University Microfilms International 300 N

INFORMATION TO USERS

This was produced from a copy of a document sent to us for microfilming. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the material submitted.

The following explanation of techniques is provided to help you understand markings or notations which may appear on this reproduction.

1. The sign or “target” for pages apparently lacking from the document photographed is “ Missing Page(s)” . If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure you of complete continuity.

2. When an image on the film is obliterated with a round black mark it is an indication that the film inspector noticed either blurred copy because of movement during exposure, or duplicate copy. Unless we meant to delete copyrighted materials that should not have been film ed, you will find a good image of the page in the adjacent frame. If copyrighted materials were deleted you will find a target note listing the pages in the adjacent frame.

3. When a map, drawing or chart, etc., is part of the material being photo graphed the photographer has followed a definite method in “sectioning” the material. It is customary to begin filming at the upper left hand corner of a large sheet and to continue from left to right in equal sections with small overlaps. If necessary, sectioning is continued again—beginning below the first row and continuing on until complete.

4. For any illustrations that cannot be reproduced satisfactorily by xerography, photographic prints can be purchased at additional cost and tipped into your xerographic copy. Requests can be made to our Dissertations Customer Services Department.

5. Some pages in any document may have indistinct print. In all cases we have filmed the best available copy.

University Microfilms International 300 N. ZEEB RD., ANN ARBOR, Ml 48106 8121868

Vargo, D onald L eslie

EFFECT OF PEROXIDE ON FREE AND ENZYME-BOUND 5- DEAZAFLAVIN

The Ohio State University Ph.D. 1981

University Microfilms International300 N. Zeeb Road. Ann Arbor, M I 48106 PLEASE NOTE:

In all cases this material has been filmed in the best possible way from the available copy. Problems encountered with this document have been identified here with a check markV .

1. Glossy photographs or pages______

2. Colored illustrations, paper or print______

3. Photographs with dark background______

4. Illustrations are poor copy______

5. Pages with black marks, not original copy______

6. Print shows through as there is text on both sides of page______

7. Indistinct, broken or small print on several pages

8. Print exceeds margin requirements______

9. Tightly bound copy with print lost in spine______

10. Computer printout pages with indistinct print______

11. Page(s)______lacking when material received, and not available from school or author.

12. Page(s)______seem to be missing in numbering only as text follows.

13. Two pages numbered______. Text follows.

14. Curling and wrinkled pages______

15. Other______

University Microfilms International EFFECT OF PEROXIDE ON

FREE AND ENZYME-BOUND 5-DEAZAFLAVIN

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate

School of The Ohio State University

Donald Leslie Vargo, B.A.

* * * * *

The Ohio State University

1981

Reading Committee: Approved By

Dr. M.S. Jorns

Dr. M.H. KLapper

Dr. R.M. Mayer Adviser Department of Chemistry To my wife, Agnes.

ii ACKNOWLEDGMENTS

My warmest thanks and deepest appreciation to my adviser, Dr* Marilyn Jorns, who suggested this problem and whose guidance and council were of inestimable value.

To my colleague, Mr. Alex Pokora, whose parallel research with oxynitrilase has been often referred to in this text,

I owe a special note of thanks. I am also indebted to Mr.

Kim Calvo for performing nonlinear least squares evaluation of my data, Dr. Charles Cottrell for Fourier transform NMR spectroscopy, Mr. Dick Weisenberger for mass spectral anal yses, Mrs. Sylvia Dahl for isolating glycolate oxidase, and

Dr. Vincent Massey for his gift of flavodoxin. I also wish to acknowledge the financial support for this research pro vided by a grant from the National Institutes of Health

(GM 22662).

iii VITA

January 3, 1949*•*•...... Born - Canonsburg, Pennsylvania

1972...... • •*...... B.A., California State College, California, Pennsylvania

1976-1979* ••• ...... Teaching Associate, Department of Chemistry, The Ohio State Univer sity, Columbus, Ohio

1979-1981 •••••• *...... Research Associate, Department of Chemistry, The Ohio State Univer sity, Columbus, Ohio

FIELD OF STUDY

Biochemistry TABLE OF CONTENTS

Page ID EL X C ATX ON ©ooo©©©©©®©©©®®®®®©©©©®®©®®®©®®®®®*®®©©©©®©* XX

ACKNOWLEDGMENTS ooo®®®®©®©®©©©®®®®®©©©®®®®®©©©©©©©©*©** XXX

VXTAa9ooo9OO90O9Qoeooocooo00eoo««oo«eo0tt*eoeftQoooaeeoa i V

L I ST OF TABLES ©oooo©®®©®®©©®®©®©®©®©®®©©®®©©©0©0©®®©©0 VXX

L 1ST OF FIGURESQooooe®©©®©©©©©©©®©©©®©®*©®©®©0©©®®©®®© VXXX

Chapter

X o INTRODUCTION ©©©ooo®®®©®®*©®*®*®©®®®®®©®®®®®©®®© 1 A. Deazaflavin as an analogue for flavin«0oo 1 B® Oxygen intermediates of flavin and deaZaflaVXno0©0©0®©©0®«®®®©0©©®©®#0®©®©®® 11

11 o MAT ERIALS AND MiiiTHODS ©ooooooo®®®®®®®®®®©®©®©®®© 17 A« Reagents and solutions©oooo©©©®®®®©®©®©©® 17 1 © Chemicals and supplierso®©®®©©©®®©®®© 17 2© Reagent purification and solutions,,®* 18

Bo Synthesis© 0 ©®©©©®©©®®©©®©©®©©©®©©©©©©®©©® 20 1 © lO-Methyl-5-deazaisoalloxazine® © ® * * © © 20 2® 3 ,10-Dimethyl~5“deazaisoalloxazine© 0® 22 3 o 5-Beazariboflavino©o®©®©®®®®®©©©©©©®® 2if if« if a, 5-Epoxy- 10-me thyl-5-deazaisoalloxazine©e®e©®©©9©«©®®oo©o©©®©©©®®© 28 5® if a, 3-Epoxy-3, IO-dimethyl-5-deaza- xsoalloxazineo©o©®®®®®®®©©®®®®®®®®®®® 29 6 ® ifa,5“Epoxy-5-deazariboflavin®...... 30 C© Preparation of glycolate oxidase and flavodoxin reconstituted with 5-deazaFMN0 31 1 © General procedure© ©eo®®®®®©®®®®®®©®®® 31 2o Glycolate oxidase© ©©ooo®®®®®®®®®®©*©© 31 3• Flavodoxxnoa©®®®®©®®©©®©®®®®®©©®®®©©® 32 Do Instruments© © ®© ©...... 33

E© Statistics© •©•••©••••©•••••••©•©•••a®®®©® 3 if

v TABLE OF CONTENTS (CON*T)

Page III. RESULTS AND DISCUSSION...... 35 A. The identification of a 4a»5-*epoxy- 5-deazaflavin...... 35 B. Reaction of 4as5-epoxy-5*-deazaisoalloxazine with iodide . • • • 61 C. Kinetic studies of the epoxidation of 3.10-dimethyl-5-deazaisoalloxazine with HOOH m aqueous h u f f o r s « 66 D. The epoxidation of 5~deazaisoalloxazine with HOOH in carbonate buffer...... • 75 E. Kinetic studies of the epoxidation of 3.10-dimethyl-5-deazaisoalloxazine with ra-chloroperoxybenzoic acid in aqueous buffersc...... 31 F. Kinetic studies of the epoxidation of 3»IO-dimethyl-5-deazaisoalloxazine with m-chloroperoxybenzoic acid in chloroform. 87 G. Studies with 5-deazariboflavin...... 90 H. Relationship of 4a} 5-epoxide to deazaFAD-X.ooeeoo.eo«eoo0 ooe....e.eo..o.o 101 I. Studies with Glycolate Oxidase...... 104 1. Reaction of 5-deazaFMN glycolate oxidase with HOOH.o...... 104 2. Reaction of 5-deazaFMN glycolate oxidase with m-chloroperoxybenzoic acid....ooo...... 106 3. Characterizing the first intermediate 108 4. Characterizing the second inter mediate...... 90. 0 ...... 0 O. 1 13 5. Characterization of the product...... 116 6. Kinetics of the reaction of EdeazaFMN with H O O H . 116 7. Theoretical absorption spectra... 121 J. Studies with Flavodoxin...... 127 LIST OF REFERENCES...... 132

vi LIST OF TABLES

Table Page

1. Reactivity of the if a, 5-Epoxides of 1O-Methyl-5-deazaisoalloxazine (III), 3.10-Dimethyl-5-deazaisoalloxazine (IV), and 5-deazariboflavin (VI) with Iodide and Thioxane in Methanol at 25 C„.... 65

2, Effect of pH on the Epoxidation of 3.10-Dimethyl-5-deazaisoalloxazine with Hydrogen Peroxideo..o..0 oo....o.o..o...oo.. 74

3. Effect of pH on the Epoxidation of 3,10-Dimethyl-5-deazaisoalloxazine with m-Chloroperoxybenzoic acid...... 8if

if. Epoxidation of 3, IO-Dimethyl-5-deazaisoalloxazine with m-Chloroperoxyben- zoic acid in Chloroform...... 88

5. Calculated composition of the reaction mixture during the recording of curve 2, Figure 3 5...... 124 6. Calculated composition of the reaction mixture during the recording of curve 3, Figure 35...... •...... 125

vii LIST OF FIGURES

Figure Page

1. Reaction of 10-methyl-5-deazaisoalloxazine with HOOH in acetonitrile...... 36

2. Reaction of 3 910-dimethyl-5-deazaisoalloxazine with HOOH in acetonitrile...... oo.o 37

3. Titration of 3,IO-dimethyl-5-deazaisoalloxazine With SUlfite. o o o a e . o e o o . e . e . o e o o o o . . o o . o . o . ooo 4^

If. Reaction of 10-methyl-5-deazaisoalloxazine with cyanide.oo.eoo.o.ooo..00.0.00.00.00.000.. 41

5. Reaction of IO-methyl-5-deazaisoalloxazine with t-butyl hydroperoxide in acetonitrile.... 43

6. Thin layer chromatography of the product formed with 10-methyl»5“deazaisoalloxazine and t-butyl hydroperoxide..o .0.0000.0.00.00.....o.oeooo... 46

