Quantitative structure-activity relationships of insecticides and plant growth regulators: comparative studies toward understanding the molecular mechanism of action.

Emphasis was put on the comparative quantitative structure-activity approaches to the exploration of action mechanisms of structurally different classes of compounds showing the same type of activity as well as those of the same type of compounds having different actions. Examples were selected from studies performed on insecticides and plant growth regulators, i.e., neurotoxic carbamates, phosphates, pyrethroids and DDT analogs, insect juvenile hormone mimics, and cytokinin agonistic and antagonistic compounds. Similarities and dissimilarities in structures required to elicit activity between compounds classes were revealed in terms of physicochemical parameters, provoking further exploration and evoking insights into the molecular mechanisms of action which may lead to the development of new structures having better qualities.


Introduction
Among various quantitative structure-activity correlation (QSAR) procedures, the Hansch approach has been most widely and effectively used, covering diverse fields of medicinal drugs and agrochemicals (1)(2)(3). It assumes that the potency of a specified biological activity exerted by a series of compounds can be analyzed by an equation composed of terms of various physiocochemical parameters assignable to the structures of the compounds.
Mathematically, the assumption is represented by Eq.
log(1/C) = awr + pa + +S .+..... + const. (1) Here, c is the concentration (or dose) of congeneric members that gives a standard response such as EC,,, LD50, etc., on a molar basis; r is the hydrophobic substituent parameter defined from oil/water (generally, 1-*Department of Agricultural Chemistry, Faculty of Agriculture, Kyoto University, Kyoto 606, Japan. octanol/water) partition coefficients P as Irn log P. -log PH where subscripts denote substituted and unsubstituted compounds (4). u is the Hammett constant for the electronic property of substituents derived from dissociation constants of benzoic acids (5). Depending on the situation, the hydrophobicity parameter of the whole molecule, log P, and the electronic parameter for aliphatic substituents such as a* (6) and a, (7) can sometimes be used in place of Fr and a. S is the steric parameter. According to the mode of steric interactions involved, the Taft E, (6,8), the Verloop STERIMOL (9), the van der Waals volume (10) or another parameter is used as the steric parameter. In some cases, the squared terms for hydrophobic and steric parameters are required to account for the optima for these effects. Factors for hydrogen bonding (11) and parameters for other intramolecular forces such as molecular refractivity (12) may be needed in certain cases. By the leastsquares method and statistical examinations, the regression coefficients a, p, and 8 are determined, specifying structural factors contributing to variations in the potency. Since the effect of structural variations is separated into components, and significant physiochemical factors are indicated quantitatively, this method may reveal molecular mechanisms involved in processes leading to the elicitation of biological activity. It is especially effective to apply this procedure to the direct comparison of structural requirements for activity between different compound classes showing the same type of activity as well as between the same type of compound classes showing different actions.
In this article, some of our recent QSAR studies on various sets of insecticides and plant growth regulators are reviewed to show the versatility of this procedure not only to elucidate the molecular mechanisms of biological activity but also to examine them by quantitative comparisons.

Insecticides Antiacetylcholinesterase Phenyl N-Methylcarbamates and O,O-Dimethylphosphates
Aryl N-methylcarbamates and O,O-dialkylphosphorothionates are two major groups of compounds to which a number of agricultural insecticides belong, the representative members being carbaryl (I) and fenitrothion (II). The phosphorothionates are usually oxidized in vivo, gated. Depending upon the position and nature of substituents, the value of Kd showed significant variations, whereas that of k2 did not, in each of the carbamates and phosphates. Thus, formation ofthe reversible complex was considered to be the step which governs the variation in overall inhibitory activity.
