PHYSICAL CHEMlCAL STUDY OF MONENSIN AND ITS ALKALI METAL [ON COMPLEXES ‘Ehesis for the Degree of Ph. D. MICHIGAH STATE UNIVERSH’Y PAUL GILMAN GERTENBACH 1975‘ "v LIMA. iii Iz’iélelfim'n State C} Uaivsrsity MY This is to certify that the thesis entitled PHYSICAL CHEMICAL STUDY OF MONENSIN AND ITS ALKALI METAL ION COMPLEXES presented by Paul Gilman Gertenbach has been accepted towards fulfillment of the requirements for Ph.D. Chemistry degree in / - , Major professor Date September 23, 1975 0-7639 am‘omc av "0A5 8.80“? BIND . news semen)"; MICIISMI ABSTRACT PHYSICAL CHEMICAL STUDY OF MONENSIN AND ITS ALKALI METAL ION COMPLEXES By Paul Gilman Gertenbach Alkali metal ion complexes of an antibiotic drug, monensin, have been studied using potentiometric, infrared and lithium-7, carbon-l3, Sodium-23 and proton nuclear magnetic resonance techniques in several nonaqueous solvents. The acidity constant of monensin was determined in methanol solution. Stability constants for the monensin- sodium ion complex in methanol were determined by titrating the acid with sodium salts. Complexation titrations with several other cations revealed a monensin selectivity order for the alkali metal ions as well as for silver and ammonium. Evidence was also found for complexation of alkaline earth ions by monensin. Infrared measurements in chloroform and methanol indicated that monensin forms a complex in neutral solutions which is different from the normal sodium complex prepared from basic solutions. Sodium-23 and lithium-7 NMR measurements in methanol solution served to confirm potentiometric evidence of monensin interaction with these ions. The proton NMR spectrum of the monensin sodium salt was interpreted. The monensin acid spectrum was monitored during sodium ion addition. Sodium ion induced conformational changes in the Paul Gilman Gertenbach monensin molecule were noted. Monensin was found to bind water molecules at two different sites. These water molecules can exchange with one another or with free water in the bulk solvent in acetone, tetrahydrofuran and chloroform solutions. Complexation of the sodium ion by monensin influences this water exchange. A new complex of monensin with sodium perchlorate or other sodium salts was isolated from several solvents and characterized using potentiometric, infrared, Raman, proton NMR, elemental analysis and thermogravimetric techniques. Analyses indicated that the complex is made up of monensin and the sodium salt and the solvent used for preparation in a 1:1:1 mole ratio. Additional proton NMR studies revealed the possibility of fast chemical exchange of sodium ions between this new complex and the normal monensin sodium salt in chloroform solutions. Evidence for conformational equilibria between the two monensin forms was also obtained. Experiments with a bimolecular lipid membrane (BLM) system indicated that monensin facilitates the transport of sodium ions through lipid barriers. Several tetrazole and glutarimide molecules which are also alkali metal ion complexing agents were not found to influence BLM permeability. The dilactam form of the synthetic macroheterobicyclic ligand cryptand 222 was shown to affect BLM ion transport while the cryptand 222 did not. PHYSICAL CHEMICAL STUDY OF MONENSIN AND ITS ALKALI METAL ION COMPLEXES By Paul Cilman Gertenbach A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1975 ACKNOWLEDGEMENTS The author wishes to thank Professor Alexander I. Popov for his vigilant and sincere concern about providing a proper introduction to chemical research. The helpful suggestions of Professor Christie G. Enke, who served as second reader, are also acknowledged. Gratitude is offered to Eli Lilly and Company whose generous gifts of monensin sodium salt made this study possible. Financial aid from the Department of Chemistry, Michigan State University, the National Institutes of Health of the Department of Health, Education and Welfare and the National Science Foundation is gratefully acknowledged. Appreciation is extended to Eric T. Roach for his help and encouragement on the nuclear magnetic resonance portions of the work. Stimulating and enlightening discussions with John G. Hoogerheide were also of benefit to the investigation as a whole. Finally, very special thanks to my parents who have always exhibited unswerving confidence in the direction of my academic and scientific efforts. ii TABLE OF CONTENTS Chapter Page I. HISTORICAL INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . l MONENSIN . . . . . . . . . . . . . . . . . . . . . . . . 2 BIMOLECULAR LIPID MEMBRANES . . . . . . . . . . . . . . 9 II. EXPERIMENTAL MATERIALS Reagents . . . . . . . . . . . . . . . . . . . . . . 15 Solvents . . . . . . . . . . . . . . . . . . . . . . l6 Monensin . . . . . . . . . . . . . . . . . . . . . . l7 MEASUREMENTS Potentiometric . . . . . . . . . . . . . . . . . . . 23 Spectroscopic . . . . . . . . . . . . . . . . . . . 24 Other Analyses . . . . . . . . . . . . . . . . . . . 26 Membrane Resistance . . . . . . . . . . . . . . . . 26 III. POTENTIOMETRIC STUDIES OF MONENSIN INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . 30 DETERMINATION OF ACID DISSOCIATION CONSTANT . . . . . . 31 DETERMINATION OF THE OVERALL COMPLEXATION CONSTANT . . . 43 INFRARED STUDY OF COMPLEXOMETRIC EQUILIBRIA . . . . . . 57 IV. NUCLEAR MAGNETIC RESONANCE STUDIES OF MONENSIN INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . 68 LITHIUM-7 NUCLEAR MAGNETIC RESONANCE STUDIES . . . . . . 68 SODIUM-23 NUCLEAR MAGNETIC RESONANCE STUDIES . . . . . . 69 iii Table of Contents (Continued) THALLIUM-ZOS NUCLEAR MAGNETIC RESONANCE STUDIES. PROTON MAGNETIC RESONANCE STUDIES SPECTRAL ASSIGNMENT . . . . . . . . . . EFFECTS OF COMPLEXATION . . . . . . . . V. STUDIES OF A NEW MONENSIN COMPLEX INTRODUCTION . . ISOLATION AND CHARACTERIZATION . EXCHANGE STUDIES VI. BIMOLECULAR LIPID MEMBRANE STUDIES INTRODUCTION . INITIAL TESTS OF THE SYSTEM . . . . . . . . MONENSIN AND OTHER COMPLEXING AGENTS VII. APPENDICES I. APPLICATION OF COMPUTER PROGRAM KINFIT AND SUBROUTINE EQUATION FOR THE DETERMINATION OF ACID DISSOCIATION CONSTANTS . . . . . II. APPLICATION OF PROGRAM STEUHR FOR THE DETERMINATION OF ACID DISSOCIATION CONSTANTS . III. APPLICATION OF COMPUTER PROGRAM KINFIT AND SUBROUTINE EQUATION FOR THE DETERMINATION OF COMPLEXATION CONSTANTS. IIX. LITERATURE CITED . . . . . . . . . . . . . . . iv 78 78 91 .108 .108 .123 .132 .132 .134 .137 .144 .152 .161 LIST OF TABLES Table l Equilibrium*Constant for the Reaction MonH + Na+ t MonNa + H+ in Methanol Solution at 25°C . 2 Equilibrium Constants for the Reaction MonH + M+ Z MonM + H+ in Methanol Solution at 25°C . . . 3 Equilibrium Constants for the Reaction MonH + M Z MonM + H+ in Methanol Solution at 25°C . . . 4 Illustration of the Analysis of Several Peaks in the MonNa Proton Magnetic Resonance Spectrum . . . . . . 5 Band Positions in the Raman Spectrum of the Solid Monensin Sodium Salt . . . . . . . . . . . 6 Elemental Analysis of New Monensin Complexes . . . . . . 7 Positions of the l and 3 Methyl Groups in Several Monensin Solutions in Chloroform-d1 . . . . . . Page 46 50 54 81 .120 .122 .130 Figure 10 11 12 13 14 LIST OF FIGURES Monensin. Nigericin A Cyclopolymethylenetetrazole, a Disubstituted Glutarimide and the Cryptand 222 The Carbon-l3 NMR Spectra of Monensin and the Monensin Sodium Salt. The Proton NMR Spectra of Monensin Sodium Salt and Monensin at 100 MHz. . The Amplifier Module of the BLM Resistance Meter Titration of 1.00 mmole Benzoic Acid in 50.00 ml of Solution with 0.0400 M Sodium Methoxide in Methanol . Titration of 1.00 mmole Monensin in 50.00 m1 of Solution with 0.0400 M Sodium Methoxide in Methanol . Titration of 1.00 mmole Monensin in 50.00 ml of Solution with 0.0400 M Tetrabutylammonium Hydroxide in Methanol. . Computer Fit of Figure 7 Data Using KINFIT and Equation(5). Titration of Various Amounts of Monensin with Varying Concentrations of Sodium Perchlorate in Several Initial Volumes of Methanol . Titration of 0.100 mmole Monensin in 50.00 ml of Solution with 0.00500 M Sodium Iodide, Sodium Chloride, Sodium Perchlorate and Potassium Chloride in Methanol Titration of 0.100 mmole Monensin in 50.00 ml of Solution with 0.00500 M Salts of the Alkali Metal Ions and Silver and Ammonium in Methanol. Titration of 0.100 mmole Monensin in 50.00 ml of Solution with 0.00500 M Salts of the Alkaline Earth Ions in Methanol vi Page 13 20 22 ‘28 33 35 37 41 45 49 52 56 List of Figures (Continued) 15 16 17 18 19 20 21 22 23 24 25 26 27 Titration of 5.00 mmoles Monensin in 50.00 mi of Solution with 0.1667 M Sodium Perchlorate in Methanol and Several Infrared Spectra. The Crystalline Structure of Monensin Sodium Salt Variation of Monensin 2025 cm_1 and 2220 cm—1 Infrared Bands with Addition of Sodium Perchlorate in Chloroform-dl . . . . Variation of Monensin 2025 cm.1 and 2220 cm_1 Infrared Bands with Addition of Sodium Perchlorate in Chloroform-d 1. Sodium—23 NMR Spectrum of Monensin Sodium Salt in Saturated Methanol Solution . Sodium—23 NMR Study of 0.200 M Monensin with Increasing Sodium Perchlorate Concentration in Methanol Sodium-23 NMR Study of 0.200 M Monensin with Increasing Sodium Perchlorate Concentration in Methanol . . . . . . . . . . . . . . . . . Sodium-23 NMR Study of 0.250 M and 0.500 M Sodium Perchlorate with Increasing Monensin Concentration in Methanol Sodium-23 NMR Study of 0.250 M and 0.500 M Sodium Perchlorate with Increasing Monensin Concentration in Methanol . . . . . . . . . . . Numbering System for the Protons and Methyl Groups of the Monensin Sodium Salt Illustration of the Effect of the Shift Reagent Eu(fod)3 on the Methyl Region of the 100 MHz Proton NMR Spectrum of the Monensin Sodium Salt in Chloroform-dl Illustration of the Effect of Irradiation at 128 Hz and 150 Hz on the Methyl Region of the 60 MHz Proton NMR Spectrum of the Monensin Sodium Salt in Chloroform-d1 Proton NMR Study of 0.250 M Monensin with Increasing Sodium Perchlorate Concentration in Chloroform-dl . vii 65 67 70 72 73 75 77 80 84 86 92 List of Figures (Continued) 28 29 30 31 32 33 34 35 36 37 38 39 The Crystalline Structure of Monensin . Proton NMR Spectra of the Bound Water Region of 0.250 M Monensin in Chloroform-dl at Various Temperatures Proton NMR Study of 0.250 M Monensin in Tetrahydrofuran—d8, Acetone-d6 and Chloroform—d1. Proton NMR Study of 0.250 M_Monensin with Increasing Sodium Perchlorate Concentration in Chloroform-dl and Acetone—d6 . . . . . . . . . . . . . . Proton NMR Study of 0.250 M Monensin with Increasing Sodium Perchlorate Concentration in Chloroform—d1 and Acetone-d6 Thermogravimetric Analysis of Monensin, the Monensin Sodium Salt and the Monensin Sodium Perchlorate Complex Prepared from Chloroform Solution . Titration of 0.100 gram Monensin Sodium Perchlorate Complex in 50.00 ml of Solution with 0.0046 M Tetrabutylammonium Hydroxide in Methanol Titration of 0.100 gram Monensin Sodium Perchlorate Complex in 50.00 ml of Solution with 0.0046_M Tetrabutylammonium Hydroxide in Methanol . . . . . The Raman Spectrum of the Solid Monensin Sodium Salt . Proton NMR Study of 0.100_M Monensin Sodium Perchlorate Complex with Increasing Monensin Sodium Salt Concentration in Chloroform-d1. Proton NMR Study of 0.100 M Monensin Sodium Perchlorate Complex with Increasing Monensin Sodium Salt Concentration in Chloroform-dl . . . . . . . The Dilactam Form of Cryptand 222 . . . . . . . . . . . viii 96 98 100 102 104 112 115 117 119 - 126 . 128 136 CHAPTER I HISTORICAL INTRODUCTION Studies of the relatively few compounds which form strong complexes with the alkali metal cations have long been of interest to chemists. Recent discoveries that many of these agents possess biological activity have spurred a parallel interest in metal ion complexation on the part of the life scientist. Of particular note have been the compounds capable of affecting the physiological balance of sodium and potassium ions. Many complexing agents of alkali metal cations have been found to be capable of transporting these ions across semipermeable barriers. The phenomenon of selective ion transport has been extensively investigated in order to provide a model for the osmotic properties of biological membranes and to thoroughly ascertain the mechanism of complex formation. The search for new metal ion ligands is carried on by many investigators. Initial characterization and detailed biological testing of these compoundsare relegated to the life scientists. However, analytical chemistry most certainly can play an important role in the careful investigation of the physicochemical properties of these drugs. In the case to be presented here, we shall consider the application of various analytical techniques to the chemistry of a family of several interrelated drugs, with emphasis on the antibiotic monensin. MONENSIN The isolation of the monocarboxylic acid antibiotic monensin, a stereoisomer of 2-[2-ethyloctahydro-3'-methy1-5'-[tetrahydro-6- hydroxy—6-[hydroxymethy11-3,5-dimethyl—ZH-pyran-Z—yl][2,2'-bifuran- 5-y1]l-9-hydroxy-8-methoxy—u,Y,2,8-tetramethyl-1,6—dioxaspiro[4,5] decane-7-butanoic acid (Figure 1), from culture filtrates of Streptomyces cinnamonensis was first reported by Agtarap,_g£_§l. (1) in 1967. A few of the chemical properties of the molecule, then called monensic acid, and its sodium and silver salts were described. Most of the details of the isolation procedure, elaboration on the chemistry and documentation of biological testing were presented at a later date by Haney and Hoehn (2). During this period other papers appeared dealing with maximization of monensin yields by variation of fermentation broths (3) and with investigation of the molecular structure of the compound by proton nuclear magnetic resonance spectroscopy and mass spectrometry (4). Column and thin layer chromatography studies (2,5) showed that monensin consists of four chemical species usually referred to as monensin A, B, C and D. The components were found to be very nearly identical, differing by no more than one -CH2- unit. At the present time the name monensin usually refers to the major factor, A. Thorough biological testing revealed that monensin was an excellent anti— coccidiostat for chickens (2,6). The antibiotic was found to have a dramatic influence on ion transport in mitochondria (7) as well as affecting ion tranSport and photophosphorolation in bacterial chromatophores (8,9). A bioautOgraphic assay for monensin in chick tissues was also developed at that time (10). :Hmcocoz IO .H muswwm Iomro OI After this period of discovery and initial chemical characterization in 1967 and 1968, many workers began to study the chemistry and biological activity of monensin using a wide variety of techniques. Menensin was found to belong to