7. Reaction of 4a»5-epoxy-3,10-dimethyl-5-deazaisoalloxazine tlV) with morpholine...... o.. 51

8. Reaction of 3*IO-dimethyl-5-deazaisoalloxazine with m-chloroperoxybenzoic acid in chloroform. 52

9. Thin layer chromatography of the products pre pared by reacting 3 SIO-dimethyl-5-deazaisoalloxazine with alkaline hydrogen peroxide and with m-chloroperoxybenzoic acid...... 54

10. NMR spectra of 4a.5-epoxy-3,10-dimethyl-5-deazaisoalloxazine (IV; and 3»10-dimethyl-5-deazaisoalloxazine (II) m GUCl^...... ooeoo.o.eoo.. 5 5

11. Infrared spectra of 4 a»5-epoxy-3,10-dimethyl- 5-deazaisoalloxazine (IV) and its parent com pound (II)... .OOO. 000...... 00000.00.... 56 12. Mass spectra of 4a«5-epoxy-3s10-dimethyl-5~deazaisoalloxazine (IV) and its parent compound

(II). . 0 0 . 0 .0 0 9 0. 0 0 0. 0 .0 0 ...... 0 0 0 0. 0 0 0...... 60 viii LIST OF FIGURES (CON'T)

Figure Page

13. Reaction of 4a,5-epoxy-3,10-dimethyl-5-deaza- lsoalloxazine with xodide..o..e.ee.o.«eoe.o.. 63

14. Reaction of hydrogen peroxide with thioxane.. 68

15. The first-order plot for the reaction of thioxane with hydrogen peroxide in methanol at 25° C q . . . . . 00000.0.0 000.09000. 0 0 0 . 0 0 ...... 68

16 • Reaction of 1O-methyl-5-deazaisoalloxazine with HOOH. ooo... .0 000.0000. ...ooo ...... 0.00 • 7^

17. Plot of kobs versus hydrogen peroxide concen tration for the reaction of 3,10-dimethyl- 5-deazaisoalloxazine with HOOH in 1«0 M sodium glycine buffer, pH 10o1 at 25° Co.o...... 72 18. Plot of the apparent second-order rate con stant as a function of pH for the epoxidation of 3,1O-dimethyl-5“deazaisoalloxazine with HOOH in aqueous solution....o.ov.ooeooooe.o.o 76

19. Reaction of 1O-methyl-5-deazaisoalloxazine with HOOH in 0 o01 M sodium carbonate buffer, pH 9.0 at 29 C. OOOO. 0000000000*0® OOO© 00000.0 77 2 0 . Reaction of 10-methyl-5"deazaisoalloxazine with HOOH in 1.0 M sodium carbonate buffer, pH 9.Oooo... 000.00 00. 0..0.0.00.0...... 0... 79 21. First-order plots for the reaction of 10- methyl-5-deazaisoalloxazine with HOOH in 1.0 M sodium carbonate buffer, pH 9.0 at 2 5^ C.ee. o.ooooooeooeoooo. 00000000000.ooo. 000 80

2 2 . Effect of the concentration of m-chloroperoxybenzoic acid on the pseudo-first-order rate constant observed for the reaction with 10- methyl-5-deazaisoalloxazine in 0.10 M sodium phosphate buffer, pH 7.0 at 25° C...... 85

ix LIST OF FIGURES (CON’T)

Figure Page

23. Plot of the apparent second-order rate con stant as a function of pH for the reaction between 3 »10-dimethyl-5-deazaisoalloxazine and m-chloroperoxybenzoic acid in 0.10 M sodium pyrophosphate buffers at 25° C...... 86 2/f. Plot of the observed rate constant against the peracid concentration for the reaction of 3 ?IO-dimethyl-5-deazaisoalloxazine with m-chloroperoxybenzoic acid in chloroform at 25 C...... e.0.0...... 89

25. Plot of -log kobs against -log Pperacid] for the reaction of 3 ,10-dimethyl-5-deazaisoalloxazine with m-chloroperoxybenzoic acid m chloroform at 25^ C...«o..o..oo....oe.o 00. 91

26. Plot of kobs/ [peracid] versus [peracid] for the epoxidation of -3 „ 10-dimethyl-5-deazaisoalloxazine with m-chloroperoxybenzoic acid in chloroform at 25° C...... 92

27. Reaction of 5-deazariboflavin with HOOH...... 93 28. Reaction of the diorthoformate derivative of 5-deazariboflavin with HOOH...... 95

29. Reaction of 5-deazariboflavin with m-chloroperoxybenzoic acxd. ....o...... 96

30. Reaction of the diorthoformate derivative of 5-deazariboflavin with m-chloroperoxybenzoic acid....o.e.e.ao»o.eo..0...e..e.o.eoa 97

31. Decomposition of ^a,5-epoxy-5-deazariboflavin m aqueous solutxon...... e.o.o..... 98

32. Reaction of /*a,5-epoxy-5-deazariboflavin with

X O d i d e o..a..o.ooofioft.ooo..9 oeooooo.e.ooo.oo.ft 99

33. Decomposition of *ta,5-epoxy-3»10-dimethyl- 5-deazaisoalloxazine (IV) in aqueous solution 103 x LIST OF FIGURES (CON'T)

Figure Page

34. Denaturation of 5-deazaFMN-reconstituted glycolate oxidase in 2% sodium dodecyl sulfate...... o...... 103

35. Reaction of 5-deazaFMN bound to glycolate oxidase with HOOH©...... ooo...... 107

36. Reaction of iodide with the first inter mediate formed with HOOH and 5-deazaFMN bound to glycolate oxidase...... 109

37. Reaction of 5-deazaFMN bound to glycolate oxidase with triiodideo.o.o...... 0.0.0.00 111

38. Effect of sodium dodecyl sulfate on the first intermediate formed with 5-deazaFMN bound to glycolate oxidase and HOOH, 0000*0090 112

39. Monitoring of during the initial reaction of 5-deazaFMN bound to glycolate oxidase in 0.1 M sodium phosphate buffer, pH 7.0, containing 0.3 mM EDTA at 15° C With 3 tllM HOOH, eeeoooooooooeoooooooeoooeooooo 119

bO. Serai-logrithmic plot of between 5-480 min after the addition of 3 mM HOOH to 5-deazaFMN bound to glycolate oxidase in 0.1 M sodium phosphate buffer, pH 7.0, con taining 0 o3 mM EDTA at 15^ C©...©...... 120

41. Theoretical spectra of the coenzyme species formed during the reaction of 5-deazaFMN bound to glycolate oxidase and 3 mM HOOH in 0.1 M sodium phosphate buffer, pH 7.0, con taining 0.3 mM EDTA at 15^ G...... 128

b2. Spectrum of 5-deazaFMN bound to flavodoxin in 0.1 M sodium pyrophosphate buffer, pH 7.5* containing 0.3 mM EDTA...... 130

xi I. Introduction

A. Deazaflavin aa an analogue for flavin

My research focuses on the reaction of hydrogen per oxide and other peroxides with free 5-deazaflavin and with 5-deazaflavin bound to two flavoproteins, glycolate oxidase and flavodoxin« 5-Deazaflavin derivatives were first synthesized in

1970 by 05Brien et al« (l) as analogues for the biologi cally important oxidation-reduction coenzymes, FMN (fla vin mononucleotide) and FAD (flavin adenine dinucleotide)

(Structure l)«,

R R

FLAVIN 5-DEAZAFLAVIN

FWW, 5-dFMN Q = r.u NH.

FAD, 5-dFAD R= CH^CHOH^-CHg-O-^-O-P-O-CHg

HO OH

Structure 1 2

The O'Brien synthesis stimulated numerous model studies and studies in which 5-deazaflavin was used as a substi tute for normal flavin to probe the mechanism of reactions catalyzed by flavoproteins# Although originally intro duced as artificial flavin analogues, 5-deazaflavin derivatives have recently been discovered as naturally occurring cofactors in various bacteria# Eirich et al,

(2) isolated a naturally occurring 8-hydroxy-5-deazaflavin derivative, factor 1*20 (Structure 2), from the methane producing bacterium, Methanobacterium bryantii.

HO. OH HO O OH OH H

Structure 2

Methanogens carry out the 8-electron reduction of carbon

dioxide to methane (Equation l)»

2fH2 + C02 — > 2H£0 + CH^ (1)

These bacteria contain a hydrogenase-transhydrogenase

system which transfers electrons from H2 to NADP using

F 1+20 as a coenzyme# F 1+20 has also been found in other methanogens (3)® A molecule, whose chromophoric portion Is identical with that of F 4-20, has been found in the prokaryotic actinomycete, Streptomyces griseus (U)e This coenzyme,

SF 420, could be substituted in both the hydrogenase and the NADP-linked oxidoreductase systems of M® bryantii with complete activity in both systems® The chromophore is a cofactor in the DNA photoreactivating enzyme from

S. griseus (5), indicating that, besides functioning in electron transfer reactions, this coenzyme may also be photochemically active in biological systems® Briistlein and Bruice (6) used 5-deazaflavin in model reaction studies to obtain evidence regarding the mechan ism of hydrogen transfer from NADH0 In these studies it was found that reduction of 5-deazaflavin involved direct transfer of hydrogen from NADH to the 5 position of re duced 5-deazaflavin (Scheme 1)®

CONK S^CONH,

Scheme 1

It is pertinent to note that similar studies are not fea sible with normal flavin owing to solvent exchange with hydrogen attached to nitrogen at the 5 position of normal reduced flavin® These studies prompted the use of 5-de azaflavin to study th© mechanism of hydrogen transfer in 4

reactions catalyzed by flavoproteins (7- 16).

