the estimated experimental error and stem from specific interactions with the particular solvent. Frequently, however, determination of the partial molar volume of the solute at infinite dilution is not feasible (low solubility, instability or adsorption, uncertainties in sample concentration). In such cases reliable calculations of volume quantities are required. A universal and easy-to use approach [11] allows empirical calculation of partial volumes of organic compounds, that are valid for aqueous solutions at 25 C. The method is based on Traube’s additivity principle, together with the concept of volume increments for atoms and/or atomic groups that allows correction for covolume, ring formation and ionization according to X X X V oc ¼ V i þ V CV  V RF  V ES : ð4Þ Vi is the volume increment for any atom or atomic group, VCV is the correction due to the covolume and VRF and VES take into account the decrease of volume caused by ring formation and ionization (electrostriction), respec-

A. Zega et al. / Journal of Molecular Structure 875 (2008) 354–363

tively. The covolume contribution, i.e. the excluded volume contribution, arising from the large size of the solute molecules relative to that of the solvent molecules, is independent of solvent conditions [25]. The volume increments for a broad spectrum of different groups are reported [11], together with the data for special increments and decrements (covolume, ring formation, ionization). Nifedipine, nimodipine and nitrendipine are non-electrolytes and their VES contributions were neglected in their treatment, but were considered for amlodipine besilate. It should be noted that the calculated partial volumes are only valid for aqueous solutions at 25 C. The validity of the additivity approach was confirmed by comparing more than 500 predicted and observed volumes of organic and biochemical compounds in aqueous solution [11], including small molecules and polymers of non-ionic and ionic nature, mono- and polyfunctional molecules, aliphatic and aromatic compounds, heterocycles of different ring size, etc. The vast majority of the calculated values were within ±2% of the experimental values. Some examples for compounds in organic solvents are also reported [11,26,27] and show quite reasonable agreement between calculated and experimental values. The deviations are discussed in terms of the solute–solvent interactions. The principle of the additivity of a functional group could also be used with the help of ACD/ChemSketch commercial program. In Table 3, the values obtained by the geometrical optimization are reported and, except for amlodipine besilate, are only 2% higher than the values obtained by applying Eq. (4). This difference could be ascribed to the 3D optimization of the structure made by the program. The large difference between V oc and V o3D for amlodipine besilate stems from the program’s inability to treat it as one compound. The volume contribution of amlodipine and benzenesulfonic acid were therefore calculated separately and then added together. In this way the correction due to the covolume, VCV = 12.4 cm3 mol1 [11], was taken into account twice, which is clearly not correct. Obviously, amlodipin besilate acts as an neutral ionpair, even in dilute solutions. In the further procedure therefore only the V oc values were discussed for this compound. The unambiguous dependence of apparent molar volumes, V/, on molality, i.e. Sm „ 0, indicates the presence of solute–solute interactions. Thus, at non-infinite dilution, solute–solute, rather than solute–solvent interactions, will be present. The slopes, i.e. Sm, presented in Table 3 are almost always negative. This is the expected behaviour for hydrophobic solutes [28] and non-electrolytes [27,29,30] in water. The usual interpretation of this behaviour is that the solute species interact with destructive overlap of their hydration spheres [30,31]. For apolar species the positive volume component of Vo, originating from hydrophobic hydration, starts to decrease as the solute concentration increases [31]. The overlap of two hydrophobic hydration co-spheres relaxes some water molecules from the solvation sphere to the bulk, giving rise to a neg-

