Huang F, Buchwald P, Browne CE, Farag HH, Wu WM, Ji F, Hochhaus G and Bodor N Receptor Binding Studies of Soft Anticholinergic Agents AAPS PharmSci 2001; 3 (4) article 30 (https://www.pharmsci.org/scientificjournals/pharmsci/journal/01_30.html).

Receptor Binding Studies of Soft Anticholinergic Agents

Submitted: April 26, 2001; Accepted: November 13, 2001; Published: November 28, 2001

Fenglei Huang¹, Peter Buchwald¹, Clinton E. Browne¹, Hassan H. Farag¹, Whei-Mei Wu¹, Fubao Ji¹, Guenther Hochhaus¹ and Nicholas Bodor¹

¹Center for Drug Discovery, College of Pharmacy, University of Florida, Gainesville, FL 32610-0497

Abstract

Receptor binding studies were performed on 24 soft anticholinergic agents and 5 conventional anticholinergic agents using 4 cloned human muscarinic receptor subtypes. The measured pK _i values of the soft anticholinergic agents ranged from 6.5 to 9.5, with the majority being in the range of 7.5 to 8.5. Strong correlation was observed between the pK _i s determined here and the pA ₂ values measured earlier in guinea pig ileum contraction assays. The corresponding correlation coefficients (r ² ) were 0.80, 0.73, 0.81, and 0.78 for pK _i (m₁ ), pK _i (m₂ ), pK _i (m₃ ), and pK _i (m₄ ), respectively. Quantitative structure-activity relationship (QSAR) studies were also performed, and good characterization could be obtained for the soft anticholinergics containing at least 1 tropine moiety in their structure. For these compounds, the potency as measured by the pK _i values was found to be related to geometric, electronic, and lipophilicity descriptors. A linear regression equation using ovality (O _e ), dipole moment (D ), and a calculated log octanol-water partition coefficient (QLogP) gave reasonably good descriptions (r = 0.88) for the pK _i (m₃ ) values.

Abbreviations:
[³ H]NMS: - N-[³ H]-methyl-scopolamine
AQC - (α-cyclopentylphenyl) 3-acetoxyquinuclidium chloride
DMPC - (α-cyclopentylphenyl) methyl-1,2-dimethylpyrrolidinium chloride
MDP - (hydroxymethyl)-3-diisopropylmethylammonium chloride-9-methylxanthene-9-carboxylate
MPC - (α-cyclopentylphenyl) 1-methylpyrrolidinium chloride
PCDT - methoxycarbonylphenylcyclopentylacetoxy-N,N-dimethyl-3-pyrrolidinium methyl sulfate
PCHA.Et - 2-phenyl-2-cyclohexen-1-carboxyl-Nα-ethoxycarbonylmethyltropinium methyl sulfate
PCHA.Me - 2-phenyl -2-cyclohexen-1-carboxyl -Nα-methoxycarbonylmethyltropinium methyl sulfate
PCHB.Et - 2-phenyl-2-cyclohexen-1-carboxyly-Nβ-ethoxycarbonylmethyltropinium bromide
PCHB.Me - 2-phenyl-2-cyclohexen-1-carboxyl-Nβ-methoxycarbonylmethyltropinium bromide
PCMS-1 - ethoxycarbonylphenylcyclopentylacetyl-N,N -dimethyltropinium methyl sulfate
PCMS-2 - methoxycarbonylphenylcyclopentylacetyl-N,N -dimethyltropinium methyl sulfate
PCPA.Et - phenylcyclopentyl-Nα-ethoxycarbonylmethyltropium methyl sulfate
PCPA.Me - phenylcyclopentyl-Nα-methoxycarbonylmethyltropium methyl sulfate
PCPB.Et - phenylcyclopentyl-Nβ-ethoxycarbonylmethyltropinum bromide
PCPB.Me - phenylcyclopentyl-Nβ-methoxycarbonylmethyltropinum bromide
PCTM - methoxycarbonylphenylcyclopentylacetoxy-ethyl-N,N,N-trimethylammonium methyl sulfate
p -F-HHSiD - p -fluoro-hexahydro-sila-difenidol hydrochloride
PMTR - phenylmalonic atropine analogues
PMTR.MeSOMe sulfonyl tropyl 3-phenylmalonate methyl chloride salt
PMTR.TR - di-(8-methyl-8-azabicyclo[3.2.1] oct-3-yl) phenylmalonate dimethiodide
PSTR.TR - di-(8-methyl-8-azabicyclo[3.2.1]oct-3-yl) phenylsuccinate dimethiodide
QSAR - quantitative structure-activity relationship
TMTR.Et - ethyl tropyl 3-thienylmalonate methosulfate salt
TMTR.iPr - isopropyl tropyl 3-thienylmalonate methoiodide salt

Introduction

Muscarinic receptor antagonists, belladonna alkaloids in particular, have been used for a long time to treat a variety of clinical conditions, such as peptic ulcer, asthma, and Parkinson's disease¹ . Muscarinic receptor antagonists have also been used as mydriatic/cycloplegic agents^2-4 and as experimental antiperspirants^5-7 . Because of the broad range of anticholinergic effects, treatment with belladonna alkaloids directed to a certain organ system almost always induces undesirable effects in other organ systems⁴ . These side effects include dry mouth, blurred vision, increased heart rate, bronchodilation, reduced bronchiolar secretions, decreased gastrointestinal motility, and reduced thermoregulatory sweating⁴ . Even topical applications of anticholinergics can lead to unwanted systemic side effects because of their absorption and drainage into the systemic circulation. In a series of attempts to separate desired therapeutic effects from toxic effects-that is, to improve the therapeutic index-several series of novel anticholinergic agents have been designed, synthesized, and tested in our laboratories since the early 1980s based on soft drug design approaches^8-10 .

Soft drugs are defined as biologically active, therapeutically useful chemical compounds (drugs) characterized by a predictable and controllable in vivo destruction (metabolism) to nontoxic moieties after achieving their therapeutic role^11,12 . Soft anticholinergics were intended for topical application, for example, to be used as antiperspirants¹³ or as mydriatic agents^14,15 . By incorporating an adequate metabolically labile moiety into their structure, they can be potent and locally active as anticholinergic agents, but with only minimal systemic anticholinergic effects due to their rapid metabolism in the systemic circulation. Thus, the overall therapeutic index is greatly improved^9,10 .