As noted by Walsh (17), various observations suggest ed that 5-deazaflavin was a good structural analogue for normal flavin in flavoproteins:

a. 5-deazaflavin is tightly bound to every apo-

enzyme tested, b. the spectral changes observed upon binding of

5-deazaflavin are similar to those observed

with the natural coenzyme,

c. 5-deazaflavin-reconstituted enzymes form com

plexes with substrate analogues, and

d. the spectral pertubation observed upon binding

substrate analogues, and the dissociation con

stants observed for the complexes, are similar

to those observed for native enzyme,,

Studies with numerous 5-deazaflavin-reconstituted enzymes

(7-17) showed that specific substrates would reduce the artifical coenzyme to the 1,5-dihydro form. This is illustrated for the reaction observed with glycolate oxi dase (12) (Scheme 2)„

+ R-C-CO, i ‘ + R-C-C02 OH

Scheme 2 In all cases the enzymic reactions were found to involve

transfer of substrate hydrogen to the 5 position of the

reduced 5”de®zaflavins similar to that observed in model

reaction studies (6sl8s19)o Many of these enzymic re

actions were found to be equilibrium reactions® Addition

of oxidized substrate (e®g® R-C-COp) resulted in the re- 0 * oxidation of the reduced enzyme, accompanied by the trans

fer of the tritium label from the reduced coenzyme to the OH substrate (R-9-CCC)D now formed as the product in the re- T * action® These studies showed that the hydrogen transfer

red from substrate during coenzyme reduction was the same

hydrogen removed from the reduced coenzyme during reoxi

dation® The results provide evidence for stereospecific

hydrogen transfer at the C-5 position of the coenzyme9

similar to that observed at the C-i* position with pyridine

nucleotides (20)® The rates of substrate reduction with

5-deazaflavin-reconstituted enzymes are considerably

slower (lO^-lCT2 ) as compared with the rates observed

with the corresponding native enzymes® This has frequent ly been attributed to the fact that the reduction poten

tial of 5-deazaflavin (-311 mV for 5-deazariboflavin) is

considerably lower than that observed for normal flavin

(-208 mV for riboflavin) (21)® The very slow rates ob

served for substrate reduction with 5-deazaflavin-reconstituted enzymes prompted the use of 5-deazaflavin in

studies to evaluate the role of flavin in oxynitrilase 6 from almonds. Oxynitrilase is an unusual flavoprotein (l mol FAD per molecular weight of 75,000) since its only known function is to catalyze the reversible condensation of cyanide with various aldehydes to form D-ot-hydroxyni- triles (22) (Scheme 3).

HCN + R$H v x R-8-H R = for mandelo- Aw o 5 nitrile

Scheme 3

Although the catalytic reaction does not involve a net

oxidation-reduction, the enzyme exhibits many properties

characteristic of flavoprotein oxidases as opposed to

other classes of flavoproteins (e.g. dehydrogenases).

Similar to the flavoprotein oxidases, chemically reduced

oxynitrilase is oxidized rapidly by oxygen to form oxi

dized enzyme plus HOOH (23). One-electron reduction of

oxynitrilase results in the formation of a stable red

anionic semiquinone, also characteristic of oxidase en

zymes. In addition, flavin bound to oxidase enzymes or

to oxynitrilase will react with sulfite to form a rever

sible covalent complex (Scheme if). R R H

0 Scheme if S O * 0 7

This reaction, first observed with glucose oxidase (2if) is

oxygen-independent and is characterized by the bleaching

of the visible absorption spectrum of enzyme-bound flavin,

accompanied by the formation of a new absorption band around 330 nm0 The spectrum of the enzyme-sulfite complex is similar to, but not identical with, that of reduced en

zyme „ Similar complexes are formed with free flavin, but

are much less stable than the complexes formed with flavin

bound to flavoprotein oxidases (23)©

Removal of FAD from oxynitrilase results in a cata-

lytically inactive apoprotein which is extremely labile

as compared with native enzyme (25)© Full activity is

restored upon reconstitution with FAD0 The results show that FAD is required for both activity and protein stabil

ization, Other studies showed that the flavin and cata

lytic sites are near each other and raised the possibil

ity that FAD might directly participate in catalysis.

The latter was taken to include any function for the co

enzyme which would depend on its redox properties. In

order to evaluate this possibility, FAD modifications

were selected which would alter the redox character of

the coenzyme with minimal structural change.

Formation of a sulfite complex with FAD bound to

oxynitrilase resulted in complete loss of activity which

could be restored upon incubation with excess carbonyl

compound, which regenerates the original uncomplexed form of the enzyme. While the observed inactivation by sul fite could be regarded as a requirement for FAD partici pation in catalysis9 an alternative explaination is that the sulfite complex interferes with substrate binding at the nearby active site. The latter could be demonstrated using benzoate, a known competitive inhibitor of oxynitrilase. (Aldehyde and HCN both react nonenzymatically with free sulfite9 preventing their use in the experiment).

In the presence of benzoate the dissociation constant for the enzyme-sulfite complex is increased. The observed value is in good agreement with the value predicted by assuming that the binding of benzoate and sulfite are competitive. The absence of a ternary complex is also supported by the kinetics observed for conversion of the enzyme-sulfite complex to the benzoate complex. The nor mally rapid formation of the enzyme-benzoate complex is limited by the rate of release of the enzyme from the sulfite compleXo Formation of the semiquinone and the fully reduced forms of FAD likewise result in inactivation with 100% activity being restored after air oxidation.

However8 these results could not be unambiguously inter preted as evidence for direct participation of FAD in catalysis since reduction might induce structural changes sufficient to block catalysis at the nearby active site.

Similar to native enzyme9 oxynitrilase containing

5-deazaFAD in place of FAD is stereospecific for D-mandelonitrile and forms complexes with inhibitors.

However, substrate turnover for EdeazaFAD is much slower relative to the native enzyme: with mandelonitrile,

V '■"50 min*”1 (or about 0,1% of native oxynitrilase), max and with vanillin cyanohydrin, vmax = 2JfO min"1 (3% of native enzyme). The decrease in activity is similar to

that observed with other deazaflavoproteins where flavin functions as an oxidation-reduction catalyst (9-11s16,17,

19)o Similar to oxynitrilase, glyoxalate carboligase catalyzes a reaction which does not involve a net oxidation-reduction reaction (1?). Apoglyoxalate carbo ligase is catalytically inactive, but full activity could be restored by reconstituting with either FAD or 5°-deaza-

FAD, indicating that the redox properties of the coenzyme

were not important in catalysis with this enzyme. How ever, the results obtained with deazaFAD oxynitrilase suggested that the redox properties of the coenzyme might be important in catalysis, consistent with results obtain ed with other FAD modifications. But studies with native enzyme provided no evidence for oxidation-reduction inter action of the flavin with substrate. The apparent dis crepancy was resolved based on results obtained in studies on the reaction of deazaFAD oxynitrilase with HOOH (25)®

This reaction causes a dramatic increase in the catalytic activity of the preparation, which could be attributed to an irreversible modification of the coenzym© (deazaFAD-X), The absorption spectrum observed for peroxide-modified

enzyme (EdeazaFAD-X; A„_,_ 347 nm and 362 nm) is quite fflclX different from that observed for untreated enzyme (Edeaza-

FAD: A max 353 nm and 415 nm), which exhibits a spectrum

similar to that observed with 5-deazaflavin bound to other

proteins (9-13916,17)o Unlike EdeazaFAD, EdeazaFAD-X ex hibits turnover numbers similar to native enzymes 179900 min” 1 with mandelonitrile (47o4% of native enzyme)9 and 5,500 min” 1 with vanillin cyanohydrin (6806% of native oxynitrilase)® Heat denaturation releases the modified

coenzyme, indicating that deazaFAD-X is not covalently bound to the protein®

Unlike EdeazaFAD, EdeazaFAD-X does not form complex

es with either sulfite or cyanide, and is not reduced by dithionite or borohydride® These latter results Indicated that deazaFAD-X is apparently an oxidation-reduction in active derivative of deazaFAD® The fact that it can re place FAD in oxynitrilase without significantly affecting catalytic activity indicates that the redox properties of the coenzyme are unlikely to be significant in catalysis®

The results also indicate that the coenzyme is important as a structural component of the active site since cata lytic activity is very sensitive to small modifications in the coenzyme® The low activity observed with EdeazaFAD indicates that protein interaction near position 5 of the coenzyme is critical for catalytic activity® The similar 11 activity observed for EdeazaFAD-X and native enzyme sug gests that the peroxide modification might be located at or near the 5 position in deazaFAD-X* My studies on the reaction of free 5-deazaflavin derivatives with HOOH were initiated in the hope that the results would provide in sight regarding the structure of deazaFAD~X0

B. Oxygen intermediates of flavin and deazaflavin

The reaction of oxidized normal flavin with HOOH has been extensively investigated in connection with studies on intermediates in the reaction of reduced flavin with oxygen* Gibson and Hastings (26) found that the reoxida tion of 195-dihydroflavin exhibited saturation kinetics with respect to the oxygen concentration (Scheme 5)®

HOOH

Scheme 5

These studies provided the first indication that the re action might involve an oxygenated flavin intermediate*

While additional kinetic evidence was later provided by

Massey et al„ (27,28), these studies with simple reduced flavins failed to produce direct physical evidence for the postulated intermediate* However, the latter was obtained in studies on the reoxidation of certain substi tuted reduced flavin derivatives (29-31)® These findings 12 prompted studies of the reaction using the corresponding oxidized flavin derivatives with HOOH and other oxygen nucleophiles (e0 gc -OH and -OCH^) which might form rela tively stable adducts analogous or identical to the tran siently formed intermediate detected during the reoxida tion reactions (Equation 2)„

FI + HOOH FI HOOH F1H2 + 02 (2)

Consistent with this hypothesis, Kemal and Bruice (32.) synthesized a q.a-hydroperoxy flavin derivative via reaction of oxidized N(5)-alkyl flavin with HOOH, which exhibited spectral properties identical to an intermediate detected during the reoxidation of reduced N(5)-alkyl flavin with oxygen (Scheme 6)e

HOOH

Scheme 6

Other studies (33) showed that N(5)-blocked oxidized fla

vins react with hydroxide, which also adds at position k & t

to form a pseudobase, similar to the reaction observed with HOOH (Scheme 7)«

CH CH.

-h OH

O h OH CH Scheme 7 Spectral studies with several external monooxygenase enzymes have provided evidence for a Jfa-hydroperoxy flavin derivative as a catalytically significant intermediate in the hydroxylation reactions catalyzed by these enzymes

(3*f-3?)o Evidence for a similar intermediate has been obtained for the bioluminescent reaction catalyzed by bacterial luciferase (38,39)» Studies with luciferase indicate that the same intermediate can be formed by mix ing reduced enzyme with oxygen, or by reacting oxidized enzyme with HOOH (1+0) (Scheme 8)e

EFIH2+02-> EFIHOOH^EFI + HOOH RCHO V EFS +RC0 0 H +Hg0 +hu

Scheme 8

Flavin hydroperoxide derivatives have also been de tected in model studies during reoxidation of reduced fla vins bearing alkyl residues at the N(1) position or at the carbonyl oxygen at position 2, and in nucleophilic addition reactions with the corresponding oxidized flavin derivatives (29-31 ,Jf1)» In the case of a N(l)-blocked oxidized flavin, addition of HOOH occurs at position 10a, while nucleophilic attack occurs at the C-2 atom of a C-2 alkoxide flavin (Scheme 9)« -ifv5VV HOOH Cfe i ^ A r N-cH3 o « 0

CH3 u 9N3 00H f Y ^ V « 3^ f V Sy V 0CH3 3

Scheme 9

The results show that the position of the hydroperoxy moi ety varies depending on the structure of the flavin deriv ative used (U19U2), suggesting that the results obtained with various flavin derivatives might not be directly applicable to the reaction of HOOH with 5-deazaflavin.

