v STUDIES ON HUMAN SERINE PROTEASES: 'I‘RYPSIN, PLASMIN, THROMBIN AND C'ls DANBEL PAUL ROMAN, JR. mega; mm: m: DEGREE m M. s. MICHEGAN STA“ UNIVERSBTY I977 LIBRARY Nld‘gqq Si???) 1 T I - :1. l .‘ l. ;z;.:‘-’f--3r ,7 a “Pa ‘3 I ‘1; I. , _.-,---. ~. v v us. I This is to certify that the thesis entitled STUDIES ON HUMAN SERINE PROTEASES: TRYPSIN, PLASMIN, THROMBIN, AND ClS presented by Daniel Paul Roman, Jr. has been accepted towards fulfillment of the requirements for Master' Ldegree in Microbiology Major professor Date December 14, 1977 0-7 639 ABSTRACT STUDIES ON HUMAN SERINE PROTEASES: TRYPSIN PLASMIN, THROMBIN AND Cls by Daniel Paul Roman, Jr. A series of substituted benzamidines were investigated as inhibitors of the hydrolysis of Na-carbobenzoxy-L-Tyrosine-para-nitrophenyl ester (Na-Z-L-Tyr-p-Np) by the human proteases trypsin, plasmin, Cl§ and thrombin. The inhibition constants (Ki) were measured and correlated with various substituent parameters to try to determine some of the fac- tors that controlled the magnitude of Ki- The most important single variable involved in the correlations for trypsin, thrombin and Cl§ was R, the Swain and Lupton resonance parameter, resulting in equations of the form log l/Ki = klR + k2 where RI and k2 are calculated constants from regression analysis. For Cl§, the inclusion of "N02 and ("N02)2. measures of substituent hydrophobicity derived from the nitrobenzene series,results in a significantly better correlation. In the case of plasmin, molar refractivity (MR) and fl (benzene series) were the most significant single variables. It does not appear that any electronic effect plays a significant role in benzamidine binding to plasmin, thus making it very different from the other three enzymes. A comparison between human trypsin and thrombin and their bovine counterparts indicates to us that differences do exist, most possibly in the hydrophobic regions of these enzymes. The equations generated for human plasmin in this work and in the work by Coats (1973) are similar, indicating that a mass synthe- sis approach to inhibitor design is unnecessary. STUDIES ON HUMAN SERINE PROTEASES: TRYPSIN PLASMIN, THROMBIN AND ClS by Daniel Paul Roman, Jr. A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTERS OF SCIENCE Department of Microbiology and Public Health 1975 ACKNOWLEDGEMENTS I am grateful to the Department of Microbiology and Public Health for their partial financial support, and to the Center for Blood Research for allowing me to use its facilities during the course of this work. I would also like to express my appreciation to Dr. Michael Cory, SRI, who did the biological correlations and to the following people: Dr. Judith Andrews, Gail Conneely, Dr. Dianne Kenney, Dr. Fabian Lionetti and Stephen Hunt for their assistance in reading the thesis and discussing the data. ii TABLE OF CONTENTS LIST OF TABLES . . . . . . . LIST OF FIGURES . . . . . . INTRODUCTION . . . . . . . . LITERATURE REVIEW . . . . . Part I: Part II: Trypsin, Thrombin, Introduction . . Trypsin . . . . . Thrombin . . . . Plasmin . . . . . Cl§ . . . . . . Enzyme Kinetics . Concluding Remarks Plasmin and Cl; Regression Analysis . . . . . . . . Introduction . . Parameters in QSAR Partition Coefficients, n . . . . Hammett o Constants . . . . . . . Swain and Lupton F and R . . . . Nys and Rekker, f . . . . . . . . Molar Refractivity . . . . . . . Use of Hansch Technique . . . . . . LIST OF REFERENCES . . . . . APPENDIX A C O O O O O O O C General Theory of Substituent Constants and Multiparameter Regression Analyses . . The Hydrophobic Parameter, fl . . . . . . Hammett, o, and Swain and Lupton, F and R Multiple Parameter Regression Analysis . iii Page vi UlKOCDQO‘obubb l-' l6 16 16 17 18 18 l9 -19 21 25 25 26 27 29 Page ARTICLE 1 Quantitative Structure Activity Relationships: Inhibition of Human Cl§, Thrombin, Trypsin and Plasmin by Substituted Benzamidines. D.P. Roman, Jr. and D.H. Bing. Manuscript submitted to Biochemistry . . . . . . . . . . . . . . . . . 31 CONCLUDINGREMARKS...................... 61 iv ([[II‘([ LIST OF TABLES Table Page LITERATURE REVIEW 1. Properties of Trypsin, Plasmin, Thrombin and Cl§ . . . . 10 ARTICLE 1 1. Substituent Constants Used in Quantitative Structure Activity Relationships . . . . . . . . . . . . . . . . 38 2. Kinetic Parameters for Reaction of Human C15, Trypsin, Plasmin, and Thrombin with Na-Z-L-Tyr-p-Np and Na-Z-L-Lys-p-Np. . . . . . . . . . . . . . . . . . . . 4o 3. Substituents Constants for X . . . . . . . . . . . . . . 46 4. Benzamidine Inhibitors Tested for Inhibition . . . . . . 47 5. Correlation Equations for Trypsin, Plasmin, Thrombin and Cl§ (All Compounds) . . . . . . . . . . . . . . . 48 6. Correlation Equations for Cl§, Trypsin, Thrombin, Plasmin (Selected Compounds) . . . . . . . . . . . . . 54 Ill. .( {.‘I Ill .IlrlllllltlII‘Fll IIAIAI‘III-ll LIST OF FIGURES Figure Page LITERATURE REVIEW Scheme 1. Reaction of a Substrate with a Serine Protease . . . 9 Scheme II. Charge Transfer System . . . . . . . . . . . . . . . 12 ARTICLE 1 Figure 1. Lineweaver-Burk Plot of the Reaction of N -Z-L- Tyr-p-Np with Human Cls, Trypsin, Plasmin and Thrombin . . . . . . . . . . . . . . . . . . . . . 42 Figure 2. Lineweaver-Burk Plot of the Reaction of N -Z-L- Lys-p-Np with Human Cl§ and Trypsin . . . . . . . 45 CONCLUDING REMARKS Figure l. Toxicity vs. Effectiveness . . . . . . . . . . . . . 63 vi INTRODUCTION The human body contains hundreds of enzymes and proteins which vary widely in their structure and function. One important class of enzymes is the serine proteases. These all have a serine residue essential for the catalytic hydrolysis of both natural and synthetic ester substrates. The following study was carried out on four enzymes; thrombin, plasmin, trypsin and Cl§ which differ in their physiological roles but are simi- lar in two important aspects. First, that they are all serine proteases, and second their hydrolytic action is almost exclusively directed toward l-lysyl and l-argininyl bonds in polypeptides. The anionic binding site which all the enzymes are known or presumed to possess is a possible starting point for a comparison of the nature of the substrate specificity of these enzymes. Much is known about the biological roles and the physical-chemical properties of these enzymes. Trypsin has been the most studied of the group and much physical data has been accumulated, culminating recently in the three dimensional structure of bovine trypsin as determined by x-ray crystallographic techniques. Biologically, this pancreatic enzyme is responsible for the activation of all the other zymogens of pancreatic tissue. The fact that trypsin will hydrolyze virtually any peptide bond adjacent to a basic amino acid is probably critical to its role as a digestive enzyme. Plasmin is functionally similar to trypsin. It is an endopeptidase and will hydrolyze a wide variety of peptide bonds adjacent to an 1 . . It‘ll {([I'l‘[[([.[ll argininyl or lysyl residue. Its main role is the dissolution of insoluble fibrin polymers; however, it also hydrolyzes a number of other naturally occurring proteins. Thrombin is an enzyme of the coagulation sequence and its major substrate is fibrinogen. Thrombin and other coagulation enzymes react on fibrinogen to give fibrin polymers and a stable fibrin clot. As with plasmin and trypsin, the bond in fibrinogen which is cleaved is an arginine bond, indicating an anionic binding site in thrombin. Cls, a subunit of the first component of complement, Cl, is again a serine protease which is responsible for the cleavage of C4 and C2, early acting components of the complement sequence. C1 initiates the comple- ment system which acts as a biological amplifier of the immune response. The action of Cl§ on C4 and C2 also results in the generation of phlogistic peptides important in immune hypersensitivity states. Cl; is inhibited by substituted benzamidines and guanidines, suggesting an anionic binding site adjacent to the catalytic site. The esterolytic activity of each enzyme was determined with a simple spectrophotometric assay procedure. The same substrate was used for each of the enzymes so that comparable inhibition constants could be accurately determined. Inhibition constants (Ki) for a series of substituted benza- midines was determined and the Ki correlated with various substituent constants. The equations gnerated were used to compare the four enzymes to determine the factors which control substrate specificity. This thesis is divided into two parts. The first part is a literature review, in which pertinent information on the nature of trypsin, thrombin, plasmin and Cls, the kinetics of the reaction involved, and on the use of the Hansch multiparameter regression analysis in correlating biological activity of inhibitors with substituent parameters is discussed. The second part deals with the correlation of Ki with substituent 3 parameters and a determination of factors important in determining the biological activity of these compounds. A comparison is made of the human and bovine systems for thrombin, trypsin and plasmin while Cl§ is compared to whole guinea pig complement to determine if any significant differences are seen between the species. LITERATURE REVIEW Part I Trypsin, Thrombin, Plasmin and Cl§ Introduction Human serine proteases comprise an important class of enzymes in the body. Their wide range of functions and specificities are important in contributing to the health or, unfortunately, to the detriment of the in- dividual. Volumes of data have been collected on trypsin and the so called "trypsin-like" enzymes. This review will be concerned with detailing some of the significant similarities and differences of four enzymes of the trypsin class, trypsin, thrombin, plasmin and Cl§. The second part of this review will be a description of the Hansch multiparameter regression analysis technique and its application, not just to drug design but, to developing ideas concerning possible structural differences in enzyme binding sites. T sin Trypsin has been one of the most widely studied enzymes in the class of serine proteases. Serine proteases are so named because these enzymes are inactivated by the reaction of a serine residue in the active site of these molecules by diisopropylfluorophosphate (DIFP). Trypsin is inhibitable by this compound (1). Another