Equation (3) is the result of analysis of the Kd (in M) from data obtained using bovine erythrocyte AChE (15) for o-, m-and p-substituted phenyl N-methylcarbamates (III). 0 III log (1/1(K) = (2) After establishing experimental conditions to determine a reliable set of kinetic parameters Kd( = k -,k,) and k2, for the inhibitory reaction (14), the molecular mechanism of enzyme inhibition was extensively investi-ir is derived from the experimentally determined 1-octanol/water partition coefficients, and the subscripts 2,3, and 4 indicate substituents at the ortho, meta, and para positions, respectively. The slopes of the X2,3 and XT4 terms suggest that the hydrophobic nature of the enzyme surface interacting with the ortho and meta positions is approximately equivalent, and higher than that of the surface interacting with the p-position. &' is the electronic parameter which, it is supposed, contains no through-resonance effect (16). The effects of ortho substituents were treated according to our recently developed procedure (17,18) in which the "ordinary" electronic effect of ortho substituents was taken as being equivalent to that of para substituents. The substituents are classified in terms of electronic effect into two groups: those in one group are more electron withdrawing and promote an attack by a nucleophile of the enzyme on the carbonyl cabon of the carbamyl group, and those in the other group are more electron-releasing and assist an electrophilic attack by an acidic group of the enzyme on the carbonyl oxygen atom. Substituents in the first group are those at the ortho position, and those which are electron-withdrawing at the para position, such as NO2, CN, and acyl. Their electronic effect is expressed by the°term. All other substituents belong to the second group, the electronic effect of which *In this and following equations, n is the number of compounds used in the regression, r is the multiple correlation coefficient and s is the standard deviation. The figures in parentheses are 95% confidence intervals. is represented by the &r<c term. The significance of these two terms in Equation (3) suggests different  mechanisms for the two groups of substituents, leading  to a common tetrahedral intermediate as shown in Figure 1. Electron-releasing ortho substituents do not follow the negative p mechanism because the acid-catalytic site of the enzyme does not fit the carbonyl oxygen atom due to hindrance exerted by these ortho substituents.
E. is the Taft-Kutter-Hansch steric parameter (8), the reference of which is shifted to that of H and F is the Swain-Lupton-Hansch field effect constant (12,19), both for ortho substituents. The coefficient values of these terms, 0.17 and 0.77, are very close to those for the alkaline hydrolysis of ortho substituted phenyl acetates (17). Thus, in support of the above discussion, the proximity effects of ortho substituents are considered to be those on the tetrahedral intermediate formation.
HB, an indicator variable for the hydrogen bonding effect of substituents, is 1 for hydrogen bonding substituents such as o-OR, m-acyl, -CN, -NO2 and -NMe2, but otherwise is zero. The significance of the term in Equation (3) indicates a specific hydrogen bond formation of these groups with a hydrogen donor on the enzyme. The hydrogen donor site is supposed to be located unsuitably for interaction with other hydrogen bonding groups such as o-NO2, -CN and m-OR. Figure  2A shows the stereospecific situation schematically.
The mechanism of the inhibition reaction of the same series of compounds against AChE prepared from fly heads was found to be quite similar to that against bovine erythrocyte AChE [Equation (4)] (20).
One of the slight differences to be noted is that the effect of hydrogen bonding substituents is represented by two indicator variable terms, HB1 for o-OR, -CN, -NO2, m-CN, -NO2, and -acyl, and HB2 for m-OR. The hydrogen donor group of fly-head AChE is suggested to be more suitably located for interaction with m-OR than that on the bovine erythrocyte AChE as shown in Figure 2B.
Carbamates are metabolically detoxified by a variety of biochemical reactions in the body of the housefly. Among these, the oxidative metabolism generally associated with the mixed-function oxidases has been shown to be of prime importance (21). Under conditions where oxidative metabolism was suppressed with piperonyl butoxide, the insecticidal LD50 (in mole/head) values against housefly were determined. QSAR analysis gave Equation (5) which resembles Equation (4) for AChE inhibition but the slope of each term is generally lower, suggesting that a detoxification mechanism(s) other than oxidation may be involved in the whole-body activity. log(l/LD,0) = 0.375Th23 -0. 0821T The insecticidal activity values of compounds with strongly electron-withdrawing substituents such as NO2 and CN are not included in Equation (5). They were considerably lower than those to be expected from their inhibitory activity against AChE, suggesting the possibility of spontaneous hydrolysis during the test period. These results were helpful in designing new compounds having higher activities. Polysubstituted derivatives having a hydrophobic OR group at the ortho position and a hydrophobic alkyl group at the meta position were synthesized and, of these, the 2-isopropoxy-5-n-and -s-butyl derivatives showed high potency as expected (20).
