Use of the Topliss scheme for the design of more effective chelating

Jul 1, 1988 - Use of the Topliss scheme for the design of more effective chelating agents for cadmium decorporation. Shirley G. Jones, Pramod K. Singh...
0 downloads 0 Views 497KB Size
Chem. Res. Toxicol. 1988,1, 234-237

234

Use of the Topliss Scheme for the Design of More Effective Chelating Agents for Cadmium Decorporation Shirley

G. Jones, Pramod K. Singh, and Mark M. Jones*

Department of Chemistry and Center i n Molecular Toxicology, Vanderbilt University, Nashville, Tennessee 37235 Received April 7, 1988 The Topliss scheme, which designates a special series of substituents for use in sequence in the search for suitably substituted aromatic compounds with biological properties superior to the parent unsubstituted aromatic compound, has been applied to the design of new chelating agents for the decorporation of cadmium from its deposits in vivo. Substituted benzylamines have been used in the syntheses of the corresponding substituted sodium N-benzyl-D-glucamine-N-carbodithioates, Le., RC6H4CH2N(CS2Na)CH2(CHOH)4CH20H (or RBGDTC): CF,BGDTC, R = 3-CF,; MeBGDTC, R = 4-CH,; MeOBGDTC, R = 4-OCH3; ClBGDTC, R = 4-C1. Examination of the behavior of these compounds as agents for the mobilization of cadmium from its deposits in vivo shows that such substitution can have an appreciable effect on the organ selectivity for cadmium removal, as well as the efficacy. Of the four compounds suggested by the Topliss scheme for pilot studies, two were much more effective than the parent compound (BGDTC), and of these the best one, sodium N-(4-methoxybenzyl)-~-glucamine-N-carbodithioate (MeOBGDTC) reduced kidney cadmium levels to 30% and liver cadmium levels to 50% of those obtained with the parent compound. Sodium N-[3-(trifluoromethyl)benzyl]-~-glucamine-Ncarbodithioate (CF,BGDTC) was also found to be superior to the parent compound in reducing the liver cadmium levels, though it was equivalent in reducing renal cadmium levels.

Introduction Of the various metals which are environmentally significant as toxic species, cadmium plays an unusual role because of the tenacity with which it remains in the human body. This is due in no small measure to the ease with which absorbed cadmium is converted into metallothionein, a cadmium-containing protein whose properties are responsible for the very long half-life of cadmium in man and other mammals (1,2). This protein keeps cadmium largely in intracellular spaces where it is not accessible to the majority of typical chelating agents (3-5). As a consequence there is currently no acceptable chelate therapy for chronic cadmium intoxication. The development of improved chelating agents for the removal of toxic metal ions has previously been carried out almost exclusively on the basis of analyses of parameters based on the coordination sphere of the toxic metal ion and its manipulation, either via alteration in stability constants via structural changes or by the elaboration of a more fully encompassing chelate structure (6). I t has thus traditionally been carried out by using methods which are quite different from those characteristically utilized by organic chemists. The present study was carried out to see if one of the schemes widely used by organic chemists could be applied to the design of more effective chelating agents for the mobilization of cadmium from its aged deposits in mammals. This is the scheme of Topliss (7, 8), which has been used to reduce the amount of synthetic work necessary to obtain optimum compounds which contain a benzene ring as an essential feature. Topliss’ scheme directs attention to a limited number of the possible substituted compounds of this type in order t~ arrive quickly at substituents which are optimal, or more nearly so than the starting material. The substituents in this scheme are selected to cover a range of electronic character and to lead the investigator through these in a systematic fashion which greatly reduces the number of syntheses 0893-228~/88/2701-0234$01.50/0