important feature in the mechanism of action of trypsin is the presence of a charge relay system consisting of aspartate, histidine and serine residues. The structural evidence for this system has 4 been shown by X-ray crystallographic data (2). The inactivation of trypsin by tosyl lysine chloromethyl ketone (TLCK), which reacts with the active site histidine, and other experiments have helped confirm the importance of an intact charge relay system (3). Human trypsin has been purified from pancreatic tissues by adsorption to Sepharose-TrasylolR columns and then subsequent elution with a pH gradient (4). Trypsin is secreted by the pancreas as the proenzyme trypsin- ogen. Trypsinogen is activated to trypsin by the hydrolysis of a single peptide bond near the N-terminus. Trypsin can undergo further hydrolysis and at least five varieties of trypsin are known. Comparisons of human trypsin with trypsins from other species indicate many similarities includ- ing molecular weight, amino acid composition, pH optimum and inhibition by several synthetic and naturally occurring inhibitors (5,6). Trypsin exhibits a marked preference for lysinyl and argininyl side chains in the hydrolysis of peptide bonds (7,8,9); however, it can hydro- lyze some neutral substrates (10). Although trypsin is relatively specific in its requirement for basic amino acid residues, its specificity is broad and it will hydrolyze virtually any peptide, amide, or ester which contains a lysinyl or argininyl side chain. Since trypsin is a digestive enzyme and is responsible for the activation of the other zymogens in the pancreatic tissue, this wide hydrolytic activity takes on a special significance. A wide variety of substrates have been used to determine trypsin activity and a number of different types of inhibitors are also known (11). The development of an active site directed titrant p-nitrophenyl-p'-guanidino- benzoate (PNPGB) by Chase and Shaw (12) has allowed the accurate deter- mination of the amount of active enzyme in any solution. The importance of an active site titrant has been shown in a recent example by Moroi g£_al. (15) which demonstrates that a-l-antitrypsin combines in a 1:1 ratio with trypsin instead of a 1:2 complex as has been described elsewhere (13,14). Thrombin Thrombin is a serine protease and both its esteratic and clotting activities can be blocked with DIFP (16,17). Thrombin's physiological role is the formation of fibrin monomers from fibrinogen following the activation of prothrombin to thrombin via interaction with thromboplastin, factor V, factor Xa and Ca++ (18). Thrombin is purified by a combination of ion exchange chromatography, ethanol fractionation and gel filtration. Combination of the above techniques with affinity chromatography on p-chloro- benzylamido-E—amino caproyl agarose or mNHz-benzamidino agarose results in the recovery of protein that is 90% active based on active site titration with PNPGB (19,20). Thrombin consists of two chains, A and B. It is the B chain that con- tains the reactive serine residue analogues to trypsin (21). Although no X-ray crystallographic data is available, there is a high degree of homology between the B chain and chymotrypsin A and B, trypsin and elastase (21). Because of this homology the existence of the charge relay system can be postulated. The fact that thrombin is inhibited by DIFP and also be tosyl lysine chloromethyl ketone (TLCK), which reacts with the trypsin active site histidine residue (22,23), is further evidence for the presence of this system. Thrombin is much more specific in its activity than trypsin or plasmin. The major physiological substrate for thrombin is fibrinogen. The formation of fibrin monomers is brought about through the cleavage of an arginine- glycine bond (24). Thrombin can also hydrolyze thrombosthenin M (25) and the two hormones secretin (26) and cholecystokinin-pancreozymin (27); all three are cleaved at a susceptible argininyl peptide bond. Although no lysine containing polypeptide is known to be a substrate for thrombin, it can hydro- lyze N-carbobenzoxy-L-lysine-p-nitrophenyl ester (28) and p-toluene sulfonyl- L-lysine methyl ester (29). Thrombin, like trypsin, can be titrated with p-nitrophenyl-p'—guanidinobenzoate (30), p-nitrophenyl-m'-guanidinobenzoate can also be used as an active site titrant (30). Thrombin can be inhibited by benzamidine and benzylamines, although they are poor inhibitors of thrombin compared to their inhibition of plasmin and trypsin (31). Plasmin Plasmin like thrombin and trypsin is a serine protease inhibitable by DIFP. It is also specifically inactivated by TLCK which reacts with a single histidine in the active center (32,33,34). Plasmin's physiological function is the lysis of blood clots via the proteolysis of fibrin. Plas- min is purified as plasminogen by affinity chromatography on lysine- Sepharose columns (35). The plasminogen obtained is then activated to plas- min by the addition of urokinase (36). The activation of plasmin, molecular weight 75,400 (37), occurs via the cleavage of a single arginyl valine bond (33,34,38). Reduction and alkylation of plasmin results in the formation of two major polypeptide chains, a heavy chain of approximately 49,000 and a light chain of 26,000. The active site serine is contained in the light chain (32,33,34). Plasmin is similar to trypsin in its specificity for lysine and arginine esters (39). Its action on S-sulfo fibrinogen is also similar to that of trypsin, since only lysine and arginine bonds are cleaved; however, whereas, trypsin will cleave 80% of the bonds of this molecule, plasmin will hydro- lyze only 50% (40,41). Plasmin can be titrated with p-nitrophenyl-p'- il ill Ill i guanidinobenzoate, and will hydrolyze p-nitrophenyl-m'-guanidinobenzoate analog as well (30). Plasmin is inhibited by substituted benzamidines (42,43). 935 C15 is a subunit of the first component of complement, Cl. Complement is a system of eleven proteins that act as biological amplifiers in immuno- logically mediated reactions and are also important in hypersensitivity disease states. Cls exists in serum as a proenzyme, which is activated upon interaction of Cl with antigen-antibody aggregates (44,45,46,47). Cls can also be activated by plasmin and trypsin (48). C4 and C2, two early acting components of the complement sequence are the only two protein sub- strates known for Cl§ (49,50,51). The action of Cl§ on C4 and C2 is highly restricted. In each molecule only one peptide bond is cleaved (49,50,52). It is important to point out that the nature of the bonds cleaved in C2 and C4 is not known, so that a comparison with trypsin, plasmin and thrombin as a protease with specificity towards basic amino acid side chains must be based on other evidence. Cl§ has been purified by a variety of methods including differential salt fractionation, ion exchange chromatography and gel filtration (53,54). More recently, affinity chromatography coupled with preparative electro- phoresis has provided relatively high yields of purified enzyme (55). Cl§ is a serine protease inhibitable by DIFP (56). Sulfonyl fluoride compounds which react with the serine hydroxyl group of serine proteases are known to irreverisbly inactivate Cl§ (57). Inhibition of Cl§ by benzamidine, pyridinium and guanidinium compounds is presumptive evidence for an anionic binding site (57,58). Further evidence for an anionic site is provided by the fact that Cl§ will hydrolyze N-tosyl-l-arginine methyl ester (60) and N—carbobenzoxy-L-lysine-p-nitrophenyl ester (Roman, D.P., this work). Although the preceeding data can be used to tentatively classify Cl§ as a "trypsin—like" enzyme, important differences must be noted. Cl§ does not react with p-nitrophenyl-p'-guanidinobenzoate (61). Phenyl methane sulfonyl fluoride which inhibits a number of proteases will not inactivate Cl§ (62). TLCK and TPCK, two reagents that react with active site histi- dines involved in the charge relay system do not inactivate Cl; (61). Cls is similar to plasmin in its activation mechanism. The conversion of C13 to Cl§ results in the formation of two polypeptide chains, a heavy chain (M.W. 70,000) and a light chain (M.W. 30,000). It has been shown that the DFP reactive serine residue is in the light chain of the Cl§ molecule (63,64,65). The properties of these enzymes are summarized in Table 1. Enzyme Kinetics Before one can investigate the nature of the active sites of enzymes, an appreciation of the mechanism of action of proteolytic enzymes and the evaluation of kinetic parameters which can be used in such studies should be developed. The four enzymes under study are all serine proteases and the hydrolysis of substrates proceeds via a double displacement mechanism. The following scheme shows the reaction of a substrate with a serine pro- tease (55) . H+ acceptor Ser Ser Ser Ser H+ acceptor\ " \H-0 \0 0H RI k1 Lg k2 I __ k3 + C-OR —+R-c-0R ——->R—c + OR —>R-C-0H (1) lol P ||\ 0 OK In the first step of the reaction enzyme (E) plus a substrate (S) combine to form an ES complex, presumably in the form of a tetrahedral intermediate (I). 10 .moehucm upon Ham Mom moumuumnom o>wumuzommumou who mZOImumlammoa can malmnmhqlnlNuz .QZIQIHABIAINIZ 000.0m cannotq asonx mwmwmnxuu 000.05 mcoz mow oz mow NU .vo szocxco soap: cannons mHU 000.5H caahuowuoamm :wmnolq :«aaxoDmmomHogo afiuouomm mmUUmQUaU 000.0m mwmzm mow mm» mm» . :mmocflnnflm 0H4 mcw>om Gannon: :flnEounB muonuo + :flunwm 000.mm :wadmcH cwmnolq mau . :«mmmu qmwwmowoo 000.m0 momZm new new we» "mmooaocH qu .mum omen: Gannon: aflfimoam Koo moooo mumzm mm» wow wow Ham qu .mna wcw>om 000.mN cflmmhua ucouufia ocwofiaoucmm MUAB mmo mononumnsm wueoflmwommm wocmzvmm .z.z msxucm Hmaoomaoz :wmuoum ocom mafiumm muouwnflscH moflummm wuflm m>fluo¢ man can someones .caEmmHm .cflmmaua mo mmwuuwmonm .H magma '(‘llll‘lllllullllf'lf'll 11 The next step is the release of the alcoholic or amino portion of the sub- strate leaving an acylated enzyme intermediate, ES'. The final step, if k3>0, is the return to free enzyme by the addition of H20 and the release of the acidic moiety of the substrate. The main feature of the reaction of the enzyme with the substrate is the participation of a catalytic triad of the enzyme which provides the necessary charge relay and nucleophilic attack groups for hydrolysis. The triad consists of aspartate, histidine and serine residues. The charge relay system shown for a-lytic protease and assumed for other serine proteases is shown in scheme II (67). Some ways to investigate the above reaction mechanisms are to use specific acylating or alkylating reagents that can react with the active site serine or histidine residues. Two popular reagents are diisopropyl— fluorophosphate and Na-tosyl-L-lysine chloromethyl ketone, serine and his- tidine, acylating and alkylating reagents, respectively. Another method for examining the reaction of these enzymes, particularly the specificity controlling features, is through the use of low molecular weight organic inhibitors. The use of competitive inhibitors can provide useful informa- tion concerning the nature of the binding sites of these enzymes. In order to carry out such studies, a knowledge of the basic kinetics of the reac- tions of the enzymes under study is necessary. The following reaction scheme is necessary for the development of the kinetic parameters which can be used for investigating the binding and ac- tive sites of the serine proteases (68,69,70). k1 k2 k3 E + S‘_——AES —"‘.