The Kd (in M) value for O,O-dimethyl-O-phenyl phosphates (IV) determined with fly-head AChE was simi-0 C O_ P*OCH3 IV larly analyzed to give Equation (6) (K. Kamoshita  The effect of ortho substituents is only the "ordinary" electronic effect as represented by the corresponding &'(para) value, and the proximity factors expressible by F and Es parameters as in Equation (3) are insignificant. The through-resonance is also insignificant for ortho substituents, probably due to the lack of coplanarity of the side chain.
The above QSAR analyses indicate that the detailed substituent effects differ between the two series of compounds. For carbamates, the electronic effect of substituents is position-specific as well as biphasic. Regiospecific hydrophobic and hydrogen-bonding factors operate in addition. For phosphates, the situation is much simpler, complex formation being dependent mostly on the electron-withdrawing substituent effect. We consider these differences to be due in part to the fact that a closer fit into the enzyme is required for the complex formation with carbamates. The core molecular mechanism is, however, common between the two series, being the nucleophilic attack of the serine OH of the enzyme against the ester moiety of insecticides.
XT2,3 iS the experimentally determined Fr value for ortho and meta substituents. The -w term for para substituents was not significant in determining the variations in activity. The a' (sigma mixed) term (18,22) is a composite of cr' for ortho and meta substituents and a-for para substituents (23). Since the p value (2.25) is close to that in the dissociation of substituted phenols (2.11) (23), and also since the cr-is better for the correlation of para substituents, the structure of the reversible complex is considered not to be the regular pentacovalent-type but to involve a part of the leaving process of phenoxides as shown in Figure 3.  Synthetic Pyrethroids and DDT-Related Compounds.
The pyrethroids are a class of insecticides from a plant source. Recently, a number of synthetic analogs have been developed that are effective not only against household but also against agricultural pest insects. DDT is a well-known synthetic insecticide. Although its use has been prohibited in a number of countries, the mode of action, as well as structure-activity studies of related compounds, has been continued in the hope of developing analogs devoid of unfavorable environmental impacts.
Although -the origins are entirely different, their mode of insecticidal action has been shown to be very similar (24). Structural characteristics of compounds have been merged in recently developed novel insecticidal structures as shown in Figure 4 as an example. The principal target site of these two classes of compounds is believed to be the axonal membrane of the nervous system of insects, inhibiting the closing mechanism of the Na+ channel (28,29). Recently, we have measured the in-CI-o\*,H-OCI  duction of hyperexcitatory symptoms in excised nerve cords of American cockroaches immersed in physiological saline solution containing various concentrations of substituted-benzyl (+ )-trans-chrysanthemates (V) and DDT types of compounds (VI-IX) in terms of the min- H3COo CH oI IX imum concentration (MEC in M) required to induc repetitive train of impulses in response to a single sti ulus (30).
The quantitative analysis has been performed for I ortho, meta, and para substituted-benzyl chrysant mates separately (31) to yield Equations (7), (8), (9). For ortho derivatives, AV, means the value of van der Waals volume relative to that of H, scaled by 0.1 to make it nearly equiscalar with the other parameters. Equation (7) indicates that the larger the van der Waals volume, the more favorable are the ortho substituents to the excitatory activity on the nerve cords among the substituents tested here. The electron-withdrawing effect of the ortho substituents weakens the activity. The activity varies parabolically with the T value, the optimum of which is around wr = 0. Equations (8) and (9) show that the optimum van der Waals volume exists at about 4.9 and 2.2 for meta and para substituents, respectively. Hydrophobicity of substituents is not favorable to the activity at the para position.