Intermediates have not been detected during the very slow reaction of reduced 5-deazaflavin with oxygen.

Although addition of HOOH has not been observed at position 5 with normal flavins, this position is suscep tible towards nucleophilic attack with sulfite, as pre viously discussed. Compared to normal flavin, 5-deaza flavin derivatives are far more susceptible towards nu cleophilic attack at position 5 and will form reversible covalent 1,5-dihydro adducts with sulfite, cyanide and hydroxide (14 ) (Scheme 10). 15 Rz Rz H

H O A B B-1, X = S0~, CN, OH B-2, X = 00H (reaction is irreversible)

Scheme 10

Concurrent with my studies, Chan and Bruice (Jf3,V+) were also investigating the reaction of 5-deazaflavin with HOOH and proposed a 5-hydroperoxy structure for the product

(Scheme 10, B-2)e Results obtained in my studies and by

Chan and Bruice showed that the reaction of 5-deazaflavin with HOOH was irreversible, unlike the reversible nucleo philic addition reactions observed with 5-deazaflavin and sulfite, cyanide or hydroxide,, The products formed in the latter reactions exhibit similar visible absorption spectra, whereas significantly different spectral proper ties ©re observed for the product formed with HOOH* In addition, elemental analysis, NMR and mass spectral data reported by Chan and Bruice were not in accord with the proposed 5-hydroperoxy structure* As will be discussed, my results provide strong evidence that the product is not a 5-hydroperoxy derivative but, rather, a novel 5-epoxy derivative (Structure 3)« 16

CHS i •

tructure 3

Comparison of the properties of the 4a,5-®poxide with those observed for 5-deazaFAD-X from oxynitrilase indicated that the latter could not be an epoxide deriva tive,, This result was somewhat surprising since analo gous covalent adducts are formed when other nucleophiles

(e.g. sulfite and cyanide) react with free 5-deazaflavin or with 5-deazaFAD bound to oxynitrilase. However, stud

ies by Pokora and Jorns (if 5) provided strong evidence for an epoxide derivative as an intermediate in the oxynitril ase reaction. These studies showed that the conversion of the epoxide intermediate to EdeazaFAD-X was facilitated

by the protein moiety, since the reaction was inhibited

by benzoate derivatives which bind near the coenzyme site.

The results suggested that formation of 5-deazaflavin-X might be observed with peroxides and deazaflavins bound

to other enzymes, provided the environments at the co

enzyme sites were similar to oxynitrilase. This prompted my comparison studies on the reaction of peroxides with

5-deazaflavin bound to glycolate oxidase and to flavodoxin. II* Materials and Methods

A, Reagents and solutions

1. Chemicals and suppliers

Barbituric acid, tert-butyl hydroperoxide, 3»4-dimethylaniline, m-chloroperoxybenzoic acid, methyl urea, p-toluene sulfonic acid monohydrate, palladium (10%) on activated carbon, trimethylorthoformate, 2,4,6-trichloropyrimidine, morpholine, 1,^-thioxane and malonic acid were purchased from the Aldrich Chemical Company, Inc*

Trichloroacetic acid, acetonitrile, chloroform, dimethyl sulfoxide and triethylamine were obtained from the Fisher

Scientific Company® Methyl N-methyl anthranilate and p-toluenesulphonyl chloride were purchased from Matheson

Coleman & Bell Manufacturing Chemists® Catalase (30,000

Sigma units/mg), sodium dodecyl sulfate and D(-)ribose were obtained from the Sigma Chemical Company. N,N-Di- methylformamide was obtained from the Eastman Chemical

Company. Hydrogen peroxide (30%), methanol and anhydrous sodium sulfite were from Mallinckrodt, Inc. Hydrazine monohydrate was purchased from the J. T. Baker Chemical

Company. Disodium dihydrogen ethylene diaminetetraacetate

(EDTA) was obtained from the G. Frederick Smith Chemical

Company. Silica gel F-25^ thin layer chromatography

17 18 plates were purchased from Merck. 5-Deazariboflavin was converted to 5-deazaFMN by Mr. Alex Pokora of our labor atory following the enzymatic procedure of Spencer et al.

(15).

2. Reagent purification and solutions

m-Chloroperoxybenzoic acids as received from Aldrich

(technical gradec approximate 80-90%)9 was purified before use to remove a m-chlorobenzoic acid contaminant. Based on the method of Schwartz and Blumbergs (46) a suspension of 1.6 g of the peracid was stirred in 20 ml of 0.2 M sodium phosphate buffer9 pH 7.5 for 1 h at 25° 0. The crystals were filtered off9 washed with water and dried under vacuum to give 1.1 g of the purified peracid. (Pur ity was determined by iodide titration (47))»

1 g4",Thioxane received from Aldrich was purified by vacuum distillation (b.p. 50° C/21 torr (48)).

Buffers were prepared at 25° C by dissolving the ap propriate salt in deionized water and adjusting the pH to within ± 0.01 pH unit with HC1 or the appropriate base.

Dilute solutions (0.1 mM) of 5-deazaisoalloxazines used for spectral studies were prepared by diluting a concentrated stock solution into the reaction mixture.

Stock solutions (10 mM) of 10-methyl-5-deazaisoalloxazine9

4 a ,5-epoxy-10-methyl-5-deazaisoalloxazine 9 4a,5-epoxy-

3,IO-dimethyl-5-deazaisoalloxazine and 5-deazariboflavin were generally prepared in dimethyl sulfoxide. A less 19 concentrate (5 mM) stock solution was prepared with 3*10- dimethyl-5-deazaisoalloxazine since this compound is less soluble,, Chloroform or acetonitrile was substituted for dimethyl sulfoxide in the case of reactions with m-chloroperoxybenzoic acid since the later will oxidize dimethyl sulfoxide,,

Methanolic potassium iodide (0*5 M) was prepared by dissolving 0*83 S in 10*0 ml of solution* This solution was prepared daily*

2*0 M potassium cyanide9 pH 9*5 was prepared by dis solving T3o0 g of the salt in 60 ml of water cooled in an ice-salt bath in a fume hood* An electrode was introduced into the stirring solution and the pH was adjusted to 9® 5 by dropwise addition of 6 N HCl* The volume was brought to 100 ml and the solution used immediately* Excess cyanide was disposed of by adding sodium hydroxide and excess ferrous sulfate solution, stirring for 1 h, then

flushing the mixture down the drain with copious amounts

of water (49)® Thin layer chromatograms were generally developed in

the following solvent systems:

1» Chloroform-ethanol (4:1) 2* Benzene-methanol (65*35) 3® 2-Butanol-ethanol-water (7:2:1)

Spots on the thin layer chromatograms were identified

either by their fluorescence (5-deazaisoalloxazines and

5-deazariboflavin) or as dark spots which quenched the 20 ultraviolet indicator embedded in the silica gel when the plate was exposed to ultraviolet light*

B0 Synthesis 1 o 1O-Methyl-5-deazaisoalloxazine

Methyl N-methylanthranilate (Compound 1) was convert ed to o-methylaminobenzaldehyde (Compound 1+) by the pro cedure of Barlin (50)® The latter was condensed with barbituric acid to give 10-methyl-5-deazaisoalloxazine (I) according to O ’Brien et al0 (1) (Scheme 11)«

NHCH, NHCHt NHCH,

COCH, CNHNHo CNHNHSOP0CH, II ° ii £ ii a 0 O O

r ^ T ' N H C H 3

"CHO 4

Scheme 11

N-Methylanthranilate hydrazide (Compound 2)»

Methyl N-methylanthranilate (Compound 1: 80 g; 0,48 mol) was added to a 500 ml round-bottom flask containing a mixture of 80 ml (l»65 mol) of hydrazine monohydrate and 80 ml mol) of water0 A condenser was attached and the mixture was refluxed for 3 h® After cooling, the crystalline product was filtered off and recrystallized 21 from benzene* Yield: 47%; m*p* 135-139° C* (Lit* (50)

141-142° C)* N-Methylanthranilate p-toluenesulphonhydrazide (Compound 5)*

p-Toluenesulphonyl chloride (42*5 g; 0*22 mol) was added with stirring over a 30 min period to a beaker con taining 36*9 g (0*22 mol) of the hydrazide (Compound 2) in 110 ml of pyridine* The temperature was maintained at

0° C throughout the addition* The mixture was then set aside for 2 h at 25° 0 before pouring it into 650 ml of water* The colorless product was filtered off and recrys tallized from ethanol* Yield: 70%; m*p* 170-173° C, (Lit*

(50) 170-172° C)0 o-Methylaminobenzaldehyde (Compound 4)*

To a solution of 48*8 g (0*15 mol) of the p-toluenesulphonhydrazide (Compound 3) in 250 ml of ethylene glycol heated to 160° C in an oil bath was added, in one portion,

50 g (0*47 mol) of anhydrous sodium carbonate* After 1 min the mixture was poured into 1000 ml of ice water*

The oily product was extracted with ether and dried over

Na2 S0^* The ether extract was used directly in the next step* It was assumed that the yield was equal to that re ported in the literature (50), i*e*, 59%®

10-Meth.yl-5-deazaisoalloxazine (I)»

Barbituric acid (Compound 5s 11*34 g; 89 mmol) was dissolved in 190 ml of water and placed in a 250 ml round- bottom flask* A reflux condenser was attached and the solution was brought to a boil* Th® ether extract of o-methylaminobenzaldehyde (Compound **) was CAREFULLY added down the condenser* After refluxing for 3 h the yellow product was filtered off and recrystallized from dimethyl-

formamide and washed with ether* Yield (based on barbi turic acid): 72%; m*p* 300° C* (Lit* ( 5 0 359° C); tic

(system O : Rj, O 0**8; (system 2)s Rf 0*52; (system 3)s

0«**0; uv/vis (0*1 M NaHCO^, pH 7*0): A max (€), 253 (28,200)

265 sh (26,900), 32** (10,000), 390 (11,200); (acetonitrile)