In parallel with our efforts directed toward the development of safer anticholinergics by soft drug approaches, considerable research has been directed toward the delineation of muscarinic receptors and receptor subtypes in the past 20 years. This was done in the hope that by understanding the function of muscarinic receptors and receptor subtypes, muscarinic antagonists selectively targeted to the particular symptom(s) can be made, and, therefore, the therapeutic index can be improved¹⁶ . Three muscarinic receptor subtypes were located and characterized by biochemical and functional studies. The M₁ receptor, which is involved in behavioral and cognitive functions, exists predominantly in the brain^17,18 . The heart is one of the rare tissues where only 1 muscarinic receptor subtype, M₂ , is present^16,19 . In secretory glands, the muscarinic receptors mediating the enhancement of secretion are of the M₃ subtype. M₃ also occurs in the smooth muscles of airways, the gastrointestinal tract, and the urinary bladder^20,21 . With advances in molecular pharmacology in the past 15 years, 5 muscarinic receptor subtypes (m₁ , m₂ , m₃ , m₄ , and m₅ ) were cloned from human tissue^22-24 . By functional studies, the cloned receptors have proved to be well correlated with the previously established M₁ , M₂ , and M₃ receptor subtypes^25,26. A tissue counterpart of m₄ has been found in the peripheral lung strip of the rabbit^27,28; however, the physiological function of M₄ has not been elucidated yet. The tissue and functional counterparts of m₅ have not been discovered yet.

Utilization of readily available cloned muscarinic receptor subtypes offers the possibility of studying the binding characteristics of muscarinic ligands in detail^25,29 and, therefore, can facilitate drug discovery efforts in the search for safer anticholinergic agents. Cloned receptors have been used to determine the potency of novel compounds and to determine the subtype selectivity of antimuscarinic agents^30-32 . Currently, several muscarinic receptor subtype-selective agents are in advanced clinical trials^33,34.

The aims of the present study were (1) to establish and validate the method of receptor binding using cloned human muscarinic receptors as a tool for the discovery of soft anticholinergics; (2) to examine the potency and subtype selectivity of the existing and newly synthesized anticholinergics; and (3) to investigate the quantitative structure-activity relationship (QSAR) of these soft anticholinergic agents.

Materials and Methods

Cloned m₁, m₂, m₃, and m₄ receptors were ordered from RBI (Boston, MA). Dissociation constants (K _D , nM) for N-[³ H]-methyl-scopolamine ([³ H]NMS) were also provided by RBI (m₁ 0.166, m₂ 0.24, m₃ 0.11, m₄ 0.06).

Pirenzipine and (±)-p -fluoro-hexahydro-sila-difenidol hydrochloride (p -F-HHSiD) were obtained from Sigma (St Louis, MO). [³ H]NMS was obtained from DuPont NEN Research (Boston, MA). Scintiverse BD was obtained from Fisher Scientific (Pittsburgh, PA). Atropine, scopolamine, propantheline, and all other reagents were from Sigma Chemicals (St Louis, MO). All soft anticholinergics were synthesized at the Center for Drug Discovery, University of Florida.

Synthesis

Ethyl tropyl-3-thienylmalonate methosulfate salt (4a) and isopropyl tropyl-3-thienylmalonate methoiodide salt (4b) (Figure 1 ).

The synthesis of 4a and 4b were performed by appropriate modifications of previously reported methods³⁵ . Physicochemical data are listed below.

Ethyl 3-thienylmalonic acid (2a)

m.p. 74~75°C; ¹ H-NMR (CDCl₃ ) δ 8.70 (brds, 1H, CO₂ H ), 7.40-7.16 (m, 3H, C4H ₃ S), 4.81 (s, 1H, CH (CO₂ R)₂ ), 4.80 (q, 21-f, OCH ₂ CH,), 1.29 (t, 3H, CH₂ CH ₃ ). Anal. for C₉ H₁₀ O₄ S; Calcd.: C% 50.46, H% 4.70, S% 14.97; Found: C% 50.38, H% 4.75, S% 15.07.

lsopropyl 3-thienylmalonic acid (2b)

m.p. 82-83°C, ¹ H-NMR (CDCl₃ ) δ 8.60 (brds, 1H, CO₂ H ), 7.39-7.16 (m, 3H, C₄ H₃ S), 5.09 (m, 1H, CH (CH₃ )₂ ), 4.76 (s, 1H, H C(CO₂ R)₂ ), 1.28 (d, 3H, CH₃ ), 1.24 (d, 3H, CH₃ ). Anal. for C₁₀ H₁₂ O₄ S; Calcd.: C% 52.62; H% 5.30, S% 14.05; Found: C% 52.33, H% 5.28, S% 14.27.

Ethyl tropyl 3-thienylmalonate oxalate salt (3a)

m.p. 112-114°C; ¹ H NMR (CD₃ OD) δ 7.40- 7.18 (n, 3H,C₄ H ₃ S), 5.10 (brds, 1H, H-3 of tropine), 4.95 (s, 1H, CH (CO₂ R)₂ ), 4.12 (q, 2H, OCH₂ CH₃), 3.80 (brds, 2H, H-1, H-5 of tropine), 2.75 (s, 3H, NCH ₃ ), 2.50-1.90 (m, 8H, tropine), 1.25(t, 3H, CH₂ CH ₃). Anal. for C₁₉ H₂₅ NO₈ S; Calcd. C% 53.38, H% 5.89, N% 3.28, S% 7.50; Found: C% 53.30, H% 5.94, N% 3.24, S% 7.50.

Isopropyl tropyl 3-thienylmalonate oxalate salt (3b)

m.p. 140-142°C; ¹ H-NMR (CD₃ OD) δ 7.50-7.18 (n, 3H, C₄ H ₃ S), 5.10 (brds, 1H, H-3 of tropine), 5.05 (m, 1H, CH (CH₃ )₂ ), 4.97 (s, 1H, CH (CO₂ R)₂), 3.80 (brds, 2H, H-1 and H-5 of tropine), 2.75 (s, 3H, NCH ₃ ), 2.50-1.95 (m, 8H, tropine), 1.30 (d, 3H, CH ₃), 1.25(d, 3H, CH ₃ ). Anal. for C₂₀ H₂₇ NO₈ S; Calcd. C% 54.44, H% 6.16, N% 3.17, S% 7.26; Found: C% 54.32, H% 6.32, N% 3.07, S% 7.14.