~—— ES' + 191:3 + P k-1 k_2 k-3 k 4 k ES' + H V_ES'H Ki — k_4 k4 ‘ I (III III ‘IV I I‘ll Ill 1 ‘ | l . .l 13 If one assumes that in the early stages of the reaction k2[ES]>>k_2[ES][P1], k3[ES]>>k3[E][P2] and [S] = [S]O where [S]O is the initial substrate concen- tration. Then application of the steady state approximation to ES yields {(k2 + k'3)[S]o + k'BKm}[E] = k'3Km[E]o + szm[S]O[E]o exp {-[(k2 + k3) s o + k'3Km]t} Km + [510 Km + [310 where [E]O = [E] + [es] + [ss']t; k'3 = k3Ki/(Ki + [H]+); Km = k_1 + k2 k1 At small values of t, if sufficiently high concentrations of [S]o are used such that (k2 + k'3)[S]o>>k'3Km; k2 + k'3 and Km can be evaluated by means of a Lineweaver-Burk plot of the following equation at various substrate concentrations, where K is the first order rate constant for the reaction k = (k2 + k'3)[s]O + k'3Km Km + [sit Measurements of the rate of appearance of P1 or P2 at large values of t so that the exponential term is very small allows one to calculate the apparent constants Km(app) = k'3 Km and kcatlElo from the following equation k2 + k'3 v = dPl = dp2 = kzk'3[S]O[E]o dt EH?” (k2 + k'3[S]o + k'3Km Vmax = (kcatlElo) is thus determined. With the kinetic constants established, it is now possible to examine effects of inhibitors on the reaction of an enzyme and its substrate. This study is to examine the effect of structural changes of synthetic inhibiors on competitive inhibition constants. Once competitive inhibition has been shown, Ki values can be readily determined. The usual method for the deter- mination of Ki (assumed to be the enzyme-inhibitor dissociation constant) is 14 to plot l/v vs [I] at two or more substrate concentrations (71). The point of intersection of these lines will be at a value of [I] = 'Kir l/v = l/Vmax- Although it is theoretically possible to substantiate that competitive inhib- ition is taking place with the above method, because of experimental error some ambiguity does remain in differentiating competitive from mixed inhibi- tion. In mixed inhibition plots of l/v vs [I] at different substrate concen- trations yields intersecting lines that intersect at a point i = -Ki, l/v = [l-(Ki/K'i)]/vmax (72). Competitive inhibition is an extreme case where K'i+w, thus giving the required intercepts. In order to further demonstrate competitive inhibition, one should make plots of s/v vs [I] at different substrate concentrations. The observed pattern of lines, parallel for competitive inhibition, delineates the mode of inhibition (72). Once competition has been shown it is only necessary to run inhibition experiments at one substrate concentration, provided Vmax has already been determined. A line drawn parallel to the x-axis at a height of l/vmax will intersect the other line at i = “Ki (71). It must be stressed that this is valid for competitive inhibitors only. K- is taken to be a measure of the dissociation constant of the enzyme- 1 inhibitor complex. Because Ki is an equilibrium constant, this value can be used to get a value for the free energy (AG) of binding of the inhibitor to the enzyme from the relation AG = -RTani. This fact also forms the basis for the extra thermodynamic approach to the study of structure activity relationships. In most cases, a biological activity (BA) such as Ki, that has a basis, can be represented as BA = f(AH, AS, AE). In particular, for Ki it is assumed that AG can be factored into polar electronic and steric parameters AGx = 5xAGhydrophobic + GxAGelectronic + 5xAGsteric- As will be seen later, the substituent parameters used to develop equations 15 describing changes in Ki are derived from changes in AG for a particular process (73). AH, AG and AE are thermodynamic state variables and can be represented by other state variables. These variables do not define the nature of the interaction, however, they do give general ideas as to the forces involved and are a starting point for more extensive investigation. Concluding Remarks The preceding sections have attempted to review some of the properties of the four enzymes under study related to: inhibitors, molecular titrants, activation mechanism and substrates, as well as, some of the kinetic fea- tures of the enzymes. Table l is a tabulation of some of the more impor- tant properties of these enzymes which can be used as a reference for the similarities and differences of these proteins. luv]. I A Illl‘. [ [I Part II Multiparameter Regression Analysis Introduction The development of quantitative structure activity relationships (QSAR) for biochemical systems is of growing importance to the biochemist and medi- cinal chemist. The separation to electronic, steric and hydrophobic forces and their contributions to specific chemical interactions are important to the elucidation of biochemical mechanisms as well as to drug design. This section will deal first with a description of some of the parameters commonly used in QSAR. The second section will discuss the rationale of such analyses and some of the systems to which this approach has been applied. Parameters in QSAR Igartition CoefficientsL n Simply stated, the partition coefficient P is the ratio of concentrations of a solute distributed between two separate phases, in this case Coctanol/ Cwater (66). The value of P gives an indication of the relative hydropho- bicity of a molecule and its derivatives. w is derived from the partition coefficient by the following relation: fix = long - logPH (67) where Px is the partition coefficient of the substituted molecule and PH of the unsub- stituted molecule. fix is therefore the value for a particular molecular fragment. Most N values remain relatively constant between one solute system and another provided there are no special intramolecular electronic or steric interactions between substituents and the parent molecule. The choice of the series from which the n values are chosen will depend on the 16 17 class of compounds being tested in a particular system. For example, are n values obtained from a nitrobenzene series or a phenylacetic acid series more representative of the n values for substituted benzamidines (e.g. should "x for be derived from or “~ )? In practice one usually 1’ c N02 CHZCOOH H2N + NHZ obtains n values for a few of the benzamidine derivatives to be tested, then following a comparison with a standard series, the rest of the n values can be taken from the appropriate standard reference (66). Hammett o Constants o constants are obtained for benzene substituents in a fashion analogous to that of obtaining n. 0 can be defined by the relation: long/KH = op (68). a values are calculated from a standard reaction series (i.e. ionization of benzoic acids) and p is assumed to equal 1.0. From this standard series the 0 values calculated can then be applied to other reactions and correlations of their effects on reaction rates or equilibria can be determined. 0 values are a measure of the electronic effect of the substituent on a particular reaction. The most common parameters are 0m and op, values for meta and para substituents, respectively. Other 0 values based on different parent compounds, CI (69), 0+ (70), 0R (69), 0‘ appear frequently in the literature. The slope p is a measure of the effect of a on a particular reaction under consideration. The value of pobs is obtained by plotting long vs 0. A positive slope indicates that in the transition state the reaction center is electron rich, whereas a negative slope indicates an electron deficient center relative to the initial state. 18 Swain and Lupton F and R (72) Since the develOpment of the Hammett relationship and the application of o constants to different sets of reactions, more than 20 0 sets have been developed to explain and correct deviations of correlations of reaction con— stants vs 0. Thus, the convenience of predicting reaction constants has been destroyed and casts serious doubts as to the validity of the Hammett equation. In order to reduce the need for so many 0 sets and to restore the general nature of the equation, two constants F and R.have been developed. F and R are a factorization of o constants into pure field and pure resonance effects. The relationship for any 0 constant in terms of F and R can be expressed by c = fF + rR, where f and r are empirical weighting factors which are independent of the substituent but depend on the series being investigated (om, op, OI, 0R, etc.), and F and R are the field and resonance constants which differ for each substituent. Nys and Rekker, f (73) The hydrophobic fragmental constant, f, is a parameter which is obtained in a manner similar to w. The relationship used for the determination of f is logP(sx) = f(s) + f(x). The main difference between this relationship and the relationship for obtaining n, logP(SH) = logP(sx) + n(x), is that f represents the contribution of a particular fragment to the total lipo- philicity; whereas n represents the lipophilicity of a group X upon substitution for hydrogen in the present compound. For example: Using w: logP(C6H5-CH2-CH2-C6H5) logP(C6H6)+ 2n(CH3) + n(C6H5) Using f: logP(C6H5-CH2-CH2-C6H5) 2f(C6H5) + 2f(an) 19 Molar Refractivity M.R. (74) Molar refractivity is a parameter based partially on the size of a mole- cule. It is