In a concentration range higher than those exhibiting repetitive responses, this class of compounds blocks nerve conduction. We have also determined neuroblocking activity in terms of the minimum effective concentration (MBC in M) in the saline solution (30). In contrast to the effect on excitatory activity, substituent effects on blocking activity were not specific to substituent positions. With the position-independent hydrophobic and steric parameters, Equation (10) was formulated for 20 benzyl chrysanthemates where definite activity indices are available (31).
Peculiar "topographical" effects of substituents have long been observed for toxicity of substituted-benzyl chrysanthemates against houseflies (32). The effect of the benzyl group attached to the benzyl ring is highest (7) in the meta while lowest in the ortho derivative. For the allyl derivatives, activity was highest in the para and lowest in the ortho isomer. Equations (7)-(9) indicate that the optimum van der Waals volume of substituents for the neuroexcitatory activity is largest at the ortho and smallest at the para position. In Figure  W  5, the activity is expressed as the function of the AV, value at each position according to Equations (7)- (9). It is easily understood that, other factors being equal, the sequence of the activity among positional isomers varies from para > meta > ortho via meta > para > ortho to meta > ortho > para with increase in the bulkiness of substituent from nitro to phenoxy. Since the neuroblocking activity is not position-specific, the peculiar topographic effects of substituents on the insecticidal activity of benzyl chrysanthemates are understood on the basis of those on the neuroexcitatory activity. For the aromatic substituent effects on the neuroexcitatory activity of DDT (VI), DDD (VII), prolan (VIII), and their analogs, Equation (11)  analogs. There is practically no difference in activity between the DDT and DDD series if the aromatic substituents are the same. The optimum AV, value is about 2.1, which is close to those of Et and OEt. This optimum is very similar to the value for the para substituent effect of benzyl ( + )-trans-chrysanthemates on the same type of activity. The aromatic moiety of the pyrethroids and DDT-type compounds may fit into the target sites at the axonal membrane with a closely related (or a common) mechanism.
For the effect of benzylic substituents examined with the methoxychlor analogs (IX), Equation (12) was formulated (33), where L is the STERIMOL length parameter (9 *Equations (11) and (12) vary slightly from the original ones (33), where some of the data were erroneously used.
with n = 17 r= 0.961 s = 0.409 Equation (12) reveals that the optimum bulkiness and the optimum hydrophobicity of substituents are located at about AV, = 6.89 and -T = 0, respectively. Equation (12) also shows that the shorter the length of benzylic substituents, the higher is the activity. Thus, thickset substituents are desirable for high activity. Although the requirements for the optimum bulkiness as well as for the optimum length are not satisfied simultaneously, such thickset substituents as CBr3, CH(Et)NO2, and C(Me)2NO2 are in fact most favorable. Holan proposed that benzylic substituents of DDT analogs would fit into the channel of a pore in the cell membrane to induce leakage of Na+ ion (34). He demonstrated that the optimum diameter of the benzylic substituents is about 6 to 6.3 A to plug the pore. Assuming the substituent with 6 to 6.3 A diameter to be a sphere, the volume is calculated as being about 73 cm3/ mole, which is very close to the optimum van der Waals volume estimated from Equation (12), which is about 72.5(= [optimum AVw + Vw (H)] x 10).
Equation (13) was derived for the insecticidal activity against the American cockroach in terms of the minimum lethal dose (MLD in mole/insect) of four series of DDT-type compounds where definite activity was determinable by injection (33).
The insecticidal activity is determined by the neuroexcitatory activity when the role of hydrophobicity (log P) in the transport process to the target was considered and separated. The above analyses indicate a number of similarities in structure-activity relationships, in particular, in parabolic dependencies of neuroexcitatory activity on steric factors between two series of compounds. They are, however, still incomplete, and the accumulation of this type of analyses systematically performed for a number of substructural features would be one of the most rational approaches in comparing the molecular mechanism of actions.