Ethyl tropyl 3-thienylmalonate methosulfate salt (4a)

m.p. 122-124°C; ¹ H-NMR (CO₃ OD) δ 7.40-7.10 (m, 3H, C₄ H ₃ S), 5.10 (m, 1H, H-3 of tropine), 4.95 (brds, 1H , CH (CO₂ R)₂ ), 4.10 (q, 2H, OCH₂ CH₃), 3.75(m, 2H, H-1 and H-5 of tropine), 3.55 (S, 3H, OCH₃ ), 3.10, 3.00 (s, 6H, N(CH₃ )₂ ), 2.60-1.80 (n, 8H, tropine), 1.15 (t, 3H, CH₂ CH₃ ). Anal, for C₁₉ H₂₉ NO₈ S₂ Calcd. C% 49.23, H% 6.31, N% 3.02. S% 13.83; Found: C% 49.50, H% 6.31, N% 3.07, S% 13.76.

Isopropyl tropyl 3-thienylmalonate methoiodide salt (4b)

m.p. 228°C (dec): ¹ H-NMR (CO₃ OD) δ 7.50-7.10 (m, 3H, C₄ H₃ S), 5.15 (m, 1H, H-3 tropine), 5.05 (m, 1H, CH (CH₃ )₂ ), 4.80 (brds, 1H, CH (CO₂ R)), 3.85(m, 2H, H-1 and H-5 of tropine), 3.18, 3.10 (s, 6H, N(CH₃ )₂ ), 2.70-1.90 (n, 8H, tropine), 1.25, 1.20 (d, 6H, CH(CH ₃ )₂ ). Anal. for C₁₉ H₂₈ INO₄ S; Calcd. C% 46.25, H% 5.72, N% 2.84, S% 6.50; Found: C% 46.17, H%, 5.70, N% 2.83, S% 6.46.

Synthesis of di-(8-methyl-8-azabicyclo[3.2.1]oct-3-yl)phenylmalonate dimethiodide (10a) and di-(8-methyl-8-azabicyclo[3.2.1]oct-3-yl)phenylsuccinate dimethiodide. (ditropinyl phenylsuccinate ester diquaternary, 10b).

The synthesis of 10a and 10b were performed by appropriate modifications of previously reported methods³⁵ . The physicochemical data are listed below.

Di-(8-methyl-8-azabicyclo[3.2.1]oct-3-yl)phenylmalonate dimethiodide (10a).

FAB-MS (w/NBA + Na) 441.2 (M-CH₃ ), 583.2 (M + I), 733.1 (M + 21 + Na). ¹ H-NMR (CDCl₃ ) δ 7.5 (s, 5H, C₆ H₅ ); 5.2 (t, 2H, 3-tropyl Hs); 4.76 (s, 1H, C₆ H₅ -CH ); 3.1 and 3.18 (two singlets, 12 H, two N⁺ (CH₃ )₂ ), 3.8 (broad doublet, 4H) and 2.8-1.65 (m, 16 H) the rest of tropyl hydrogens.

Di-(8-methyl-8-azabicyclo[3.2.1]oct-3-yl)phenylsuccinate dimethiodide (ditropinyl phenylsuccinate ester diquaternary, 10b).

FAB-MS (w/Gly) 455.2 (M-CH₃ ). ¹ H-NMR (DMSO-d₆ ) δ 7.38 (s, 5H, C₆ H₅ ); 5.0 (m, 2H, 3-tropyl Hs); 4.2 (m, H, C₆ H₅ -CH ); 3.18, 3.15, 3.6, and 3.0 four singlets each for 3H (2 N(CH₃ )₂ ), 3.9-1.2 (multiplets, 20H, rest of tropine Hs).

Stability Studies of Soft Anticholinergics in the Receptor Environment

On the basis of previous studies^9,10 , the most labile soft anticholinergics from each series of soft anticholinergics were chosen to test the integrity of the compound during incubation with the esterase enzyme inhibitor. AQC, PMTR.Et, PCMS-1, PCDT, and PCTM were chosen to test the stability in the receptor media.

Method

The incubation buffer was phosphate-buffered saline (PBS), with 10 mM NaF as esterase enzyme inhibitor. The incubation mixture (1 mL) contained 500 µL diluted receptor membrane (prepared in incubation buffer), 100 µL of 100 mM soft anticholinergic (dissolved in buffer), and 400 µL buffer. Incubation was carried out at room temperature (about 23°C) for 60 minutes. At the end of incubation, 0.1 mL of aliquots were removed, and 0.9 mL 5% DMSO in acetonitrile was added, vortexed, and centrifuged at 1000 rpm for 15 minutes. Aliquots of the supernatant (20 µL) were injected into HPLC for analysis. The HPLC system was adopted from previously reported studies³⁶ .

Radioligand Binding Assay

Binding studies were performed with [³ H]NMS following the protocol from RBI. The binding buffer, pH 7.4, consisted of 0.15 M NaCl, 1.5 mM KH₂ PO₄ , and 2.7 mM Na₂ HPO₄ . NaF 10 mM was added to the buffer as an esterase inhibitor. The assay mixture (1 mL) contained 100 µL diluted membranes (receptor proteins, final concentration: m₁ 25 µg/mL, m₂ 42 µg/mL, m₃ 15.9 µg/mL, m₄ 20 µg/mL). Final concentrations of [³ H]NMS for the m₂-m₄ binding studies were 0.5 nM and 1 nM for m₁ . Specific binding was defined as the difference between the [³ H]NMS binding in the absence and presence of 1 µM atropine. Incubation was carried out at room temperature for 60 minutes. The assay was terminated by filtration through a Whatman GF/B filter (presoaked with 0.5% polyethyleneimine). The filter was then washed 3 times with 10 mL ice-cold binding buffer, transferred to vials, and added with 10 mL of Scintiverse liquid. Finally, detection was performed on a Packard 31800 liquid scintillation analyzer (Packard Instrument, Downer Grove, IL).

Data Analysis

To obtain the Hill coefficients, n , data from the binding experiment were fitted to the following equation: % [³ H]NMS bound = 100 - [100χⁿ / k / (1 + χⁿ /k )]. Next, they were fitted to the % [³ H]NMS bound = 100 - [100χⁿ / IC₅₀ / (1+χⁿ /IC50)] equation in order to obtain the IC₅₀ values. Here, χ denotes the concentration of the tested compound (in a series concentration). K _i was derived by the method of Cheng and Prusoff ³⁷ : K _i = IC₅₀ /(1 + L/K _D ), where L is the concentration of the radioligand, IC₅₀ is the concentration of drug causing 50% inhibition of specific radioligand binding, and K _D is the dissociation constant of the radioligand-receptor complex. Data were analyzed by a nonlinear least-squares curve-fitting procedure using the Scientist software (MicroMath, Salt Lake City, UT).