needed to arrive at superior compounds. The first step is the comparison of the 4-chloro compound ClBGDTC with the parent compound BGDTC and the determination whether it is less, equally, or more active than the parent compound. For each of these cases the next derivatives to be examined are specified and the process is repeated. The ability of sodium diethyldithiocarbamate (DEDTC) to act as an antagonist for acute cadmium intoxication in mice was observed by Gale and his co-workers (9). While it was subsequently shown that this compound enhanced the cadmium concentration in the brain (IO),it was also shown that this could be prevented by appropriate substitution of polar groups in the dithiocarbamate (11). The preparation and use of sodium N-methyl-D-glucamine-Ncarbodithioate (MGDTC) showed that this type of compound possessed low toxicity and was able to remove aged cadmium from its renal and hepatic deposits in mice (12). Kojima and his co-workers then prepared sodium Nbenzyl-D-glucamine-N-carbodithioate (BGDTC) and demonstrated that it was superior to the N-methyl compound (MGDTC) (13-16). In the search for a compound superior to the N-benzyl compound it became apparent that the Topliss procedure offered one possible route to such a compound which obviated the necessity of preparing a very large number of such compounds to screen for activity. The present study reports the results of the use of such procedures to develop a compound for the removal of cadmium from its aged intracellular deposits which is superior to sodium N-benzyl-D-glucamine-N-carbodithioate in efficacy. The reaction sequence used in the syntheses is shown in Figure 1.

Experimental Section Preparation of Substituted N-Benzyl-D-glucamines3a-d. The general procedure of Kagan et al. (17)was used with some modification. In a typical experiment 140.0 mmol of the substituted benzylamine la-d (Figure l),138.7 mmol of CY-D-(+)0 1988 American Chemical Society

Effective Chelating Agents for Cadmium Decorporation

1a.d

4550°C

4a-d

3a-d

a, R = 3-CF3; b, R = 4-CH,; c R = 4-KH3: d. R = 4-C1

Figure 1. Synthetic scheme used for the preparation of the substituted sodium N-benzyl-D-glucamine-N-carbodithioates (4a = CF,BGDTC; 4b = MeBGDTC; 4c = MeOBGDTC; 4d = ClBGDTC). glucose (2), and 7.5 mL of water were stirred mechanically under argon a t 60 "C until gel formation took place. The gel was dissolved in methanol (300 mL) with a little warming when necesary, 1.60 g of Pt02 catalyst was added, and the mixture was then subjected to hydrogenation (50 psi) a t 40-45 "C. After 24 h, the hydrogenation was stopped, and the reaction mixture was treated with 250 mL of methanol and heated on a steam bath to dissolve the product. The catalyst was separated by filtration, and the fitrate was concentrated. Cooling and standing overnight afforded a white crystalline solid (3a-d). The corresponding yields and m p were as follows: 3a (75%), mp 105-107 "C; 3b (79%), mp 153-156 "C; 3c (66%), m p 140.5-142 "C; 3d (47%), mp 122-130 "C. The products were further characterized by spectral analyses ('H NMR and IR). Preparation of Carbodithioates 4a-d. The method of Kojima et al. (13) was modified slightly. In general, to the stirred solution prepared on a steam bath from the substituted glucamine 3a-d (30.0 mmol) and a mixture of 25-40 mL of methanol and 200 mL of dioxane was added a solution of NaOH (30.5 mmol) in water (8-10 mL) a t 22 "C ( 5 min) under argon. After 30 min, CS2 (10 mL) in dioxane (50 mL) was added to the clear reaction mixture while being stirred at 5-10 "C over 30 min and then further stirred overnight. The yellowish white solid which formed was separated, washed with ether (3 X 50 mL), and dried in air and then in vacuo to give the product as a white amorphous solid. The analytical sample was prepared by recrystallization from ethanol. The products were stored under argon a t 0 "C. The yields and physical properties of the newly prepared compounds are described below, and the overall reaction sequenca is shown in Figure 1. The sodium N-benzyl-D-glucamine-N-carbodithioate (BGDTC) was prepared and characterized as described by Kojima e t al. (13). Sodium N- [3-(trifluoromethyl)benzyl] -~-glucamine-N-carbodithioate (CF,BGDTC) (4a-1H20): yield, 51%; mp 151-153.5 "C dec; 'H NMR (D20)6 7.45-7.28 (m, 4 H), 5.70 (d, J = 16.1 Hz, 1 H), 5.05 (d, J = 16.1 Hz, 1H), 4.36-4.32 (dd, J = 13.9, 3.8 Hz, 1 H), 4.25-4.22 (m, 1 H), 7.74-7.69 (dd, J = 13.7, 8.2 Hz, 1 H), 3.64-3.56 (m, 3 H), 3.49-3.43 (m, 2 H); IR (Nujol) 3350 s (br), 3175 s, 1610,1440,1400,1335 s, 1290,1255,1200 s, 1150,1230 s, 1100, 1080,1065,990,970,950,930,810,795,760,710,660cm-'. Anal. Calcd for Cl5Hl9F3NNaO5S2.1.OHZO:C, 39.56; H, 4.65; N, 3.08; S, 14.08. Found: C, 39.18; H, 4.49; N, 2.95; S, 14.37. Sodium N-(4-methylbenzyl)-~-glucamine-N-carbodithioate (MeBGDTC) (4b.1H20): yield, 65%; mp 192-195 "C dec; 'H NMR (D2O) 6 7.13 (d, J = 8.0 Hz, 2 H), 7.05 (d, J = 8.0 Hz, 2 H), 5.67 (d, J = 15.6 Hz, 1 H), 5.03 (d, J = 15.6 Hz, 1 H), 4.31-4.26 (dd, J = 13.5, 4.2 Hz, 1 H), 4.24-4.20 (m, 1 H), 3.76-3.74 (dd, J = 13.4, 7.9 Hz, 1H), 3.66-3.57 (m, 4 H), 3.51-3.45 (m, 1H), 2.19 (s, 3 H); IR (Nujol) 3325 s (br), 3150 s, 1600, 1510, 1435, 1400, 1365,1320,1290,1255,1195,1150s, 1120,1100,1070 s, 1040,990, 970, 935, 855, 840, 800, 775, 705 cm-'. Anal. Calcd for C15H22NNa05S2.1.0H20: C, 44.87; H, 6.03; N, 3.49; S, 15.97. Found: C, 45.10; H, 5.96; N, 3.34; S, 16.34. Sodium N-(4-methoxybenzyl)-~-glucamine-N-carbodithioate (MeOBGDTC) (4c.1.0H20): yield, 81%; mp 197-199 "C dec; 'H NMR (D2O) 6 7.11 (d, J = 8.5 Hz, 2 H), 6.86 (d, J = 8.5 Hz, 2