a general group of antibiotics known as the nigericins. Other members of this family of compounds produced by various strains of Streptomyces include dianemycin, X206, and the title compound nigericin. Nigericin had been isolated and characterized previously (ll-13) and as can be seen from Figure 2, its structure is indeed quite similar to that of monensin (14-17). The structure of dianemycin was published in 1971 (18). Nigericin was known to transport alkali metal cations across lipid barriers (l9) and sheep red blood cells (20). It was also found to be useful for coccidiosis control in poultry (21) and shares many of the chemical and biological properties of monensin. In 1969, Pressman and Haynes (22) reported a physicochemical study of monensin, nigericin and the cyclodepsipeptide antibiotic valinomycin as well as several other compounds. Using several analytical techniques they examined the complexing ability of monensin and reported a marked specificity for the sodium ion. Pressman also commented on the pH dependency of conformation and complexation phenomena and suggested that the monensin free acid is linear while the metal ion complex is cyclic (23). Crystallographic studies revealed the crystal structures of the silver, sodium, potassium and thallium salts of monensin (24). The solid complexes were shown to be cyclic with the metal ion coordinated by six oxygens in a nearly octahedral array (see Figure 16). This insulation of the ionic charge and the hydrophobic exterior of the complex results in a high solubility of monensin salts in organic solvents. afloaumwfiz .N muswfim Preparation of the various solid complexes of monensin was found to be dependent on the radius of the metal ion. No lithium, cesium or ammonium complex crystals have been prepared (24). Monensin was shown to be active as an ion transporter in model membrane systems made up of simple water-chloroform barriers (25). A study of the mass spectrometry of the molecule and its factors and salts showed markedly different fragmentation patterns for the free acid and salts (26). Some of the most definitive physicochemical work on monensin has been done by Simon and coworkers in Switzerland. They have conducted infrared studies which have shown the monensin acid and the sodium salt to be cyclic in solution (27). Electrochemical measurements with ion specific electrodes allowed the first estimates of alkali metal ion complex formation constants in basic methanol solutions which were 7.0 X 105 for sodium ion and 9.5 X 104 for potassium ion. Careful observations of model liquid membrane transport showed a monensin selectivity order of Na+>K+$Li+$Rb+$Cs+. Further X-ray crystallographic studies compared the monensin structure specificity relationship to that of related antibiotics in the nigericin family (28). Simon's group (29) also pointed out interesting differences between the structure of solid monensin salt and the acid. Although the two molecules are very similar, subtle changes in the hydrogen bonds, which hold the otherwise planar structure in its cyclic form, cause significant conformational changes on complexation. Other important differences were found in the location and the number of bound water molecules in each solid. Solid monensin free acid holds one water molecule directly in its center’hydrogen bonded at all three atoms (see Figure 28). The salt, or complex form, however was always found to crystallize with two water molecules, now attached at the periphery of the molecule (see Figure 16). These facts led to the first model for a complexation mechanism. The hydrogen bonded water molecule was thought to function as one of the water molecules in the hydration sphere of the metal ion during the initial complexation step. Displacement of the bound water as the H3O+ ion from the center would complete the dehydration and deprotonation necessary to form the complex. In 1971 Simon and coworkers presented the best available values for the thermodynamic parameters of monensin and nigericin interactions with sodium and potassiumions in methanol solution (30,31). Using solutions of the antibiotics which were made basic to ensure complete deprotonation, they reported AH°, AG° and AS° of complexation as determined by a microcalorimetric method. These results led to values of the formation constants for the various complexes from the antibiotic anion and the metal cation. For monensin, the formation constants of the sodium and potassium complexes are l X 106 and 4 X 104 respectively. For nigericin the computed values are 8 X 103 for sodium and 4 X 105 for potassium. In 1971, Haynes, Pressman and Kowalsky (32) reported the use of 3Na nuclear magnetic resonance spectrometry to study the complexation of the sodium ion by monensin, nigericin, valinomycin and several other ionophores. They monitored the 23Na chemical shift and line width as they titrated monensin and nigericin sodium complexes with excess sodium ion. Shifts were found to be a direct linear function of the fraction of complexed cation or a weighted average of the chemical shift of free and complexed forms. By back extrapolation to the 1:1 complex, they were able to report a monensin sodium shift in methanol ranging from 1.5 to 3.7 ppm upfield from a 2.0 M_aqueous sodium chloride reference. The nigericin sodium 1:1 complex shift was found to be between 3.8 to 5.0 ppm. Line widths in excess of 300 to 400 Hz at half-height in chloroform precluded measurements of chemical shifts in that solvent. The 23Na quadrupole coupling constants for the two drugs were also given. Pressman's latest contribution (33) to monensin chemistry has been the value for the acidity constant of the free acid in a 90% ethanol-water solvent mixture which he obtained by titration of the acid with tetramethylammonium hydroxide. Using this method, he has found a pKa of 7.95 for monensin and 8.45 for nigericin. In a series of three papers (34-36), Cussler has shown a new potential use of monensin in possibly nonbiologically related applications (37). By constructing artificial membranes which incorporate monensin, he has devised a scheme in which a pH difference across the barrier can cause the selective transport of sodium ion against its own concentration gradient. An exact theoretical description of the system predicts the observed behavior. He also found that monensin accelerates diffusion of several solutes in mixed solvents (38). Huang (39,40) has developed a detailed kinetic theory to explain antibiotic ion carrier behavior of the nigericin group. Transport properties of monensin have also been recently examined in chloroplasts (41) and in conjunction with calcium accumulation in rat liver mitochondria (42). Sodium efflux in barnacle muscle fibers has also been found to be sensitive to monensin addition (43). Investigation of biological properties has been of increasing interest, particularly to the Eli Lilly Company which now markets monensin sodium as an anticoccidiostat under the trade name Coban. New analytical methods for microbiological assay of monensin in chick rations have been proposed (44-47). Monensin has been investigated as an additive for increasing food utilization by ruminants in sheep (48). Studies have led to better understanding of the anticoccidial activity in chickens (49-51) as well as approval for general use in feeds by the Food and Drug Administration (52,53). Monensin has also been evaluated as a coccidiosis preventative in rabbits (54) and is currently under intensive investigation for use in improving feed efficiency and weight gain in feedlot cattle (55—58). A biosynthetic pathway for monensin has been proposed (59) and various metabolites and by-products have also been examined (60,61). CONCLUSIONS The questions associated with monensin solution chemistry have most certainly generated a deserved amount of interest among chemists and life scientists alike. However, detailed studies of the physical chemical properties of monensin and its complexes in pure solvents are still lacking. BIMOLECULAR LIPID MEMBRANES Even though the first successful attempt to make a bimolecular lipid membrane (hereafter designated as BLM, both singular and plural) was reported in 1962 (62-64), the literature now abounds with thousands of references to a legion of different experiments using this model. Several recent monographs present excellent reviews of both the 10 theoretical and the more important practical aspects of the system and also offer very complete narratives of BLM history (65-67). From an oversimplified point of view, the BLM model consists only of two layers of lipid molecules in a back to back arrangement with their polar head groups exposed on each side to some aqueous solution. This membrane has been found to be an unexpectedly good mimic of many natural membrane systems and measurements of its physical properties have provided meaningful insights into the function of such systems. Of particular interest for our study are the methods by which BLM have been used to examine membrane permeability changes to alkali metal cations. The electrical resistance of a BLM in aqueous electrolyte solution is normally very high due to low solubility of charge carrying ions in the hydrophobic membrane phase. With proper experimental technique, this resistance can be measured and monitored as a function of addition of various compounds to the BLM or to the electrolyte solution. It has been found that a large number of agents are capable of lowering the extremely high BLM resistance to one closer to that of natural systems. The most well known chemical to show this effect is the previously mentioned valinomycin. Valinomycin is a cyclic complexing agent which forms strong complexes with alkali metal cations. It is large enough to "bury" the cationic charge within its interior and thereby increase the solubility of the cation in the lipid phase. Several different types of experiments with valinomycin were designed to study this effect in BLM (68-70). Many other compounds have since been found to complex alkali metal ions and carry them across BLM in a very similar manner. 11 Among the more important agents are the enniatins (71) and the actins (72). The polyether crown compounds of Pederson (73) have also been found to cause resistance changes (74). Although nigericin group antibiotics have been found to increase the permeability of natural membranes (7) and transport ions through bulk organic phases (25), their effect on BLM has remained unclear. Reports have varied from "failure of nigericin to change the electrical resistance of black lipid membranes" (75) to those in which increased conductance is reported (76). Reference (65) contradicts both reports and states that "compounds of the nigericin group, including monensin, do not appear to have been tested on BLM." Several other compounds which could have potential as BLM resistance altering agents do not seem to have been tried in the system. One of these classes is the tetrazoles (77,78), a series of drugs whose well known physiological activity as convulsants has often been investigated in terms of membrane permeability changes (79-81). The tetrazoles (Figure 3) have been shown to possess a variable activity depending on the length and variety of various attached side chains (82,83). The molecules are known to be complexing agents (84) and have been found to form weak complexes in solution with both sodium (85) and lithium ions (86). The glutarimides (Figure 3) share many properties with the tetrazoles (78,87,88) and vary from convulsant to anticonvulsant depending on the nature of the substituent groups R and R2 (89-91). These 1 molecules were also found to form weak alkali metal ion complexes in solution (92,93). Finally, a very new class of strong alkali metal ion complexing agents has been investigated by Lehn and 12 Figure 3. A Cyclopolymethylenetetrazole, A Disubstituted Glutarimide and the Cryptand 222 l4 coworkers (94,95). These compounds, called cryptands (Figure 3), form selective cation complexes based on the size of the coordinating cavity (93,96,97). Neither biological nor membrane activity has as yet been reported for cryptands or cryptate complexes. CONCLUSIONS The BLM model has in the past been shown to be useful as an indicator of possible membrane permeability altering activity. However,thorough studies of the effect of monensin, tetrazoles and glutarimides, and cryptands on BLM ion transport properties do not seem to be available. CHAPTER II EXPERIMENTAL MATERIALS REAGENTS Tetrabutylammonium hydroxide (TBAH) (Eastman Kodak) was used as received as were deuterium oxide (Columbia Organic, 99.5%) and tetramethylsilane (Aldrich). Benzoic acid (Matheson Coleman and Bell) was dried at 90°C for 48 hours. Sodium perchlorate (G. F. Smith), sodium methoxide (Matheson Coleman and Bell), sodium chloride (Matheson Coleman and Bell), sodium iodide (Matheson Coleman and Bell),potassium chloride (Matheson Coleman and Bell), rubidium iodide (Alfa), cesium iodide (Alfa) and ammonium perchlorate (G. F. Smith) were dried at 110°C for at least 72 hours. Lithium perchlorate (Fisher) was dried at 190°C for several days. Beryllium sulfate (Eberbach), magnesium perchlorate (Allied), calcium chloride (Fisher), strontium perchlorate (K and K), barium perchlorate (G. F. Smith) and silver perchlorate (G. F. Smith) were also dried at 110°C for more than three days. The shift reagent Eu(fod)3 (Alfa) was used as received as were thallium bromide (Alfa), thallium acetate (Alfa), thallium nitrate (Alfa) and tetraphenylarsonium chloride (G. F. Smith). Phosphatidyl choline (egg lecithin, Applied Science) was received in the form of an ethanol solution. The solvent was evaporated at room temperature and the resulting solid was dissolved at about 1% in n-octane (Aldrich) which had been previously deoxygenated by bubbling nitrogen through it for 2 hours. The lecithin solution was stored at —5°C. Cholesterol (Sigma) was recrystallized from 15 16 ethanol. The product was suspended in n-octane and refluxed at 126°C for 6 hours during which time oxygen was constantly bubbled into the solution. The resulting solution, which appeared to be saturated with oxidized cholesterol, was also stored at -5°C. Valinomycin (Calbiochem), dicyclohexyl—lS-crown-6 (E. I. DuPont), 4,7,13,16,21,24- hexaoxy-l,lO—diazabicyclol8.8.8]hexacosane (cryptand 222) (E. M. Laboratories), pentamethylene tetrazole (Aldrich), glutarimide (Eastman Kodak), 3,3-dimethylglutarimide (Aldrich) and 3-ethy1- 3-methylglutarimide (Pfaltz and Bauer) were used as received. Several other tetrazole compounds and the cryptand dilactam synthetic precursor were obtained from syntheses in progress in our laboratories. The proteins protamine (Sigma) and