Guinea pig ileum assay (pA2 value)

Standard guinea pig ileum method^35,38 was used to determine the pA ₂ values of the soft drugs. Dose response curves were plotted, and pA ₂ values were calculated using a Schild plot.

Quantitative structure-activity relationship (QSAR)

All structures were optimized using the Am₁ advanced semi-empirical quantum chemical method³⁹ on a Silicon Graphics Origin 2000 server with the Sybyl molecular modeling program (Tripos, St Louis, MO). Keywords used were as follows: Am₁ PRECISE POLAR CHARGE = 1. Twenty-five descriptors were studied in our search for possible relevant parameters: molecular weight (MW); van der Waals molecular volume (V ) and surface area (S ) together with the corresponding ovality (O ) as a shape descriptor⁴⁰ ; effective van der Waals molecular volume (V _e ) and surface area (S _e ) together with the corresponding ovality (O _e ) calculated with a new procedure and a slightly different van der Waals radii set⁴¹ ; Am₁ calculated dipole moment (D ), average polarizability (α ), ionization energy (I ), and heat of formation; HOMO-LUMO energies (E _HOMO , E _LUMO ); absolute electronegativity (χ ), calculated as the negative of the average HOMO and LUMO energies, and absolute hardness (η ), calculated as half the HOMO-LUMO difference; calculated log octanol-water partition coefficient (log P _o/w ) using BLOGP⁴⁰ , QLogP^41,42 , and MLOGP⁴³ ; calculated log water solubility (log W ) using BLOGW⁴⁴ ; Am₁-calculated partial atomic charges on the quaternary nitrogen (q _N ) atom and the sp² carbon and oxygen atoms of the ester moiety (q _C= , q _O= ); the distance between the quaternary nitrogen atom and the sp² carbon of the ester moiety (d _C-N ); and the inaccessible solid angle around these atoms (Ω_C, Ω_N ) as a measure of steric hindrance⁴⁵ .

Results

Stability studies showed that all tested compounds retained at least 90% integrity during the 60-minute incubation period. Table 1 presents the mean pK _i ± SEM values obtained for each compound in the receptor binding studies. The pK _i values obtained by us for atropine, scopolamine, p -F-HHSiD (m₃ selective agent), and pirenzepine (m₁ selective agent) were in good agreement with published data^25,26,29,31 . The Hill coefficients, n , for the above compounds were not significantly different from unity, indicating that drug-receptor interactions obeyed the law of action and that binding took place at only 1 site. This further validates the method used here to evaluate the binding of the soft anticholinergics. However, the Hill coefficients obtained for some soft anticholinergics were significantly different from unity. Theoretically, n is an integer that reflects the number of molecules binding to a specific drug receptor. Normally, the binding of classical antagonists to muscarinic receptors is well described by the simple Langmuir isotherm, indicating a Hill coefficient close to unity⁴⁶ . Low Hill coefficients are often attributed to either recognition by the antagonist of more than 1 receptor site or conformation, or to interaction of the antagonist with a second binding site on the receptor molecule, which causes a negative cooperative effect on the first site^{30, 47,48} . None of these seems to apply in our situation. It is possible that small amounts of the inactive metabolite generated from the hydrolysis of soft drugs interfered with the binding of the parent compounds at the extremely low concentrations used in these studies (10^-4 to 10_-11 M), causing the Hill coefficients (n ) to deviate from unity. In preliminary experiments, for several soft anticholinergics bound to m₃ , we observed n values below 0.8 before adding the esterase inhibitor NaF. Addition of NaF boosted the Hill coefficients to values around 0.8. Nevertheless, the exact reasons as to why n was significantly different from unity still needs further investigation. It should be pointed out that the soft anticholinergics concentrations used in the receptor binding studies were much lower than those used in the stability studies because of the HPLC detection limit. Even if the Hill coefficient (n ) strongly influences the B_max (maximum binding) value, it has very little effect on the estimation of IC₅₀ if a sigmoidal E _max model is used⁴⁹ . Therefore, pK _i values, which are derived from IC₅₀ ³⁷ , should represent valid potency estimates even under the present experimental conditions.

A validation of the receptor binding measurements on cloned muscarinic receptors is the correlation of pK _i values with the pA ₂ values determined by the guinea pig ileum contraction method. The correlation between pA ₂ and different receptor subtype pK _i are fairly good as characterized by the r ² values (n = 18): 0.803, 0.734, 0.813, and 0.781 for m₁ , m₂ , m₃ , and m₄ , respectively. Even if differences are relatively small, the strongest correlation was observed between pA ₂ and pK _i (m₃ ) values (Figure 3 ), which is in agreement with the fact that M₃ mediates smooth muscle in airway and gastrointestinal tract contraction³⁴ . Determination of the pA ₂ value for guinea pig ileum contraction has been a classical functional study for anticholinergic affinity toward the M₃ receptor. For soft anticholinergics, the pA ₂ values obtained from these studies were generally comparable to the pK _i values obtained from m₃ binding studies, even though in most cases, the pA ₂ values were somewhat lower than the corresponding pK _i values of m₃ binding. Relative values for each tested compound were essentially the same for either method. Because the receptor-binding assay allows faster screening, it has an advantage over pA ₂ value determination as a method to measure relative potency in anticholinergic compounds. Furthermore, the latest research indicates that muscarinic receptors in guinea pig ileum are heterogeneous with a major M₂ receptor population (~80%) and a minor M₃ population (~20%). The function of the minor M₃ population is clearly related to contraction, but the function of the predominate M₂ population is yet unclear⁵⁰ . Therefore, m₃ receptor-binding data provide a more reliable estimate of intrinsic activity toward the M₃ receptor.

Twenty-four soft anticholinergic agents were included in the present study. They can be divided into 2 major groups: compounds containing at least 1 tropine moiety within their structure (Table 2 and Table 4 ) and compounds that contain no such moiety (Table 3 and Table 5 ). Compounds in the first group were designed by using the inactive metabolite-based soft drug approach^11,12 . Those in the second group, except PCDT and PCTM, were designed by using the soft analogue approach^11,12 .