Chem. Res. Toxicol., Vol. 1, No. 4, 1988 235 Table I. Cadmium Levels following Chelate Treatment of Mice N, liver, kidney, brain, mOUD mice Dpm DDm DDb 5 42.7 f 4.4 20.8 f 1.9 55.9 f 17.2 Cd only 5 31.7 f 6.8' 11.8 f 5.28 39.0 f 9.5 Cd + BGDTC Cd + 5 19.2 f 2.8*td 12.6 f 3.p 35.4 f 5.5 CF3BGDTC 36.8 f 3.1 Cd + 5 28.9 f 3.gCse 11.1f 2.7h MeBGDTC Cd + 5 16.6 f 5.2'~~ 3.9 f l.liJ 36.9 f 11.9 MeOBGDTC Cd + ClBGDTC 5 28.8 f 6.7cf 11.4 f O.ah 31.4 f 4.7 controlsk 5 4.3 f 2.1 ppb 4.7 f 2.1 ppb 5.2 f 5.3 a-Cd-iSignificanceof differences: from Cd only: a, 0.01 < p 5 0.025; b, p 5 0.0001; c, 0.001 < p 50.005; g, 0.05 < p 50.10; h, 0.001 < p I0.005; i, p I0.0001. d-fsjSignificanceof difference from BGDTC (sodium N-benzyl-D-glucamine-N-carbodithioate): d , 0.001 < p 5 0.005; e, 0.025 < p 5 0.05; /, 0.05 < p 50.10; j , 0.01 < p I0.025. kCadmium levels in the normal (untreated) animals are all in ppb. H), 5.62 (d, J = 15.4 Hz, 1H), 4.99 (d, J = 15.4 Hz, 1H), 4.28-4.24 (dd, J = 13.6, 4.3 Hz, 1 H), 4.41-4.16 (m, 1 H), 3.76-3.68 (m, 4 H), 3.64-3.54 (m, 4 H), 3.50-3.44 (m, 1H); IR (Nujol) 3350 s (br), 3150,1610,1515,1400,1340,1300,1260 s, 1210,1180,1160,1120, 1100 s, 1070, 1025, 990,975, 955 s, 925,820, 800, 770, 730,670 cm-'. Anal. Calcd for Cl5HzzNNaO6S2~1.OHz0: C, 43.15; H, 5.79; N, 3.36; S, 15.36. Found: C, 42.71; H, 5.86; N, 3.29; S, 15.66. Sodium N-(4-chlorobenzyl)-~-glucamine-N-carbodithioate (ClBGDTC) (4d.1.0H20): yield, 35%; mp 168-170 "C dec; 'H NMR (D2O) 6 7.22 (d, J = 8.4 Hz, 2 H), 7.04 (d, J = 8.4 Hz, 2 H), 5.61 (d, J = 15.9 Hz, 1H), 4.98 (d, J = 15.9 Hz, 1H), 4.29-4.25 (dd, J = 13.7,4.2 Hz, 1 H), 4.20-4.16 (m, 1 H), 3.69-3.65 (dd, J = 13.7, 8.3 Hz, 1 H), 3.61-3.51 (m, 4 H), 3.46-3.40 (m, 1 H); IR (Nujol) 3310 s (br), 3110 s, 1590, 1380, 1275,1240, 1180, 1140 s, 1100,1085s, 1050 s, 1000,970,955,940,920,780,770,710,640 cm-'. Anal. Calcd for Cl4Hl9C1NNaO5S2-1.0H20; C, 39.86; H, 5.02; N, 3.32; S, 15.20. Found: C, 40.43; H, 5.29; N, 3.31; S, 15.42. Animal Studies. Of 35 male ICR mice (Harlan Industries, Indianapolis, IN), 30 of them weighing 31 f 2 g were loaded with cadmium via a series of ip injections amounting to 1, 3, 3, and 3 mg of CdC12.2.5H20/kgin 0.5 mL of 0.9% saline on consecutive days. After a 3-day interval they were randomly divided into six groups of five each. One group which had not been given cadmium served as a normal control and was given 0.9% saline for all injections. One group served as the cadmium control group and was given 0.9% saline instead of chelating agent. Each of the other groups was given an ip injection of one of the chelating agents a t a level of 1 mmol/kg each day for 5 consecutive days. Each injection contained the appropriate amount of chelating agent in 0.5 mL of 0.9% saline. A milky solution was obtained and used with 4b. Two days later the animals were sacrificed by cervical dislocation and dissected. Weighed amounts of each animal's liver, kidney, and brain were digested in pure nitric acid on a heating block (80 "C), taken to dryness, redissolved in 1% nitric acid solution made with pyrogen-free deionized water and analyzed for cadmium using a Perkin-Elmer Model 403 atomic absorption spectrometer. The liver and kidney analyses were done in the flame mode and the brain and normal control liver and kidney in the flameless mode. All of the compounds described in this paper have L D N values in excess of 5 mmol/kg. All animals were kept in an AALAC approved animal care facility during the course of the experiments and were provided with free access to food and water. The statistical analysis of the data was carried out by using standard analysis of variance methods.