histone (Aldrich) were used as received and stored at -5°C. SOLVENTS Methanol (Baker-analyzed reagent) was refluxed over granulated calcium hydride (Baker) for 24 hours and fractionally distilled. Water content was found by Karl Fischer titration to be less than 70 ppm. Chloroform-d1 (Norell, 99.8%), tetrahydrofuran-d 8 (Norell, 99%), acetone-d6 (Diaprep-Aldrich, 99.5%) and methanol-d 4 (Diaprep-Aldrich, 99.5%) were used as received. Many solutions involving the use of these solvents were prepared in the dry box. Chloroform (Mallinckrodt), acetone (Mallinckrodt), methanol (Matheson Coleman and Bell), ethanol (Baker), 1,2-dichloroethane (Fisher), benzene (Baker-analyzed reagent), ethyl ether (Mallinckrodt), hydrochloric acid (Baker-analyzed reagent) and petroleum ether needed in various drug purification and preparation steps were used as received. l7 MONENSIN Monensin was received as the sodium salt (QA 166H Lot Numbers 910 AD3 and 024 FC4, hereafter MonNa) through the generous gift of the Eli Lilly Company. The salt was dissolved to saturation in boiling methanol and the solution filtered. The filtrate was cooled and the salt reprecipitated upon addition of a small quantity of water. This procedure was repeated twice and eliminated most of the brown impurity first noticed as a straw yellow color in solution. The salt was further purified by recrystallization from a 1:1 ether-petroleum ether mixture and dried at 110°C for 48 hours. The acid form of the molecule (hereafter designated as MonH) was prepared as follows: A concentrated solution of the salt was prepared in chloroform and shaken with an equal volume of aqueous 0.1 M hydrochloric acid and the phases separated. The chloroform phase was evaporated to dryness at room temperature. The yellow solid product was then dissolved in acetone and reprecipitated with water. The now white solid was dried for 24 hours at 30°C under vacuum, then dissolved in boiling methanol. The solution was filtered and MonH was reprecipitated from the cooled filtrate by the addition of water. After drying, the acid was recrystallized from a 1:1 ether-petroleum ether mixture and again dried under vacuum for 24 hours at 30°C. Elemental analysis of the sodium salt and acid forms gave the following results: Calculated for C Na, C, 62.70%; 36H61011 H, 8.87%. Found, C, 61.93%; H, 8.88%. Calculated for C36H62011'H20, C, 62.77%; H, 9.36%. Found, C, 62.63%; H, 9.28%. 18 Melting points of 269°C and 117°C for the salt and acid respectively agreed with reported values of 267-269°C (2) and 117-122°C (29). Mass spectra of the two solids showed fragmentation patterns similar to those previously observed (26). Infrared spectra, taken in chloroform solutions, agreed with the available literature spectra (2) and showed a diagnostic shift of the carbonyl band from 1563 cm—1 to 1704 cm.1 in the salt to acid transformation. Carbon—l3 NMR spectra, shown in Figure 4, revealed a dramatic variation in carbon environments and were especially useful in showing a change from 188.2 ppm to 177.8 ppm in the carbonyl chemical shift, which is the characteristic difference between a salt and free carboxylic acid. Tentative assignments of all of the 36 resolved carbon signals were made to the various carbon atoms in each molecule. Proton NMR spectra at ambient probe temperature in chloroform-d1 (Figure 5) also showed many differences between the acid and the sodium salt, the most obvious one being the appearance of a previously unreported broad water peak at approximately 6 ppm in the spectrum of the acid which is totally absent in the spectrum of the salt. On the basis of these measurements, it was concluded that the pure free acid had been prepared. No attempt was made to separate the four related factors of either the sodium salt or the acid. The acid was found to be stable in the solid form but quite susceptible to degradation in solutions. 19 Figure 4. The Carbon-l3 NMR Spectra of Monensin and the Monensin Sodium Salt 20 ms; geode— _Jl_ I omou 5% $5.08 z_m2mzo_2 2.: ask— moo Z_meZO.2 21 Figure 5. The Proton NMR Spectra of Monensin Sodium Salt and Monensin at 100 MHz 22 252ng i % .5620: $5.08 Z.meZO_2 23 MEASUREMENTS POTENTIOMETRIC Potentiometric measurements were made using an air tight cell constructed to accomodate a Beckman 41263 glass pH electrode, a Sargent-Welch S-30080-150 saturated calomel reference electrode and an extended tip of a 50 m1 burette. The electrodes were initially soaked in methanol for 48 hours and at least for 5 hours between runs. Even this precaution still did not yield potentials whose absolute values were completely reproducible, but this fact has been accounted for in data interpretation where only changes in potential are used. In some experiments somewhat more stable potentials were obtained by replacing the aqueous saturated potassium chloride solution in the SCE by a saturated methanolic solution of this salt. Measurements were taken at ambient temperature with a Heath EU-302A servo-digital pH/voltmeter allowing sufficient time between readings for the stabilization of the potential. A constant pressure of nitrogen was maintained in the cell at all times and the magnetic stirrer used for mixing was stopped during measurements. Voltage drift and reading error combined to give potential readings with an uncertainty of‘: 0.002 V and volumes were read with an uncertainty of :_0.02 ml. A CDC-6500 computer was used extensively for data processing in conjunction with a revised version of a program initially developed by Briggs and Steuhr (98) for the determination of equivalence points and pKa or the FORTRAN IV KINFIT program (99). More detail on computer data analysis is presented in the appendices. 24 SPECTROSCOPIC Perkin Elmer 237B and 457 grating infrared spectrophotometers were used to obtain infrared spectra. A Barnes Engineering fixed pathlength (0.109 mm) potassium bromide cell was used for all infrared studies. The frequencies were calibrated using polystyrene reference peaks. Raman spectra were obtained on the Spex Ramalog 4 Laser- Raman system employing either a Spectra—Physics Model 164 Argon—Ion Laser or the Coherent Radiation Model 52 Ion Laser. The 5145 A green line was used for excitation with the third monochromator removed and the photomultiplier operating in the pulse counting mode. Samples were either packed or injected into 1.6-1.8 X 90 mm melting point capillary tubes and sealed. Lithium-7 NMR spectra were obtained on a Varian Associates DA—60 spectrometer operating in the continuous wave mode at 23.3 MHz with a field of 14.09 kG. The instrument was frequency locked to a 4.0 M_aqueous solution of lithium perchlorate held by Teflon spacers in a 1 mm o.d. melting point capillary positioned in a Wilmad 506 pp 5 mm o.d. polished NMR tube. Chemical shifts were measured at approximately 30°C with respect to lock with a Hewlett Packard 5245L frequency counter. Sodium-23 NMR spectra were obtained at ambient probe temperature on a highly modified Nuclear Magnetic Resonance Specialties MP-lOOO superconducting solenoid spectrometer operating at 60 MHz with a field of 53.3 kG. A Nicolet 1083 computer was used for time averaging and to drive the frequency synthesizer. Chemical shifts were measured with respect to a 3.0 M aqueous solution of sodium 25 chloride in a Wilmad 520-2 coaxial reference tube inserted in either a Wilmad 506 pp 5 mm o.d. or a 513A—5 pp 8 mm o.d. polished NMR tube. Uncertainties in chemical shift values were estimated individually for each spectrum and varied with sweep width and signal line width. Thallium-205 NMR spectra were run on a modified Varian Associates HA-lOO NMR spectrometer operating at 57.7 MHz with a field of 23.4 kG. Samples were run in the Wilmad 506 pp tube. Carbon-13 NMR spectra were obtained using the Bruker HFX—lO spectrometer operating at 22.6 MHz with a field of 21.1 kG. The instrument operated in the normal Fourier transform mode under control of Nicolet 1083 and 290 computers with complete proton decoupling at 90 MHz and an 84 MHz hexafluorobenzene internal lock and TMS internal reference. Samples were run in Wilmad 513-3 pp 10 mm o.d. polished NMR tubes. Proton NMR spectra were run on three instruments. The Varian T-60 spectrometer was used in the normal mode and also for some decoupling experiments. The Varian A56/60D spectrometer with temperature controller was also used with calibrated chart paper and sideband calibrated sweep widths. The Varian HA-lOO spectrometer was used for precision work with either a tetramethyl- silane (TMS) or chloroform frequency lock. All chemical shifts were ultimately referenced to TMS and are given in ppm downfield. Mass spectra were run on the Hitachi Perkin Elmer RMU-6 spectrometer. W“: 1...!“ 26 OTHER ANALYSES Elemental analyses were done by Chemalytics Incorporated of Tempe, Arizona. Melting points were determined on the Fisher-Johns melting point apparatus. Analysis for water was accomplished with a Photovolt Aquatest II automatic Karl Fischer titration apparatus. MEMBRANE RESISTANCE All BLM experiments were performed inside of a Faraday cage which consisted of a wooden frame covered by a double layer of copper screening in electrical contact and then covered with light shielding black cloth. Inside, a lattice of Fischer Flexframe and support stands held steady by lead weightswere used for the attachment of four pieces of apparatus. A monocular zoom observation microscope (Model ZMM-l, Titan Tool Supply) was mounted at the front of the lattice to be used for inspection of BLM thinning. A spotlight illuminator (Model LS, Unitron Instrument) was used to direct a reflected light from the BLM at an appropriate angle to the microscope. In the center of the arrangement, a sturdy support held the cell in which BLM were formed and above this was the amplifier system to which electrodes were attached. The BLM were formed on a precision drilled 1 mm diameter hole in a 10 ml Teflon cup (Chemplast). This cup stood in the center of a quartz cell built by Mr. Keki Mistry of the department's glass shop. The entire volume of the cup and quartz cell, when filled with electrolyte solution, was about 25 ml. Addition and subtraction of material from either side of the membrane after formation was accomplished using a 25 ul repeating sampling syringe (Oxford). Membranes were initiated by delivering lipid 27 solutions with a 50 ul fixed needle syringe (Hamilton) and repeating dispenser (Hamilton) by the well known brush technique. This entire arrangement is quite similar to the common description of the BLM set-up in the literature (65). Electrodes, inserted on either side of the Teflon cup, were of the fiber junction reference type (Beckman 39270) and were always left soaking in 0.1_M electrolyte when not in use. The two electrode connections led directly to the instrument designed by Mr. Martin Rabb of the department's electronics staff for the measurement of membrane resistance. The instrument consists of two parts: the amplifier, and the power supply, meter and controls. The amplifier module was used inside of the cage in close proximity to the BLM and housed the circuit shown in Figure 6. Basically, this circuit consists of an inverting amplifier which produces an output voltage which is the product of the input voltage and the ratio of the feedback resistor and the input resistor. The left third of the circuit serves as a voltage divider to provide an accurate stable reference input voltage. The 2000 variable resistor allows for setting a required 7.5 ma current through the reference zener diode at the measure link. The 1 KO variable resistor in the divider chain allows accurate setting of a 1.00 V input at point A. The divider and array of input resistors cause the output of the first operational amplifier to be -l.00 V for decade electrode resistances from 104 to 1090. The operational amplifier also connects to a 1 MO resistor to help reduce bias current and a trim pot to help 9.50th noun: mocmumfimom sum osu mo oasmoz poawaaea< one .0 muswfim » LOfiOE 0* 240.30” 28 .11)))1|I. x23: bmm. go. \ mm 33.52.. v. oi . , o an." m . [$1128. \ .zmcaou I. 2. IJ))\II. + + 22 e >2 t SH V >ouf I 4 3;- Zuni 13.31 .2. .l + rllz2)1lll4 on .2. So '5? < 50> xo. $2.9. x. goo. \ >.n . 2; go. 4— Lp r(((|.l an...“ 2. go: III&(3. lJ\\ oouv _ 2. mm .5531 H e :1. “01980.0 > WI. ._._‘ -~-".. r.-' 29 compensate for offset voltage. The 5.1 V zener diode, 1 K0 resistor and switching diode limit the output voltage of the operational amplifier to 5.1 V if the electrode terminals are open circuited. The 51 pF capacitor reduces the amplifier bandwidth for improved stability. The second operational amplifier acts as an inverter with a gain of 1. Voltage is limited to 2.4 V to protect the 1.00 V full scale meter connected to the output. The bottom 10 KO input resistor and the 1 MO resistor of the main operational amplifier feedback loop are switched into the circuit to allow checking and setting of the amplifier to zero offset. Cables led out of the cage to the power supply module which also contained the meter display for direct reading of membrane resistance along with the switches for changing input resistors. The instrument could measure resistances to a‘: 5% tolerance in the range of 104 to 109 ohms. This was a far greater precision than necessary considering the experimentally observed reproducibility of BLM resistances. Further details on this device are found in reference 100. CHAPTER III POTENTIOMETRIC STUDIES OF MONENSIN xii-fl INTRODUCTION Potentiometric measurements on monensin solutions were initiated primarily to study quantitatively the three equilibria shown below: + .