dependent on the molecular weight of the compound of interest, which can also be used in regression analyses. Molar refractivity can be considered as a measure of the bulk steric effect of a molecule involved in intermolecular reactions in contrast to Es which is more of an intramole- cular steric effect. Use of the Hansch Technique The Hammett equation has been shown to be an important relationship in predicting the substituent effects on rates and equilibria in organic re- actions (75). However, for biological systems, although limited success has been seen with the Hammett equation, the need for other substituent parameters to explain variance in the data has become apparent. From the work of Meyer (76), Overton (77) and others (78,79), evidence has been pro- vided that the relative hydrophobic nature of substituents is important. A linear combination of two constants to give an equation: (Eq. 1) logl/BR = K10 + K2“ + K3; where BR is some biological response (150, inhibi- tion constant KI, LD50, etc.) and K1, K2 and K3 are constants obtained by least squares has resulted in better correlations than the use of 0 con- stants alone (80). However, even this has not seemed to be enough. The hypothesis that a parabolic relationship existed between BR and n led to an important and generally useful equation: (Eq. 2) logl/BR = K1 logP2 + K2 logP + K30 + K4 (80). In retrospect, the form of equation g_translates into the following scheme which has been carried out on a 3 compartment system by Hansch: Drug A A A Scheme I H20 lipid ' H20 20 This scheme represents the movement of a Drug A from an aqueous compart- ment 1 into a lipid or other hydrophobic compartment such as a cell membrane or protein-binding site followed by passage back into the aqueous phase. If a drug is too hydrophobic it will never get into compartment 2 and if too lipophilic the drug will enter compartment 2 and become trapped there. Therefore, a balance of lipophilicity and hydrophilicity is important in obtaining maximally active compounds. There are many examples where biological activity is related paraboli- cally to n which one should expect according to Scheme I (81,82,83,84,85). However, what one logically expects to find in biological systems is not always true and there are also hundreds of examples where a linear relation in n is sufficient to explain the variance in the data and the addition of 1T2 does not improve the correlation (81). Multiparameter regression analyses are not difficult to perform pro- vided one has access to a large computer and the appropriate programs. A strict mathematical treatment of the solution of such equations is beyond the scope of this review and the reader is referred to a book by Daniel and Wood (86). The basic technique is based on the familiar linear regression analysis where equations of the form y = mx + b are fitted. The theory is the same in that the coefficients of the independent variables are found by minimizing the sum of the squares of the deviations, the differences between each observed value yj and the fitted value Yj. LIST OF REFERENCES 10. ll. 12. 13. 14. 15. 16. 17. 18. 19. LIST OF REFERENCES Jansen, E. F., M. D. F. Nutting, R. Jang, and A. K. Balls. 1949. J. Biol. Chem. 179, 189. Krieger, M., L. M. Kay, and R. M. Stroud. 1974. J. Mol. Biol. 22, 209. . Beeley, J. G., and H. Neurath. 1968. Biochemistry Z, 1239. Johnson, D., and J. Travis. 1976. Anal. Biochem. 12, 573. Roberts, R. C., and J. Travis. 1969. Biochemistry 2, 2884. Mallory, P. A., and J. Travis. 1964. Biochemistry 22, 2847. Neurath, H., and G. W. Schwert. 1950. Chem. Rev. 22, 69. Baird, J. B., E. F. Curragh, and D. T. Elmore. 1965. Biochem. J. 22, 733. Baines, N. J., J. B. Baird, and D. T. Elmore. 1964. Biochem. J. 22, 470. Sanborn, B., and G. E. Hein. 1968. Biochemistry 1, 3616. Walsh, K. A. 1970. Methods Enzymol. 22, 60. Chase, T., and E. Shaw. 1967. Biochem. Biophys. Res. Comm. 22, 508. Johnson, D., R. Pannell, and J. Travis. 1974. Biochem. Biophys. Res. Comm. 21, 584. Pannell, R., D. Johnson, and J. Travis. 1974. Biochemistry 22, 5439. Moroi, M., and M. Yamasaki. 1974. Biochem. Biophys. Acta 222, 130. Gladner, J. A., and K. Laki. 1956. Arch. Biochem. Biophys. 22, 501. Miller, K. D., and H. van Vunakis. 1956. J. Biol. Chem. 222, 227. Seegers, W. H. 1973. Ser. Haemat. 2, 36. Hixson, H. F., and A. H. Nishikawa. 1973. Arch. Biochem. Biophys. 154, 501. 21 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 22 Thompson, A. R., and E. W. Davie. 1971. Biochem. Biophys. Acta 250, 210. Magnusson, S. 1971. The Enzymes, Vol. III (Boyer, P. D., Ed.), Academic Press, NY, p. 278. Baird, J. B., and D. T. Elmore. 1968. FEBS Letters 2, 343. Shaw, E., W. Mares-Guia, and W. Cohen. 1965. Biochemistry 2, 2219. Bailey, K., and F. R. Bettelhein. 1955. Biochem. Biophys. Acta 22, 495. Cohen, 1., Z. Bohak, A. de Vries, and W. Katchalski. 1969. Europ. J. Biochem. 22, 388. Mutt, V., S. Magnusson, J. E. Jorpes, and E. Dahl. 1965. Biochemistry 4, 2358. Mutt, V., and J. E. Jorpes. 1968. Europ. J. Biochem. 2, 156. Curragh, E. F., and D. T. Elmore. 1964. Biochem. J. 22, 163. Kezdy, F. J., L. Lorand, and K. D. Miller. 1965. Biochemistry 2, 2302. Chase, T., and E. Shaw. 1969. Biochemistry 2, 2212. Markwardt, F., H. Landmann, and P. Walsmann. 1968. Europ. J. Biochem. Q, 502. Groskopf, W. R., L. Summaria, and K. C. Robbins. 1969. J. Biol. Chem. 244, 3590. Summaria, L., B. Hsieh, W. R. Groskopf, K. C. Robbins, and G. H. Barlow. 1967. J. Biol. Chem. 242, 5046. Groskopf, W. R., B. Hsieh, L. Summaria, and K. C. Robbins. 1969. J. Biol. Chem. 244, 359. Deutsch, D. G., and E. T. Mertz. 1970. Fed. Proc. 22, 647. White, W. F., G. H. Barlow, and M. M. Mozan. 1966. Biochemistry 2, 2160. Barlow, G. H., L. Summaria, and K. C. Robbins. 1969. J. Biol. Chem. 249, 1138. Robbins, K. C., L. Summaria, B. Hsieh, and R. J. Shah. 1967. J. Biol. Chem. 242, 2333. Sherry, S., N. Alkjaersig, and A. P. Fletcher. 1966. Thromb. Diath. Haemorrh. 22, 18. Blomback, B., M. Blomback, B. Hessel, and S. Iwanaga. 1967. Nature 215, 1445. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 23 Iwanaga, S., P. Wallen, N. T. Grondahl, A. Henschen, and B. Blomback. 1967. Biochem. Biophys. Acta 147, 606. Markwardt, F., P. Walsmann, and H. G. Kazmirowski. 1969. Pharmazie 24, 400. Markwardt, F., P. Walsmann, and H. Landmann. 1970. Pharmazie 22, 551. Lepow, I. H., L. Wurz, 0. D. Ratnoff, and L. Pillemer. 1964. J. Immunol. 22, 146. Lepow, I. H., and L. Pillemer. 1955. J. Immunol. 22, 63. Lepow, I. H., O. D. Ratnoff, and L. Pillemer. 1956. Proc. Soc. Exp. Biol. Med. 22, 111. Lepow, I. H., G. B. Naff, E. W. Todd, J. Penskey, and C. F. Hinz. 1963. J. Exp. Med. 117, 983. Ratnoff, O. D., and G. B. Naff. 1967. J. Exp. Med. 125, 337. Patrick, R. A., S. B. Taubman, and I. H. Lepow. 1970. Immunochemistry 7, 217. Budzko, D. B., and H. J. Muller-Eberhard. 1971. Immunochemistry 2, 227. Stroud, R. M., K. F. Austen, and M. M. Mayer. 1965. Immunochemistry 2, 219. Stroud, R. M., M. M. Mayer, J. A. Miller, and A. T. McKenzie. 1966. Immunochemistry 2, 163. Muller-Eberhard, H. J. 1969. Adv. Immunol. 2, 1. Haines, A. L., and I. H. Lepow. 1964. J. Immunol. 22, 456. Assimeh, S. N., D. H. Bing, and R. Painter. 1974. J. Immunol. 222, 225. Baker, B. R., and M. Cory. 1971. J. Med. Chem. 22, 805. Bing, D. H., M. Cory, and M. Doll. 1974. J. Immunol. 222, 584. Bing, D. H. 1969. Biochemistry 2, 4503. Westfall, S., G. H. Wirtz, and J. Canady. 1972. Immunochemistry 2, 1221. Roman, D. P., J. A. Andrews, and D. H. Bing, this work. Bing, D. H., personal communication. Bing, D. H., J. L. Mernitz, and S. E. Spurlock. 1972. Biochemistry .22, 4264. Sakai, K., and R. M. Stroud. 1973. J. Immunol. 111, 296. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 24 Barlas, T., G. K. Scott, and J. E. Fothergill. 1973. Biochem. Soc. Trans. 2, 1219. Fothergill, J. E. 1974. Workshop on Complement - Abstract - 2nd International Congress of Immunology. Leo, A., C. Hansch, and D. Elkins. 1971. Chem. Rev. 12, 525. Fujita, T., J. Isawa, and C. Hansch. 1964. J. Am. Chem. Soc. 22, 5175. Hammett, L. P. 1970. Physical Organic Chemistry, 2nd Edition, McGraw Hill, NY. Taft, R. W., and I. C. Lewis. 1958. J. Am. Chem. Soc. 22, 2436. Brown, H. C., and Y. Okamoto. 1958. J. Am. Chem. Soc. 22, 4979. Biggs, A. I., and R. A. Robinson. .2222, 388. Swain, C. G., and E. C. Lupton. 1968. J. Am. Chem. Soc. 22, 4328. Nys, G. G., and R. F. Rekker. 1974. Europ. J. Med. Chem. 2, 361. Moore, W. J. 1962. Physical Chemistry, 3rd Edition, Prentice Hall, NJ, p. 561. Jaffe, H. H. 1953. Chem. Rev. 22, 191. Meyer, H. 1899. Arch. Exp. Pathol. Pharmakol. 22, 109. Overton, E. 1899. Vierteljahresschr. Naturforsch. Ges. Zeurich. 22, 88. Meyer, K. H., and H. Hemmi. 1935. Biochem. Z. 222, 39. Hansch, C. 1971. Drug Design, Vol. I (Ariens, E. J., Ed.), Academic Press, NY, p. 271. Hansch, C. 1969. Accounts Chem. Res. 2, 232. Hansch, C. 1972. Adv. Chem. Ser. 114, 20. Hansch, C., A. R. Steward, S. M. Anderson, and D. Bentley. 1968. J. Med. Chem. 22, 1. Penniston, J. T., L. Beckett, D. L. Bentley, and C. Hansch. 1969. Mol. Pharmacol. 2, 333. McFarland, J. W. 1970. J. Med. Chem. 22, 1192. Higuchi, T., and S. S. Davis. 