Results The results obtained when these compounds were used t o remove c a d m i u m from its deposits i n mice are shown in Table I. All of the compounds are found t o be able t o remove c a d m i u m from its deposits i n both t h e liver a n d

236 Chem. Res. Toxicol., Vol. 1, No. 4, 1988

Jones et al.

Table 11. Theoretical and Experimental Substituent Sequences ranking factor rank order 4-OCH3(-0.27); 4-CH3 (-0.17); 4-H (0.00); electronic (a) 4-C1 (0.23); 3-CF3 (0.43) hydrophobic (T) 4-OCHs (-0.02); 4-H (0.00); 4-CHa (0.56); . 4 4 1 70.70); 3-CF3 (0.88) steric (MR) 4-H (0.00): 3-CFo (4.00): 4-CHo" (4.7): 441 . , . (4.8); 4:0CH3"(6.5) liver (% Cd lost) 4-H (26%); 4-CH3 (32%); 4-C1 (33%); 3-CF3 (55%); 4-OCH3 (61%) kidney ( % Cd lost) 3-CF3 (39%); 4-H (43%); 4-C1 (45%); 4-CH3 (47%); 4-OCH3 (81%) I ,

"

kidney under conditions where the majority of such cadmium is tightly bound in metallothionein. Of the four compounds evaluated, two (CF,BGDTC and MeOBGDTC) are clearly superior to the parent compound in removing cadmium from the liver and one (MeOBGDTC) is quite superior also in removing cadmium from the kidneys. All of the compounds are thus equal or superior to the parent compound in their overall ability to mobilize cadmium.