359

ative change in volume [32]. For hydrophilic ionic species the volume of water molecules in the solvation shell is smaller, due to the effect of electrostriction and the smaller hydrogen bonded network of water molecules in the solvation sphere than in the bulk (the so-called structure breaking effect). The overlap of the co-spheres of two ionic species relaxes some solvation water to bulk, so that overall structure is increased, giving rise to a positive volume change [31]. It should be emphasized that the negative volume contribution due to hydrophobic interactions has been observed with some amino acids (leucine and its isomers, methionine), indicating that their side chain length increases the hydrophobicity [31]. However, no evidence exists in the literature for an analogy between hydrophobic hydration in aqueous solutions in non-aqueous solution (= solvophobic solvation). Even in a structured solvent like methanol, despite the ability of its molecules to undergo self-association, this behaviour was not observed [33]. The structure here is less complex than that of water, although the details are not known unambiguously. The negative dependence of V/ on molality of linear hydrocarbons in methanol [27] was ascribed to the attractive solute–solute short-range interactions. This could also be assumed for the observed dependence in our systems. All compounds (Fig. 1) could self-associate by intermolecular forces (dipole–dipole, even hydrogen bonding, dipole induced dipole. . .). The interactions become significant at higher concentrations. On the other hand, low solubility in water, methanol and even ethanol could be ascribed to competition between strong solvent– solvent interactions and weaker solvent–solute interactions. In spite of much investigation, however, the concentration dependence of apparent molar volume in non-aqueous media remains poorly understood at the molecular level. The experimental volumes of all the compounds are very close to those calculated for dichloromethane and DMSO (Table 3). In acetone the experimental values are slightly lower and in ethanol higher than those predicted by calculation. The basic assumption of the additivity of every group, i.e. that each group in a molecule behaves and interacts with the surrounding solvent independently of other groups, is clearly a crude approximation. The partial molar volume is a derivative of the chemical potential with respect to pressure. Strictly speaking it is a property of the entire system and cannot be attributed to one molecule or atomic group. Thus, the values of group contributions have a physical meaning only in the limit of relatively weak interactions with solvent. From this point of view the agreement between calculated and experimental values of the partial molar volumes, estimated at infinite dilution in solvents with very different properties (Table 3), is surprisingly good. Thus, the additivity principle, and even the ACD/ChemSketch program, provide reliable predictions of volume quantities of physical relevance, even for more complex species in non-aqueous media. As already noted, V oc will be considered in further discussion only

A. Zega et al. / Journal of Molecular Structure 875 (2008) 354–363

than that between that hydroxyl and the NH group. The predominating acceptor function of 1,4-DHP derivatives in ethanol solution results in the volume increase. Significant negative differences between the experimental and calculated molar volumes at infinite dilution were found for nifedipine, nitrendipine and nimodipine in acetone (25 < D/cm3 mol1 < 3). In water solutions, negative differences, D < 0, were observed for compounds containing groups that can act as donors (or both donors and acceptors equivalently) [12]. In acetone, nifedipine, nitrendipine and nimodipine can act as H-bond donors only (NH. . ..O@C). For these solutions only one m(C@O) band is observed, at 1712 cm1. In DMSO and dichloromethane solutions of 1,4-DHP derivatives, m(C@O) bands are observed between the positions of those observed in ethanol and acetone solutions. In dichloromethane, due to the electronegativity of the chlorine atom, there is a weak hydrogen bond interaction between the NH group of 1,4-DHP and the dichloromethane molecule, which causes slight increases of electron density of the dihydropyridine–carbonyl conjugated system in 1,4-DHP. As a result, the m(C@O) band is observed a little higher, at 1695 ± 2 cm1. A similar effect was observed in DMSO, due to the hydrogen bond between the NH proton of 1,4-DHP and the (S@O) group. The acidity (AN) and basicity (DN) of the solvent significantly affect the solvent induced frequency shift [15,35]. For various systems a linear dependence of m(C@O) on AN was observed and the fit to the experimental data was improved by adding the solvent DN as a second variable [15,35]. The correlation between the m(C@O) I band of nifedipine and the solvent AN (Fig. 4) shows that the m(C@O) is shifted to lower frequencies with an increase in the ability of the solvents to accept electrons, the solute acting as a Hbond acceptor, as has been found for various compounds in organic solvents [15,34]. No linear dependence was found in the present system, indicating that AN values