316 (8,200), 398 (11,900); (methanol): 319 (9,000), 395

(12,200); (chloroform): 320 (8,300), **00 (11,600), **20 sh

(99300); nmr (DMSO-dg): **o0** ppm (s, 3H, NCH3 ), 7*50-8*22 ppm (m, **H, ArH), 9*00 ppm (s, 1H, CH), 11*08 ppm (s, 1H,

exchangeable, NH)*

2* 3»10-Dimethyl-5-de®zaisoalloxazine

The procedure was similar to that described for 10- methyl-5-*deazaisoalloxazine (I) except N-methylbarbituric acid (Compound 6) was substituted for barbituric acid in the final step* Compound 6 was synthesized according to the method of BLltz and Wittek (52) (Scheme 12)*

NHCH

4 6 n

Scheme 12 23

N-Methyl barbituric acid (Compound 6)e

A solution containing 15 g (0*20 mol) of methylurea and 2 5 g (0*23 mol) of malonic acid in 36 ml of glacial acetic acid was placed in a 250 ml round-bottom flask equipped with a condenser and a dropping funnel and heated to 60-70° C* While maintaining this temperature 30 ml

(0„32 mol) of acetic anhydride was added through the drop ping funnels with stirring9 over a 30 min period,, After

1 h ah additional 15 ml of acetic anhydride was added9 followed 1 h later by another 15 ml addition* Over the next 3 h the temperature was gradually raised to 90° C and held at this temperature for 3 h* Solvent was removed by evaporation under reduced pressure and the resulting syrup was solidified by the addition of 75 ml of ethanol and cooling to 0° C® Yields 43%; m.p, 132° C* (Lit* (52)

132° C)* 3 «1Q-Dimethyl-5-deazaisoalloxazine (II)*

By the method described in detail for lO-methyl-5-deazaisoalloxazine (I)9 N-methyl barbituric acid (Compound 6) was substituted for barbituric acid (Compound 5) in the final step of the synthesis and was refluxed with an equi- molar amount (assuming a 59% yield) of o-methylaminobenzaldehyde (Compound 4 ) 0 The product was filtered off and recrystallized from dimethylformamide and washed with ether* Yields 39%; m®p* >300° C0 (Lit* (51) 327° C); tic

(system 1)s R^ 0,68; (system 2 ) s Rf 0*63; (system 3): Rf 24 0,50; uv/vis (0«1 M Na^PO^, pH 7„0):Amax (e), 257 (24,800),

3 2 4 (11,400), 391 (12,000); (acetonitrile): 317 (9,700),

4 0 0 (12,400); (methanol): 320 (10,200), 397 (12,300);

(chloroform): 308 sh (9,700), 319 (9,800), 400 (11,600),

420 sh (9,200); nmr (CDCl^): 3.46, 4®17 ppm (s, each 3H,

NCH3 ), 7®50-8o00 ppm (m, 4H, ArH), 8*91 ppm (s, 1H, CH); ms: m/e 241 (M+)0

3® 5-Deazariboflavin

5-Deazariboflavin was prepared following the proce dure of Ashton et a lo (53)® 3 ,4-0im©thylaniline-N-D-ribopyranoside (Compound 8) was made via a modification of the method of Karrer and Meerwein (54)® Compound 8 was con densed with 6-chlorouracil (Compound 9), synthesized according to Cresswell and Wood (55), to form 6-(N-D-ribityl-3,4-xylidino)uracil (Compound 10). This compound was then converted to 5-deazariboflavin (V) (Scheme 13)®

CH (CHOHLChLOH

OH OH

CH2(CH0H)3CH20H CHCH^HCHCHCHa2CHCHCHCH2 CH2(CHOH)3CH2OH N\^Nn .O nn ^ n^ o

Scheme 13 25

5 »4-Dime thylaniline-N-D-ribo-pyranoside (Compound 7)«

D(~)R±bose (6®0g; 40 mmol) was mixed with 1 ml of water to form a syrup which was added to a 100 ml round- bottom flask containing 4® 8 g (40 mmol) of 3 94*“dimethylaniline in 50 ml of methanol® A condenser was attached and the mixture refluxed for 3 h» The solvent was evap orated under reduced pressure and the product was then washed with ether® Yield! 63%? m®p® 117-118° C dec®; uv

(methanol): Xmax ( O g 243 (l2g600)s 290 (19800); tic (sys tem 1)! 0®32.o The above procedure was repeated at a 6-fold larger scale and the product from this reaction was used in the next step®

N-D-Ribityl-3.>A-xylidine (Compound 8)® In a 2000 ml glass hydrogenation vessel containing

24®0 g (95 mmol) of 3 94“dimethylaniline-N-D-ribopyranoside

(Compound 7) in 950 ml of absolute methanol was added 5«0

g of 10% palladium on carbon® The vessel was then sealed

in its high pressure jacket and heated at 40-50° C for

3 h under 45 atmospheres of hydrogen® The solution was

filtered and the solvent was removed by evaporation under

reduced pressure® The residue was washed with water and

recrystallized from ethanol® Yield: 60%; m®p® 140-142° C®

6-Chlorouracil (Compound 9)®

2 s4 j6 -Trichloropyrimidine (25 g; 0®13 mol) was added

to 220 ml of 2®5 N NaOH in a 500 ml round-bottom flask® 26

A condenser was attached and the mixture refluxed for 16 h.

After cooling, the mixture was acidified to litmus with

12 N HC1® The product was filtered off and washed with water® Yields 78%; m,p» >300° C; uv (0®1 N NaOH): Amax

( O , 221 (7,200), 280 (9,300).

6-(N-D-Ribityl-3<>4°x;yiidino)uracil (Compound 10)®

In a 2 5 0 ml round-bottom flask equipped with a con denser 13®5 g (53 mmol) of N-D»ribityl-3,4-xylidin@ (Com pound 8) and *j.o0 g (27 mmol) of 6-chlorouracil (Compound

9) were added to 100 ml of water and refluxed for 16 h®

The solution was cooled in an ice bath and made alkaline to litmus with 2® 5 N NaOH, which produced a heavy precipi tate® The suspension was stirred at 0° C for 1 h, then filtered to remove unreacted 6-chlorouracil® The fil trate was acidified at 0° C with 6 N HC1 to pH 3, and the solvent then removed by evaporation under reduced pressure®

The residue was dissolved in 15 ml of boiling methanol, filtered, and the solvent removed by evaporation® The re sultant residue was recrystallized from water® Yieldg

32%; m®p® 172-I8*f° Co (Lit® (53) 183-185° C). uv (meth anol): Amax (e), 279 (21,200); tic (system l): Rf 0®11®

The next step in the synthesis was conducted using product collected by repeating the above procedure for a total of six times® Unreacted N-D-ribityl-3 ,*f-xylidine, recovered during workup of the reaction mixture, was used again when the procedure was repeated® 27

2.5 « 4 ^-Bjs-O-me thoxymethylene-1 -deoxy-1- (3... 4-dihydro~7^8~ dimethyl-2«,4°°dioxopyrimido f4«5-bl quinolin- 10(2H)yl)-D~rib- itol (Compound 11)®

In a 100 ml round-bottom flask equipped with a stirring bar and a condenser, a mixture of 3*54 S (9*7 mmol) of 6 -(N-D-ribityl-3,*t-xylidino)uracil (Compound 1 0 ), 3 0 ml (0 * 2 7 mol) of trimethyl orthoformate and 0 * 1 8 g (0 , 9 5 mmol) of p-toluenesulfonic acid monohydrate was refluxed under nitrogen for 4 days0 The lime-green crystals were

filtered off and washed with acetone* Yield: 43%; m®p®

260° C dec, (Lit® (53) 276-278° C); uv/vis (methanol)

Xmax (€), 227 (39*800), 265 (32,400), 332 (12,000), 402

(1 4 ,5 0 0 ); tic (system l): R^. 0 *6 6 ®

5 -Deazariboflavin (V)®

A suspension of 0®50 g (1®1 mmol) of Compound 11 in

11 ml of 1 N HCl was heated with stirring on a boiling water bath for 45 min® The suspension was then cooled, neutralized to litmus with a saturated solution of NaHCO^, filtered off and washed with water, acetone and ether®

Yield: 100%; m®p® 239-241° C dec® (Lit® (53) 289-292° C;

(51) 291° Cj (1 ) 286-288° C); uv/vis (methanol): Xmax («),

2 7 0 (2 6 ,3 0 0 ), 3 3 2 (1 0 ,0 0 0 ), 4 0 0 (1 1 ,2 0 0 ); (acetonitrile):

269 (2 7,2 0 0 ), 331 (9 ,7 0 0 ) 5 3 9 8 (1 1 ,4 0 0 ); (chloroform):

273 (1 8,3 0 0 ), 335 (9 ,3 0 0 ), 4 0 0 (1 0 ,5 0 0 ), 419 sh (9 ,4 0 0 );

tic (system 1 ): Rf 0®12; nmr (DMSO-dg): 2*33, 2*45 ppm

(s, each 3H, ArCH^), 3o63-5»14 ppm (m, 11H, ribityl-H), 28

7.83, 7.93 ppm (s, each 7H, ArH), 8.8l ppm (s, 7H, CH),

7 7.03 ppm (s, 1H, exchangeable, NH); ms: m/e 376 (M++ 7).