Discussion The mobilization of cadmium from its intracellular deposits is difficult to achieve. Within 24 h subsequent to the injection of cadmium into a rodent, the majority of the element is present in intracellular sites (31, and at the end of 3 days, most of this intracellular cadmium is present as the complex with metallothionein (1,2,4). These facts are responsible for the inability of many typical water soluble chelating agents to affect such deposits (5). Dithiocarbamates (18, 19) and 2,3-dimercapto-l-propanol (BAL) (20, 21) have both been shown to be capable of removing cadmium from such deposits. The behavior of dithiocarbamates in this respect is very strongly dependent upon the structure of the compound, especially the balance between polar and nonpolar groups (11, 19). The most effective compounds which remove cadmium from the liver and the kidney without carrying it to the brain are those containing both nonpolar and nonionizing polar groups. The removal of cadmium from its deposits in the liver is critical because this is a large organ in comparison with the kidneys and contains a very large percentage of the total body burden of cadmium. Both of the structural types of chelating agent which are effective in mobilizing cadmium cause large increases in the biliary excretion (20, 22). This is preferable to methods which increase the urinary excretion of cadmium because of the great sensitivity of the kidney to cadmium passing through it. The fact that all of the compounds prepared are equal or superior to the parent compound in removing cadmium from its aged deposits together with the variety of the electronic properties of the substituents used suggests that steric and hydrophobic factors are at least as important, if not more important, than electronic ones in determining the relative order of efficacy of these compounds. The distance between the substituents on the benzene ring and the dithiocarbamate group which is involved in the coordination of the cadmium is so large that little variation is expected in the values of the stability constants of the cadmium complexes, though such structural changes can have a significant effect on the membrane permeability. From an examination of the values of the substituent constants for electronic (a), hydrophobic (a),and steric (MR) interactions (23) shown in Table I1 and the corresponding rank order of these compounds for the reduction in the cadmium levels in the liver and the kidney, one notes that the methoxy group stands at an extreme (high or low)

for each series. From this one would expect that the combination of factors which would give superior compounds for the removal of cadmium from the liver or kidneys of rodents is a negative u, a negative r,and a large positive value for MR. These experiments demonstrate that the Topliss procedure can lead to chelating agents of significantly enhanced efficacy. In the present case the compounds studied were limited by the availability of suitable starting compounds, the water solubility of the products, and the toxicity of the resulting compounds. Thus preliminary studies on the 3,4-dichloro derivative showed that it was too toxic for serious consideration. The parent compound (BGDTC) is a compound of very modest toxicity with an LD50 for ip administration in mice of 11.1 mmol/kg (15) and the LD50 of MeOBGDTC estimated in the course of the current study with a minimum number of mice (ip) was found to be approximately 10 mmol/kg. The superior properties of the 4-methoxy-substituted (MeOBGDTC) compound are probably related to the greater ability of this compound to penetrate cellular membranes and react rapidly with cadmium present as cadmium metallothionein in the cells in the liver and kidneys of the experimental animals. While cadmium containing metallothionein is present in the nucleus and the cytosol of liver and kidney cells, the exact source of the cadmium which is mobilized by these agents is not presently known.

Acknowledgment. We acknowledge with thanks the support of this work by the National Institutes of Environmental Health Sciences through Grant ES 02638-07.