Insect Juvenile Hormone Mimics
Insect juvenile hormone (JH) mimetic compounds are divided roughly into two classes, terpenoid and nonterpenoid types. A representative of the first class, isopropyl (2E, 4E)-1 1-methoxy-3, 7,11-trimethyl-2,4dodecadienoate, named methoprene (X), exhibits high CH3O 3~~x activity on Aedes aegypti (yellow-fever mosquito) but it is not always so on other insect species (35). For Tenebrio molitor (yellow mealworm), the most active member is N-ethylamide of dodecadienoic acid (XI) (36). N-Ethyl-1,2-bis(isobutylthiolcarbamoyl)ethane (XII) compound of high aromatic content, shows a moderate activity on the same insects (37). The compounds in the nonterpenoid class are not always very active but are novel in structure. It is expected that they provide possibilities to develop useful compounds without the deficiencies like poor field stability and costly synthesis of the compounds having integrity of the terpenoid structure.
2,4-Dodecadienoates and related compounds are a class having,the highest total number of compounds tested so far for activity. Using the data published by Henrick et al. (38), analyses were performed for activities on A. aegypti and T. molitor, quantitatively in terms of physicochemical parameters, to explore the similarities as well as dissimilarities between the species of the structural effects on activity (39). The structural information obtained could be transposed to other, nonterpenoid types of compounds if the site of action or the receptor is the same.
The structure of the series of compounds shown in Figure 6 varies at both ends, X and Y, of the chain. To express the molecular shape, the steric parameters shown by Figure 7 were defined. Lx is the length of the X end along the bond axis (C1-X), and W, is the width in the direction in which the longest chain of the X substituents extends in the staggered confonnation. The maximum length of the whole molecule was expressed by the D parameter, which is the summation of the D, and Dy. The DX is the length of the X moiety along an axis which passes through the C1 and C11 atoms in the fully extended conformation and the Dy is that of the Y moiety. The D parameter is accordingly the length of the molecule after the common C,-C,, part is subtracted. The Bx parameter shown by Figure 8 manifests the bulkiness toward the carbonyl group of ao-substituents in the alcohol moiety of ester and thiolester derivatives. The values of these steric parameters are calculated geometrically based on the CPK models using the STERIMOL program (9).
The hydrophobicity of the X end, ,, is expressed by the ,r value of the C(Me) = CH-CO-X moiety, taking the conjugation of X with the oa,,1-unsaturated carbonyl group into consideration. The wYr term is the hydrophobicity of the Y-C,,-C10 moiety since some compounds have substituents at the A and/or B positions defined in Figure 6.
After separate analyses for two sets of compounds where either the X or Y end is held constant, Equation   pI,, is the logarithm of reciprocal of the I,, value, which is the molar concentration for A. aegypti larva and mole/ pupa for T. molitor to produce 50% inhibition of metamorphosis. Significance as well as the overlaps of the coefficient values of the log P and its squared term in Equations (14) and (15) indicate the similarity of the transport processes. Other common terms are D, B,, and Ibr, I'b is an indicator variable for compounds having a branch at any position in the X moiety of ketone derivatives. Its negative coefficient may suggest that the branch obstructs the proper fit in the receptor. Contrary to this, the branch parameter of esters, B,, has a positive coefficient. The esters appear susceptible to hydrolysis, and the site of the enzymatic attack of those having an a-branch becomes sterically hindered. It is interesting that the D term, which is the summation of Dx and Dy, is significant. This suggests that the X and Y ends are located at the site of action parallel in terms of the D. and Dy axes, irrespective of the conformation of the middle part of the molecule. On this basis and minimizing the strains between bonds, a model-building study suggested an extended active conformation like that depicted by Figure 7.