4. if a, 3- Epoxy- 7 O-me thyl -5-deazais oall oxazine (III)

(Scheme 74 )«

i m

Scheme 74

A suspension of 273 mg (0,94 mmol) of 7O-methyl-5- deazaisoalloxazine (I) in 75 ml of acetonitrile contain ing 0*75 ml (5.3 mmol) of triethylamine and 0.75 ml (6.7 mmol) of 30% hydrogen peroxide was stirred at 2 5 ° C for

3.5 h 0 The product was filtered off and washed with ethyl alcohol and ether® Yield: 30%; m.p. 238-240° C dec.; uv

(0.7 M Na^PO^, pH 7.0): Xmax (c), 3 2 9 (76,000); (aceto

nitrile): 322 (76,000); (methanol): 323 (76,400); tic

(system 7): Rf 0*63; (system 2): R^ 0 .5 2 ; (system 3): Rf

0.53; nmr (DMSO-dg): 3.54 ppm (s, 3H, NCH^), 5*20 ppm

(s, 7H, CH), 7*06-7.96 ppm (m, 4H, ArH), 7 7.30 ppm (s, 7H, exchangeable, NH). 29

5o 4&»5-,Epoxy-3, 1O-dimethyl-5 -deazaisoalloxazine (I¥)

(Scheme 1 5 )®

c h 3

Scheme 15

A suspension of 736 mg (3.05 mmol) of 3,10-dimethyl-

5 -deazaisoalloxazine (II) in 5 ml of chloroform contain ing 0*65 g (3o77 mmol) of m-chloroperoxybenzoic acid was stirred at 25° C for 150 min® The solvent was then evap orated and the residue dissolved in dimethylformamide at

60° C to form a near-saturated solution* After cooling to room temperatures ether was added to the dimethylform amide solution until the first sign of precipitation®

Further cooling in an ic© bath gave recrystallized pro duct® Yield: 20%; m®p® 219-220° C; tic (system 1): Rf

0 *6 8 ; (system 2 ): 0*63; (system 3 )* 0 *5 0 ; uv/vis

(0*1 M Na^PO^j pH 7*0): Xmax ( O , 331 (15,500); (ace to- nitrile): 3 2 4 (1 6 ,3 0 0 ); (methanol): 3 2 6 (1 6 ,1 0 0 ); nmr

(CDCI3 ): 3.22, 3.75 ppm (s, each 3H, NCH3 ), 5.18 ppm

(s, 1H, CH), 7*14-7.84 ppm (m, 4H, ArH); ms: m/e 257 (M+)*

Anal* Calcd for C ^ H ^ N ^ O ^ : C, 60*70; H, 4®31; N,

16.33® Found: C, 60*71; H, 4-37; N, 16.33. Alternately, 296 mg (1*23 mmol) of 3*10-dimethyl-5- deazaisoalloxazine (II) was dissolved in a solution con taining 7 ml of acetonitrile, 0*7 ml (5*0 mmol) of triethylamine and 0*7 ml (6*3 mmol) of 30% hydrogen peroxide*

After stirring for 15 min at 25° C the reaction was com plete and the product began to precipitate® The crystals were filtered off after cooling to -20° C and washed with water, ethanol and ether® Yield: 59%*

6* 4a,5-Epoxy-5-deazariboflavin (VI) (Scheme 1 6 )®

C H_(C H 0 H )_C Ho0H

Scheme 16

To a solution of 1*16 g (6*7 mmol) of m-chloroperoxybenzoic acid in 2*2 ml of dimethylformamide was added

37*53 mg (0*1 mmol) of 5-deazaribo flavin (V)* The mixture was vortexed for 1 min at 25° C* The reaction was monitored by the bleaching of the yellow color of the sol ution® After 3-4 min the pale yellow mixture was poured into 40 ml of ice-cold ether* The white precipitate was collected by centrifugation and washed with ether* (The epoxide is highly susceptible to moisture* Collecting the cold crystals on a funnel results in loss of product, possibly due to moisture condensation on the crystals)* 31

Yield: 58%; m.p. 178° C dec.; uv (methanol): Xmax (O,

336 (16,500).

C. Preparation of glycolate oxidase and flavodoxin recon

stituted with 5-deazaFMN.

1o General procedure

All solutions were prepared with deionized water.

Buffers for the glycolate oxidase preparation are sodium

phosphate, while potassium phosphate is used for the

flavodoxin preparation unless otherwise specified. All

buffers contain 0.3 mM EDTA and are used at 2-if° C unless

otherwise specified. Dialysis tubing was boiled in approx

imately 1 mM EDTA, pH 12 for 5 min and washed with copious

amounts of water.

2. Glycolate oxidase

Native glycolate oxidase from pig liver was prepared

in our laboratory by Mrs. Sylvia Dahl according to the

procedure of Schuman and Massey (56). Preparation of the

apoenzyme and its reconstitution with 5»deazaFMN were per

formed according to the method of Jorns and Hersh (H).

The preparation of the apoprotein involved dialysis of

native glycolate oxidase (0.50 ml; A ^ ® = ^.^8) versus

five changes of 250 ml of 3 M KBr in 0.1 M buffer, pH 6.3*

The excess KBr was removed by further dialysis against three changes of 0.01 M buffer, pH 7.0 (two changes with

250 ml of buffer and a final change with 500 ml of buffer). The apoenzyme was removed from the sac and centrifuged for

20 min at 43*000 g to remove a small precipitate of de natured enzyme® The clear supernatant was transferred to a teflon-stoppered microcuvette and the absorption spec trum was recorded ( A ^ O - 0.022)® The sample was mixed with 20 (ii of 5s7 mM aqueous 5“deazaFMN and the mixture incubated for 7 min at 37° 0® A second 20 fil addition was made and incubation continued for another 7 min®

After a third 20 fil 5-deazaFMN addition and 4 additional min of incubation the enzyme was dialyzed against three

25 ml changes of O J M buffer, pH 7®0® The 5-deazaFMN- reconstituted glycolate oxidase was removed from the sac, centrifuged as above, and the absorption spectrum recorded®

The 5“deazaFMN enzyme exhibited maxima at 400, 334 and 280 nm (A2®®/A^00 = 6®9)® The enzyme was stored at 0° C pro tected from light®

3® Flavodoxin

Native flavodoxin from Megasnhera elsdenii was a gift from Dr® Vincent Massey (University of Michigan)®

The apoprotein was prepared as described by Wassink and

Mayhew (57)® In this procedure 50 //I of a 55% solution of trichloroacetic acid was added to native flavodoxin

(2®50 mg) dissolved in 0®50 ml of 0®1 M buffer, pH 7.0.

After 5 min the mixture was centrifuged for 10 min at

10,000 go The yellow supernatant was removed and the white pellet suspended in 0.50 ml of a 5% trichloroacetic 33 acid solution containing 0«3 mM EDTA, After centrifuging as before, the white pellet was dissolved in a minimal volume (0o3 ml) of 0,1 M buffer, pH 7»0 and dialyzed against three 100“fold volumes of the same buffer.

The apoprotein was reconstituted with 5-deazaFMN by dialyzing it against four TO ml changes of 5.4 t*M 5-deaza-

FMN in 0,1 M buffer, pH 6,8, followed by three 10 ml changes in 0, 1 M buffer, pH 7,0, The reconstituted pro tein exhibited maxima at 402, 344 and 280 nm (A2^ / A ^ 2 =

5.7). D, Instruments

A Beckman Model 25 Spectrophotometer was used to ob tain UV/Visible spectra. Infrared spectra were taken on a Perkin-Elmer 467 Grating Infrared Spectrophotometer, A

Varian EM-360L (60 MHz) spectrophotometer was used for most NMR spectra, except in the case of 5-deazariboflavin and Its epoxide where spectra were recorded by Dr, Charles Cottrell using a Bruker HX-90 (90 MHz) spectrophotometer equipped for Fourier transform spectroscopy. Mass spec tral analyses were obtained by Mr, Dick Weisenberger, and elemental analyses were performed at Scandinavian Micro- analytical Laboratory, Denmark, Melting points were ob tained with a Thomas Hoover capillary melting point apparatus and corrected values are reported, pH measure ments below pH 10,7 were made with a Radiometer Model 26 pH meter equipped with a Radiometer GK 2301C glass-calomel combination electrode. Measurements above pH 10,7 were made using narrow range pH paper (pHydrion controls) from

Micro Essential Laboratory, Brooklyn, Centrifugations were made with a Sorvall RC-5 Superspeed Refrigerated

Centrifuge,

E, Statistics

Values accompanied by uncertainties (i.e., N ± n) re present the mean and sample standard deviation (s), re spectively. Computer curve fittings were performed by Mr.

Kim Calvo using the method of nonlinear least squares with subroutine programs GRADLS or FDERIV (69»76)« III. Results and Discussion

A. The identification of a i+a „5-epoxy-5-deazaflavin

The spectral course of the reaction of 1O-methyl-

5-deazaisoalloxazine (I) with HOOH in acetonitrile in the presence of triethylamine is shown in Figure 1. The re action is characterized by a bleaching of the absorption band of the starting compound at 398 run accompanied by a bathochromic shift of the band at 316 nm. No reaction is observed in the absence of triethylamine, indicating that the reaction requires a basic catalyst. A similar spec tral course is observed when the reaction is conducted using other organic solvents (e.g.9 dimethylfonn amide9 dimethyl sulfoxide and methanol) in place of acetonitrile or when 3 910-dimethyl-5“de®zaisoalloxazine (n) iS used in place of 10-methyl-5-deazaisoalloxazine (I) (Figure 2).

In the following discussion the compounds formed with HOOH and compounds I and II will be referred to as compounds

III and IVs respectively. The following results indicate that compounds III and IV are not in equilibrium with their respective parent compounds:

a. Neutralization of the triethylamine in the reaction

mixture at the time of maximum product formation

does not regenerate the parent compound. In a 1 ml

35 36

I.C

1.2

UJO z

3 0 0 4 0 0 nm

Figure 1* Reaction of 10-methyl-5-deazaisoalloxazine with HOOH in acetonitrile: Curve 1 is the initial spectrum of 10-methyl-5-deazaisoalloxazine (9*8 x 10“^ M) with triethylamine (0,33 M) at 25° Cj curves 2-6 were recorded 8, 20, 35* 60 and 120 min, respectively, after the addition of HOOH (8*7 x 10“2 M). The inset shows the first-order plot for the reaction at 398 nm. 37 1.6

too- 080- 0.60-

0 40

0.20 1.2 0.10 O.OB 0 0 6 Ui o z < CD o 0-8

0.4

0.0 300 400 500 nm

Figure 2* Reaction of 3,10-dimethyl-5-deazaisoalloxazine with H00H in acetonitrile: Curve 1 ia the initial spec trum of 3»IO-dimethyl-5-deazaisoalloxazine (1,0 x 10“^ M) with triethylamine (0.33 M) at 25° C; curves 2-5 were re corded 1 5 8 359 60 and 110 min, respectively, after the addition of H00H (8.7 x 10"2 M). The inset shows the first-order plot for the reaction. mixture containing 0.1 mM of compound I or II,

8.7 x 10~2 M of HOOH and 7.0 x 10”2 M of triethyl amine in acetonitrile, the formation of III or IV, respectively, is complete after about 6 h at 25° C.

Addition of 5 of glacial acetic acid (17-4 M) at this time did not cause a reversal of the spec tral course, while addition of acetic acid during the formation of III or IV abruptly ended the re action.

Removal of free HOOH does not reverse the reaction.

The addition of 1,170 Sigma units/ml of catalase to a reaction mixture of III or IV in acetonitrile

(see above) did not regenerate starting compound.

Addition of catalase during the formation of III or

IV stopped the reaction. (Further evidence to show that catalase is active under these conditions will be presented later).