1970. J. Pharm. Sci. 22, 1376. Daniel, C., F. S. Wood. 1971. Fitting Equations to Data, Wiley- Interscience, NY. APPENDI X A APPENDIX A General Theory of Substituent Constants and Multiparameter Reggession Analyses In Part II of the Literature Review I dealt with a qualitative descrip- tion of the substituent constants that were used in this study and the method employed for the calculation of the correlation equations. I am now going to present a somewhat more quantitative approach that should help to explain where these constants are derived and to explain why these variables can be used to explore the nature of the inhibition of serine proteases. The parameters to be discussed will be W, the hydrophobic parameter and the electronic parameters 0, F, and R. The rationale behind the correlation of Ki with substituent constants is based on the fact that interactions in biological systems are subject to the same laws of thermodynamics which govern simple organic reactions. The substituent constants used in this work are derived from simple organic sys- tems and reflect the effect of substituents on equilibrium constants. The inhibition constants, Ki, obtained for each of the benzamidine derivatives studied is defined as the dissociation constant for the enzyme-inhibitor complex. Therefore, l/Ki is the association constant for the same complex. Ki is an equilibrium constant and is related tolAG by the thermodynamic relationshiplAG = -RTani. While the interactions which effect AG for biological systems are more complex than those of simple organic systems, these interactions are chemical interactions. A substituent on benzamidine 25 26 affects the nature of its chemical interaction whether it is a simple organic or a complex biological equilibrium. The Hydrophobic Parameter, n The parameter n, as with the case of o is best examined in extra thermo- dynamic terms. Consider the following equilibria for the reaction of movement of a solute between two solvent systems where H and L stand for hydrophilic and lipophilic solvents, respectively. L x - ¢ - H H - ¢ - H (2) H L X - ¢ - H X - ¢ - H (a) The free energy change associated with these reactions will be for (2) AG = 2.3RT log(H-2-H)L = 2.3RT logPH (2) (H-o-H) for (2) AG = 2.3RT log(X-2-H)L = 2.3RT 1°9Px (g) (X-¢-H) where (H-Q-H)L and (X-¢-H)L are the partition coefficients PH and Px' (H-¢-H) (X-¢-H) H respectively. It should be noted that PH and Px are not the thermodynamic partition coefficients since those are the ratios of mole fractions (XaL/XaH) = P', however, the ratio of concentration is more more commonly used and is therefore used to develop N values. Now subtracting Eq. 2 from Eq. 2 gives AGX - AGH = 2.3RT long/PH . (2p AGx - AGH = long - logPH (22 2.3RT is then defined as long - logPH and is a measure of the substituent- contribution to the difference of the change in free energy in the two molecules as they are partitioned. From thermodynamics it is known that 27 free energy changes are additive and this combined with observation leads to the fact that if polysubstituted molecules are used the value of "total for a molecule will be “total = Xvi. This is strictly only for an ideal case; however, this relation is often times true in practice. Hammett, o, and Swain and Lgpton, F and R The derivation of o is very similar to that of B. It is based on dif- ferences in the free energies of reaction of unsubstituted and substituted compounds. AG x 2.303RT long AGH 2.303RT logKH Therefore AGX " AGH = = long/KH 2.303 where o is the substituent constant and p is the reaction constant, a measure of the sensitivity of the equilibrium or reaction to changes in the substi- tuent. The O values defined by Hammett represent the effect of each sub- stituent on the ionization of benzoic acids under standard conditions. The reaction constant, p, is set equal to 1.00 and o = long/KH. The determina- tion of any two rate or equilibrium constants can then be used to determine p and the rates of reaction of other substituted compounds can be predicted. This relationship is also very valuable in determining the nature of the electronic characteristics of the transition state. A consideration of a plot of log K vs. 0 shows us that if p is positive the rate will be increased by electron withdrawing substituents and when p is negative the rate will increase with electron donating substituents. A number of other sub- stituent constants have been derived to explain deviations which occur by utilizing the original Hammett constants. In an effort to restore the 28 generality of using substituent constants to explain variations in rate data and to try to combine all of the various types of constants seen into one general set of constants, Swain and Lupton devised the para- meters F and R. The parameters, F and R, are defined to be measures of the field and resonance potentials of a substituent. The calculation of F and R rests on the assumption that any 0 set can be represented as a linear combination of two other oi sets and that one can ignore any factors other than field and resonance for a substituent kept three or more atoms distant from the reaction center. This leads to the equation oi = fF + rR where f and r are weighting factors dependent on reaction series and F and R are the field and resonance terms dependent on the substituent. For the calculation of F values the parameter 0' was used initially and the correlation of o' = aom + b0p was done. The F values (which are calculated values of 0') could then be determined from F = a0m + bOP. The term 0' was used as a measure of pure field effect since it is based on the ionization of 4—substituted bicyclo [2.2.2] octane-l-carboxylic acids. In this series, there is no resonance because there is no inter— vening conjugation or unsaturation between the substituent and the reac- tion center. To obtain R values the equation OP = aF + R was used. Before the application of the above equation, field effect may be removed from o by assuming that for the trimethylammonium ion substituent R = 0.00. pThe value a, the weighting factor for field effects, is readily calculated [op(trimethylammonium ion) = aF] and from this the rest of the R values may be obtained. eque C. ef.. the We a. ll? of NO lee l WEE 29 Multiple Parameter Regression Analysis (31) The following section will deal with the theory and derivation of general equations for the least squares technique. In least squares analysis the co- efficients of the equation are obtained by minimizing the sum of squares of the differences between the observed value yj and the expected value, Y- The 3' following least squares solution will be for three independent variables. The estimated equation is Y = b0 + lel + b2X2 + b3X3 (A) Since bo is always of the form h0 = y - E :ixi; for K constants (2) 1: We need only find the values of b1, b2 and b3 since upon substitution of (2) into (2) we get (2). Y - Y = b1(Xl"Xl) + b2 (XZ-XZ) + b3 (X3’X3) (2) The solution for the coefficients are obtained by minimizing the sum of squares of the residuals [rr] 1 Y. is fitted value N Z (yj-Yj)2 y- is observed value (2) j: 3 Substituting for - (from (2)) J Y N z i=1 Now differentiating (2) with respect to each bi and setting equal to zero leads to the following sets of equations: bl[ll] + b2[12] + b3[13] = [1y] b2[21] + b2[22] + b3[23] = [2y] (g) b3[31] + b2[32] + b3[33] = [3Y1 where [n] a 2(xlj-21)2; 30 [21] E [12] 2(le‘il)(X2j'§2)i [1Y1 53(X1j'§1)(yj-§); etc. to illustrate this solution drr _ _- _ _' - -- - -- '- EBI-— 22[(Yj Y) b1(X1j X1) b2(x2j x2) b3(X31 x3)](xlj x1) where (xlj-xl) = dbl(x1j-xl) dbl Continuation of this process leads to the equations (2). All that is left is to solve the simultaneous linear equations which may be done by the method of determinants. Quantitative Structure Activity Relationships: Inhibition of Human C15, Thrombin, Trypsin and Plasmin by Substituted Benzamidines by Daniel P. Roman, Jr. and David H. Bing (Manuscript submitted to Biochemistry) 31 ABS TRACT The inhibition constants for a series of substituted benzamidines were measured and correlated with various substituent parameters to try to determine some of the factors that control the magnitude of Ki. The most important single variable involved in the correlations for trypsin, thrombin and C15 is the Swain and Lupton resonance parameter, resulting in equations of the form 1/Ki = klR + k2 where R1 and k2 are calculated constants. For Cl§ the inclusion of 1TNO2 and (RN02)2 measures of sub- stituent hydrophobicity derived from the nitrobenzene series, results in a significantly better correlation. In the case of plasmin, molar refractivity (MR) and benzene series were the most significant single variables. It does not appear that any electronic effect plays a signi- ficant role in benzamidine binding to plasmin, thus making it very dif- ferent from the other three enzymes. A comparison between human trypsin and thrombin and their bovine counterparts indicates that dif- ferences exist, most probably in the hydrophobic regions of these enzymes. The equations generated for human plasmin in this work and in the work by Coats (1973), are similar, indicating that a mass syn- thesis approach to inhibition design is unnecessary. 32 C15, trypsin, plasmin and thrombin are all serine proteases, and all appear to possess an anionic binding site (Perlmann and Lorand, 1970; Muller-Eberhard, 1968). C15, a subunit of the first component of complement, Cl, is the enzyme responsible for the hydrolysis of C4 and C2, two early acting components in the complement sequence. complement is a group of eleven serum proteins which act as biological amplifiers in the immune response. Thrombin (EC 3.4.4.13) is a proteolytic enzyme whose main physiological activity is the conversion of fibrinogen to fibrin which can then react to form a fibrin clot. Plasmin (EC 3.4.4.14) is another serum protease whose primary function is the dissolution of the fibrin clot. Plasmin has also been shown to activate C15 (Ratnoff and Naff, 1967), and to catalyze formation of prekallikrein activating fragments from activated Hageman factor (Austen, 1971). Trypsin (EC 3.4.4.4), one of the most studied of the serine proteases, is a pancreatic digestive enzyme, which is responsible for the activation of the other digestive enzymes in the pancreas. These enzymes are known to hydrolyze basic and certain aromatic amino acid esters, in particular, Na-Z-L-Tyr-Np, Na-Z-L-Lys-p-Np and Na-Z-L-Lys-p-Np (Kezdy 22 22,, 1965; Silverstein, 1973; Bender 22 22., 1965; Martin 22 22., 1958). Extensive data exists on the inhibition of the bovine enzymes, thrombin, trypsin and human plasmin by substituted benzamidines (Mark- wardt 22 22,, 1968, 1969, 1970). Baker and his coworkers compiled *The terminology used for the complement proteins is that sug- gested in the Bull. Wld. Hlth. Org. "Nomenclature of Complement," Immunochemistry Z, 137, 1970. Other abbreviations used in this paper are: Na-carbobenzoxy-L-Tyrosine-p-nitrophenyl ester, Na-z-L-Tyr-p-Np; Na-carbobenzoxy-L-Lysine-p-nitrophenyl ester, Na-Z-L-Lys-p-Np. 33 34 considerable data on the inhibition of whole guinea pig complement by substituted benzamidines (Baker and Erickson, 1969; Baker and Cory, 1969a, 1969b, 1971a, 1971b). Coats (1973) has reported multiple para- meter regression analyses of these data, and has compared trypsin, thrombin and plasmin suggesting that differences in the anionic binding sites of these enzymes do exist. Hansch 22 22, (1974) have developed quantitative structure activity relationships (QSAR) from the data de- rived