PMTR.Et, PMTR.cHx, and PMTR.Hx are soft analogues of atropine^{35, 51} . These 3 compounds were obtained by replacing the OH groups of the methatropine moiety with different ester groups. Because atropine does not show muscarinic subtype selectivity, it is not expected that these compounds will exhibit subtype selectivity. The rank order of potency is generally m₁ = m₃ >> m₂ . Studies of these 3 compounds indicated that substitution of the OH group with an ester in methatropine tends to reduce potency. This will be further discussed in the QSAR section of this paper. PMTR.MeSOMe is a similar structure containing a sulphone group, and it has moderate potency.

DMPC, MPC, AQC, and MDP (Table 3 and Table 5 ) were intended for the suppression of perspiration. MDP is an analogue of propantheline and, hence, a possible anti-ulcer agent as well. All of these compounds were designed using the soft analogue approach, and they displayed higher potency in receptor-binding studies (pK _i > 8.5) than most of the soft drugs designed using the inactive metabolite-based approach (pK _i < 8.5). None of these compounds showed muscarinic receptor subtype selectivity. PCTM and PCDT are analogues of glycopyrrolate. Slight subtype selectivity was observed for PCTM (m₃ /m₂ binding ratio of around 8).

TMTR.Et (4a ), TMTR.iPr (4b ), PMTR.TR (10a ), and PSTR.TR (10b ) are newly synthesized soft anticholinergics. Their syntheses proceeded smoothly following appropriate modifications of previously established methodology³⁵ (Figure 1 , Figure 2 ). TMTR.Et and TMTR.iPr are soft anticholinergics similar to PMTR.Et, PMTR.cHx, and PMTR.Hx, but with a thienyl moiety replacing the phenyl in an attempt to enhance potency. However, this replacement did not significantly increase the potency. PMTR.TR and PSTR.TR (Figure 2 ) contain 2 tropine moieties in their structures. Binding results indicated that the increase of the number of tropine moieties did not enhance the potency either.

PCPA.Me, PCPA.Et, PCPB.Me, PCPB.Et, PCHA.Me, and PCHA.Et are soft anticholinergics with the "soft spot" (the metabolically labile moiety) introduced at the quaternary nitrogen head. Of these compounds, only PCHA.Me and PCHB.Me showed muscarinic receptor subtype selectivity (m₃ /m₂ ).

Quantitative structure-activity relationships (QSARs)

Because the compounds included in the study contain quaternary nitrogens and the QLogP method was not previously tested for estimation of the log distribution coefficient (log D ) of such compounds, the possibility of extending this model for such predictions was evaluated. QLogP uses only 2 parameters to estimate the log octanol-water partition coefficient, log P _o/w = 0.032V _e - 0.723N , where V _e is the calculated effective van der Waals molecular volume and N is a mainly additive parameter assigned by a fully automated algorithm that has integer values and is related to hydrogen bond acceptor ability^41,42 . Because the distribution of charged molecules is pH- and counterion-dependent⁴² , experimentally determined log D values of permanently charged molecules are less reliable than experimental log partition coefficients (log P ), which by definition describe the partitioning of neutral molecules. Nevertheless, as data in Table 6 indicate, the extension of QLogP could be reasonably well performed by assuming a contribution of N = 9 for the positively charged quaternary nitrogen functionality of these molecules. It is reassuring that this N = 9 value is the same as the one used earlier in the extension of the QLogP model to describe the distribution coefficient of quaternary pyridinium compounds⁴² , and this can be regarded as an additional proof of the consistency of this method.

Because of the structural diversity of the compounds included in the present study, general size, shape, lipophilicity, solubility, electronic, and steric parameters were employed in the present QSAR analysis. Our initial efforts to correlate activity with physicochemical parameters for all soft anticholinergics included in this study did not prove very successful: None of the included parameters was found to be a good descriptor of the measured receptor binding data (pK _i ). Size descriptors (eg, V, S, MW) proved again to be the best, but they were far less relevant than previously found for pA ₂ data⁹ . This is true even despite the relatively good correlation found between pA ₂ and pK _i data (Figure 1 ). For example, V _e accounts for close to 65% of the variance in pA ₂ (r ² = 0.63, n = 18), but for only about 16% of the variance in m₃ pK _i (r ² = 0.16, n = 29) (Figure 2 ). Note, however, that compounds with available pK _i data contain larger structural variety than those with available pA ₂ data. Inclusion of second- or higher-order terms gave no significant improvement.

However, when only the soft anticholinergics with 1 or 2 tropine moiety in their structures were included, significantly improved correlation was found between activity and geometric descriptor, such as volume (V _e ), surface area (S _e ), and ovality (O _e ). Significant correlation was also found between activity and electronic descriptor(s) such as dipole moment (D ). It is not unreasonable to separate the soft anticholinergics with a tropine moiety in their structure from the other soft anticholinergics in the present study. In the compounds containing the tropine moiety, 3 carbon atoms separate the quaternary nitrogen head from the ester function. In contrast, in the compounds of Table 5 , there is a significantly shorter separation of only 1 (or sometimes 2) carbons. It is very likely that for the corresponding 2 series of soft anticholinergics, there are 2 distinct receptor-ligand complexes in terms of conformation, since the spatial structure of muscarinic receptors was considered to contain 4 binding sites⁵² .

In order to establish a quantitative relationship between physicochemical parameters and activities (pK _i s), the correlations between pK _i (m₁ to m₃ ) and the 25 physicochemical parameters included were examined with the S-Plus program (Mathsoft, Seattle, WA). Parameters showing statistically significant correlations with the pK _i values were selected, and the intercorrelations among these parameters were further examined. Of the parameters showing strong intercorrelation (r > 0.8), only 1 was kept for the final study. For example, strong intercorrelation was found among O, S, V, O _e , S _e , and V _e ; therefore, only O _e was kept for the final study as it showed the strongest correlation with the pK _i values. For the final QSAR study, 8 physicochemical descriptors were selected with this procedure (Table 7 ), including a geometric descriptor (O _e ), a lipophilicity descriptor (QLogP), a water-solubility descriptor (BLOGW), and 5 electronic descriptors (D, I, χ, q _C =, and q _O =).

Linear regressions were obtained by using the stepwise regression mode of the S-Plus program. Following the usual methodology, only those parameters that improved (when added) or decreased (when deleted) the goodness-of-fit in a statistically significant manner were kept in the final regression model.