References (1) Webb, M. (1979) "The metallothioneins". In The Chemistry,

Biochemistry and Biology of Cadmium (Webb, M., Ed.) pp 195-266, Elsevier/North Holland Biomedical, Amsterdam. (2) Webb, M. (1986) "Role of metallothionein in cadmium metabolism". In Cadmium (Foulkes, E. C., Ed.) pp 281-337, Springer Verlag, Berlin. (3) Shaikh, Z. A., and Lucis, 0. J. (1972) "Biological differences in cadmium and zinc turnover". Arch. Enuiron. Health 24,410-418. (4) Cantilena, L. R., Jr., and Klaassen, C. D. (1982) "Decreased effectiveness of chelation therapy with time after acute cadmium poisoning". Toxicol. Appl. Pharmacol. 63, 173-180. (5) Waalkes, M. P., Watkins, J. B., and Klaassen, C. D. (1983) "Minimal role of metallothionein in decreased chelator efficacy for cadmium". Toxicol. Appl. Pharmacol. 68, 392-398. (6) Turowski, P. N., Rodgers, S. J., Scarrow, R. C., and Raymond, K. N. (1988) "Ferric ion sequestering agents. 18. Two Dihydroxamic acid derivatives of EDTA and DTPA". Znorg. Chem. 27, 474-481. (7) Topliss, J. G. (1972) "Utilization of operational schemes for analog synthesis in drug design". J. Med. Chem. 15, 1006-1111. (8) Topliss, J. G. (1977) "A manual method for applying the Hansch approach to drug design". J. Med. Chem. 20, 465-469. (9) Gale, G. R., Smith, A. B., and Walker, E. M., Jr. (1981) "Diethyldithiocarbamate in the treatment of acute cadmium poisoning". Ann. Clin. Lab. Med. 11, 476-483. (10) Jones, S. G., Basinger, M. A., Jones, M. M., and Gibbs, S. J. (1982) "Polarity and antidotal efficacy of dithiocarbamates in acute cadmium intoxication". Res. Commun. Chem. Pathol. Pharmacol. 40, 155-164. (11) Jones, S. G., and Jones, M. M. (1984) "Structure-activity relationships among dithiocarbamate antidotes for acute cadmium chloride intoxication". EHP, Enuiron. Health Perspect. 54, 285-290. (12) Shinobu, L. A,, Jones, S. G., and Jones, M. M. (1984) "Sodium N-methyl-D-glucamine dithiocarbamate and cadmium intoxication". Acta Pharmacol. Toxicol. 54, 189-194. (13) Kojima, S., Kaminaka, K., Kiyozumi, M., and Honda, T. (1986) "Comparative effects of three chelating agents on distribution and excretion of cadmium in rats", Toxicol. Appl. Pharmacol. 83, 516-524. (14) Kojima, S., Kaminaka, K., Kiyozumi, M., and Honda, T. (1986) "Effects of chelating agents on biliary and urinary excretion and

Effective Chelating Agents for Cadmium Decorporation tissue distribution of cadmium in rats”. Toxicol. Lett. 34, 41-46. (15) Kojima, S., Kiyozumi, M., Honda, T., Kaminaka, K., Oda, Y., and Senba, Y. (1987) “Effects of N-benzyl-D-glucaminedithiocarbamate on distribution and excretion of cadmium in rats“. Toxicology 45, 93-102. (16) Kojima, S., Kiyozumi, M., Honda, T., Senba, Y., Kaminaka, K., and Ohnishi, M. (1987) “Studies on poisonous metals. XVIII. Effects of several dithiocarbamates on tissue distribution and excretion of cadmium in rats’’. Chem. Pharm. Bull. 35,3838-3844. (17) Kagan, F., Rebenstorf, M. A., and Heinzelman, R. V. (1957) “The preparation of glycamines”. J. Am. Chem. SOC.57, 3541-3544. (18) Shinobu, L. A., Jones, S. G., and Jones, M. M. (1984) “Sodium N-methyl-D-glucamine dithiocarbamate and cadmium intoxication”. Acta Pharmacol. Toxicol. 54, 189-194. (19) Gale, G. R., Atkins, L. A., Smith, A. B., Jones, S. G., and Jones,

Chem. Res. Toxicol., Vol. 1, No. 4, 1988 237

M. M. (1987) “Amphipathic dithiocarbamates as cadmium antagonists N-cyclohexyl-N-sulfonahalkylderivatives”. Res. Commun. Chem. Pathol. Pharmacol. 58, 371-391. (20) Cherian, M. G. (1980) “Biliary excretion of cadmium in rat. IV. Mobilization of cadmium from metallothionein by 2,3-dimercaptopropanol”. J. Toxicol. Environ. Health 6, 393-401. (21) Rau, W., Planas-Bohne, F., and Taylor, D. M. (1987) “Influence of several chelating agents on the distribution and binding of cadmium in rats”. Human Toxicol. 6, 451-458. (22) Cikrt, M., Basinger, M. A., Jones, S. G., and Jones, M. M. (1986) “Structural effects in the dithiocarbamate-induced biliary excretion of cadmium”. J. Toxicol. Enuiron. Health 17,429-439. (23) Norrington, F. E., Hyde, R. M., Williams, S. G., and Wootton, R. (1975) “Physicochemical-activity relations in practice. 1. A rational and self-consistent data bank”. J. Med. Chem. 18, 604-607.