- + - +. = [H liflon ] (l) MonH + Mon + H , Ka [MonH] ‘(2) Mon- + M+ Z MonM; Kf = fig]:— [Mon ][M l + t 1_u____1 (3) MonH + M I MonM + H+; K = H MonM [MonH][M 1 where (l) is the acid dissociation of monensin, (2) the formation of the monensin metal ion complex and, (3) the overall metathesis reaction of the acid molecule complexing a metal ion and releasing a proton. In this simplified treatment, dehydration of the monensin molecule or the participation of the metal salt anion in the above equilibria are neglected. Previous studies gave MonH acidity constant values only in mixed solvent systems (2,33). Estimates of Kf had been made by studying equilibrium (2) alone by using excessively basic solutions which ensured complete acid deprotonation (27,30,31). Equilibrium (3) has never been studied and indeed monensin had been assumed not to be a complexing agent in neutral solutions. It should be noted that if the above equilibrium scheme is correct, the 30 31 equilibrium constant for complexation of metal ions by the free acid should be equal to the product KaKf' Neither the sodium salt nor the acid form of monensin are appreciably soluble in water, consequently all studies were carried out in nonaqueous solvent systems. Methanol was chosen as the primary solvent for the study because of previous work done with the drug in this solvent and also due to its suitability for potentiometric measurements. DETERMINATION OF ACID DISSOCIATION CONSTANT Before attempting a titration of the MonH molecule with a strong base, a standardization titration was designed using benzoic acid (hereafter, HBz) which has a Ka value somewhere near that anticipated for MonH. The titration curve of 1.00 mmole of HBz in 50.0 ml of solution with 0.0400 M sodium methoxide in methanol is shown in Figure 7. A smooth sigmoidal curve with a distinct equivalence point was observed. A similar titration of 1.00 mmole of MonH with 0.0400 M sodium methoxide gave the curve shown in Figure 8. It can be seen that the results indicate that MonH is a much stronger acid than HBz. This would seem to be incorrect and indeed the MonH result is fortuitious. The effect of the sodium ion in solution has been to introduce another reaction causing complexation of sodium ion by Mon— and a consequent increase in the acid dissociation constant. For this reason, sodium methoxide or any other base with a complexable cation cannot be used to study only the acid-base reaction. Figure 9 shows a similar titration of 1.00 mmole of MonH with 0.0400 M_tetrabutylammonium hydroxide. In this case, the cation 32 Figure 7. Titration of 1.00 mmole Benzoic Acid in 50.00 ml of Solution with 0.0400 M_Sodium Methoxide in Methanol 33 35: m2340> code 8.8 8.8 8.9 o — 1 u q q l § (A) '3'3'9 'SA 'IVIlNEIlOd \ -- a... é 34 Figure 8. Titration of 1.00 mmole Monensin in 50.00 ml of Solution with 0.0400 M Sodium Methoxide in Methanol 35 code .r is: 8mm ms=jo> 8.8 8.9 q — d 0 m ICON... N H V __I llOOfl... A .S .8 Iloofi .0 1. w i 36 Figure 9. Titration of 1.00 mmole Monensin in 50.00 m1 of Solution with 0.0400 M Tetrabutylammonium Hydroxide in Methanol 37 35: m2340> 00.8 00.8 00.8 00.8 00.9. 1 1 1 1 1 o o 00000 ooooooo .00 o o o o o o o o oet‘ o oo o o ,1 § l I :82. (A) '3'3'3 'SA 'IVIlNaiOd l 8 <1; P 1... com... 38 is too large to fit into the monensin cavity. The new results, as expected,indicate that MonH is a weaker acid than HBz. Several such titrations at different concentrations were performed to confirm the reproducibility of the data. Three different methods were used in attempting to calculate a Ka value for MonH using the data from the above experiment. The first is the most well known and involves the use of the equation given below. Neglecting autoprotolysis and activity corrections, Meites and Thomas (101) show that the value of the acidity constant can be obtained from the titration curve with the expression Va + + fca v + Vb + [H ] (4) K =[H1 a a Va + (l-f)°a v + v ' [H 1 a b where C8 is the analytical acid concentration, Va is the initial volume of acidic solution, Vb is the volume of base titrated into is the base concentration and f = V C /V C , the b b b a a fraction of the equivalent volume of base added. Potentials observed solution, C far past the equivalence point were assumed to arise only from excess base in solution thus allowing the establishment of a voltage to pH relationship. The electrodes were assumed to have a‘ Nernstian response and calculations gave pH values in the titration buffer region. However, use of these values in equation (4) gave values of the pKa for H82 in the area of 8.9. These were in poor agreement with the literature value of 9.27 (102). 39 A second approach involved the use of all of the titration data rather than just those in the buffer region. The exact equation for the titration of a weak acid with a strong base with uncomplexable cation is the cubic given below. (5) [n+13 + ((3b + Ka)[H+]2 + (KaCb - KaCa - KS)[H+] — KaKS = o The symbols are defined as before with K5 the autoprotolysis constant. By putting the analytical solution for a cubic equation into the KINFIT program subroutine (detailed in Appendix I), it was possible to fit the observed data to the above equation and obtain a Ka value. In this analysis, a Nernstian electrode response was again assumed, but the intercept of a Nernstian pH versus voltage plot was programmed as an adjustable parameter to give the best fit to the data. The technique proved to be extremely sensitive to the initial estimate of the adjustable Ka parameter, often causing convergence premature to reasonable data fits. After many trials, the best fit obtained using this approach is shown for HBz in Figure 10 and indicates a pKa of about 7.4 which is unacceptably low. This method was therefore discarded. Fortunately, a third method was successful in reproducing the literature pKa for HBz. After investigating two other recent applications of computer analysis of multiparametric titration curve fitting (103,104), a useful method developed by Briggs and Steuhr (98) was found. These authors derived a linear equation based on exact mole balance relationships which simultaneously determines pKa and equivalence point values and also takes into consideration the presence of any excess of strong acid or base in the titrated 40 Figure 10. Computer Fit of Figure 7 Data Using KINFIT and Equation (5) 41 is: m2340> 8.3 8.8 8.8 8.9 o 1 1 l 1 8... d m m 3330.8 4 u . a... :83... notomno . on“! 1 $1 A .1 8a.. .3 O O O O 0% II 8.1—. T.— < ) 4 A ‘ ‘ l.\ 42 sample. They present equation (6) 1 (6) V + R - AV - -aH+(V + R - AV) Ka + Ve where V is the volume of base added, AV is the volume of this base neutralized by any strong acid present, R = (CH+ - COH_)Vt/Cb, Vt is the total volume, Ve the equivalence volume and aH+ the hydrogen ion activity. The equation was originally used in a least squares fit program which also allowed for aqueous activity corrections. A copy of this program was generously supplied by J. E. Steuhr and modifications necessary for the HBz or MonH system in methanol were made as detailed in Appendix 11. Using this modified Steuhr equation a pKa of 9.25 i 0.05 for HBz in methanol was calculated. This is in excellent agreement with the literature value and has been corrected for a small amount of basic impurity in the methanol. The same technique applied to MonH gave a pKa value of 10.15 110.05. The only previously determined values were 6.65 in 66% dimethylformamide-water (2) and 7.95 in 90% ethanol-water (33) mixtures. These lower values would be expected in solvent mixtures of higher dielectric constant . CONCLUSIONS The results indicate that MonH is a relatively weak acid in methanol and in neutral solution it is mainly in the associated form. Consequently the Mon- anion is a relatively strong base in methanol solutions. 43 DETERMINATION OF THE OVERALL COMPLEXATION CONSTANT It was thought that perhaps equilibrium (3) could be monitored simply by adding some alkali metal cation to a solution of MonH and measuring the concentration of released hydrogen ions. Various initial concentrations of MonH were titrated with various concentrations of the salt sodium perchlorate in methanol solution. Several of the titration curves obtained are shown in Figure 11. As can be observed, a gradual release of protons is indicated. Calculations reveal that the concentration of released hydrogen ions is nearly what would be expected for a 1:1 replacement of a proton of monensin with a sodium ion. It was desired to treat these data to give an estimate of the equilibrium constant for the case of the sodium ion in equilibrium (3). Since the K8 for MonH is very small, it was assumed that the concentration of Mon- ion is negligible. With this assumption and neglecting activities and solvent autoprotolysis, we get _ [MonNal[H+] _ [H+]2 K " + ' + + [MonHllNa ] (Ca - [H l)(CNa+ - [H ]) (7) where CNa+ is the analytical metal ion concentration. The quadratic solution to the hydrogen ion concentration was inserted in the KINFIT program and the Nernstian pH versus voltage intercept was again adjusted as well as was the value of K. Details of this program are shown in Appendix III. The values of K obtained for various sodium perchlorate titrations are shown in Table l. Intercept values were found to vary slightly depending on the time 44 Figure 11. Titrations of Various Amounts of Monensin with Varying Concentrations of Sodium Perchlorate in Several Initial Volumes of Methanol O , 0.0975 mmoles, 0.00502 M, 50.00 ml I , 0.195 mmoles, 0.0100 M, 70.00 ml . , 0.488 mmoles, 0.0400 M, 50.00 ml 0 , 0.975 mmoles, 0.0400 M, 50.00 ml Solid lines are the computer fitted data. 45 is: £238 8.8 8.8 88 8o. o e _ _ _ _ ll .L n .. I 1 HM 1.88 I Al .. m . I. q o o o > u 2.. c c . . . .8o..m ... w ... .I a A a .s 8... . .S __ .3 3 .. 8...“. 46 Table 1. Equilibrium Constant for the Reaction MonH + Na+': MonNa + H+ in Methanol Solution at 25°C MonH (mmole) NaClO4 (M) K 0.0977 0.00100 0.96 i 0.29 0.0975 0.00502 0.22 i 0.07 0.195 0.0100 0.79 i 0.04 0.488 ' 0.0400 0.53 i 0.11 0.975 0.0400 0.17 i 0.07 1.50 0.0411 0.27." 0.09 2.00 0.0411 0.50 i 0.31 5.004 0.1667 0.18 i 0.14 47 alotted between various titrations. The solid lines shown in Figure 11 are computer fitted data. As can be seen by examining Table l, the K for monensin acid complexation of sodium ion is very small compared to that of the anion. Previous studies of MonH complexation ability had always been done by first adding excess base giving Mon-—sodium ion formation constants of about 106 in methanol. Using the previously determined Ka and the calculated K values, it should be possible to estimate K from the K/Ka ratio. These calculations, however, give K f f values of about 109 whichrnxaconsiderably higher than the reported values. This fact gave the first indication of the possibility that the complex formed in neutral methanol solution is not identical to that obtained under basic conditions. Shown in Figure 12 are the results of a similar experiment with sodium iodide, sodium chloride, sodium perchlorate and potassium chloride. Also shown are the results of titration of a methanol blank into a MonH solution which, as expected, gives no potential change other than that due to dilution effects. An identical computer analysis of these data gives the K values tabulated in Table 2. As seen, variation of K with sodium counterion is not significantly outside of the experimental error of K for the titrations shown in Table 1. Potassium ion appears to form a complex quite similar in strength to that of sodium. Further experiments showed that monensin is capable of complexing all of the alkali metal ions. Figure 13 illustrates the results obtained in titrating MonH with lithium, sodium, potassium, 48 Figure 12. Titration of 0.100 mmole Monensin in 50.00 m1 of Solution with 0.00500 M Sodium Iodide, Sodium Chloride, Sodium Perchlorate and Potassium Chloride in Methanol 49 :5: m2040> 00.0w 00.00 00.0w 00.0w 0 v _. w “ o8.. .002 «H A! Jul .mZ 4r _0¥ «w an... H U u n m n stilODOF a . o: 1 a .. or; u; 00:. ll 09:. C) C) C) C) O 20.10 t (A) 'H'O'S 'SA ‘lVllNElOd 50 Table 2. Equilibrium Constants for the Reaction MonH in Methanol Solution at 25°C MonH (mmole) 0.0980 0.0980 0.0975 0.0980 Titrant (M) 0.00502 NaI 0.00507 NaCl 0.00502 NaClO4 0.00499 KCl 0.71 0.48 0.22 0.42 + 1+ I+ I+ |+ + + M Z MonM + H 0.11 0.09 0.07 0.04 51 Figure 13. Titration of 0.100 mmole Monensin in 50.00 ml of Solution with 0.00500 M Salts of the Alkali Metal Ions and Silver and Ammonium in Methanol 52 00.0.. 3.): 92340) 88 .008 I r L‘I w .6002 T 00.8 ‘+ ll. A Ox w 3m T .0 > ‘oofz W 10.0: 00.0— 08. Iowa— (A) 308 'SA 'IVIlNBlOd 53 rubidium and cesium as well as with ammonium and silver. Anion variation was necessary due to solubility requirements. It is seen that a definite selectivity scheme has emerged. Unfortunately, the use of equation (7) and the KINFIT program does not give the equilibrium constant for lithium, rubidium, cesium, ammonium or silver ions. Although relatively small values of total residuals are calculated,convergence of the program.at a best value of K and the intercept does not occur. However, if a more exact equation is used,such as +3 +2 (8) (K - 1)[H ] + (KaK - Ka - KCa - KCM+)[H ] + (KaCa - 2KKaCa + KCaCM+) [11+] + KaKCaZ = 0 which adds the consideration of the Mon- concentration and therefore uses the constant Ka, estimates of K are obtained. Moreover, it is found that using this cubic equation as is also shown in Appendix III, all of the sodium ion data can be reinterpreted and give values of the equilibrium constant identical to those calculated previously with equation (7). In Table 3 are the values of K for the alkali metal ions and silver and ammonium. The values indicate a selectivity sequence of Na+WK+>Ag+>Rb+>NH +>Li+tCs+ which is in some agreement 4 with results obtained in basic solutions and correlates extremely well ‘with ionic size parameters with lithium ion too small and cesium or ammonium ions too large for coordination in the monensin cavity. Another extension of the original experiment is obtained by titrating MonH with salts of alkaline earth ions. These titrations are shown in Figure 14 and are the first indication ever obtained 54 + + Table 3. Equilibrium Constants for the Reaction MonH + M : MonM + H in Methanol Solution at 25°C Titration of 0.100 mmole MonH with 0.00500 M Salts Salt K 1.10104 (9.3 i 1.2) x 10‘7 NaClOa (2.2 i 0.7) x 10‘1 KCl (4.1 i 0.4) x 10'1 RbI (3.3 i 0.8) x 10'6 CSI (3.0 i 0.3) x 10'7 10140104 (1.5 i 0.5) x 10'6 -2 AgClO4 (6.6 i 8.4) x 10 g. 55 Figure 14. Titration of 0.100 mmole Monensin in 50.00 ml of Solution with 0.00500 M Salts of the Alkaline Earth Ions in Methanol :5: m2340> 00. 0? 00.00 00.0w 00.0.. 0 . -|_| lrllllol 1 N...o_ 0.3 .08.. i .8615 T 18.. d O I. 3 N u o8.. V .I A S .1 8... New a w om: 57 of MonH complexation of ions of the second group. Interpretation of these data has not as yet been successful since divalent cations will not fit into the previous equilibrium scheme. Future work on the complexation mechanism involved is expected to yield alkaline earth K values. CONCLUSIONS The monensin acid molecule has been found to form only a weak complex with sodium ion in methanol with an equilibrium constant slightly less than 1. Values of Kf calculated from K/Ka ratios do not agree with previously published values. Although the complex is only a weak one, the acid in methanol solution still exhibits a selective complexation scheme both for alkali metal and alkaline earth cations as well as silver and ammonium. Calculations of K values for alkali metal ion complexes show a good correlation with ionic size. INFRARED STUDY OF COMPLEXOMETRIC EQUILIBRIA Since the titrations described above had indicated varying degrees of proton release, it was desired to monitor this deprotonation in the infrared spectrum. Monensin has an easily observable carboxylic acid band at 1704 cm"1 in solution while the normal sodium salt has a carboxylate band at 1563 cm-1. It was found that in chloroform solutions, the 1704 cm.1 band could easily be shifted to the salt position by an addition of ammonium hydroxide. Conversely, the 1563 cm.1 salt band can be caused to move to 1704 cm.1 by an addition of hydrochloric acid to the chloroform solutions. Shown in Figure 15 is a titration curve of MonH with sodium perchlorate in methanol and |-- Figure 15. 