An important factor in determining the activity is the length of the X-substituents, Lx in Equation (14) for A. aegypti, whereas it is the width parameter W, in Equation (15) for T. molitor. The significance of the squared terms suggests that a receptor wall exists in the L., direction ca. 5 A, the optimum value, distant from the Cl atom in the A. aegypti receptor and it is located in the W,. direction ca 3.8 A distant from the L., axis in the T. molitor receptor. The L, and W, parameters derived similarly to L., and W. were found to be insignificant, showing receptor wall existing only in the direction of the DY axis at the Y end. For the C7 enantiomeric effect, the (R)-( -)-isomer always shows lower activity in A. aegypti than the (± +)-counterpart.
) A spatial wall is thought to exist closer to the methyl group at the C7 atom in the interaction site, and this is expressed by the indicator variable term I( -), which takes the value of 1.0 for the (R)-( -)-and zero for the (±+)-isomer. Another term which reflects the species difference is the position-specific -r.r term in Equation (15). The region of the T. molitor receptor with which the X moiety is in contact is considered to be hydrophobic in nature, and interacts more strongly with compounds having a hydrophobic X moiety.
The I,R is an indicator variable that takes the value of 1.0 for the amides and ketones and is otherwise zero. Its significance with the positive sign in Equation (15) for T. molitor may reflect the resistance of the amides to and the impotency of the ketones against hydrolytic attack by the enzyme. The insignificance of the INR term in Equation (14) seems to reflect the weaker potency of A. aegypti hydrolytic enzyme. The IN in Equation (14) is zero for the others. Its physicochemical meaning is obscure but the negative coefficient shows that the activity of amides is uniformly lower in A. aegypti. Another indicator variable term IOR in Equation (14) takes 1.0 for the compounds whose Y moiety is alkoxy and otherwise zero. This effect seems to be due to a hydrogenbonding type interaction with an acidic group on the receptor, enhancing binding. With a most favorable combination of various factors, the very potent activity of isopropyl 1 1-methoxy-3,7,1 1-trimethyl-2,4-dodecadienoate(methoprene X) on A. aegypti and that of the corresponding N-ethylamide (XI) on T. molitor are understandable. Based on these results, an inclusive "mode of action" map was drawn and shown in Figure 9A for A. aegypti and 9B for T. molitor. The affixes explain the roles or meanings of the parameters incorporated into the correlations. The models help in understanding the overall resemblance as well as the species difference of the mode of action.
It is worthwhile to test the validity of the receptor models and/or the possibility of transposing the structural information to other classes of compounds. Thus, in Figure 10A, the CPK model of the nonterpenoid Nethyl-1,2-bis(isobutylthiolcarbamoyl)ethane (XII) was accommodated to a model which shows inclusively the receptor contour of both insect species with that of methoprene. The carbamoylethane that fits well into the model reportedly exhibits a high morphogenetic activity on T. molitor (36), but the activity data on A. aegypti is lacking. The compound is expected to show some activity on A. aegypti, if the N-ethyl group does not interact so badly with the I( -) wall. Figure 10B  activity on T. molitor (37). It has a conformationally rigid benzene ring at the center of the molecule. If one assumes the same target site, both ends should be parallel, and thus an extended conformation as shown by Figure 10B is adopted. The approaches could be extended to other insect species as well as to other diverse classes of compounds. In these phases, the mode of action models or the receptor maps will act as a guide. Over the dissimilarities of the modes of action between species, the structural essentials which confer the JH activity through species would be more clearly indicated.

Cytokinin-Active Adenine and Urea Derivatives
The term cytokinin refers to a class of plant hormones that promote cell division and growth in certain plant tissues. They are often involved in cell differentiation and organ formation, enhancement of seed germination, and resistance to senescence. 6-[(E)-4-Hydroxy-3methyl-2-butenylamino]purine(zeatin) and 6-(3-methyl-2-butenylamino)purine are naturally occurring representatives of this class. N,N'-Diphenylurea, isolated as a cell-division factor of coconut milk (40), is in another class of cytokinins which possess seemingly very different structures from but the same activity as the N6substituted adenines.