Table 8 presents the obtained regression equations for pK _i (m₁ ), pK _i (m₂ ), and pK _i (m₃ ), and Table 9 gives a correlation matrix between selected physicochemical descriptors and pK _i values. QSAR for pK _i (m₄ ) was not performed at this time because the physiological significance of m₄ has not been clearly established yet. For all 3 pK _i s, O _e was a major factor. Ovality, O _e , is a parameter that describes the overall size and shape of the molecules^40,41 . It is defined as the ratio between the area of the surface of the molecule and that of the minimum surface corresponding to the volume of the molecule (that is, the surface of a sphere with a volume that equals the volume of the molecule). Hence, it represents a size and shape descriptor, as it depends on the shape of the molecule and, for common organic molecules, it also scales with size. Our previous 2 structure-activity relationship studies already indicated an important role played by molecular size in this class of soft anticholinergics^{9, 53} . In the present study, molecular size as measured, for example, by volume is still highly correlated with muscarinic binding activity in the analyzed 3 receptor subtypes; however, ovality (O _e ) seems to be somewhat more relevant. This suggests that, in addition to size, molecular shape might also influence activity in this series of soft anticholinergics.

For m₁ , the best QSAR equation found was pK _i (m₁ ) = 0.69 - 5.16(±1.23)O _e + 33.57(±13.86)q _C = - 20.12(±6.92)q _O =. This indicates that besides geometric descriptors, electronic properties-namely, the partial atomic charges on the sp² carbon (q _C =) and oxygen (q _O =) atoms of the ester moiety-also influence the binding affinity of the soft anticholinergic to the m₁ receptor. For m₂ , the QSAR equation obtained was pK_i (m₂) = -3.68 - 3.27(±1.19) O_e + 37.57 (±13.32)q_C= - 17.84(±6.27)q_O=. This is similar to that obtained for m₁.

For the m₃ receptor binding data, details on the stepwise improvement of the linear regression for the soft anticholinergic agents with a tropine moiety in their structure are included in Table 8 . The pK _i (m₃ ) were correlated in decreasing order with ovality (O _e , r = 0.65), dipole moment (D, r = 0.53), and calculated lipophilicity (QLogP, r = 0.39). Lipophilicity has been long recognized as an important factor determining antimuscarinic activity^54-56 . Because O _e and D are essentially perpendicular (show very little intercorrelation: r = -0.038, Table 9 ) and both of them are well correlated with pK _i (m₃ ), the combination gives a reasonably good description:

pK _i (m₃ ) = 22.63 - 6.66(±1.32)O _e - 0.10(±0.02)D

n = 18, r = 0.85, SE = 0.42, F = 21.61

This clearly indicates that for the compounds included here, both size/shape (O _e ) and electronic properties (D ) influence receptor binding. Addition of a lipophilicity parameter, QLogP, (eq. 7, Table 8 ) provides some improvement in the correlation (r = 0.88, SE = 0.39) and is statistically justified (at a p < 0.06 level), but because on the included data O _e and QLogP are intercorrelated, the improvement is less significant.

Discussion

It has to be mentioned again that unlike in our previous studies on guinea pig ileum pA ₂ s^{9, 53} , in this study size descriptors do not account for a majority of the variance in activity; in addition, the role of electronic properties seems to be more important. Part of the reason for this deviation might be the fact that pA ₂ values are not as closely related to the contraction mediated by the M₃ receptor subtype as previously thought. The latest research indicates that muscarinic receptors in guinea pig ileum are heterogeneous with a major M₂ receptor population (~80%) and a minor M₃ population (~20%). M₃ has been clearly shown to mediate contraction, but the role M₂ plays in smooth muscle contraction is not completely clear at present. Its role might be related to the inhibition of the relaxation of the smooth muscle^{33, 50} . Therefore, p K _i (m₃ ) values, which are solely related to the m₃ receptor subtype, could deviate from pA ₂ values, which might be related to mediation by both M₂ and M₃ receptor subtypes.

The electronic character of the molecules has already been considered as one of the most important parameters related to the antagonist activity of anticholinergic agents^54,57,58 . This is particularly true for the present set of molecules. The presence of a free hydroxyl group in methatropine and methscopolamine and the presence of an ester group near the tropine quaternary nitrogen head in several compounds will certainly influence the overall binding ability of the compounds to the muscarinic receptor. In an aqueous environment, the effect of electrostatic interactions may less important, but in a secluded receptor-ligand complex, the electronic effects of various substituents may easily alter the interaction of the ligand with the receptor. Many investigators have noted that esters with an OH group in the acyl side chain were generally associated with greater antimuscarinic activity. Recanatini and co-workers⁵⁹ proposed that the presence of the OH group might influence the conformation of the molecule so that the group might interact with additional binding sites at the receptor. Other investigators related such increases to the change of the electronic properties of the overall molecule caused by the electronic withdrawing properties of the OH group⁵⁸ or to its ability to interact by electrostatic interaction or hydrogen bonding with the receptor⁵⁷ . Our results seem to support the latter suggestions because electronic properties have been shown to be major factors determining antimuscarinic activity in our present QSAR study.

Conclusion

In conclusion, the receptor binding assay based on cloned human muscarinic receptors was proved to be a valid screening method to determine the potency of soft anticholinergic agents. Two soft anticholinergics were shown to have moderate muscarinic receptor subtype selectivity. A QSAR study performed on soft anticholinergics containing tropine moiety within their structures showed that the pK _i values are related to geometric properties (O _e ), eletronic parameters (q _C =, q _O= , and D ), and lipophilicity (QLogP).

References

1. Brown JH, Taylor P. Muscarinic receptor agonists and antagonists. In: Goodman LS, Gilman A, Hardman JG, Gilman AG, Limbird LE, eds. Goodman & Gilman's The Pharmacological Basis of Therapeutics. New York: McGraw-Hill; 1996:141-160.

2. Birkhimer LJ, Jacobson PA, Olson J, Goyette DM. Ocular scopolamine induced psychosis. J Fam Prac. 1984;18(3):464-469. [PUBMED]

3. Fraunfelder FT. Mydriatics and cycloplegics. In: Drug Induced Ocular Side Effects & Drug Interactions. New York: Lea & Febiger; 1989:445-458.

4. Katzung B. Cholinoceptor-blocking drugs. In: Katzung B, ed. Basic & Clinical Pharmacology. Stamford, CT: Appleton & Lange; 1998:105-117.