58 Titration of 5.00 mmoles Monensin in 50.00 ml of Solution with 0.1667 M_Sodium Perchlorate in Methanol and Several Infrared Spectra 59 ..:2. m230> q— 00.0¢ . 00._00 00._0N . ED 009 000.. FI o.d...mm ._ _. 'UUJIO ‘0- .. .i o8.o : + § E -= o (A) 'a'os 'SA ‘IVIiNEliOd 60 parallel infrared spectra taken at various points along the course of the experiment. As is illustrated, no change in the position or intensity of the 1704 cm-1 band is noted. Since this result was somewhat unexpected and much of the region of interest was obscured, further experiments were carried out in a better infrared solvent. It was found that MonH is capable of solubilizing sodium perchlorate in chloroform up to 1:1 mole ratios. The 1704 cm“1 MonH band was monitored in chloroform—dl solutions with increasing sodium perchlorate concentration. Again, even at 1:1 sodium ion to MonH mole ratios, no change in the intensity or position of the 1704 cm"1 band was observed nor did any trace of the 1563 cm_1 band appear. It was concluded, therefore, that if the carboxylic proton_i§ lost by the acid molecule, some other proton from the hydrogen-bonded pattern near this area takes its place (see Figure 16). This is to say that in the addition of a sodium ion to monensin in neutral solution, some other proton must be more acidic than the one on the carboxylic group. This more acidic hydrogen ion must belong to one of the several hydroxyl groups. Consequently, the complex may have a lower energy state in which the negatively charged oxygen atom is somehow closer to the positive alkali metal ion than the carboxyl oxygen. The resultant zwitterionic structure may reflect a charge stabilized complex, which would give the molecule .an even more hydrophobic exterior. In any case, this behavior is a second indication that the acid-sodium complex formed in neutral solution cannot be of the same structure as the ordinary salt. 61 Figure 16. The Crystalline Structure of Monensin Sodium Salt Closed circles are carbon atoms, open circles are oxygens with the water oxygens dotted and dashed lines indicate hydrogen bonds. 62 63 Other areas of the spectrum monitored during the sodium perchlorate addition in chloroform-d gave further evidence of a l complexation mechanism different from the one in basic solution. In the O—H stretching region, MonH showed a broad band at about 3300 cm.1 and a sharper band on its shoulder at 3520 cm-1. As the sodium concentration was increased, the sharper band gradually disappeared and the 3300 cm.1 band increased slightly in intensity. Again,the 1:1 spectrum bore little resemblance to the MonNa infrared spectrum in chloroform-d . These spectral changes must result from 1 the removal of the water molecule from the center of the monensin ring. Also noted in the above experiment was the gradual appearance of two new bands at 2025 cm.1 and 2220 cm”1 which were also present in the spectrum of the pure MonNa salt. Their growth (Figures 17 and 18) was such that at 1:1 sodium perchlorate to MonH mole ratios, this area of the spectrum was identical to the spectrum of the salt in chloroform-d1. CONCLUSIONS The carboxylic acid proton on monensin, which is easily removed in basic solution, is not the proton released by sodium complexation in neutral methanol or chloroform solutions. Other information frxnn the infrared spectrum indicates that this new type of complex is only somewhat similar in structure to the salt. This fact explains some of the discrepancy between the apparent Kf value obtained :hl neutral solution and the values previously determined in basic solution. CHAPTER IV NUCLEAR MAGNETIC RESONANCE STUDIES OF MONENSIN 64 Figure 17. Variation of Monensin 2025 cm.1 and 2220 cm-1 Infrared Bands with Addition of Sodium Perchlorate in Chloroform-d1 65 —1 0.040. 2025 cm p o 73 l -1 2220 cm A BSORBANC E a 8 0.600 —- 0 0.400 0.800 1.200 MOLE RATIO NaCIo4 TO MONH 66 Figure 18. Variation of Monensin 2025 cm.1 and 2220 cm.1 Infrared Bands with Addition of Sodium Perchlorate in Chloroform-d1 67 -1 2025 cm -1 2220 cm 1 l L . . s 1.00 3.00 5.00 7.00 9.00 MOLE RATIO MONH TO NaClO, INTRODUCTION Nuclear magnetic resonance (NMR) has been proven to be a very powerful tool for the examination of various types of drug interactions. Alkali metal ion NMR has already been mentioned in connection with the investigation of the monensin sodium complex (32) and both 7L1 and 23Na NMR have been used for studying tetrazole (86), glutarimide (92), crown (105) and cryptand (93,97) complexation. Proton magnetic resonance (PMR) has been of tremendous use in the study of conformational changes which take place in drug molecules during complexation in solution. The actins have been particularly well studied in this regard by Prestegard and Chan (106,107). The technique has also been used extensively in the study of valinomycin solution chemistry (108-114). Data obtained in the potentiometric experiments indicate that alkali ion NMR will be useful for MonH complexation studies. The many differences observed between the MonH and MonNa PMR spectra displayed in Figure 5 show that a detailed PMR study will also be of interest. LITHIUM-7 NUCLEAR MAGNETIC RESONANCE STUDIES Solutions of lithium perchlorate were made 0.02 M_in methanol with concentrations of MonH ranging from 0.02 to 0.20 M, Even with the drug to lithium ion mole ratio of 10 to l, a shift of only about 2 Hz from the normal resonance of uncomplexed lithium ion at 68 69 0.02 M concentration was observed. This value, while experimentally significant, indicates an exceedingly weak interaction unless the chemica1.shiftsof solvated and complexed lithium ion are quite similar. CONCLUSIONS This result correlates with the data obtained in the potentiometric titrations and does confirm the evidence which shows that lithium ion may be only weakly complexed by MonH, probably because of ionic size considerations. SODIUM-23 NUCLEAR MAGNETIC RESONANCE STUDIES Preliminary measurements were made on a saturated solution of MonNa in methanol in an effort to locate the 23Na signal for MonNa. A very broad peak was obtained (Figure 19) with a width at half- height of about 700 Hz. Because of the linewidth, it was impossible to determine the 23Na chemical shift. Several experiments were carried out in which increasing concentrations of sodium ion, in the form of sodium perchlorate or sodium tetraphenylborate, were added to MonNa solutions. As high added salt concentrations were reached, a sodium signal of reasonable linewidth allowing position measurements did eventually appear and some shifting due to averaging of the 23Na environment by fast exchange was noted. However, this experiment failed to produce reliable and reproducible data and was abandoned in favor of the experiments described below. TWO studies of MonH-sodium ion interaction using 23Na NMR were conducted in methanol solution. In the first, the concentration 70 Bouu< .haoumwmeom can monouowmu owfiuoazu 85.nom msoonwm m.o.H m 00 cofiufimou mmumofiwnfl .GOHunaom Hocmnuoz wmumusumm ca uamm anwvom Sansone: mo annuoomm mzz mmlaswvom _ NI com _ .s. 0.5mH. 71 of MonH was held constant at 0.200 M while the concentration of sodium perchlorate was gradually increased. Shown in Figures 20 and 21 are the results of monitoring the 23Na signal until it was lost due to line broadening. As can be seen, sodium perchlorate alone has a concentration dependent chemical shift, but the behavior is definitely altered by the presence of MonH. A quantitative interpretation of these data was not attempted due to the superimposition of the sodium perchlorate shift, but evidence for the sodium ion-MonH interaction in neutral solution was definitely obtained. A more direct and interesting approach to the problem involved the use of two constant sodium perchlorate concentrations of 0°500.H and 0.250 M’in methanol. In each case, MonH was gradually added and the 23Na signal monitored out to the line broadening limit of about 0.75 to l. The data, shown in Figures 22 and 23, indicate marked 1 23Na chemical shifts due only to the drug addition. The observation of potentially larger shifts as would be obtained with ligands such as the cryptands is precluded by line broadening. Due to this inability to go further in drug to sodium mole ratio and the fact that, considering the error involved, data of the Figure 23 yield only straight lines, an interpretation based on calculated chemical shifts as the weighted average of a complexed and free position is not possible. This quantitative treatment of the data would have allowed the calculation of a complexation equilibrium constant as was accomplished for other drugs (85). 72 4 50 --L E g: _- NaCIO4 <1 \ ALONE 3.50 ‘— 0.200M MONH 3.00 -— l l J l I I I T 0.100 0.300 0.500 0.700 0.900 MOLE RATIO MONH TO NaCIO4 Figure 20. Sodium—23 NMR Study of 0.200 M Monensin with Increasing Sodium Perchlorate Concentration in Methanol 73 5.00 O 4.50 ——- E __ a. NaCIO4 / 3: ALONE / 4 3.50 +- 0.2ool\_fl MONH 3.00 '1 l l l l 1 l I l 100 3.00 5.00 7.00 9.00 MOLE RATIO NaCIO4 TO MONH Figure 21. Sodium-23 NMR Study of 0.200 M Monensin with Increasing Sodium Perchlorate Concentration in Methanol 74 Figure 22. Sodium-23 NMR Study of 0.250 M and 0.500 M Sodium Perchlorate with Increasing Monensin Concentration in Methanol 2.50 —— 2 0.250 |\_I| NaC|04 n_ _L. 0. <1 35° “ 0.500 M NaCIO4 4.00 —- l l l | j o 5.00 10.0 15.0 MOLE RATIO NaCIO4 TO MONH 76 Figure 23. Sodium-23 NMR Study of 0.250 M and 0.500 M Sodium Perchlorate with Increasing Monensin Concentration in Methanol 77 .QOoZ 0... 120.2 O....<~. m..0.2 o8. oomo 8.8 8.3 8.8 J) .082 a 82 WM%MWMMMMMMM . . M M I . m .982 .2 83 WNW M mum H IIOmIN (de) V 78 CONCLUSIONS Measurements on the 23Na spectrum of MonNa were somewhat inconclusive due to line broadening and other effects. The 23Na NMR study of the acid MonH however, definitely confirms the potentio— metric and infrared results indicating a fairly strong MonH-sodium ion interaction in methanol. THALLIUM—205 NUCLEAR MAGNETIC RESONANCE STUDY The monensin anion has been found to complex thallium ion (115). Several attempts were made to find the signals of various thallium salts in methanol and other solvents. Neither thallium acetate, nitrate or bromide were sufficiently soluble in methanol solutions to give detectable signals. It was found that MonH will solubilize thallium salts, but still no signal was detected. CONCLUSIONS Experiments in which MonH was added to thallium solutions to give 205Tl chemical shifts were not successful due to lack of sufficient thallium solubility in solvents which could be used for the study of MonH equilibria. PROTON MAGNETIC RESONANCE STUDIES SPECTRAL ASSIGNMENT Figure 5 shows the PMR spectra of MonNa and MonH. The spectra are somewhat complex since each species has over 60 hydrogen atoms. The complete PMR spectra of MonNa or of MonH have not been intepreted in the literature, although some partial assignments of the resonances of small fragments and derivatives are given in the first reports on the drug (1,5). The MOnNa spectrum in chloroform-dl is slightly better resolved than 79 that of the acid and this spectrum was chosen for an assignment attempt. The protons were numbered as shown in Figure 24. The entire PMR spectrum of MonNa at 0.25 M concentration in chloroform-d1 was run at the narrowest available sweep widths at both 100 MHz and 60 MHz. Sweep widths were calibrated by the side band technique and all peak positions were located within the limits of resolution and peak shape. Naturally, many peaks remained unresolved. By inspecting peak intensities, possible spin-spin coupling constant (J) values, chemical shifts and shapes, decisions were made as to whether a singlet, doublet, triplet, quartet and so forth were possible. The values of these peak positions were then recorded, an average J value calculated, and the center of the pattern either located or calculated. This center frequency was then used to predict the center frequency in the spectrum taken at the other frequency and J values used to predict the peak positions. These peak positions were compared with the actual peaks observed in the spectrum and a visual inspection of peak intensities and shapes also made. This procedure showed when an erroneous choice was made. A good fit however, did not guarantee that a true multiplet had been located since the errors in peak positions and overlap of patterns often allowed for several possibilities. 'An illustration of this type of analysis is shown in Table 4 for a singlet, doublet, triplet and quartet which are found in the MonNa 100 Hz spectrum. Final assignment of a particular peak or multiplet to a particular group or set of protons then proceded by examining its chemical shift and intensity with respect to the structure of 80 Figure 24. Numbering System for the Protons and Methyl Groups of the Monensin Sodium Salt 81 Table 4. Illustration of the Analysis of Several Peaks in the MonNa Proton Magnetic Resonance Spectrum At 100 MHz At 60 MHz Observed Predicted Observed Peak J Center Center Peak Peak Position Frequency Frequency Position Position 152.1 - 152.1 91.3 91.3 91.3 116 68 68 123 7 119.5 71.7 75 75 137 80 81 142 5 142 85.2 85 85 146 90 90 159 90 90 167 98 99 176 9 172 103 107 107 185 116 116 82 the molecule. It was here where various previous fragment assignments and general PMR references (116-119) were of assistance. No meaningful integrated PMR spectrum was Obtained. Two other techniques were helpful in assigning various areas of the spectrum. Five of the nine MonNa methyl groups show up as a large clump of only partially resolved peaks in the chemical shift region less than about 1 ppm. Even at narrow sweep widths, these peaks cannot be assigned. Figure 25 shows this area of the PMR spectrum in chloroform-d solution as well as the effect of adding 1 first a small amount, then a larger amount of shift reagent Eu(fod)3. As can be seen, aside from the broadening effect of the diamagnetic europium, the pattern has been selectively shifted downfield making some of its fine structure visible. Additional resolution was provided by diluting this final solution which lowered the Eu(fod)3 concentration and sharpened the peaks somewhat. The actual positions of the peaks were estimated from their positions in the original unshifted spectrum and will be presented below. Another useful technique in assigning methyl doublets and the protons to which they were coupled was nuclear magnetic double resonance or proton decoupling. Figure 26 shows a portion of a normal MOnNa 60 MHz PMR spectrum and the effect of irradiation of proton signals at 128 Hz and 150 Hz from TMS respectively. Two separate doublets are caused to collapse indicating methyls coupled to single protons. Using the information obtained from the 100 MHz and 60 MHz narrow sweep spectra, shift reagent studies, various decoupling experiments and general trial and error methods and calculations 83 Figure 25. Illustration of the Effect of the Shift Reagent Eu(fod)3 on the Methyl Region of the 100 MHz Proton NMR Spectrum of the Monensin Sodium Salt in Chloroform—d1 NO SHIFT FIRST ADDITION FINAL ADDITION Figure 26. 