Compounds III and IV, isolated from preparative- scale experiments as described in Methods, are stable when dissolved in various HOOH-free organic solvents (e.g., dimethylformamide, acetonitrile, dimethyl sulfoxide, methanol and chloroform). A relatively slow decomposition is observed upon ad dition of base such as triethylamine, i.e., the same condition required to form the product. This does not constitute a major problem during product 39 formation in organic solvents since the rate of pro

duct formation is generally much faster than its de

cay in all of the solvents examined® This is also

evident from Figures 1 and 2 where the isosbestic

point is maintained over the entire reaction®

Further incubation of the products under these con

ditions does result in a slow bleaching of the ab

sorption band near 325 nm« As will be discussed,

the same products can be formed in aqueous solution,

but the yields are generally much lower owing to the

fact that the decomposition of the products is com

petitive with the rate of their formation®

Previous studies (11*) with other nucleophiles (e®g®9 sulfite9 cyanide and hydroxide) showed that 5-deazaisoalloxazine is susceptible towards nucleophilic attack at position 5 yielding 1,5-dihydro derivatives (Scheme 10,

B-1, p« 1 5 )® This raised the possibility that an analo

gous 1,5-dihydro adduct was formed with HOOH (Scheme 10,

B-2, p® 15)® As illustrated in Figures 3 and k for the

reaction of 5-deazaisoalloxazine with sulfite and cyanide, these reactions also result in bleaching of the absorption band of the starting compound near 390 nm, but this is accompanied by a hypsochromic shift of the band near 320 nm, unlike the bathochromic shift observed for the per oxide reaction (Figures 1 and 2)® In aqueous solution compounds III and IV exhibit absorption maxima (329 and ZfO

1.0

UJ o z < CD cc o

0.5

300 400 500 nm

Figure 3. Titration of 3,lO-dimethyl-5-deazaisoalloxazine with sulfite: Curve 1 is the initial spectrum of 3, lO-dimethyl-5-deazaisoalloxazine (f*.0 x 10”^ M) in 0.1 M sodium phosphate buffer, pH 7*0 at 25° curves 2 and 3 were recorded immediately after adding sodium sulfite to concentrations of 3*^1 mM and 15*25 mM, re spectively. 41

0.50

od 0.25

0.00 300 4 00 500 nm Figure k* Reaction of 1O-methyl-5-deazaisoalloxazine with cyanide: Curve 1 is the initial spectrum of 10- methyl -5-deazaisoalloxazine (9.8 x 10-5 m) in 1.0 ml of methanol at 25° C; curve 2 was recorded immediately after the addition of 10 fil of 2 M potassium cyanide, pH 9.5j curve 3 was recorded after 3 min. No further changes were observed. U2

331 nm, respectively) which appear at longer wavelengths as compared with 1,5-dihydro adducts ( Amax <310 nm).

Furthermore, the reactions observed with sulfite, cyan ide and hydroxide are all equilibrium reactions, in con trast to the irreversible reaction observed with HOOH. These results suggested that the product formed with HOOH might not be a 7,5-dihydro adduct. The spectral course of the reaction for 10-methyl-

5-deazaisoalloxazine with t-butyl hydroperoxide in aceto nitrile is shown in Figure 5® Similar results are ob tained with 3 910-dimethyl-5-deazaisoalloxazine. As ob served with HOOH, a basic catalyst is also required for the reaction with t-butyl hydroperoxide. The reaction with t-butyl hydroperoxide is much slower than with HOOH owing, in part, to the lower pK© of the latter. Con sequently, higher concentrations of both t-butyl hydroper oxide and the basic catalyst are required in order to achieve comparable reaction rates. The reaction with t-butyl hydroperoxide remains isosbestic for about 7 0% of the reaction as judged by the residual absorbance near

ZfOO nm when the isosbestic point is lost (Curve k in Fig ure 5). The subsequent course of the reaction (Curves

5-8 in Figure 5) results in a further bleaching of the residual absorbance at **00 nm accompanied by a progressive loss of the absorption by the product near 325 nm. The decomposition of the product under the conditions required for the reaction observed with t-butyl hydroperoxide and 43 1.5 -

r\-3

r \ \ - 6

Hi o z 1-7 < CD a: o CD CD <

0.5

300 4 0 0 500 nm

Figure 5, Reaction of 10-methyl-5-deazaisoalloxazine (1,2 x 10-5 with t-butyl hydroperoxide (0,96 M) in acetonitrile: Curve 1 is the initial spectrum at 25° C 'I* curves 2-8 were recorded 10, 25, 50, 105, 180, 300 and 4 2 0 min, respectively, after adding triethylamine (0,36 M), kk the very high reagent concentrations seriously interfered with attempts to isolate pure product from this reaction.

However, thin layer chromatography studies in several solvent systems showed that the product formed ?/ith t-butyl hydroperoxide exhibits identical chromatographic properties with product III, formed with 10-methyl-5~deazaisoalloxazine and HOOH (Figure 6),

The high concentrations of t-butyl hydroperoxide used in these experiments raised the possibility that the re sults might be attributed to the presence of a small amount of HOOH as a contaminant in the t-butyl hydroper oxide, In order to evaluate this possibility it was necessary to develop a procedure which would result in the destruction of HOOH in solutions of acetonitrile. In order to monitor the latter reaction it was also necessary to develop an assay for HOOH in acetonitrile. Previous studies (58) showed that iodide is oxidized to iodine by

HOOH (Equation 3)®

HOOH + 2 1“ + 2 H+ I2 + 2 H20 (3) I2 + I" ^ I3 (*f)

The iodine reacts with excess iodide to form the highly absorbant triiodide ion (Equation *f) (€^^ = 25,000 in methanol (59))® The reaction of 0,1 mM HOOH with 0,1 M methanolic potassium iodide proceeds at a reasonable rate

^ 2 * * (k2 = 5®5 x 10 J M s ), However, an extremely slow Figure 6* Thin layer chromatography of the product formed with 10-methyl-5-deazaisoalloxazine and t-butyl hydroperoxide: Pure 10-methyl-5-deazaisoalloxazine is shown in spot A. The product isolated from a suspension —p reaction of 10-methyl-5-deazaisoalloxazine (5»2 x 10 M) with hydrogen peroxide (0,40 M) and triethylamine (0,32 M) in acetonitrile (see Methods) is shown in spot B, This product is contaminated with a small amount (8%) of the starting material, an estimate based on the absorbance of the sample at 398 nm, A reaction mixture containing 10-methyl-5-deazaisoalloxazine (9®4 x 10“^ M), t-butyl hydroperoxide (4*7 M) and triethylamine (0,34 M) in acetonitrile at 25° C was allowed to react for 62 min. At this point the solution contained 23% of the parent compound plus product which had reached a maximum con centration, An aliquot (50 jil) was then withdrawn for chromatography (Spot C), SOLVENT 1 SOLVENT 2 SOLVENT 3

O O O O O ' o o o O O

o o

ABC ABC A B C

Figure 6 hi

reaction was observed with 8.9 x 10“2 M HOOH in aceto nitrile,, Addition of 8.5 x 10"2 M glacial acetic acid to the acetonitrile reaction mixture resulted in a very- rapid reaction as judged by the increase in absorbance at

358 nm, attributable to the formation of triiodide. Al though catalase is not stable in acetonitrile, the

following results show that it can be used to destroy HOOH

in this solvent. Bubbling is observed briefly upon the

addition of catalase (1,170 Sigma units/ml) to a solution

of 8.9 x 10“2 M HOOH in acetonitrile. Formation of tri iodide is not observed upon subsequent addition of potas

sium iodide and acetic acid. In a separate experiment

catalase (1,170 Sigma units/ml) was added to a solution

of 9.8 x 10-5 io-methyl-5“deazaisoalloxazine and 87 mM

HOOH in acetonitrile. There was no change in the absorp

tion spectrum of the chromophore when 71 mM triethylamine

was added 1 min after catalase, indicating that formation

of compound III was prevented by catalase. In similar

studies it was found that catalase would rapidly destroy

HOOH in the presence of 0.97 M t-butyl hydroperoxide.

(Higher concentrations of t-butyl hydroperoxide did inter

fere and were avoided). Solutions of 0.1 mM 10-methyl-

5-deazaisoalloxazine in acetonitrile containing 0.97 M

t-butyl hydroperoxide were treated for 1 min with catalase

(1,170 Sigma units/ml) before adding 0.70 M triethylamine.

Catalase did not affect the reaction with t-butyl hydroperoxide as evidenced by comparing the initial veloc- ity of the reaction (measured by the decrease in A^95) with a control sample lacking catalase,. Addition of cat

alase to a reaction mixture containing 0®1 mM 10-methyl-

5-deazaisoalloxazine, O097 M t-butyl hydroperoxide and

87 mM HOOH, followed by the addition of 0„70 M triethyl amine resulted in a initial rate identical to that ob served for samples where HOOH was not included,, If cat alase was omitted from the reaction mixtures containing

HOOH and t-butyl hydroperoxide, a very rapid reaction was observed upon the addition of triethylamine, character istic of the much faster reaction of 10-methyl-5-deazaisoalloxazine with HOOH under these conditions„

The proceeding results indicated that the reaction

observed with t-butyl hydroperoxide is not due to a HOOH

contaminanto The fact that the same product can be formed

by reacting 1O-methyl-5-deazaisoalloxazine with either

HOOH or t-butyl hydroperoxide indicates that the product

cannot be a 5-hydroperoxy derivative as proposed by Chan

and Bruice (Scheme 10, B-2, p® 15)© Nucleophilic addition

of peroxide at position ifa (Structure if) is also excluded

by the results®

Structure if However, the results are compatible with a reaction re sulting in the formation of a novel 4a,5-epoxy derivative

(Scheme 1?)0

Scheme 1?