by Baker 22 22, (1969a, 1969b, 1971a, 1971b) and a fairly defin- itive direction is indicated in making more pharmacologically active compounds. The following study was undertaken to examine the inhibi- tion of these enzymes by a series of substituted benzamidines and to relate the inhibition constants to the physical-chemical properties of the substituents. MATERIALS AND METHODS Organic Compounds. Na-Z-L-Tyr-p-Np was obtained from Nutritional Biochemical Corp. Na-Z-L-Lys-p-Np was prepared from Na-carbobenzoxy- N5-carbo-t-butyloxy-L-lysine-p-nitrophenyl ester (Sigma Chemical Co.) according to previously described procedures (Bender 22 22,, 1965a). All of the above esters were determined to be greater than 99% homo- geneous as indicated by the presence of less than 1% p—nitrophenol in 10‘3 M solutions of these compounds at pH 8.1, 410 nm. Benzamidine, 3-aminobenzamidine and 3—nitrobenzamidine were purchased from Aldrich Chemical Co., 4-Nitrobenzamidine was prepared according to previously described procedures (Pinner and Gradenwitz, 1897). The remainder of the benzamidine compounds were the generous gift of Dr. Michael Cory, SRI, Menlo Park, CA, and the preparation of the compounds has been de- scribed (Cory, 1971). Enz es. Highly purified human C15 was prepared by affinity chroma- tography by the method of Assimeh 22.22, (1974). Human plasmin and human thrombin were the gift of Dr. Robert Rosenberg, Beth Israel Hospital, Boston, MA. Plasmin was purified on lysine-Sepharose as previously de- scribed (Deustsch and Mertz, 1970). Thrombin was purified by the method of Rosenberg 22_22. (1970). Human trypsin was kindly donated by Dr. James Travis, Department of Biochemistry, University of Georgia, Athens, GA. This protein was purified from human pancreas by affinity chromatography of TrasylolR—Sepharose (Johnson and Travis, 1975). 35 36 Protein concentrations were measured by the method of Lowry 22_22, (1951), and bovine serum albumin used to determine a standard curve. Enzymatic Analysis. All readings were made on a Gilford Model 240N spectrophotometer. The esterolytic activity of the samples was deter- mined using Na-Z-L-Tyr-p-Np as described by Bing (1969). A stock solu- tion of 10'3 M Na—Z-L-Tyr-p-Np was made in acetone and 0 to 50 ul of the solution was added to a solution of .005 ionic strength Tris-HCl con- taining 0.09 M NaCl, pH 8.1 containing the protein to be studied. The final volume of the assay solution was 1.0 ml. The production of p- nitrophenol was measured at 410 nm and recorded continuously on a Heath SR255B recorder. For the assay using Na-Z-L-Lys-p-Np, a stock solution of 10’3 M Na—Z-L-Lys-p-Np was made in 50% acetone-H20 and 10 to 50 pl of this solution was added to a 1.0 ml sample of protein in 0.02 ionic strength acetate solution pH 6.0 containing 0.085 M NaCl. For the human trypsin assay with both substrates, 50 mM CaCl2 was included in the assay buffer. Production of p-nitrophenol was measured at 340 nm and recorded continuously on a Heath Model SR255B recorder. A blank with no protein was run with each assay to correct for the rate of spon- taneous hydrolysis of Na-Z-L-Tyr-p-Np. The rate of spontaneous hydrolysis was less than 1% of the enzyme catalyzed hydrolysis of Na-Z-L-Lys-p-Np. All analyses were carried out between 25 and 26°C. The molar extinction coefficient of p-nitrophenol at 340 nm, pH 6.0 was determined to be 5.85 x 103. Inhibition was measured at a substrate concentration of 3 x 10'4 with the inhibitor concentration adjusted to give approximately 10 to 80% inhibition. Plots of l/v vs. inhibitor concentration were made. Ki values were calculated using the slopes of these plots and previously 37 determined Vmax (Dixon, 1953). Ki is assumed to be an estimate of the dissociation constant for the enzyme-inhibitor complex. Competitive inhibition was shown by the methods of Cornish-Bowden (1974) and Dixon (1953). Plots were made of s/v vs. inhibitor concentration at three different substrate concentrations, and l/v vs. inhibitor concentration at two substrate concentrations. Plots of s/v vs. I for the four en- zymes are linear indicating competitive inhibition. In order to investigate the nature of the factors responsible for the magnitude of Ki exerted by some substituent x, the following general equation was used: log l/Ki = k + k'xl + k"x2 + k"'x3 where k, k', k", and k"' are constants which are fixed for a given system and derived from multiparameter regression analysis and x1, x2 and x3 are various substituent parameters. Table I is a list of the parameters that were used in the biological correlations. The QSAR were done on the SRI CDC 3600 computer with a multiparameter regression analysis program. 38 Table 1. Substituent Constants Used in Quantitative Structure Activity Relationships 0, cm, op - Electronic Parameters - Hammett Constantsa Swain and Lupton Parametersb F - Field Effect R - Resonance Effect n, "N02 - Hydrophobicity (Benzene, Nitrobenzene Series, respectively)c F - Hydrophobic Fragmental Constant developed by Nys and Rekkerd M.W. - Molecular Weight M.R. = PE - Molar Refractivity, Polarizabilitye aHammett (1970) waain and Lupton (1968) CLeo 22 gr. (1971) dNys and Rekker (1974) eHansch 22H22. (1974) RESULTS AND DI SCUSS ION Reaction of Human Trypsin, Plasmin, Thrombin and C15 with Na-Z-L- gyr-p-Np. The rate of the reaction of the four enzymes with 3 x 10’4 M Na-Z-L-Tyr-p-Np was proportional to enzyme concentration. The reaction was linear for at least three to five minutes depending on the amount of enzyme used. The deviation from linearity beyond five minutes was probably due to a buildup of product and competitive inhibition between Na-Z-L-Tyr and Na-Z-L-Tyr-p-Np. The kinetic parameters for the reaction of C15, trypsin, plasmin and thrombin are shown in Table II. Figure 1 shows the Lineweaver-Burk plots of the reaction of each of the enzymes with Na-Z-L-Tyr-p-Np . Reaction of Human C15, Plasmin, Trypsin and Thrombin with Na-Z-L- Lys-p-Np. As in the case of Na-Z-L-Tyr-p-Np, the reaction of the four enzymes with 3 x 10’4 M Na-z-L-Lys-p-Np was linear with enzyme concen- tration. The assay of these enzymes was attempted at pH 8.1 in order to get a better comparison of the relative affinities and kcat values of C15, thrombin, trypsin and plasmin for Na-Z-L-Tyr-p-Np and Na-Z-L- Lys-p-Np. However, at pH 8.1 the lysine ester is very unstable and the spontaneous rate of hydrolysis was almost equal to hydrolysis in the presence of enzyme. At pH 6.0 the enzymatic rate of hydrolysis was linear for five to six minutes and the spontaneous rate of hydrolysis of Na-Z-L-Lys-p-Np was negligible. The values for Km(app)' 39 38: Am mm >385 35.0.: :kumum>.n..nmm 4O om. mm. mm.H on.m m.H Ho.vH mac one. II pH.m mm.m ov.o mm.a geneouse wow u: mo.m owe. mmm. mv.m seammam m.om H.mm mo.m m.em M.Rm mmm. camoxue Iaummwe Aos\eas\z omwwe Admmmwwv Aanwwwv ms\efi5\z omwwv Mummwaxv wesuem x > x x > as mzumumsononuuez ozumwumeuoumnez mznosmmqnouuuez eem o2ndnu>euqunuez cuss caneoune poo .CHEmMHm .cwmmuxe .mHU c055: mo noduooom you mumuwfiouom OHDOCAM .N «Home Figure l. 41 Lineweaver-Burk plot of the reaction of Na-z-L-Tyr-p-Np with human C15, trypsin, plasmin and thrombin. Assays were performed as described in Methods. Ordinate, recip- rocal of initial rate in min mmole'l; abscissa, recipro- cal of concentration of substrate 1. mole'l. 42 7 mac: ._ m- .3. up Or a 2.95:9... ll 2520.51... z.5m<._a or x m: o N. n .A no... *- 9 W m D. W w 0 1. 3 cc 0. ... 06 N... Figure 1. 43 Vmax and kcat for C15 and trypsin and Km(app) and kcat for thrombin (Kezdy 22_22,, 1965) and plasmin (Silverstein, 1973) are shown in Table II. Figure 2 contains plots for trypsin and C15 for their reac- tion with Na-Z-L-Lys-p-Np. The hydrolysis of some synthetic neutral and cationic substrates by bovine trypsin (Bender 22 22,, 1965; Martin 22 22,, 1958), human plasmin (Silverstein, 1973), thrombin (Kezdy 22_22,, 1965) and C15 (Bing, 1969) have been reported. This part of the study was designed to accumulate Km(app)r Vmax and kcat values for the hydrolysis of Na-Z- L—Tyr-p-Np and Na-Z-L-Lys-p-Np by the human enzymes, trypsin, plasmin, thrombin and C15. Table III contains the values of Ki obtained for each of the four enzymes. Table IV contains the values for W, “N02, f, F, R and MR which were used in the correlations. In the correlations obtained for all the enzymes, except plasmin, R, the Swain and Lupton resonance parameter (1968) was the most signi- ficant single variable. Only thrombin produced an equation that was significant at the .95 level of the F test with R alone. All the other enzymes were dependent on more than one term to give equations that are considered significant at the .90 level or better. The best sin- gle variable equation for plasmin was a correlation in MR. Correla- tions with n and D gave equations which were similar in their F test values. D, is a dummy variable used to determine if 4-position isomers had a special effect on the inhibition of the enzymes compared to 3-position isomers. D equals 1 for 4-position and zero for 3- position substituents. Table V is a summary of the most significant equations obtained for each enzyme. Figure 2. 44 Lineweaver-Burk plot of the reaction of Na-Z-L-Lys-p-Np with human C15 and trypsin. Assays were performed as described in Methods. Ordinate, reciprocal of initial rate in min. mmole’l; abscissa, reciprocal of concen- tration of substrate in l. mole-1. 45 C15 5. ° 2 >. C ... O O 4 O O Q G N ‘- ,_sa1ow w NIIN 9_0L x All. Figure 2. 10 1/8 x 10‘5 L MOLE "‘ 46 :3: mm mm 030 Avhma .mnmav mexmm nan mhzn 35m: .flm Mic. someomm om. om. man. He. oo.o oo.o oo.o memmouv me. me. map. Hm. 80.0 08.8 oo.o memeoum me. me. ems. em. Hm.n om. s~.u omzne mm.~ mm.~ Ame. me.a sm.- mm. OH. axomouosxoemeeum em. em. eve.a mm.H ma. mm. em. umoooum No.u mm.u new. ov.- mm. mm. mm. moooum HH.H mm. mom. m~.H ma. mm. me. mmoum mm. mm. mom. on. ma.. vo.- RH.- mznm mm.. mm.. ems. oo.- me. so. he. Nozum m~.: m~.u ems. mo.u we. so. we. Nozl. mH.H NH.H mH.H ov.H ma.- «8.: so.. maneum.m Ho.a NH.H mH.H ov.H o~.u mo.- v~.- essene.m om.m Ho.~ oo.m me.m Ho.u mo.. 80.- assemmnm em.H om.H em.m m.a mo.u mo. mo. deemeeum em. mo.- ems. «em. Hm.- on. ma. omznm 08.- em.- mmm. RH.H AH.- we. mm. umtm me.- m~.H- «em. Hm.- mo.