85 Illustration of the Effect of Irradiation at 128 Hz and 150 Hz on the Methyl Region of the 60 MHz Proton NMR Spectrum of the Monensin Sodium Salt in Chloroform-dl 86 N: s. c: N18. =2. ..<.2m.02 -==: 87 supplemented by PMR literature whenever possible, tentative proton assignments were made. As a general rule,'assignments became more tenuous as analysis progressed downfield due to degradation of the resolution of peaks and the increase in the number of possible protons assignable to a particular chemical shift area. The analysis began at the high field end of the 100 MHz spectrum. For convenience, in the discussion below, all chemical shift values are given in Hz from TMS with no units included. Coupling constants are also in Hz. The shift reagent study indicates the presence of three doublets at 79 and 85, 80 and 86 and 81 and 87, each with a J of 6. These are assigned to the methyl groups 7, 8 and 9 as a group. The coupling constant for the methyl group in methyltetrahydrofuran which the 7 methyl environment resembles has been given as 6.6 (120). The analogous values for the methyl and dimethyl tetrahydropyran do not seem to be available, although a study of substituted methoxy tetrahydropyrans (121) does seem to indicate that similar coupling constants should be expected for the 8 and 9 methyls. Both the 60 MHz and 100 MHz spectra bear out the fact that the 102, 95 and 88 peaks are a triplet with J = 7. Since there is only one methyl group which may appear as a triplet, this is assigned to methyl 6. Reference (115) gives the shift of a CH3-CH2-C- methyl as between 0.9 and 1.1 ppm and the 1:3:1 intensities are approximately correct in both spectra. Peaks at 91 and 98 definitely form a doublet with J = 7 which was clearly confirmed by decoupling experiments. This doublet has been assigned to methyl 4. Its chemical shift would be expected to be similar to the dimethyltetrahydropyran analogues and methyls 8 88 and 9, but a slight downfield shift may be caused by its proximity to the 32 O—H group. The doublet moves significantly with addition of the shift reagent which may indicate its nearness to the acidic part Of the molecule in contrast with the 7, 8 and 9 methyls whose shift positions were little changed. The peaks at 116 and 123,and 122 and 128 have been found to be doublets with J values of 7 and 6 respectively. Both move quite rapidly during shift addition. Using trends predicted for methyl groups neighboring ethers and carboxylic acids (116), these doublets are assigned to methyls l and 3 respectively. Decoupling experiments also provided confirmation that these four peaks are indeed doublets. The sharp peak at 152 is definitely a singlet due to methyl 5. It is the only methyl which can appear as a singlet and its position agrees with that of a CH3-C- methyl (115). In one of the most poorly resolved areas of the spectrum there seems to be a possibility of two J = 5 triplets at 137, 142, 146 and 152, 157 and 162. Most of this region is obscured in both the 100 MHz and 60 MHz spectra, but these triplets have been tentatively assigned to protons 9 and 10 and 11 and 12. A triplet with J = 7 at 180, 187 and 194 may be due to protons 16 and 17. The peaks at 159, 167, 176 and 185 most likely represent a pure quartet with J of 8. As the region beyond the methyl groups is considered, agreement between the 60 MHz and 100 MHz spectra becomes a much more important criterion for deciding on a particular multiplet. All of those given here can be assumed to be in substantial agreement. There is only one pg£g_uncoupled quartet possible in the spectrum, and although other seemingly pure quartets were 89 naturally observed, this grouping was assigned to protons l8 and 19. Irradiation in this area produces a change in the signal for the 6 methyl and the coupling constants of 7 and 8 are within experimental error of one another. Peaks at 190, 197, 205 and 212 have been assigned as a quartet with J = 7 and are due to proton 1. The best decoupling of the 1 methyl signal is obtained in the center of this region and the coupling constant value for the 1 methyl doublet is identical. Several peaks in the 205 to 245 area contain what are thought to be coupled quartets. For example, the first multiplet is at 207, 215, 223, 231 and 239. If proton 5 gave a signal at 223, it could be split by methyl 4 with a coupling constant of about 8 to give a l:3:3:l quartet at 211, 219, 227, and 235. If each of these peaks were split by some adjacent hydrogen, say 6 or 4, each signal in the quartet would be further split to a doublet. In this case, if the second coupling constant is also on the order of 8, a quintet rather than an octet is obtained with a l/2:2:3:2:1/2 intensity pattern at 207, 215, 223, 231 and 239. This type of analysis also seems to indicate that there are coupled quartets at 211, 218, 227, 234, and 242 and also at 205, 213, 221, 229 and 237 both with J values of 8. Irradiation at 223 does effectively decouple the signal assigned to the 4 methyl and seems to provide some validity for this type of reasoning and the assignment of the first multiplet to proton 5. The other two quintets probably belong to protons 21, 26 and 29 with the likelihood that 26 and 29 are nearly coincident. These quartets would be split in a somewhat more complex fashion, but since resolution here could not distinguish a six or twelve peak multiplet, no detailed analysis is possible. 90 The peaks at 269, 263, 256 and 251 appear to be grouped as a quartet with an average J of 6. This quartet is most likely due to proton 3 which was confirmed by the decoupling of the 3 methyl doublet caused by irradiation in the 263 to 256 region. The doublet and quartet both have the J = 6 values. Signals at 315, 316, 324 and 326 can be analyzed as arising from a doublet of doublets with J values of l and 9. This signal is due to proton 2 and has been Observed in monensin fragments. Proton number 4 should also show up as a doublet of doublets and the peaks at 336 and 324 tend to point to a doublet with J = 12 and another J less than 1. The sharp singlet at 340 is due to the 2 methyl or methoxy group. This is the only MonNa peak which has been unequivocally assigned in the literature (1) and falls exactly where a methoxy signal is classically expected. The signals at 354 and 356 are tentatively assigned to a pure doublet of J = 2 belonging to protons 7 and 8, while apparent doublets at 382 and 388 as well as 400 and 406 with J values of 6 are ascribed to protons 22 and 23,and 27 and 28. Here again, poor resolution and the increased number of possible protons which may give rise to each signal make assignments more tentative. A sharp peak at 394 is thought to be a singlet due to protons 33 and 34. It is downfield far enough to be on a carbon bound to oxygen and rapid exchange of the hydroxy proton 31 would cause it to be a singlet. The 397 and 409 peaks seem to fit as a pure doublet with J = 12 which is assigned to proton 20. There is some evidence to indicate 91 that a 389, 400, 411 grouping may be a triplet with a J of 11 due to proton l3. Coupling constants of this magnitude for a proton in a bridgehead position coupled to ring protons are shown to be reasonable by the assignment and J values for similar protons in some MonNa fragments (l). A broad peak past 420 cannot be meaningfully resolved at any sweep width. Its position and width give some reason to believe that it may be due to the hydroxyl protons 30, 31 and 32. The two MOnNa water molecules do not appear in the spectrum at room temperature as will be seen later. An explicit assignment of all of the peaks in the MonH spectrum has not been made. As Figure 5 shows, many similarities exist. The most important difference in the MonH spectrum is the previously unreported broad peak at about 6 ppm which has been assigned to the bound water molecule of the acid. This assignment was confirmed by an experiment in which MonH in concentrated methanol solution was reprecipitated twice with deuterium oxide. The product gave a PMR spectrum in which the water peak had lost 50% of its intensity. Likewise,the infrared spectrum of this compound showed that intensities of the 3300 cm.1 and 3520 cm-1 bands also decreased by about 50% and new bands characteristic of deuterium oxide appeared at 2450 cm.1 and 2600 cm-1. EFFECTS OF COMPLEXATION As is shown in Figure 27, the l, 3 and 5 methyl signals undergo induced shifts in 0.25 M MonH as sodium perchlorate is added in chloroform-d1 solutions. Approximately linear shifts with different lepes are observed as the complex is formed indicating definite molecular conformation changes as MonH 92 Figure 27. Proton NMR Study of 0.250 M Monensin with Increasing Sodium Perchlorate Concentration in Chloroform-d1. Methyl 3 0, methyl 1 I , and methyl 5 A . INDUCED SHIFT AT 100 MHZ (H2) 800- 600‘ 4004 2.001 I I 93 0 -200I- 4.00 I i I I 0 0250 0500 0J50 L000 MOLE RATIO NaCIO, TO MONH 94 and the sodium ion interact. The limiting shifts at 1:1 MonH to sodium perchlorate mole ratios do not appear at the frequencies of the same signals in the MonNa PMR spectrum. Carbon-l3 NMR and PMR had already predicted conformational differences between acid and salt forms. This evidence indicated conformational changes also occur in the formation of the sodium complex with MonH in neutral solution. The frequency of the bound water signal has been found to be solvent dependent, even though, as shown in Figure 28, the water molecule itself seems to be somewhat isolated from the solvent environment. Shifts of 6.03 ppm in chloroform—d 4.83 ppm in 1, tetrahydrofuran-d and 4.98 ppm in acetone-d6 were observed for 8 0.250 M MonH solutions. Attempts to obtain a reasonable PMR spectrum in methanol-d4 were unsuccessful due to the interference of the normal methanol hydroxyl peak. The deuterium oxide experiment did indicate that the water molecule was labile in methanol which explains the interference of the OH signal as the OD exchange of methanol-d4 with the water available from MonH. The variation in peak position in the other three solvents is probably also due to exchange of the hydrogen bonded water molecule with free water in the solvent. All three peaks are in a position downfield of the shift of pure free water. In tetrahydrofuran-d the peak is closest to that of free 8 water (4.72 ppm). The bound water signal in acetone-d6 is slightly further downfield and in chloroform-d1 it is very distant from the free position. The positions are compared to the shift of pure liquid water at 4.72 ppm (122). The same relationship noted can also be observed by comparison with the infinite dilution shifts of water in each solvent 95 Figure 28. The Crystalline Structure of Monensin. Closed circles are carbon atoms, open circles are oxygens with the water oxygen dotted, dashed lines indicate hydrogen bonds and A is a possible position for associated exchangeable water. 96 97 which are 2.42 ppm for tetrahydrofuran (122), 2.40 ppm for acetone (123) and 1.99 ppm for chloroform (124). This evidence would tend to support a simple model of free and bound exchange based primarily on water solubility in each solvent. Lowering the temperature produces a downfield shift of the water peak in all solvents as is shown in Figure 29 for the chloroform-d1 case. In addition, the peak splits to give a broader line at lower field and an upfield sharper peak. Both peaks migrate further downfield at approximately constant separation with further decrease in temperature. Figure 30 graphically exhibits this behavior in the three solvents. This splitting seems to indicate that the water molecule can be attached to MonH at two different sites. The separations and temperatures of coalescence, shown in Figure 30, indicate that the'exchange is easiest in tetrahydrofuran-d8 and most difficult in chloroform-d1. The further downfield shift reaffirms the previous model of exchange with water in the solvent in that the decrease in temperature seems to favor an equilibrium towards the bound water molecules with fewer and fewer free water molecules in the solvent. There is little indication of the signals ever reaching a frequency characteristic of completely bound water molecules before line broadening and solvent freezing make measurements impossible. Figures 31 and 32 show the behavior of the bound water peak in acetone-d6 and chloroform-d1 at ambient temperature as sodium perchlorate is gradually added to the MonH solutions. In both cases, a steady decrease in peak intensity is observed and the peak disappears as the 1:1 ratio is approached. However, in acetone-d6 solution a downfield shift is noted while in chloroform-d1 the peak moves upfield. 98 .n3ono Oman ma xmon monouowou Bhowouoaso m .06 one mounumuoeama .Hvlauomouoazu a. namaonoz.m.omm.o wo cowwum moan: venom one we wuuooem mzz nauoum .mN shaman T551 .3 J. on 0..I 8 ON m I E0 73135 99 Figure 30. Proton NMR Study of 0.250 g Monensin in Tetrahydrofuran-dS, Acetone-d6 and Chloroform-d1. Signal separations and coalescence temperatures are . , tetrahydrofuran-d8, 18 Hz, 20°C; C], acetone-d6, 15 Hz, 10°C; and ‘ , chloroform—d1, 8 Hz, 0°C. 100 6.; mmDh 115 00.0? 00.00 00.0w 00.0w 0 _ _ _ c + _ _ _ _ o o oo 000 #1000... coo 0000 co. co . . c. .1 8: o o o o o o oo llooué o o o o o o 5 [Icons 0 o o 000 (A) 'H'O'S 'SA 'lVllNElOd Figure 35. 116 Titration of 0.100 gram Monensin Sodium Perchlorate Complex in 50.00 ml of Solution with 0.0046 ! Tetrabutylammonium Hydroxide in Methanol. Complex prepared from methanol solution and soaked in water. 