1715 20 1710 10 1705 0

1700

-10

1695 1690

-20

1685

-30 12

16

20

24

28

32

36

Δ/cm3 mol-1

where the additivity of the group without any additional optimization is applied. The difference, D ¼ V o  V oc , between experimental and calculated values for a given compound reflects the deviation of the actual volumetric behaviour of the solute molecules from that of a hypothetical molecule in which each group interacts with the solvent independently of the other groups. Thus, D values for polyfunctional molecules or ions are an approximate measure of the volume effects caused by intramolecular interactions, which take place between different centres, either directly (e.g. hydrogen bonding, charge induction, steric hindrance) or through the solvent (e.g. overlapping of the hydration co-spheres, solvent induced conformational changes) or electronic and nuclear polarization. The sign and magnitude of D depend on the nature of the interacting groups, their number, and relative distance between them [12]. The resulting differences depend on the solvent (Table 3): D is >0 in ethanol and <0 in acetone for all the compounds. In DMSO and dichloromethane the calculated values are within ±0.3–0.4% of the experimental values. In water solution it has been found [12] that D > 0 for compounds with at least one group that may act only as an acceptor of a hydrogen bond. Orientation of water molecules around polar centres should be a key factor in determining the volume effect associated with the overlapping of the solvation spheres. Molecular structures of the model compounds shown in Fig. 1 reveal that each compound contains only one proton donor, the dihydropyridine NH group. In contrast, a number of proton acceptors, carbonyl and nitro groups, are present in each compound. Only in ethanol solution can these compounds act as a H-bond donor and/or acceptor. The assumption of two types of hydrogen bond in ethanol solutions is confirmed by IR spectroscopy where two bands were observed for the m(C@O) of 1,4-DHP derivatives in ethanol. The two bands can be attributed to the two types of solute–solvent interaction between 1,4-DHP derivatives and ethanol. The m(C@O) (I) of 1,4-DHP derivatives in ethanol may be assigned to the hydrogen bond between the electron pair donor group of 1,4-DHP derivatives and the AOH group of ethanol. This interaction shifts the m(C@O) (I) of 1,4-DHP derivatives in ethanol to lower wavenumbers than those of 1,4-DHP in non-alcohol solvents. On the other hand, the NH group on the dihydropyridine ring in 1,4-DHP derivatives may donate a proton to the oxygen of ethanol. In this case, the electron density of the dihydropyridine–carbonyl conjugated system in 1,4-DHP increases. Therefore, m(C@O) (II) is shifted to higher wavenumbers. A similar result has been observed for the m(C@O) of methyl 4-hydroxybenzoate in a number of alcohols [34]. A considerable range of differences, D ¼ V o  V oc , was observed for ethanol solutions of nifedipine, nitrendipine and nimodipine, (14 < D (cm3 mol1) < 26). This strongly suggests that the interaction between the ethanolic hydroxyl and the carbonyl and/or ether group is more favourable

ν(C=O)I/cm-1

360

40

AN

Fig. 4. Plot of m(C@O) (m) and the differences between experimental and calculated values of partial molar volume at infinite dilutions, D ¼ V o  V oc , (n), of nifedipine versus the solvent acceptor number (AN). Lines are drawn to guide the eye.

A. Zega et al. / Journal of Molecular Structure 875 (2008) 354–363

do not take into account the steric effects of the sites of solute–solvent interaction. Consequently, there is no comparable solute–solvent interaction between different (though similar) compounds in the same solvent because of steric differences of the solutes and solvents. This was also found for carbonyl containing compounds in various organic solvents by NMR spectroscopy [36]. The difference, D, shows an inverse dependence on AN (Fig. 4). It can be assumed that, in solvents with low ability to accept an electron pair, the molar volume at infinite dilution is lower than predicted and in good electron pair acceptor solvents a higher value is to be expected. For solvents with intermediate AN values (20), e.g. DMSO and dichloromethane, the difference is almost in the range of experimental error. The deviations correlate well with the wave numbers of carbonyl stretching vibration m(C@O), and thus with the strength of the H-bond and of other interactions that influence the stretching vibration m(C@O).

361

The structures of complexes of nifedipine with one solvent molecule (acetone, DMSO, dichloromethane and ethanol) are shown in Fig. 5, while the corresponding frequencies of the carbonyl stretching (computed and experimentally determined by FTIR) are listed in Table 5. There is about 40 cm1 discrepancy between the experimental and computed values of the m(C@O) stretching band frequencies for nifedipine but much better agreement between the measured and calculated shifts of the carbonyl stretching frequencies.. For this purpose we have chosen the experimental value of 1708 cm1 as the value of free carbonyl stretching in ethanol solution. In order to allow comparison with experiment we considered in the computational study the average between the symmetric and the asymmetric carbonyl stretching. In all the spectra, except in those in ethanol, a single peak was observed for symmetric and asymmetric carbonyl stretching. Agreement between experimental and calculated frequency shifts is

a

b

c

d

Fig. 5. Complexes of nifedipine with one solvent molecule: (a) acetone, (b) DMSO, (c) CH2Cl2 and (d) ethanol, calculated at the B3LYP/6-31G(d) level.