5. Macmillan FSK, Reller H, Synder FHT. The antiperspirant action of topically applied anticholinergics. J Invest Dermatol. 1964;42:363-377.

6. Shelley WB, Horvath PN. Comparative study on the effect of anticholinergic compounds on sweating. J Invest Dermatol. 1951;16:267-274.

7. Stoughton RB, Chiu F, Fritsch W, Nurse D. Topical suppression of eccrine sweat delivery with a new anticholinergic agent. J Invest Dermatol. 1964;42:151-155.

8. Ji F, Huang F, Juhasz A, Wu W, Bodor N. Design, synthesis, and pharmacological evaluation of soft glycopyrrolate and its analog. Pharmazie. 2000;55(3):187-191. [PUBMED]

9. Juhasz A, Huang F, Ji F, Buchwald P, Wu WM, Bodor N. Design and evaluation of new soft anticholinergic agents. Drug Dev Res. 1998;43:117-127.

10. Kumar GN, Bodor N. Soft anticholinergics. Curr Med Chem. 1996;3:23-36.

11. Bodor N, Buchwald P. Soft drug design: general principles and recent applications. Med Res Rev. 2000;20(1):58-101. [PUBMED]

12. Bodor N. Novel approaches to the design of safer drugs: Soft drugs and site- specific chemical delivery systems. In: Testa B, ed. Advance in Drug Research. Orlando, FL: Academic Press; 1984:255-331.

13. Bodor N, Woods R, Raper C, Kearney P, Kaminski JJ. Soft drugs. 3. A new class of anticholinergic agents. J Med Chem. 1980;23(5):474-480. [PUBMED]

14. Hammer RH, Wu W-M, Sastry JS, Bodor N. Short acting soft anticholinergics. Curr Eye Res. 1991;10:565-570. [PUBMED]

15. Kumar GN, Hammer RH, Bodor NS. Soft drugs--XVI. Design, evaluation and transdermal penetration of novel soft anticholinergics based on methatropine. Bioorg Med Chem. 1993;1(5):327-332. [PUBMED]

16. Caulfield MP. Muscarinic receptors--characterization, coupling and function. Pharmacol Ther. 1993;58(3):319-379. [PUBMED]

17. Hammer R, Giachetti A. Muscarinic receptor subtypes: m₁ and m₂ biochemical and functional characterization. Life Sci. 1982;31(26):2991-2998. [PUBMED]

18. Watson M, Yamamura HI, Roeske WR. A unique regulatory profile and regional distribution of [3H]pirenzepine binding in the rat provide evidence for distinct m₁ and m₂ muscarinic receptor subtypes. Life Sci. 1983;32(26):3001-3011. [PUBMED]

19. Waelbroeck M, Tastenoy M, Camus J, et al. Binding and functional properties of antimuscarinics of the hexocyclium/sila-hexocyclium and hexahydro-diphenidol/hexahydro-sila-diphenidol type to muscarinic receptor subtypes. Br J Pharmacol. 1989;98(1):197-205. [PUBMED]

20. Doods HN, Mathy MJ, Davidesko D, van Charldorp KJ, de Jonge A, van Zwieten PA. Selectivity of muscarinic antagonists in radioligand and in vivo experiments for the putative m₁, m₂ and m₃ receptors. J Pharmacol Exp Ther. 1987;242(1):257-262. [PUBMED]

21. Mutschler E, Lambrecht G. Selective muscarinic agonists and antagonists in functional tests. Trends Pharmacol Sci. 1984;5 (suppl.):39-44.

22. Bonner TI, Buckley NJ, Young AC, Brann MR. Identification of a family of muscarinic acetylcholine receptor genes. Science. 1987;237(4814):527-532. [PUBMED]

23. Bonner TI, Young AC, Brann MR, Buckley NJ. Cloning and expression of the human and rat m5 muscarinic acetylcholine receptor genes. Neuron. 1988;1(5):403-410. [PUBMED]

24. Peralta EG, Ashkenazi A, Winslow JW, Smith DH, Ramachandran J, Capon DJ. Distinct primary structures, ligand-binding properties and tissue-specific expression of four human muscarinic acetylcholine receptors. Embo J. 1987;6(13):3923-3929. [PUBMED]

25. Buckley NJ, Bonner TI, Buckley CM, Brann MR. Antagonist binding properties of five cloned muscarinic receptors expressed in CHO-K1 cells. Mol Pharmacol. 1989;35(4):469-476. [PUBMED]

26. Bolden C, Cusack B, Richelson E. Antagonism by antimuscarinic and neuroleptic compounds at the five cloned human muscarinic cholinergic receptors expressed in Chinese hamster ovary cells. J Pharmacol Exp Ther. 1992;260(2):576-580. [PUBMED]

27. Dorje F, Levey AI, Brann MR. Immunological detection of muscarinic receptor subtype proteins (m₁-m5) in rabbit peripheral tissues. Mol Pharmacol. 1991;40(4):459-462. [PUBMED]

28. Lazareno S, Buckley NJ, Roberts FF. Characterization of muscarinic m₄ binding sites in rabbit lung, chicken heart, and NG108-15 cells. Mol Pharmacol. 1990;38(6):805-815. [PUBMED]

29. Dorje F, Wess J, Lambrecht G, Tacke R, Mutschler E, Brann MR. Antagonist binding profiles of five cloned human muscarinic receptor subtypes. J Pharmacol Exp Ther. 1991;256(2):727-733. [PUBMED]

30. Barbier P, Renzetti AR, Turbanti L, et al. Stereoselective inhibition of muscarinic receptor subtypes by the eight stereoisomers related to rociverine. Eur J Pharmacol. 1995;290(2):125-132. [PUBMED]

31. Wess J, Lambrecht G, Mutschler E, Brann MR, Dorje F. Selectivity profile of the novel muscarinic antagonist UH-AH 37 determined by the use of cloned receptors and isolated tissue preparations. Br J Pharmacol. 1991;102(1):246-250. [PUBMED]

32. Waelbroeck M, Lazareno S, Pfaff O, et al. Stereoselective recognition of the enantiomers of phenglutarimide and of six related compounds by four muscarinic receptor subtypes. Br J Pharmacol. 1996;119(7):1319-1330. [PUBMED]

33. Eglen RM, Waston N. Selective muscarinic receptor agonists and antagonists. Pharmacol Toxicol. 1996;78:59-68. [PUBMED]