With the aim of elucidating structure-activity relationships as well as at developing highly active analogs, a large number of related compounds have been synthesized in both series of cytokinins. To reveal the correspondence of structural requirements between two series of compounds, the QSAR technique has been utilized (41). Equation (16) (16) is the molar concentration at which the 50% of the maximum callus yield is obtained in the tobacco callus bioassay, and C in Equation (17) is the minimum molar concentration at which the activity is detected in the tobacco pith assay. Wnlar is the maximum width perpendicular to the bond axis of the N6-substituents, as shown in Figure 11. Since the bulkiness of ortho and meta substituents on the Nf6-benzyl moiety is not reflected by the WnaX parameter as depicted by Figure 12 arately defined as W,,,,7. Lo and Lp are the length parameters for ortho and para substituents of diphenylureas and r,,, is the hydrophobic parameter of the meta substituents. W and L parameters were calculated by the STERIMOL program (9) for the fully extended conformation.
The significance of the (Wr,ax)2 term in Equation (16) shows that there is an optimum steric condition (ca. 5.2 A) for activity in terms of the maximum width at the N6-substituents of adenylate cytokinins. The negative coefficient of the W ,,,7 term indicates that, the thicker the ortho and meta substituents, the lower the activity. For N,N'-diphenylurea derivatives, Equation (17) indicates that the position-specific steric and hydrophobic effects of the aromatic substituents are of major importance in determining the activity. At the ortho and para positions, substituents length is unfavorable, as shown by their negative coefficient. No particular length effect is observed for meta substituents, but their high hydrophobicity enhances activity. The hydrophobicity of the whole molecule is, however, not significant in both series, suggesting that transport in the tobacco tissue is not a process important for activity within the compound sets analyzed. Variations in activity are thus governed mainly by variations in the interaction with the target site.
The electron-withdrawing effect expressed by the positive sign of the Taft's a* term in Equation (16) for N6-substituents of adenines is considered to be operative at the N6-H. A similar electronic effect indicated by the Hammett a term in Equation (17) suggests an electronic interaction via the NH group on the urea bridge. The basic partner group of the interaction could be common with that for the adenylate cytokinins, if one assumes the same receptor. The site of interaction for the substituted benzene moiety of diphenylureas is then better considered to correspond to the heterocyclic moiety of the adenylate cytokinins, since the positionspecific hydrophobic effect expressed by the rnl, term is not observed for the N6-substituents in adenine derivatives.
On the basis of these results, the modes of binding for both the adenine and urea cytokinins to the site(s) of action are schematically represented in Figure 13. The stippled lines in the figure represent the steric interaction sites or the spatial walls deduced from the steric parameters incorporated into Equations (16) and (17), and the smooth line facing the NH groups is the electron-donating site. The striped circle is the hydrophobic region where the meta substituents of the diphenylureas come on when they fit with the receptor, as suggested by the wr,,. term in Equation (17). The combined binding model displays the structural correspondence and/or similarity between the two series of cytokinins. Since the identity of their sites of action has recently been indicated by a kinetic approach (43)

Cytokinin-Agonistic and Antagonistic Pyrrolo[2,3-d]pyrimidines
Antagonists of a biologically active compound play an important role in studying its bioregulatory mechanisms and mode of action. In this respect, quite a few structural classes of cytokinin antagonists, anticytokinins, have been developed recently (44)(45)(46)(47)(48), all of which possess similarities in structure to N6-substituted adenylate cytokinins. Among these, N4-substituted 4-amino-2-methylpyrrolo [2,3-d] Both N4-alkyl and N4-phenyl derivatives exhibit activity, agonistic or antagonistic depending on the structure. The steric features were thought to be responsible for the activity change, the maximum width (W,,,,.r) and the thickness (W,,) upward and rectangular to the W,,,, being most plausible. The definition of steric parameters are shown in Figure 11. Equation (18) XVI since their activity varies from agonistic to antagonistic with the transformation of the N4-side chain (47). They are accordingly a class of compounds structurally congeneric but having different types of activity.