Table 5 B3LYP/6-31G(d) harmonic frequencies associated with carbonyl stretching of nifedipine and one solvent molecule Calculated a

Dry Acetoneb DMSO CH2Cl2 Ethanol

Experimental

m(C@O) I

m(C@O) II

Frequency shift relative to the in vacuo value

Averaged frequency shift

m(C@O) I

dd

1770 1760 1750 1770 1741

1775 1771 1772 1774 1772

0 10 20 0 29

0 7 11.5 0.5 16

1712 1698 1697 1686 (1708c)

4 10 11 22

0 4 3 1 3

No frequency scaling was applied. Units: m(C@O), cm1. a Nifedipine in vacuo. b Note that the calculated acetone carbonyl stretching frequency in the complex was 1811 cm1 and was well separated from the nifedipine carbonyls. c Free carbonyl i.e. not involved in hydrogen bonding with ethanol; 1708 cm1 is considered to be the free carbonyl stretching frequency. d Relative to 1708 cm1.

362

A. Zega et al. / Journal of Molecular Structure 875 (2008) 354–363

good, especially given that only one solvent molecule was considered. Similar trends – shift of the m(C@O) stretching band towards lower frequencies and increase of D with increasing AN – could also be observed for nitrendipine and nimodipine (Tables 3 and 4). In these cases no computation of harmonic frequencies has been performed. Nitrendipine possesses only one ACH2A group more in the side chain than nifedipine but the ANO2 group is moved to the meta position. The empirically calculated volume increment for a methylene group is 16.1 cm3 mol1 [11], whereas the experimentally obtained differences Vo (nitrendipine)  Vo (nifedipine) are 29.6, 23.1, 21.2, 18.9 in acetone, DMSO, dichloromethane and ethanol solutions, respectively. It can be assumed that the position of the ANO2 group and the longer side chain, that causes steric hindrance in the solute–solvent interactions, contribute to the higher volume increment. The same is true for the nimodipine solutions. At first glance, amlodipine besilate exhibits quite extraordinary behaviour, although the experimental and calculated volume values agree within experimental error (Table 3). In the IR spectra in ethanol the first band at 1680 cm1 is rather weak while the second, at 1691 cm1, coincides well with the frequencies observed in DMSO and dichloromethane. The peaks are red shifted (Table 4) and are perceptibly broader, as observed with other compounds. This suggests that a stronger hydrogen bond is formed in all solvents (except in acetone) than is found with the other compounds. However, this conclusion contradicts the previous discussion on the interplay of hydrogen bond strength and volume properties and cannot easily be rationalized. þ Amlodipine besilate contains ASO 3 and ANH3 groups  (Fig. 1). The ASO3 group can only act as a hydrogen bond acceptor. The electric field produced by the neighbouring, opposite charge of ANHþ 3 may induce solvent electrostriction. These two competitive processes would be expected to lead to no change in the volume in ethanol. DMSO possesses a relatively high DN and therefore strong cation solvation would be expected, accounting for the solubility of amlodipine besilate in DMSO. Here also no difference in the volume could be ascribed to the interplay of solvation and electrostriction. Due to its low solubility, the apparent molar volumes of amlodipine besilate could not be studied in acetone or dichloromethane. Certain amino acids and other charged molecules exhibit similar behaviour [12]. Investigations on amlodipine by X-ray crystallography, small-angle X-ray scattering [37] and high-resolution differential scanning calorimetry [38] suggested essential differences in solvent interactions for amlodipine and other 1,4-DHP analogs. The solutes examined interact with the solvents in different ways. The former act mainly as hydrogen bond donors in acetone and as hydrogen bond acceptors in ethanol. The IR spectra indicate similar solute–solvent interactions in DMSO and dichloromethane, but not the volumetric prop-

erties. DMSO has a high donor number, high polarizability and large electrostatic factors (e, l), and is expected to interact with the solute in quite a different way from dichloromethane which has a small dipole moment but high polarizability and good electron donor properties. Interactions of solvent and solute depend critically on the electron distribution between donor and acceptor atoms in these substances, which in turn are determined by the bonds chemical nature and bond angles of neighbouring donor and acceptor atoms. In the case of aromatic nuclei, different substituents cause different charge distributions in the benzene ring with, consequently, different interactions with the solvent [12]. A plausible guess is that easier packing into the solvent structure, for geometrical and steric reasons, may be responsible for the lower difference between experimental and calculated partial molar volume at infinite dilution in dichloromethane and DMSO solutions.