- «0. ea.. mmznm 00.0 80.0 ema. 00.0 00.0 00.0 00.0 m camocncmmdozv : o Amcmucwmv = mum .2 m Mm. Mm ml N x Hoe museumeoo muemeuaumnsm .m wanes 47 Table 4. Benzamidine Inhibitors Tested for Inhibition HZN/ANHZ Log 1/Ki 2_ Trypsin Plasmin Thrombin 225_ H 5.11 3.28 3.14 3.21 3-NH2 5.20 3.99 5.18 3.24 3-Br N.T. 2.13 N.T. 4.01 3-MeO 4.87 3.53 3.42 3.77 3-Phenyl 5.36 3.52 2.00 3.64 3-Benzyl 5.20 3.86 3.15 3.19 3,4-di-Me 5.08 2.48 2.96 3.93 3,5-di—Me 4.83 3.57 4.17 4.09 4-N02 3.87 2.44 2.63 1.03 3-NO2 4.10 2.95 2.69 2.79 2-Me 3.77 3.47 2.48 2.77 3-CF3 3.87 4.38 2.34 2.30 3-COOH 2.32 3.37 N.T. 3.22 3-C02Et 3.97 N.T. 2.59 N.T. 3-phenoxypropoxy 4.97 4.67 4.28 4.95 4-Me0 4.06 N.T. N.T. No Inh. 3-CH20H 4.52 2.61 2.22 2.01 4-CH20H 4.08 3.17 1.97 No Inh. 4-OH N.T. No Inh. No Inh. No Inh. 4-EtO N.T. No Inh. N.T. No Inh. 48 Table 5. Correlation Equations for Trypsin, Plasmin, Thrombin and C15 (All Compounds) Tr sin e e 5 5 log l/Ki = 4.24 - 1.33 R 17 .70 .50 1.25 = 4.54 - 1.39 o 17 .70 .49 1.23 = 4.70 - 1.58 F 17 .70 .49 1.23 = 4.20 + .l70(nN02)2 17 .70 .49 1.23 = 4.32 - 1.230 + .15(nN02)2 17 .63 .65 1.44 = 4.46 - 1.26 r + .13(UN02)2 17 .65 .62 1.34 = 4.05 - 1.23 R + .16(nN02)2 17 .62 .67 1.51 21_5 log 1/Ki = 2.91 - 2.25 R 15 .78 .61 1.38 = 2.70 + .451:N02 - 2.1 R 15 .64 .78 2.07 = 2.75 + .97nN02 - 1.95 R - 15 .60 .82 2.32 .33(nN02) Plasmin log 1/Ki = 3.49 - .79 0 16 .65 .54 1.17 = 2.99 + 2.93 M.R. 16 .64 .47 1.20 = 3.16 + 3.00nN02 16 .64 .47 1.20 Thrombin log l/Ki = 2.73 - 2.52 R 15 .62 .75 2.59 n, number of data prints 3, standard deviation from regression line r, correlation coefficient P, value F statistic 49 Thrombin was the only enzyme studied that was significantly corre- lated with only one variable. Addition of other terms did not signifi- cantly improve the equation in R alone. In fact, no other equation, single or multivariable, produced a more significant equation as judged by the F statistics. The correlation indicated that inhibition of Cl§ by benzamidines was dependent on both R and NNOZ. R is the most important single para- meter. A combination of "N02 and R was still insufficient to explain the variance of the data, however, the addition of (15102)2 resulted in an equation that was significant at the .95 level of the F test. The data on the inhibition of Cl§ by 4-substituted benzamidines deserves additional comment. No inhibition of C1§ was observed for 4-OH, 4-0Me and 4-OBt. Only 4-N02 and 4-CH20H produced measurable levels of inhib- ition and the expected values of Ki for these compounds were much smaller than those observed. Since only these two Ki values for 4-substituted benzamidines were available, a correlation of D with log l/Ki was not significant. Two explanations of the data are possible. First, inter- action might affect the binding of the para compounds. An anomaly exists in the fact that the 3,4-dimethyl compound is a very good inhib- itor. The other explanation is that the positive charge of the amidine is destroyed by resonance through the ring as shown: C C H2N/ + \H2 H2!‘( +\NH2 50 Resonance destruction of the positive charge most probably accounts for the failure of 4-hydroxy benzamidine to inhibit C1§ as well as plasmin and thrombin. The other 4-substituted benzamidines inhibit plasmin, trypsin and/or thrombin to varying degrees, but all 4-substituted com— pounds are poor inhibitors of C15. Therefore, we propose a steric interaction is interfering with the binding of these compounds to Cl§. Baker and Erickson (1969) have reported that 4-substituted benzamidines are not inhibitory against whole guinea pig complement. Further investi- gations with 4-halogen and 4-CF3 substituted compounds could be done to determine the possible limits of substituent size in the para position of benzamidines for the inhibition of Cl§. The trypsin data did not correlate well with any of the variables tested. The equations for trypsin given in Table IV were judged to be the most significant. They are presented with some reservation since none were judged to be significant at even the .90 level of the F test; however, some interesting points can be made. The most important sin- gle variable equation for trypsin is R. However, the equations in F and o are very close in terms of the F test, and it is difficult to assess what kind of electronic interaction is influencing the inhibi- tion. One more interesting equation is one that involves ("N02)2' A plot of log l/Ki vs. (flN02)2 gives a parabolic curve in which log l/Ki increases with the absolute value of "N02‘ In the case of trypsin, the reason for this can possibly be explained based on the nature of enzyme specificity. Trypsin is a digestive enzyme which will hydrolyze virtually all x-lys-x' or x-arg-x' bonds; thus, the location for bind- ing x must be able to accomodate a wide range of lipo- or hydrophobic groups. If the hydrophobic pocket that binds benzamidine is long 51 enough to accomodate the substituent, then the walls of the crevice may show different degrees of hydrophobicity. If one wall is hydrophobic it could interact better with a hydrophobic substituent thus giving an increase in binding of such molecules. If the other wall is relatively hydrophilic, polar groups could interact there just as readily. The differential hydrophobicity in the region directly to the rear of the benzamidine (relative to the active site serine) (Krieger gt_§l., 1974) then explains the dependence of log l/Ki on ("N02)2' since then binding will be expected to increase with the absolute value of “(N02). Similarly, the equations generated for plasmin (see Table IV) were insufficient to explain the variance of the data. Examination of the equations tells us something about the binding procket of plasmin. First, the best equation was a function of MR. An equation as a func- tion of n was similar to the MR equation in terms of the F statistic. This similarity is not unusual since MR and n display a good deal of collinearity. This means that their values are dependent on some of the same factors such as, molecular weight and electronic character, and that these values increase or decrease in a parallel manner. If these equations are indeed indicative of the nature of the binding site of plasmin and no electronic terms are involved, then we suggest that plasmin is different from the other three enzymes studied and Ki is dependent on only hydrophobic or molecular size parameters. The one final equation for plasmin is one that involves the dummy variable, D. This indicates that 4—position isomers affect the inhibition in a negative manner, as was the case with Cl§. It is clear that R is the single most important factor which can affect the magnitude of Ki in three of the four enzymes. Two questions 52 can be asked concerning the effect of R on log l/Ki' First, why should R be significant? The resonance interaction of a meta substituent with the amidine center will be minimal because classical resonance structures do not accomodate this interaction. The second question to be asked is why should log l/Ki show a negative dependence on R, indicating that the bet— ter electron donors give the better benzamidine inhibitors. This in- creased density in the ring and the charged moiety will tend to deloca- lize the positive charge on the amidine group and the interaction of this charge by the anionic site in the enZyme will be less favorable. Since the proposed mechanism of inhibition involves formation of an ion pair between the cationic benzamidine and the anionic site, the delocalization of the positive charge should result in a decrease in inhibition. This is the opposite of what was observed. The resonance effect and the negative dependence on R can be ex- plained as follows. The mechanism for classical resonance from a meta substituent does not allow for a direction of the electrons to the ami— dine center but rather a resonance distribution of the electron pair around the benzene ring. The increased electron density in the benzene ring might allow for the formation of a n-w interaction between the enzyme and the benzamidine. Evidence for this type of interaction is found in bovine trypsin. East and Trowbridge (1968) report the observation of a difference spectrum on the binding of benzamidine to bovine trypsin indi- cating some perturbation of the w+n* transition for the benzene ring. The formation of a n-n interaction between the benzamidine ring and the enzyme could elicit a secondary electronic effect on the amidine group by feeding electron density into the enzyme and thus withdrawing it from the amidine center. This would make the amidine more positive and increase the ionic bond strength. S3 The low correlations with all compounds for trypsin and plasmin and to a lesser extent Cl§ were somewhat distressing. However, it is our feeling that some other variable, perhaps one unknown at this time, is necessary to explain the data. For example, calculations of electron den- sity around the ring or charge transfer through fl-fl interaction from the benzamidine ring and its substituents to the enzyme surface could be important. This is not unrealistic considering the nature of the depen- dence of the equations on R, and suggests that present variables are insufficient to explain dependence of structure on log l/Ki except at a very rudimentary level. In order to determine if significant equations could be obtained and if there might be some effect that we have not considered, compounds with large deviations between observed and calculated log l/Ki were eliminated from the correlations. Table VI lists the various equations obtained for each of the enzymes and the compounds that were eliminated from the compu- tation. The equations generated without these compounds gave better correlations and the significance of the equations also increased. Tryp- sin, thrombin and Cl§ gave equations with F test values of .975, .975 and .90, respectively in which the most significant variable was R. Plasmin remained distinct from the other enzymes in that the two best equations, significant at the .90 level, were functions of MR or n. Coats (1973) and subsequently Hansch (1974) investigated quantitative structure activity relationships for bovine thrombin and trypsin, human plasmin and the whole guinea pig complement system. With the data pre- sented in this paper, it is now possible to make a comparison between the human and bovine enzymes. The equations generated in this study indicate that the human enzymes are different from the corresponding bovine enzymes. 