117 is: m5540> 00.8 00.00 00.0w 00.0w 0 _ _ _ _ t _ _ 1 _ Of 1:88 o o o o [I 00:. oo oo oo oo .oo 83 .0 II o o o o o o O. [1.805. ‘ SA 'IVIlNE .LOd (A) 'B'O'S 118 with equivalent weights roughly corresponding to the formula MonH°NaC104'S where S is an attached solvent molecule. It was of some interest to obtain positive proof that the perchlorate anion is indeed incorporated in the new solid complex. Efforts were made to determine whether perchlorate as perchloric acid might escape in the solvent evaporation during complex preparation. No evidence of perchloric acid was found in trapped portions of the evaporated solvent. However, if any of the solid complexes were dissolved in methanol, several drops of methanolic tetraphenylarsonnium chloride were found to cause a flocculant white precipitate diagnostic of the presence of the perchlorate ion. In an effort to further confirm this evidence and perhaps begin to explore the perchlorate ion surroundings, Raman spectroscopy was employed. Shown in Figure 36 is the extremely compressed Raman spectrum of solid MonNa. As can be seen, the spectrum is quite complex below about 1500 cm.1 as would be expected from the large number of carbon- carbon and carbon-hydrogen bonds. (Table 5 lists the exact positions of the various bands as obtained from this and other better resolved spectra.) The large group of peaks from 2800 cm”1 to 3000 cm-1 and the one from about 1400 cm.1 to 1500 cm-1 are typical of MonNa, MonH and the new complex from chloroform in the solid form. There are very few differences between the MonNa and MonH spectra. The complex, however, shows two very strong bands at 1862 cm-1 and 1961 cm_1. These have not been assigned. It also has a unique pattern of bands at 462 cm-1, 611 cm.1 to 634 cm.1 and 927 cm-1. These have been attributed to perchlorate ion vibrations on the basis of literature values of 460 cm-1 9 119 000w 000w F:50 o8 uamm aaflvom camcmcoz vfiaom ocu mo muuoomm amamm one .an muswfim 7:5 88 000w 120 Table 5. Band Positions in the Raman Spectrum of the Solid Monensin Sodium Salt (cm-1) 512 m 849 w 1151 1140 520 w 860 w 1158 1446 533 w 873 m 1192 1453 545 w 888 m 1197 1458 563 m 911 w 1206 1464 569 w 922 m 1227 1470 574 w 942 m 1248 2817 587 m 970 m 1268 2855 621 m 978 m 1286 2873 640 m 980 m 1293 2876 652 m 1004 m 1317 2887 665 w 1026 m 1323 2894 690 m 1045 m 1329 2916 695 m 1063 w 1335 2933 713 w 1074 w 1356 2938 764 s 1083 w 1368 2965 791 s 1095 w 1380 2975 803 w 1110 m 1393 2987 819 w 1128 w 1405 2998 840 m 1136 m 1420 121 626 cm_1 and 932 cm.1 (125). The region of the 1110 cm"1 band was obscured. Several of the new solid complexes have been subjected to elemental analysis for carbon and hydrogen. Results of these determinations are summarized in Table 6 for the sodium perchlorate complex as isolated from methanol, ethanol and chloroform and also the sodium iodide and chloride complexes from methanol. Also shown are the calculated carbon and hydrogen compositions of various formulations of the complex based on combinations of MonH and the salt and solvent used. By comparison with the calculated compositions of MonH and MonNa which are also shown, it is quite obvious that none of the new complexes are simply the acid MonH or normal salt MonNa. Choosing a composition with or without associated water or solvent solely on the basis of the elemental analysis data would be quite tenuous. The addition of these relatively small molecules simply does not significantly change the percentage of carbon and hydrogen. In four cases, all of the various calculated compositions are in fair agreement with the values observed. In the sodium perchlorate complex as isolated from chloroform, a rather high observed result has occurred. However, in this case, the analysis itself is somewhat suspect. It is known from other determinations that chloroform is indeed present in the complex, but additional analysis of the solid for elemental chlorine gave a value of 4.02%. If associated chloroform had been detected, the value would be approximately 15%. Indeed, the 4% figure corresponds to the presence of the sodium perchlorate alone, as does the carbon and hydrogen analysis. The results indicate that the chloroform was simply missed in the analysis, perhaps due to its volitility. 122 mw.w cm.m «0.5 cm.o mm.m mw.m ~©.n mN on.~o mm.mo mn.mq mo.wq @N.mm Hm.qm ww.mm 0N mzcoz 0mm.mcoz MHQmUONm oaomzmcoz m c H030 Gaomzmcoz N c 0 m 0H0mzmcoz qoaumzmcoz q q EquOHOHSU Boum meano anmz am.m cm.w mo.w mm.w mq.w mm.m NH.w mm.m ww.n No.5 mN No.nm mm.wm om.nm mN.mm ¢O.wm 0N mw.mm mm.qm mm.mm Hm.qm mm.mm 0N m0m:00Nm Homzmaoz acmmu Humzmcoz 0mm Homzmcoz Homzmaoz Hocmnumz Scum xoaaaoo Homz mmmmummUONm Odomzmaoz SONmommu Odomzmaoz N c 0 m oaomzmcoz qodomzmcoz q on Hocmsum Soup meoEoo oHomz momeanu camcwaoz 3oz nw.m om.n m@.n He.m om.n 3N nH.m oo.w mm.n mw.n H¢.m mN mo.Hm HH.Nm MMmm0 Hoamzuoz aoum meano Hmz c m0mm00~m oaumzmcaz m q :0 m0 0H0mzmcoz N w 0 m oaumzmaoz soaomzmaoz 0m>mmmm0 c Hosmnumz aoum xMHQSOU oaomz mo mfimhamc< Hmucoaoam .o mHan 123 CONCLUSIONS All analyses led to the conclusion that a new monensin-sodium complex has been prepared. Indications are that in addition to a molecule of MonH, molecules of sodium salt and the solvent used for preparation are also involved. Neither the exact stoichiometry nor the structure of the new complex have been determined, however it is assumed that the sodium ion must be at least near the center of the MonH moiety due to the ionic solubilizing effect of MonH in hydrophobic solvents. EXCHANGE STUDI ES Studies of possible exchange involving the new complex and the normal salt were begun using the sodium perchlorate preparation from chloroform and again employed PMR spectroscopy. Most spectra were run in chloroform-d1 at varying temperatures. Four temperature dependent spectra were considered. The temperature dependent spectrum of the MonH water region in chloroform-d has been shown previously (Figures 29 l and 30). The MonNa spectrum exhibits no bound water peak at temperatures above 30°C, but does develop a distinct peak at 9.6 ppm at about -20°C which at -70°C seems to split to two peaks with a separation of about 18 Hz. The spectrum of a 1:1 solution made up of MonH and sodium perchlorate in chloroform-d and the spectrum of the solid new complex 1 as isolated from chloroform and then redissolved in chloroform-d1 were found to be in every way identical with the exception of the associated solvent bands. Since the exact molecular weight and, therefore, the concentrations of the solutions of the new solid complex were unknown, all preparations of the new complex for this study were made in solution and not isolated. 124 The PMR spectrum of the complex does not show any peaks downfield from 5 ppm at normal probe temperatures. However, as the solution is cooled to 0°C, a broad peak does appear at about 5.8 ppm. In addition, at -40°C, where this new peak is quite sharp, another peak begins to grow in at 10.4 ppm. The 5.8 ppm peak is thought to be due to bound water in the complex and seems to split to give two peaks with a separation of 14 Hz at about -70°C. Efforts to assign the 10.4 ppm peak have so far been unsuccessful. Exchange experiments centered around the anticipated interaction between the 9.6 ppm signal of the salt and the 5.8 ppm signal of the complex. It was found that a 1:1 mixture of the complex and salt produced only one peak at about 7 ppm which appeared even at the ambient probe temperature of approximately 30°C but had a somewhat temperature dependent position. This led to the suspicion that fast sodium ion exchange from one environment to the other might be creating an average chemical shift for the water molecule and motivated the study below. Solutions were made up in chloroform-d which were 0.100 y_in 1 complex with concentrations of MonNa ranging from 0 to 0.150 5, Although the complex itself showed no water peak at ambient temperature, addition of 0.010 N MonNa caused a peak to appear in the 7 ppm region and further MonNa addition caused this peak to grow and gradually shift downfield towards the MonNa position. Repeating the experiment at -10°C yielded a very similar pattern which was transposed upfield. These results are illustrated in Figures 37 and 38 which show the observed chemical shifts as a function of complex to salt mole ratio \I 11.1. v 1‘ 125 Figure 37. Proton NMR Study of 0.100 g Monensin Sodium Perchlorate Complex with Increasing Monensin Sodium Salt Concentration in Chloroform-d1 A (PPM) 126 5.00 —.—— 6.00 ambient 7.00 I I 390 ‘0 theoretical | L J l l l 0 0.05 0.10 0.15 CONCENTRATION MONNa (m) 127 Figure 38. Proton NMR Study of 0.100 !_Monensin Sodium Perchlorate Complex with Increasing Monensin Sodium Salt Concentration in Chloroform-dl 128 theoretical A E _ O. D. V 4 7.00 " ambient 8.00 ‘— l 1 l I 1 1 1 1 4] i l I l 7 2.00 4.00 6.00 8.00 10.0 MOLE RATIO COMPLEX TO MON Na 129 and added salt concentration. The analogous experiment in acetone-d6 could not be carried out due to solubility limitations. As mentioned previously, a shift due to the averaging of the sodium ion environment was expected. This behavior should be predicted by the equation below: (9) Gobs = X(5.8 ppm) + Y(9.6 ppm) where 5.8 and 9.6 ppm are the temperature independent positions of the exchanging peaks for the complex and salt respectively and X and Y are the respective fractions of each form (32,125). Using analytical concentrations of the complex and MonNa present in the chloroform-d1 solutions and the observed temperature independent chemical shifts for the complex and salt alone in solution, it was possible to calculate values for a hypothetical shift of the peak with increasing concentrations of MonNa. This curve is shown in both of the figures and can be seen to be in general agreement with the shapes of the experimental plots. However, the predicted behavior is far from accurate. The equation uses temperature independent shifts and also assumes that the fraction of each form as calculated from analytical concentrations is temperature independent. The observed plots tend to indicate that the fractions of each form are temperature dependent, with different behavior at -10°C than at ambient probe temperature. At very high complex to MonNa mole ratios or at low salt to complex ratios, the positions of the observed and theoretical peaks should agree. Visual extrapolation indicates that they do not, with the ambient result falling almost 1 ppm too far upfield. Even an unknown effect of the unassigned 10.4 ppm peak could not cause the observed chemical shift to move upfield. ll} ‘I’ ‘l. 130 The only explanation which seems to be reasonable is that the fractions of complex and salt are affected by their own interaction, perhaps exclusive of sodium ion exchange. This interaction, possibly of the conformational equilibrium type, would of course have a temperature dependent equilibrium constant. Calculations show that the -10°C data can be made to fit equation 9 if an equilibrium constant for interaction is incorporated. However the actual values obtained for such a constant cannot be meaningfully interpreted considering that no value of an equilibrium constant could ever suitably account for data at the ambient temperature. Yet the results are of interest in that they do point towards exchange which is temperature dependent and possible interconversion of monensin forms. - Other evidence obtained in this same experiment by monitoring the methyl region confirms this idea of conformational equilibration. The peaks which have always been most sensitive to conformational changes are those due to methyls 1 and 3. Their shifts are tabulated below. Table 7. Positions of the 1 and 3 Methyl Groups in Several Monensin Solutions in Chloroform-d1 Chemical Shift from TMS at 60 MHz (Hz) Methyl 1 Methyl 3 Complex 66 82 59 75 1:1 Complex-MonNa 69 79 62 72 MonNa 68 75 72 79 131 The shifts in this range at -10°C are identical to those above, within experimental error of locating the peaks. Examination of the complex and 1:1 mixture reveals that the doublets in each case are both centered at 74 and 67 Hz respectively. The complex however, has a coupling constant of 16 Hz while the 1:1 mixture has a reduced J value of 10 Hz. This change in the direction of the 7 Hz value for MonNa is reasonably close to an average value of the constants calculated for a complex and salt mixture. Again, conformational changes giving rise to an average spectrum of both conformers seems to be indicated. CONCLUSIONS Average chemical shifts and coupling constants observed for the water molecules and methyl groups in the complex-MonNa mixtures indicate the possibility of the occurence of fast sodium ion exchange between the two types of monensin complexes or perhaps fast interconversion between the two monensin forms. CHAPTER VI BIMOLECULAR LIPID MEMBRANE STUDIES INTRODUCTION The bimolecular lipid membrane (BLM, singular and plural) system has often been used to test the ionic transport capabilities of various new synthetic and naturally occurring complexing agents. It was of interest to us to determine the ion transport properties of monensin as well as of some other compounds such as the tetrazoles, glutarimides and cryptands referred to previously. INITIAL TESTS OF THE SYSTEM It was found that the most stable BLM were formed with a 1:1 mixture of oxidized cholesterol and lecithin dissolved in normal octane. The electrical resistance of these BLM is usually slightly greater than 109 ohms with aqueous 0.10 M_sodium chloride solutions. It was noted from the very outset of the experiments that many variables seemed to affect the formation of BLM and measurements of their properties. Among the more important factors to be considered were the mechanical stability of the system and the electrical shielding of the apparatus. Temperature also appeared to be a crucial variable although temperature control of the system was not maintained. Age of the various lipid solutions seemed to be of the utmost importance. Older lipid solutions often did not make BLM or made membranes of low or irreproducible resistance. Even with many precautions, on occasion, the formation, the stability or the resistance of a BLM would not follow the expected pattern. 