34. Grimm U, Moser U, Mutschler E, Lambrecht G. Muscarinic receptors: Focus on presynaptic mechanisms and recently developed novel agonists and antagonists. Pharmazie. 1994;49(10):711-726. [PUBMED]

35. Hammer RH, Amin K, Gunes ZE, Brouillette G, Bodor N. Novel "soft" anticholinergic agents. Drug Des Deliv. 1988;2(3):207-219. [PUBMED]

36. Brouillette G, Kawamura M, Kumar GN, Bodor N. Soft drugs. 21. Design and evaluation of soft analogs of propantheline. J Pharm Sci. 1996;85(6):619-623. [PUBMED]

37. Cheng YC, Prusoff WH. Relationship between the inhibition constant (ki) and the concentration of inhibitor which causes 50 per cent inhibition (IC50) of an enzymatic reaction. Biochem Pharmacol. 1973;22:3099-3180. [PUBMED]

38. Van Rossum JM, Hurkmans JH, Walters CJ. Cumulative dose response curves. Arch Int Pharmacodyn Therap. 1965;143:299-330.

39. Dewar MJS, Zoebisch EG, Healy EF, Stewart JJP. Am₁: a new general purpose quantum mechanical molecular model. J Am Chem Soc. 1985;107:3902-3909.

40. Bodor N, Gabanyi Z, Wong C-K. A new method for the estimation of partition coefficient. J Am Chem Soc. 1989;111:3783-3786.

41. Bodor N, Buchwald P. Molecular size based approach to estimate partition properties for organic solutes. J Phys Chem B. 1997;101:3404-3412.

42. Buchwald P, Bodor N. Octanol-water partition: searching for predictive models. Curr Med Chem. 1998;5(5):353-380. [PUBMED]

43. Moriguchi I, Hirono S, Liu Q, Nakagome I, Matsushita Y. Simple method of calculating octanol/water partition coefficient. Chem Pharm Bull. 1992;40:127-130. [PUBMED]

44. Bodor N, Huang MJ. A new method for the estimation of the aqueous solubility of organic compounds. J Pharm Sci. 1992;81(9):954-960. [PUBMED]

45. Buchwald P, Bodor N. Quantitative structure-metabolism relationships: steric and nonsteric effects in the enzymatic hydrolysis of noncongener carboxylic esters. J Med Chem. 1999;42(25):5160-5168. [PUBMED]

46. Hulme EC, Birdsall NJ, Burgen AS, Mehta P. The binding of antagonists to brain muscarinic receptors. Mol Pharmacol. 1978;14(5):737-750. [PUBMED]

47. Hulme EC, Berrie CP, Birdsall NJ, Burgen AS. Two populations of binding sites for muscarinic antagonists in the rat heart. Eur J Pharmacol. 1981;73(2-3):137-142. [PUBMED]

48. Nathanson NM. Molecular properties of the muscarinic acetylcholine receptor. Annu Rev Neurosci. 1987;10:195-236. [PUBMED]

49. Hochhaus G, Derendorf H. Dose optimization based on pharmacokinetic-pharmacodynamic modeling. In: Derendorf H, Hochhaus G, eds. Handbook of Pharmacokinetic/Pharmacodynamic Correlation. Boca Raton, FL: CRC Press; 1995:79-120.

50. Eglen RM, Reddy H, Watson N, Challiss RA. Muscarinic acetylcholine receptor subtypes in smooth muscle. Trends Pharmacol Sci. 1994;15(4):114-119. [PUBMED]

51. Hammer RH, Gunes E, Kumar GN, Wu WM, Srinivasan V, Bodor NS. Soft drugs--XIV. Synthesis and anticholinergic activity of soft phenylsuccinic analogs of methatropine. Bioorg Med Chem. 1993;1(3):183-187. [PUBMED]

52. Waelbroeck M, Camus J, Tastenoy M, et al. Stereoselectivity of procyclidine binding to muscarinic receptor subtypes m₁, m₂ and m₄. Eur J Pharmacol. 1990;189(2-3):135-142. [PUBMED]

53. Kumar G, Huang MJ, Hammer R, Bodor N. Soft drugs. 17: Quantitative structure-activity relationships of soft anticholinergics based on methatropine and methscopolamine [letter]. J Pharm Sci. 1994;83(1):117-118. [PUBMED]

54. Banerjee S, Lien EJ. Quantitative correlations and reexamination of the importance of hydrophobic and steric factors in anticholinergic drug receptor interactions. Pharm Res. 1990;7(7):746-750. [PUBMED]

55. Nilsson BM, Sundquist S, Johansson G, et al. 3-Heteroaryl-substituted quinuclidin-3-ol and quinuclidin-2-ene derivatives as muscarinic antagonists. Synthesis and structure-activity relationships. J Med Chem. 1995;38(3):473-487. [PUBMED]

56. Nordvall G, Sundquist S, Johansson G, Glas G, Nilvebrant L, Hacksell U. 3-(2-Benzofuranyl)quinuclidin-2-ene derivatives: Novel muscarinic antagonists. J Med Chem. 1996;39(17):3269-3277. [PUBMED]

57. Feriani A, Gaviraghi G, Toson G, et al. Cholinergic agents structurally related to furtrethonium. 2. Synthesis and antimuscarinic activity of a series of N-[5-[(1'-substituted-acetoxy) methyl]-2-furfuryl]dialkylamines. J Med Chem. 1994;37(25):4278-4287. [PUBMED]

58. Xu R, Sim MK, Go ML. Synthesis, antimuscarinic activity and quantitative structure-activity relationship (QSAR) of tropinyl and piperidinyl esters. Chem Pharm Bull (Tokyo). 1998;46(2):231-241. [PUBMED]

59. Recanatini M, Tumiatti V, Budriesi R, et al. Synthesis, muscarinic blocking activity and molecular modeling studies of 4-DAMP-related compounds. Bioorg Med Chem. 1995;3(3):267-277. [PUBMED]

60. Huang F. Design, Synthesis, Pharmacokinetic, and Pharmacodynamic Evaluation of a New Class of Soft Anticholinergics [PhD thesis]. Gainesville: University of Florida; 1999.

61. Hansch C, Leo A, Hoekman D. Exploring QSAR. Hydrophobic, Electronic, and Steric Constants. Vol. 2. Washington, DC: American Chemical Society; 1995.