5. Conclusions In this work the partial molar volumes at infinite dilution, Vo, of nifedipine, nitrendipine, nimodipine and amlodipine besilate in acetone, DMSO, dichloromethane and ethanol have been determined. The experimental values were compared with the values, V oc , obtained by using Traube’s additivity principle [11]. Deviations from the experimental data were discussed in terms of the solvents properties and solute–solvent interactions, particularly hydrogen bonding. A qualitative discussion about the contribution of the hydrogen bonding to the partial molar volumes was confirmed by the FTIR spectroscopy, supported also by the calculation of harmonic frequencies using medium–high quantum chemical calculations. For uncharged compounds a connection between solvent properties, defined by acceptor number, AN, and the volume differences, D ¼ V o  V oc , was found which correlated well with the m(C@O) stretching vibration and AN [15,35]. m(C@O) shifts towards lower values of wave numbers with increasing value of AN, and D values increase with increasing AN. The D values shift from negative values in acetone, with the lowest AN (12.5), to values close to zero in DMSO and dichloromethane with moderate AN (19.3, 20.4), to markedly positive values in ethanol. The compounds act as hydrogen bond donors in solvent with low AN (acetone) and as strong hydrogen bond acceptors in a solvent with high AN (ethanol). The investigated solvents can mimic to some extent the environment of drug production and delivery, and the environment from which the drug reaches the active sites (aqueous solution, membrane interior, membrane headgroup region and receptor binding site). It can be assumed that, for the interaction with the solvent environment, hydrogen at the N-position is crucial and can only act as a hydrogen bond acceptor (acetone, DMSO, dichloromethane), whereas in protic solvents (ethanol) more groups could

A. Zega et al. / Journal of Molecular Structure 875 (2008) 354–363

be involved and the compounds act as a hydrogen bond donor and/or acceptor. Further, the interactions of amlodipine besilate with the investigated solvents are at variance with those of the other compounds. A similar difference in their interactions with the membrane has also been reported [37,38]. Volumetric measurements proved to be a very useful method for investigating the interactions of charged and uncharged 1,4-DHP with different solvents and correlate well with results from FTIR spectroscopy and calculation. Our approach can be extended to considering several solvent molecules, performing thermal averaging and calculating the vibrational spectrum beyond the harmonic approximation. The methods are developed and ready to be used [39–41]. Calculation of antagonistic activity of this class of drugs using molecular simulation by treatment of the receptor site, ionic channel and solvent molecules at atomic resolution remains a challenge for computational biophysics. Acknowledgments The authors are grateful to Mrs. Lucija Rus, M.Pharm., for performing density measurements. Financial support of the Slovenian Research Agency is gratefully acknowledged (P1-0201, P1-0189 and P1-0012). References [1] D.A. Williams, W.O. Foye, T.L. Lemke, Foye’s Principles of Medicinal Chemistry, Lippincott Williams & Wilkins, 2002. [2] M.J. Eisenberg, A. Brox, A.N. Bestawros, Am. J. Med. 116 (2004) 35. [3] R.P. Mason, I.T. Mak, M.W. Trumbore, P.E. Mason, Am. J. Cardiol. 84 (1999) 16L. [4] I. Mak, P. Boheme, W. Weglicki, Circ. Res. 70 (1992) 1099. [5] G. Sobal, E.J. Menzel, H. Sinzinger, Biochem. Pharmacol. 61 (2001) 373. [6] C. Napoli, M. Chiariello, G. Palumbo, G. Ambrosio, Cardiovasc. Drugs Ther. 10 (1996) 417. [7] X.-Q. Zhu, B.-J. Zhao, J.-P. Cheng, J. Org. Chem. 65 (2000) 8158. [8] M.E. Ortiz, L.J. Nunez-Vergara, C. Camargo, J.A. Squella, Pharm. Res. 21 (2004) 428. [9] C. Lopez-Alarcon, H. Speisky, J.A. Squella, C. Olea-Azar, C. Camargo, L.J. Nunez-Vergara, Pharm. Res. 21 (2004) 1750. [10] Y.-Z. Mao, M.-Z. Jin, Z.-L. Liu, L.-M. Wu, Org. Lett. 2 (2000) 741. [11] H. Durchschlag, P. Zipper, Prog. Colloid Polym. Sci. 94 (1994) 20. [12] L. Lepori, P. Gianni, J. Solution Chem. 29 (2000) 405. [13] G. Eaton, M.C.R. Symons, Chem. Soc. Faraday Trans. I 84 (1988) 3459. [14] G. Eaton, M.C.R. Symons, P.P. Rastogi, J. Chem. Soc. Faraday Trans. I 85 (1989) 3257. [15] G. Rezaei Behbehani, M. Hamedi, F. Hoseinpour Rajabi, (accesed 01.03.07).