54 Table 6. Correlation Equations for C15, Trypsin, Thrombin, Plasmin (Selected Compounds) Plasmin Eliminate 3-NH2, 3,4-diMe, 4-N02, 3-CF3, 3-Br log l/Ki 3.2 + .3511 3.00 + .32 M.R. Tr sin Eliminate 3-Phenyl, 3-Benzyl, H, 4-MeO, 3-COOH log l/Ki = 4.1 - 1.3 R 915 Eliminate 3-NH2, 2-Me, 3-CH20H log l/Ki = 3.01 - 3.04 R Thrombin Eliminate 2-Me, 4-CH OH 2 log l/Ki = 2.87 - 2.55 R It! 11 ll 11 12 13 IU) .32 .32 .25 .61 .45 [H .81 .81 .88 .80 .87 55 The major factors controlling the inhibition of human trypsin and thrombin seem to be electronic in nature; however, “N02 or P appears to be impor- E tant in the inhibition of the bovine enzymes. Coats (1973) also found resulted in a significant im- that the combination of cm with “N02 or PE provement in the correlation of the data and therefore electronic effects are important. We do not question the contribution of am to inhibition by benzamidines since we have found R to be the most important single variable for human trypsin and thrombin. However, we do question the validity of the equations generated by Coats (1973) for the following reasons. First, the range of cm constants that were used by Coats (1973) in his correlations for trypsin was narrow (.19 units), and five of the eight Om constants had the same value. The second objection is that the formation of a two variable equation for trypsin is at the limit of data points per variable. Topliss and Costello (1972) have suggested that at least five to six data points per parameter should be available in order to avoid chance correlations. A comparison between our equations and the ones calculated by Coats (1973) for human plasmin are very similar, if one ignores the equation in am. We have the same objections to this equation as for the trypsin equations. The similarity of the equations generated in this paper to those developed by Coats (1973) help to point out the applicability of multiparameter regression analysis. Markwardt 33 El. (1968, 1969, 1970) evaluated a number of alkoxy benzamidines different from the substituted benzamidines we have studied. Even so, the equations l_to §_below show that the same direction for inhibitor synthesis is indicated. This also shows that a mass synthesis approach to inhibitor design is not really necessary. For example, the more hydrophobic inhibitors will, to a 56 point, be the best. Coats (1973) saw that the inhibition for 3-OC7H11 benzamidine did fall off. Although this is an isolated case, we agree that this may be near the upper level of hydrophobicity acceptable and that even more hydrophobic groups would eventually lead to the inclusion 2 of 0 term. Plasmin (from Coats 1973): log l/Ki = .140 + 3.64 n=8 s=.l4 r=.80 .l *log l/Ki = .ZPE + 3.60 =8 s=.l3 r=.85 '_ Plasmin (this work): log l/Ki = .35“ + 3.2 n=ll s=.32 r=.81 .3 log l/Ki = .32 MR + 3.00 n=ll s=.32 r=.81 '1 A comparison of the CIE data with the data for the inhibition of guinea pig complement is made with some reservations as we are analyzing one enzyme, and guinea pig complement consists of at least eleven proteins. However, an inspection of two of the equations for C1§ and whole comple- ment showed them to be very similar with respect to the values of their coefficients. Cl§ (this work): log l/Ki = 2.91 + .45nN02 - 2.1R n=15 s=.64 r=.78 |U1 Whole guinea pig complement (Coats 1973): log l/ISO = 2.56 + .26nN02 - 1.230m n=25 s=.29 r=.83 .6 Combination of this observation with other data that shows C1§ to be in- hibited by a wide variety of benzamidines (Bing, 1969) supports the conclu- sion that a main site of action of whole complement inhibition by benzami- dines is in the Cl§ molecule. *Note: We have multiplied the PE coefficient by ten to make it more equiscalar with our MR values since we have multiplied our MR values by .l to put then in a range with w. 57 We have shown that when a subset of all compounds tested is used in the biological correlations that three enzymes, trypsin, thrombin and C1§ are very similar while plasmin is different. However, when all of the compounds are included, the correlations, even though somewhat less signi- ficant, do begin to point out differences among the enzymes studied. We believe that the decreased significance is a result of some unknown para- meter which is required for good correlations. The usefulness of the equations lies in the fact that they do indicate a direction for the synthesis of new compounds that could show selective specificity for the enzymes studied. ACKNOWLEDGEMENTS The authors would like to thank Ms. Sandra Spurlock and Dr. Judith Andrews for their interest, help and criticism of our work, and Ms. Rachelle Rosenbaum for typing the manuscript. 58 LI ST OF REFERENCES LIST OF REFERENCES Assimeh, S. N., Bing, D. H., and Painter, R. (1974), J. Immunol. 113, 225. Austen, K. F. (1971), Hopkins J. Med. 128, 57. Baker, B. R., and Cory, M. (1969a), J. Med. Chem. 12, 1049. Baker, B. R., and Cory, M. (1969b), J. Med. Chem. 13, 1053. Baker, B. R., and Cory, M. (1971a), J. Med. Chem. 14, 119. Baker, B. R., and Cory, M. (1971b), J. Med. Chem. 14, 808. Baker, B. R., and Erickson, E. H. (1969), J. Med. Chem. 12, 408. Bender, M. L., Kezdy, F. J., and Feder, J. (1965), J. Am. Chem. Soc. 81, 4953. Bender, M. L., Killheffer, J. V., and Roeske, R. W. (1965), Biochem. Biophys. Res. Comm. 12, 161. Bing, D. H. (1969), Biochemistry _8_, 4503. Coats, E. A. (1973), J. Med. Chem. 16, 1102. Cornish-Bowden, A. (1974), Biochem. J. 137, 143. Deutsch, G. E., and Mertz, E. T. (1970), Science 170, 1095. Dixon, M. (1953), Biochem. J. 55, 170. East, E. J., and Trowbridge, C. G. (1968), Arch. Biochem. Biophys. 125, 334. Hammett, L. P. (1970), Physical Organic Chemistry, 2nd Ed., McGraw-Hill, NY. Hansch, C., Leo, A., Unger, S. H., Kim, K. H., Nikaitani, D., and Lien, E. J. (1973), J. Med. Chem. 16, 1207. Hansch, C., and Yoshimoto, M. (1974), J. Med. Chem. 11, 1160. 59 60 Johnson, D., and Travis, J. (1976), Anal. Biochem. 23, 573. Kezdy, F. J., Lorand, L., and Miller, K. D. (1965), Biochemistry 4, 2302. Krieger, M., Kay, L. M., and Stroud, R. M. (1974), J. Mol. Biol. 83, 209. Leo, A., Hansch, C., and Elkins, D. (1971), Chem. Res. 11, 525. Lowry, O. H., Rosebrough,, N. J., Farr, A. L., and Randall, R. J. (1951), J. Biol. Chem. 193, 265. Markwardt, F., Landmann, H., and Walsmann, P. (1968), Eur. J. Biochem. 6, 502. Markwardt, F., Walsmann, P., and Kazmirowski, H. G. (1969), Pharmazie 24, 400. Markwardt, F., Walsmann, P., and Landmann, H. (1970), Pharmazie 25, 551. Martin, C. J., Bolukow, J., and Axelrod, A. E. (1958), Biochem. Biophys. Acta 21, 430. Muller-Eberhard, H. J. (1968), Advanced Immunology, Vol. 8, Dixon, Jr., F. J., and Kunkel, H.C., Eds., Academic Press, NY. Nys, G. G., and Rekker, R. F. (1973), Chim Ther. 8, 521. Nys, G. G., and Rekker, R. F. (1974), Eur. J. Med. 2, 361. Rosenberg, R. D., and Waugh, D. F. (1970), J. Biol. Chem. 245, 5049. Silverstein, R. M. (1973), Thrombosis Res. 3, 729. Swain, C. G., and Lupton, E. C. (1968), J. Am. Chem. Soc. 20, 4328. Topliss, J. G., and Costello, R. J. (1972), J. Med. Chem. 15, 1068. Walsh, K. A., and Wilcox, P. E. (1970), Methods in Enzymolggy 12, 31. CONCLUDING REMARKS CONCLUDING REMARKS The data presented in this thesis demonstrate: 1) The human enzymes, trypsin, plasmin, thrombin and C15 hydrolyze both neutral (Na-Z-L-Tyr-p- Np) and cationic (Na-z-L-Lys-p-Np) substrates and are inactivated by PMNAB; 2) The binding of neutral and cationic substrates for hydrolysis most likely occurs in the anionic binding site: 3) The binding sites of the human enzymes are different from one another as determined by QSAR techniques; 4) Bovine trypsin and thrombin differ from their human counterparts; 5) The main site of action of substituted benzamidines on whole guinea pig complement is probably on the C15 subunit of this enzyme; 6) The most common variables presently used in QSAR are not sufficient to determine an accurate description concerning the factors responsible for influencing the binding of substituted benzamidines by the four enzymes studied. Multiparameter regression analysis is a useful alternative technique for the design of active site directed reversible and irreversible inhib- itors. Most of the investigations on developing these types of inhibitors have relied on luck, intuition and educated guesses. The studies in this thesis have shown that after analyzing a small number of inhibitors, a direction for further inhibitor design can be suggested. QSAR develop- ments have raised the design of inhibitors to a point somewhere over the educated guess phase. Further developments in this area, such as, 61 62 the use of new parameters (e.g. charge transfer complex calculation, or molecular orbital-charge density calculations) offer the promise of a more rigorous absolute approach to inhibitor design. We have also used the data obtained in the QSAR analysis to support our contention that the binding site of neutral aromatic and cationic substrates are identical. We have investigated the use of Na-Z-L-Lys- p-Np in the determination of K1 by substituted benzamidines. The values of K1 for both Na-Z-L-Tyr-p-Np and Na—z-L-Lys-p-Np agree well within experimental error. This of course should be predicted since Ki is independent of the substrate used provided that the same mechanism of inhibition is seen. This is further evidence that the lysine and tyro- sine esters bind at the same location. We cannot stress the importance of the QSAR technique enough. It is true that it appears that the present state of the art is insuffi- cient to analyze specific systems, such as enzyme active sites. How- ever, refinement and development of this technique to include new con- stants and approaches will be very important in expanding our knowledge concerning the nature of enzyme specificity. From a practical stand- point, QSAR will be important in designing new effective drugs for use in disease states. It is possible to perform dual QSAR.on drugs and examine both effectiveness and toxicity. Figure l is an example where one can plot toxicity and effectiveness vs. some parameter and allow one to maximize the effectiveness while arriving at an acceptable level of toxicity. This analysis is critical to drug design whether it be by intuition, luck, guesswork or QSAR. 63 ? Biological Activity ll" \ \ Substituent Parameter Figure 1.