The identity of their receptor has been shown kinetically by the method of Lineweaver and Burk (49), where the reciprocal of the growth response in terms of the tobacco callus yield was plotted against the reciprocal of the concentration of added cytokinin (50). Figure 14 shows an example for an antagonistic cyclopentyl derivative, where the family of straight lines possesses a common intercept, fulfilling the requisite for competitive inhibition. The question that immediately arises is how their interaction with the common receptor leads to different biological results.  adenylate cytokinins in the preceding section, providing us with an insight into the bulkiness of the receptor cavity into which the alkyl substituents must fit. The nPh is an indicator variable for phenyl derivatives. Its significance means that a factor which uniformly enhances activity of the aromatic congeners is operating, although its physicochemical basis is unknown but probably due to the difference of the substituent shape. The electronic and hydrophobic factors are common through both the agonists and antagonists. The (* term is suggestive of an electronic interaction at the common N4-imino hydrogen atom with a basic site on the receptor surface. The antagonists presumably scramble for the site with agonists. Another common term, ar*, with positive coefficients, shows the importance of hydrophobicity, probably in the transport process. Equation (18) explains the potency of activity irrespective of the quality of activity, agonistic or antagonistic. Figure 15 shows the parabolic dependence of the cytokinin agonistic and antagonistic activities of alkyl derivatives on the WRv. value, and Figure 16  coincide with and provide evidence for the hypothetical concept for hormonal action that agonist binding causes a conformational change of an otherwise inactive receptor to the active form and that antagonists are species that bind similarly to the receptor but do not cause the effective conformational change. In the present case, the interaction at the WR ,, region is responsible not only for the binding but also for the quality of activity, i.e., a conformational change leading to the active species. The interaction at the WPh region has the same role also in the phenyl series of compounds, the agonists having the values larger than ca. 4.0 AA.
Within the congeneric phenyl derivatives, Equation (18) indicates that compounds having as large as possible -n and u* values and as small as possible WVj,.h values should be highly active as anticytokinins. The p-CF:, and p-i-Pr derivatives, the highest active members of the class, were thus derived.

Concluding Remarks
The above examples show that much invaluable information has been derived from quantitative comparisons of physicochemical factors determining structureactivity relationships. As stated earlier in this article, they are broadly classified into two types. In the first type, the comparisons were made between different classes of compounds having the same type of activity such as antiacetylcholinesterase carbamates vs. phosphates, pyrethroids vs. DDT-related compounds and adenylate vs. diphenylurea cytokinins. In the second type, structural factors were compared within a single type of compound; between subsets of a series of N4substituted pyrrolopyrimidines exhibiting cytokininagonistic and antagonistic actions and between activities against different insect species of a series of dodecadienoate juvenoids. In each example, similarities and dissimilarities in molecular mechanisms of action between counterparts were clearly demonstrated. Accumulation of QSAR results within each comparative study is expected to lead to the revelation of novel empirical rules determining structure-activity relationships as well as to reinforcing the validity of molecular mechanisms.
The information obtained from the QSAR analyses is helpful in designing compounds of optimized structure having the most favorable activity by modifications of substructures. In fact, a number of attempts have been made to apply this procedure to structural optimization to derive useful bioactive compounds with varying degrees of success (3,(51)(52)(53)(54)(55). The comparative QSAR studies not only promote such structural optimization procedures but are also useful in providing indications for developing other classes of compounds showing the same type of activity and having novel structural features, by integrating empirical rules governing structure-activity profiles. In cases where molecular shape is important, the "receptor-contour-maps" may be of use.
The examples in this article were selected only from the areas of insecticides and plant growth regulators. Similar comparative QSAR studies for different compound series having the same type of activity have also been performed in other fields of pesticides which include agricultural fungicides (56) and photosynthesisinhibitory herbicides (57). Recently, this type of QSAR analysis has been applied to environmental toxicological problems (58). We hope that comparative studies could also be accumulated for environmental problems so that the Hansch approach would integrate the methodologies for structure-property relationships which have been carried out independently in the field of drug and pesticide pharmacology and in the field of environmental toxicology.