363

[16] J.A. Berson, E. Brown, J. Am. Chem. Soc. 77 (1955) 447. [17] R. Teraoka, M. Otsuka, Y. Matsuda, Int. J. Pharm. 184 (1999) 35. [18] J.M.G. Barthel, H. Krienke, W. Kunz, Physical Chemistry of Electrolyte Solutions: Modern Aspects, Springer, Steinkopff, Darmstadt, 1998. [19] Gaussian 03, Revision B.03, M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, J.A. Montgomery, Jr., T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P. Hratchian, J.B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y. Ayala, K. Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S. Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J. Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, J.A. Pople, Gaussian, Inc., Pittsburgh, PA, 2003. [20] G. Schaftenaar, J.H. Noordik, J. Comp. Aided Mol. Des. 14 (2000) 123. [21] P.W. Atkins, Physical Chemistry, seventh ed., Oxford University Press, 2002. [22] D. Lin-Vien, N.B. Colthup, W.G. Fateley, J.G. Grasselli, The Handbook of Infrared and Raman Characteristic Frequencies of Organic molecules, Academic Press Inc., San Diego, CA, 1991. [23] K. Nakamoto, M. Margoshes, R.E. Rundle, J. Am. Chem. Soc. 77 (1955) 6480. [24] X.C. Tang, M.J. Pikal, L.S. Taylor, Pharm. Res. 19 (2002) 477. [25] S.E. Harding, J.C. Horton, S. Jones, J.M. Thornton, D.J. Winzor, Biophys. J. 76 (1999) 2432. [26] L. Bernazzani, V. Mollica, M.R. Tine, Fluid Phase Equlibria 203 (2002) 15. [27] E.F.G. Barbosa, S.M.C. Sousa, M. Soledade, C.S. Santos, I.M.S. Lampreia, Phys. Chem. Chem. Phys. 3 (2001) 556. [28] G. Caron, J.E. Desnoyers, J. Colloid Interface Sci. 119 (1987) 141. [29] P. Bernal, A. Bunn, J. Logan, J. McCluan, J. Solution Chem. 29 (2000) 651. [30] P. Bernal, J. McCluan, J. Solution Chem. 30 (2001) 119. [31] A.K. Mishra, J.C. Ahluwalia, J. Phys. Chem. 88 (1984) 86. [32] W. Zielenkiewicz, O.V. Kulikov, I. Kulis-Cwikla, J. Solution Chem. 22 (1993) 963. [33] H. Piekarski, A. Pietrzak, J. Mol. Liq. 121 (2005) 46. [34] W.R. Fawcett, in: P. Politzer, J.S. Murray (Eds.), Theoretical and Computational Chemistry: Quantitative Treatments of Solute/Solvent Interactions, vol. 1, Elsevier, Amsterdam, 1994, p. 183. [35] Q. Liu, D. Fang, J. Zheng, Spectrochim. Acta A 60 (2004) 1453. [36] R. Nyquist, R. Streck, G. Jeschek, J. Mol. Struct. 355 (1996) 113. [37] R.P. Mason, S.F. Campbell, S.D. Wang, L.G. Herbette, Mol. Pharm. 36 (1989) 634. [38] R.P. Mason, M.F. Walter, N.W. Trumbore, E.G. Olmstead Jr., P.E. Mason, J. Mol. Cell Card. 31 (1999) 275. [39] J. Stare, J. Mavri, Comput. Phys. Commun. 143 (2002) 222. [40] J. Mavri, J. Grdadolnik, J. Phys. Chem. A 105 (2001) 2039. [41] J. Mavri, J. Grdadolnik, J. Phys. Chem. A 105 (2001) 2045.