132 133 Due to the fact that the system often behaved in a rather chemically ill-defined manner, all experiments took on a unique character and much repetition was an absolute necessity. Basically, the only justifiable conclusions regarding the testing of various ion transport compOunds were that they either changed the BLM resistance or did not. Although quantitative estimates of resistance changes were available, they were, for the most part, not reproducible and only continued indications of resistance changes in many experiments allowed the qualitative conclusion that ion transport was occurring. Initial testing of the system to observe its response to ionic permeability changes was accomplished with aqueous solutions of valinomycin, dicyclohexyl-l8-crown-6 and the triodide ion. Several methods are often used for the introduction of modifying agents: direct addition to BLM via insertion in BLM forming solutions or addition to electrolyte before or after BLM formation. The second method was found to be the most reproducible and easiest and was used in all experiments. Valinomycin in approximately 10-6'M_concentrations in 0.10 M sodium chloride was found to change the BLM resistance by several orders of magnitude, often lowering it from approximately 109 ohms to approximately 105 ohms. The crown compound in more concentrated solutions of 10-3 M in the same electrolyte also showed this same resistance change characteristic of ion transport. Solutions of triodide ion made up with iodine and iodide in 0.10 M_sodium chloride also lowered the BLM resistance 3 to 4 orders of magnitude. This effect is suspected to be due to the delocalization of the ionic 134 3— and on higher aggregates such as I5- and 17- which causes higher solubility of the ion in the lipid phase (66). charge on I CONCLUSIONS The BLM system responded to the addition of compounds known to cause ion transport through membranes. However, BLM experiments were found to be influenced by a host of variables, some of them unknown, and data were often found to offer only qualitative evidence of ionic transport. MONENSIN AND OTHER ION COMPLEXING AGENTS Although monensin has a reported water solubility of only about 0.1 mg/ml (1), this concentration (10-4 M) is more than Sufficient for BLM testing. Saturated solutions of MonNa and MonH were prepared in aqueous 0.10 M_sodium chloride and BLM formed in these solutions usually did show a decreased resistance. The change from the normal BLM resistance was not as great as the 4 orders of magnitude drop observed for valinomycin and crown, but usually exceeded a change of 2 orders of magnitude. Thus transport of sodium ions by monensin as had been previously observed in bulk solutions (25), natural (7) and synthetic membranes (34-38) also seems to occur through BLM. Various tetrazoles were also tested in the system. Solutions of 0.10 M trimethylenetetrazole, pentamethylenetetrazole (PMT), hexamethylene- tetrazole and heptamethylenetetrazole were each made up in 0.10 M sodium chloride and BLM were formed in these solutions. No changes of the BLM resistances were ever observed in these experiments. Replace- ment of the sodium chloride with a 0.10 M potassium chloride electrolyte did not effect the BLM resistances. 135 Several BLM were formed from normal octane solutions which also contained either the protein histone or protamine in addition to the cholesterol-lecithin lipid mixtures. This produced BLM with the usual resistance (greater than 109 ohms) which were designed to be better models of the true biological membrane. Again, addition of tetrazoles had no detectable effect on the 109 ohm BLM electrical resistance. Similar tests were made with the convulsants glutarimide, 3,3-dimethyl- glutarimide and 3-ethyl-3-methylglutarimide with the same result. None of these compounds showed any resistance changes. Although all of these molecules are weak ion complexing agents and are also highly biologically active, it would seem that they are not capable of transporting ions through BLM. Tetrazoles are known to be surface active (87) and have been shown to effect ion transport in isolated biological preparations (126). However, the complex which tetrazoles form with sodium ion is not of the charge insulating type, such as that of valinomycin or monensin, and the ionic solubilizing effect is not present. Glutarimides are also surface active drugs, but again, the weak complex with sodium ion does not change its solubility in the lipid phase. The cryptand 222 (Figure 3) was also tested in the system, again using 0.10 M sodium or potassium chloride solutions. The ligand was found not to cause any resistance change at any concentration up to 10 mg/ml. Even though the cryptand is a strong complexing agent, it is quite basic in aqueous solutions. At pH 7, the nitrogen atoms are protonated and alkali ion complexes are not formed. However, experiments with the synthetic precursor to cryptand 222, the dilactam shown below, 136 Figure 39. The Dilactam Form of Cryptand 222 O OJV O\___/ 0 did cause a resistance drop of 2 to 3 orders of magnitude. The dilactam also forms alkali metal ion complexes and has been found to be much less basic in solution than cryptand 222 (127). Thus the metal ion is insulated and little charge remains on the outside of the molecule allowing for increased lipid solubility and BLM permeability changes. CONCLUSIONS Both MonNa and MonH have been found to facilitate ion transport across BLM. Tetrazoles and glutarimides, although they do form ion complexes, do not influence BLM permeability to sodium and potassium ions. The cryptand 222 does not affect BLM resistance in neutral solution, however the dilactam form is able to complex and transport ions through the BLM. APPENDICES APPENDIX I APPLICATION OF COMPUTER PROGRAM KINFIT AND SUBROUTINE EQUATION FOR THE DETERMINATION OF ACID DISSOCIATION CONSTANTS APPLICATION OF COMPUTER PROGRAM KINFIT AND SUBROUTINE EQUATION FOR THE DETERMINATION OF ACID DISSOCIATION CONSTANTS As described in Chapter III, one approach to the determination of acid dissociation constants was to fit the observed titration data to the exact cubic equation for a weak acid-strong base system. These calculations were accomplished using the CDC 6500 computer and program KINFIT (99) by manipulation of SUBROUTINE EQN. This appendix details the manner in which the analysis was performed and lists the input necessary for proper operation of the program. Equation (5), + 3 (5) [H 1 + ((2b + Ka)[H+]2 + (Kacb — Kaca - KS)[H+] - KaKS = o with symbols as previously defined, can be solved by several analytical methods. A trigonometric solution was found to be the most useful. In the general cubic equation 3 2 (10) y + py + qy + r = 0 if y is allowed to have the value x - p/3, the equation is reduced to the form (11) x3 + ax + b = 0 137 138 where (12) a = -§-(3q - p2) b = $7019 - 9pq + 27r) Then by computing the value of the angle 0 in the equation below, 3 + ('57) Q) (13) cos¢ =-— MIC!" x is found to have the values a 9+2° 3 cos(3 40 ) (l4) 2 --3 cos-3 2 --3 A o - 3 3, 3 cos(3 + 120 ), 2 This solution was inserted into the SUBROUTINE EQN as shown in the listing. Thus a value of the calculated hydrogen ion concentration was obtained. Since the hydrogen ion concentration is not the observed variable, it was necessary to convert this value to a voltage, which was accomplished via the equation + (15) E - -0.059l6 log10 [H ] + Vb which is the general form of the Nernst equation with V serving to b incorporate all constant terms. The equation appears in the program as (16) EE = -(CONST(5)*ALOGlO(ABS(DD))) + U(1) where EE should be the observed voltage, CONST(5) is the RT/nF term at 25°C with n.= 1 and Vb has been made the unknown, U(1). The value of U(1) is the intercept of a Nernstian pH versus voltage plot and .,14|l|‘ ii‘ (I (III [[1141 (ill ‘I ll). ‘7. italll‘ I I 139 was allowed to vary to give the best fit to the data. As mentioned previously, since the absolute values of the observed potentials are arbitrary, the value of U(1) is also arbitrary and does indeed vary slightly from run to run. The other unknown parameter is the value for the acid dissociation constant, U(2). The variables, titrant volume and observed voltage, were assigned as shown. Values of Vb and Ka are varied in the program to give the best fit at each volume of the calculated voltage to that observed. Five constants were necessary for these calculations and were named as illustrated in the listing. Some improvements in the fit were noticed by allowing the initial basic titrant concentration also to be an unknown. The format and structure of the control card and data deck for KINFIT have been exhaustively described previously (99,125,93) and detailed instructions as to their operation will be omitted. The normal ordering was always used with the control card first, giving the maximum number of iterations allowed, number of constants and convergence tolerance. A title card followed, after which came the card containing the values of the constants and fourth,the initial estimates of the unknowns. Voltage versus volume data completed the deck with relative variances attached in the usual manner. This particular subroutine was found to be extremely sensitive to initial guesses of the Ka and V values, often converging very near b the original estimates. Therefore, variation of the estimated values produced changes in the calculated values and also variation in the value of the total residuals calculated at convergence. This phenomenon 140 may have been due to local minima in the function. In any case, poor fits and poorer estimates of the known HBz Ka value caused this method to be abandoned. J. il 1| 141 MDZHHZOU N zmaemm mmogm > m> mm may Amvamzoo o Hz 4 om>mmmmo may ANVxx o mamqu zH mszO> Hz may 0 ~um<>oz ez m> mm may AHVa mm< mzzozxz: mes o Nuxzaoz A«monHon .Aomvoa .AooHVAN .AH~.o~veom> .AH~.omvm .AOOHvsa a.Aooavom.Ao0HVmom.Acexx.Aooa.evxez.Ao~v=.AOOH.avx onmzmzHa Hmzoo.amoz.wum>.»a.w.was.z.g.yo.e.emx.g<>oH~ m.oe.s~.m.am.om.mom.Hon.mwaxm.xx.mwaH.mmm.m<.H.emme.xesfi .xoz.emoz.mosz.m om>mmmmo may mm museum mm Ao.m\ ez eznz.>mnz.zg.Hmooa.3x.zu.fimx.ox.ax g.Aomvxm .Aomvw .Aomvx .Aomvm .Aomvmm .Aomvao> onmzmzHa A\m.mmu mmeszm.x0H.HHmm.xm.m.mmu ommo.xe.fl.mm u mzmemo .xav Hmm .xa .omHV a no azmHonmmoomoN .xo .omav amn mmamo .x0 .m.oam u m>mo .xo .ozHV H0 oz< m> .xm mus mo mmsg<> emmm may mosz zH\Aom«AHvao>v u no >H\Ao>«omzov u mo m+ avao> u aHvs om\A>a*Amoonmovv n m mo\3x : moo mo\Ava a mo AAvam:v««0.0H I Ava gm\amm : AHV>V a Avam oaamo.o u gm AHvso>+o> u >9 mm .H n H mom on oueHz oo.H a mo oflnmm.m a 3M. AcH~.Homm .om .mzma Aoau .HommV Amfl.oommV ANH.oomm.>o AHH.oomm:AvaV u HHva mm .H u H cam on oo.o n stm ox u m> 2m u HM mm\Amz:m«sz:muozbm«mzsmv 0M mm\Amzbquz=mumzpm.mev xx sz:m«sz:muoz:m«ms n ma mazHazoo AHvx«AHVx+oz:m u exam Agidfimzam u 95.... HHvx+szsm a «saw Hva+mzsm n mzbm mm .H u H mam on o o a exam 0 o u mzsm 0.0 u exam 0 o n mzsm mazHazoo AHVH«Hva: u HHvx >onAHVH u HHV» mm .H u H cam on >ma u H>mo >nz u >2 Azz.om.aozv EH H+HHz u HHz 32 .H u <2 womH on Hoo.o u >99 o u 902 2H u >mo zH u >qu o~+mo.H u zH now 000 mazHHZOU mow 149 mm .H u H omeH 00 00.0 a 020m HH~z0m«Hz0mv\mepm«mz0mv00000 u 000 00zH0200 >>«xx+mzpm a 0:00 >>0>>+~z00 : N200 xx«xx+Hz0m a Hz0m 0> : HHO> u >> 0x : AHvx : xx 00 .H u H onMH 00 00.0 n mz0m 00.0 n 0200 00.0 - H200 00\mz0m u 0» 00\0z0m u 0x H0\H0m*00.00H0 a >00 HHH:000\~z0000000 a 0m mazHHzoo 00000¢0 0H >00+>0 a >0 >00xm0.0: u >00 H~0H 00 00 HH>00.HH.>000 0H >0 a >0: >00 u >002 mHOH 00 00 H>00z.00.>000 0H 0HmH 00 00 Azz.00.0020 0H HH000H000\~z0m n >00 mszHzou 0.00200 u «:00 HmeuHHvav a 0 00 .H a H 0H0 00 00.0 n 0200 00\Hz0m u H00 005500 0mma 0HmH QOMH HNOH mHOH 000 000 150 0000000000000000000000 020 0000 H000 .H000000 00H0> 00zH0200 0H2 HmoH .H000\H00«>00 a 0 0>\H00«0>v u < 000 AmmH .H00000 HO0H .H0000Hv 00H03 00m HO0H .H0000 .m> HHOH .H000<00 00H03 H000 Homm .H000000 00H03 HHHI000\020000000 u 000 mazHHzoo H0000 +0200 n 0000 H000 : HH000 u 0 00.H n H 000H 00 0u020m 00zHHzoo <0 0 .HHVM0 .HH000 .HH000> HHHH .H000000 00H03 H000uHH000 n 0 00 .H u H 000H 00 00\mz0m u H400 00zH0200 HH000+0200 a 0:00 HHHVNx00H000< u HH000 0m0H 00 00 H0.H.HH.HH0~xV 0H 000a 000a 000a 0m0H 151 00H.H NmH.H 00.0H 00.0 0.0 m0H.H NNH.H 00.mH 00.0 mmm.0 me.H HNH.H 00.0H 00.0 000m 0mH.H 0HH.H 00.ma 00.m 0N ~0~.H 00H.H 0HH.H 00.0w 00.NH 00.0 0.mN 00H.H m0H.H 00H.H 00.0H 00.HH 00.m 0m00.0 00H.H HOH.H 00H.H 00.0H m0.0H 0m.~ 00.0w 0nH.H 0MH.H m00.H No.0H 00.0 00.~ mN.H: 0000000000000000000000 mzoz 0n.m~.m H Nmm APPENDIX III APPLICATION OF COMPUTER PROGRAM KINFIT AND SUBROUTINE EQUATION FOR THE DETERMINATION OF COMPLEXATION CONSTANTS APPLICATION OF COMPUTER PROGRAM KINFIT AND SUBROUTINE EQUATION FOR THE DETERMINATION OF COMPLEXATION CONSTANTS The computer was again employed in fitting voltage versus volume data to theoretical equations in calculation of the overall monensin complexation constant. The KINFIT program was used in a manner analogous to that explained in Appendix I. Additional information necessary to fit the complexometric titration data is included here as well as listings of the two SUBROUTINE EQNS used. As mentioned previously, equation (7) + 2 (7) K 3 [H 1 + + (ca — [H ])(CNa+ — [H 1) is easily solved for hydrogen ion concentration in the quadratic below. 2 -(1 0> 00 000 20000200 20H0<00200200 02<00H0 000 20000200 00000> H 00>00000 000 H0000 0000HH 2H 00000> 02<00H0 000 HHvxx 000 0000 000 ~n0<>02 02000200 00H00HHH000 HH<:00>0 000 2000 0000> 20 000000020 > 0> 00 000 HH00 000 02202020 000 ~n02002 H0002<0000 2H 200000000000 2H020200000<0000 20.000000 00H00 H0u00<00 00u00<0H 002H0200 000H .HH.0.0.0.~0 00 00 0200002 0000 H0H000200 .20H000 .20H00 .000 0H000x.00~00<>0H0 .H00000 .HooHvHN .HH~.0~0000> .HH~.0~00 .200HO0H 0 .200H000 .200H0000 .00000 .000H.00000 .20000 .200H.000 20H0200H0 00200.0002.000>.00.0.000.0.0.00.0.000.H<>0H~ 0.00.HN.0.00.00.000.HH00.00000.0x.000H.000.0<.H.0000.x03H .0000H.0.0.02002.0<>02.0002.002H0.000.03H.00000.00<0H.02000 200000 200 02H0000000 000000 UL) 0000000000000000000000 .004 H0H02H000<0H A0H0200.0H0200000<00< 155 020 200000 00200200 200000 0020 0200 200000 00200200 200000 00000-00n00000 0000+0A00000000000000000000200vuu00 0000000 00000000 000 00 000000 00 A000000.0-00««0000000+0u000000.000:0u00 2000000200200 200 0 000 00 00 0000000000vuu0 0 02 0000 0000 000 00 0 Auu0ANVDV+Am0ANVDvum 0000 0 000 00 02000000000 000 00 0 0000 n 0.0u0 0000 0000000 0 000 00 02000000000 000 00 0 0000000 0000000000 00000000 000 00 20000000 00000000000 000 20 A20000200+00000000000002000000000u0 000000 0200000 0000 00 2000000200200 200 00000 000 00 0 000000200+000000\00000200u00 000000 0200000 0000 00 2000000200200 0000 000 00 00 MDZHHZOU N 156 OOOOOOOOOO HHHHHHHHHH mH0.H aHO.H oao.H 0N0.H mm0.a m~0.H 0m0.H 000.H 0m0.H 000.H 00:00.0 00:00.0 00:00.0 00:00.0 00:00.0 00:00.0 00:00.0 00:00.0 00:00.0 00:00.0 00000.0 000m0.0 000No.0 00~H0.0 00000.0 00m00.0 00m00.0 00N00.0 0~000.0 00000.0 OOOOOOOOOO HHHHHHHHHH oooooooooooooooooooooo mao.a mHO.H 0H0.H HNO.H mN0.H Nm0.H 000.H Hm0.H 000.H 000.0 oamm0.0 m0:moao.m qoqo m> mm was Amvemzoo m=q<> nm>mMmmo mme ANVxx mamqu zH mzsgo> ez mma Num<>oz UUUUUUU azo mus ANVD meqo> zH emmommHzH > m> mm map 000: mm< mzzozmza mme Nuxzsoz A«aozuHm .Aomvoa .Aoofivgu .AHN.o~Veom> .AHN.o~vm .AOOHVDH m .Aooavom .Aooavmom .Aqvxx .Ao00.00x93 .Aova .Aooa.0vx onmzmzHa Hmzoo.emoz.aom>.wa.w.sna.2.a.ea.e.amx.q<>oHN m.oa.q~.m.:m.om.mom.Haxn.mwsxm.xx.mweH.mmm.m<.H.amma.x930 .xoz.emoz.szHx.m<4.asH.mm0es.mm mm>mmmmo amp mm aqsomm mm Ao.m\ HzamHHH mo