In Scheme 17* R00 is shown as the reactive species, con sistent with the requirement of a basic catalyst observed for these reactions,* Initial nucleophilic attack of R00”

at position 5 to form a 1,5-dihydro adduct as an intermed iate is consistent with the known susceptibility of 5-deazaflavins toward attack at this position,* Similar alka line epoxidation reactions with HOOH and t-butyl hydroper

oxide have previously been observed for compounds contain ing a carbon double bond conjugated to an unsaturated electron-withdrawing group (60-64), similar to the double bond between positions 4® and 5 in 5-deazaisoalloxazine

(Scheme 18),

o ii CH3 ,c -c h 3 — /» HOOH/OK / \ c h 3 h

Scheme 18

As detailed in Methods, the products formed by react ing either 10-methyl- or 3,10-dimethyl-5-deazaisoalloxazine with HOOH could be isolated. Results obtained by 50 NMR, mass spectral and elemental analysis (to b© present ed later) of these products are consistent with the pro posed epoxide structure,. However, since the latter would

represent the first example of a deazaflavin epoxide de rivative, it was desirable to obtain independent evidence for the structure© As previously discussed, attempts to

isolate preparative amounts of the product formed with

t-butyl hydroperoxide were not successful© Therefore,

two alternative approaches were attempted© The first of

these approaches, which did not prove useful, involved an

attempt to prepare a derivative of the product formed with

HOOH© In these experiments compound IV was reacted with

various amines (e©g©, morpholine, piperidine., aniline and

N-methylaniline) since epoxides are known to undergo nu-

cleophilic addition reactions© The spectral course for

the reaction of 8©3 3C 10”^ M compound IV in 1©15 M morpho

line in methanol is typical of the reactions observed with

the other nucleophiles (Figure 7)a A thin layer chroma

tography study of the reaction mixture after 21 h showed

that multiple products had formed© Unlike alkaline epoxidation reactions with HOOH and

t-butyl hydroperoxide, which are restricted to a rather

narrow class of unsaturated compounds, epoxidation of car

bon double bonds with organic peroxy acids is a general

reaction which has been observed with a much broader spec

trum of unsaturated compounds (65)© These reactions are 300 4 0 0 5 0 0 n m

Figure 7* Reaction of *fa,5-epoxy-3, IO-dimethyl-5-deaza isoalloxazine (IV) with morpholine? Curves 1-7 were re corded 0,5 min, 1, 3» 5* 7, 10 and 21 h, respectively, after adding morpholine (0®1 ml of 11.5 M morpholine) to the epoxide (0®1 mM) in 0.90 ml of methanol at 25° C 400 500 nm Figure 8. Reaction of 3,IO-dimethyl-5-deazaisoalloxazine with m-chloroperoxybenzoic acid in chloroforms Curve 1 is the initial spectrum of 3 910-dimethyl-5-deazaisoalloxazine (1*1 x 70“** M) in chloroform at 23° 0; curves 2-5 were re corded 0.5# k*59 12 and 30 min, respectively, after adding 0 »k2. M m-chloroperoxybenzoic acid. 53 frequently conducted with m-chloroperoxybenzoic acid be cause of its stability and commercial availability®

Therefore* studies were initiated to determine whether reaction of 5-deazaisoalloxazine with m-chloroperoxybenzoic acid would yield the same product as observed for the reaction with H00Ho Figure 8 shows the spectral course for the reaction observed with 3»1O-dimethyl-5- deazaisoalloxazine and 0*,^2 M m-chloroperoxybenzoic acid in chloroform, where a basic catalyst is not required®

It is apparent that the spectral course for this reaction is very similar to that observed for the reaction with

HOOH (Figure 2)® Thin layer chromatography studies in dicated that the product formed with 3 slO-dimethyl-5-deazaisoalloxazine and m-chloroperoxybenzoic acid was identi cal to the product (IV) formed with HOOH (Figure 9)®

Similar results were obtained in studies with 10-methyl-5- d eazaisoalloxazine ® The product formed with 3®10-dimethyl-5-deazaisoalloxazine and m-chloroperoxybenzoic acid was isolated from a preparative-scale reaction (as described in Methods) for direct comparison of its properties with the product formed with 3 9IO-dimethyl-5-deazaisoalloxazine and HOOH® The melting point of the purified product formed with the peracid (219-220° C) was identical to that of the product isolated from the reaction with HOOH® This value remained unchanged in a mixed melting point determination® The SOLVENT I SOLVENT 2 SOLVENT 3 ^

O O o o **• " *•» O 0 0 0 0 o

A B A B A B

Figure 9. Thin layer chromatography of the products prepared by reacting 3,10-dimethyl-5-deazaisoalloxazine with alkaline hydrogen peroxide (Spot A) and with m-chloroperoxybenzoic acid (Spot B). The products were prepared as described in Methods• A small amount of starting compound (indicated by dashed circles) was present as a contaminant in both samples. 55

CH,

Figure 10® NMR spectra of 4&95-epoxy-3,10-dimethyl-5-deazaisoalloxazine (IV) and 5» IO-dlmethyl-5-deazaisoalloX” azine (II) in CDCl^ obtained with a Varian EM-360 L (60 MHz) spectrophotometer® Peak "s" is due to the sol vent* 56 visible absorption spectra of both products were also identicale Figure 10 shows the proton NMR spectrum of

3 t10-dimethyl-5-deazaisoalloxazine and the spectrum of the product (IV) formed with H00Ho An identical spectrum was obtained for the product formed with peracid® The

NMR spectrum of the starting compound (II) exhibits a singlet at 8 ®91 ppm due to the hydrogen at C(5), a multiplet between 7°50-8000 ppm attributable to the if aro matic protons, and two singlets at 3®*f6 and if„17 ppm which are due to the methyl groups at N(10) and N(3)„ Compared with the NMR spectrum of the starting compound, the major difference in the NMR spectrum of the product (IV) is the position of the peak attributable to the hydrogen at C(5), which is shifted upfield and appears as a singlet at 5ol8 ppm, indicating that substitution has occurred at this position® A 5-hydroperoxy derivative, as proposed by

Chan and Bruice, should show peaks due to the exchangeable protons at N(1) and on the hydroperoxy moiety (Structure

5). r 2 ©

Structure 5

No evidence for these protons was obtained when the spec trum of the product was scanned from 0-20 ppm® Also, 57 the spectrum was not affected when 1^0 was added to the sample® Peaks due to exchangeable protons were also not

detected by Chan and Bruice who argued that their absence could be explained by exchange with H20 in their sample®

This explaination cannot account for the absence of these peaks in my sample since the spectrums recorded in an hydrous CDCl^g showed no evidence for a peak® The

infrared spectrum of the product did not show evidence

for an -OH group, which would have been expected in the case of a 5~hydroperoxy adduct (Figure 71)®

Results obtained in elemental analysis for the pro duct formed with 3 9IO-dimethyl-5-deazaisoalloxazine and

HOOH (IV) were identical to values observed for the pro duct formed with m-chloroperoxybenzoic acid® The observed values (C, 60®71; H, Zj.®37? N s 16®33) are in good agreement with the values calculated for a i*a9 5-epoxy derivative

(Calcd® for C ^ H ^ O y C, 60®70; H, 4®31; N, l6®33)o

Chan and Bruice (k3) reported an elemental analysis for the product formed with 8-cyano-39IO-dimethyl-5-deazaisoalloxazine and HOOH (Found: C„ 58®3*fJ H, 3<>9k% N, 19o09)®

These workers claimed that the results were consistent with a 5*=hydroperoxy derivative (Calcd® for W i a W

C, 56®00; H, 2f®03; W, 18®66)® However, values calculated for the corresponding A-a,5-epoxy derivative (Calcd® for

C ^ H ^ N ^ O j : C, 59®57? H, 3®57; N, 19*85) show that their observed elemental analysis could not be readily used to 58

CH, : | f 1 " | • » i i 1 1 i i

| m C H 3 .

1 i I :■ otCtCH \ H o C 3 • — . - .... - .. 1,.. 1------1______1______I___ • » -L______1______1- - i - - 1 ...

Figure 11. Infrared spectra of Zfa,5-epoxy-3,10-dime thyl- 5-deazaisoalloxazine (IV) and its parent compound (II): The epoxide was formed by reacting II with hydrogen per oxide as described in Methods® The IR spectra (KBr pellet) were recorded on a Perkin-ELmer i+67 Grating Infrared Spectrophotometer® 59 discriminate between these two structures.

Mass spectral data obtained for 3 * 10-dimethyl-5-deazaisoalloxazine and for the product (IV) formed with HOOH are shown in Figure 12. The latter is identical to the spectrum obtained for the product formed with m-chloroperoxybenzoic acid. The parent compound exhibits a base peak at m/e 2*j.18 which corresponds to the molecular ion. The mass spectrum of the product shows prominent peaks at m/e

2 5 7 9 2.1fl and 200 (base peak). The peak at 257 is attrib uted to the molecular ion formed from the ij.a9 5-epoxy de rivative.

In addition to 5-deazaisoalloxazine derivatives , Chan and Bruice also examined the reaction of HOOH with 1,5-dideazaisoalloxazine derivatives. Their data suggested that analogous reactions are observed with both classes of com pounds. However, the NMR spectrum of the product formed with 3,1O-dimethyl-1,5-dideazaisoalloxazine (VII) and HOOH showed only a single nonexchangeable proton at the C(i) position. Since this result would not be compatible with a 5-hydroperoxy structure (VIII), analogous to that pro posed for a 5-deazaisoalloxazine derivatives, Chan and

Bruice argued that the product exists as tautomers (IX, X of Structure 6).

VII VIII IX X Structure 6 60

Figure 12. Mass spectra of k&35~epoxy-3? 1O-dimethyl-5-deazaisoalloxazine (IV) and its parent compound (II): The epoxide was formed by reacting 3»10-dimethyl-5-deazaiso alloxazine (II) with hydrogen peroxide as described in Methods. These workers then proposed tautomers, analogous to IX and

X 9 as alternative structures for the product formed with

5-deazaisoalloxazine and HOOHc However9 these tautomers

are excluded by the same arguments used to exclude the

originally proposed 5°bydroperoxy adduct® The NMR spec

trum reported by Chan and Bruice for the product formed with HOOH and 3 P1O-dimethyl-1,5-dideazaisoalloxazine is

consistent with a 4a 25*=epoxy structure* provided the ab

sence of peaks expected for exchangeable protons in

structures IX and X is not due to exchange with water in their sample® Additional evidence in favor of a ^as5-

epoxy structure for this product is provided by the mass spectral data reported by Chan and Bruice (i®e0* prom

inent peaks at m/e 256* 2^0 and 199 (base peak))® These

peaks appear at positions which are shifted by one unit as compared with results obtained for 4a s5-epoxy-3310“°

dimethyl-5-deazaisoalloxazine* which is expected since

the latter contains N instead of CH at position 1 «

Be Reaction of 95°°epoxy-5<°deazaisoalloxazine with iodide

The following results indicate that reaction of **a95-

epoxy-3* 10-dimethyl-5-deazaisoalloxazine with iodide re

sults in the regeneration of the original parent compound

(II) plus stoichiometric amounts of triiodide. Formation of triiodide was monitored by the increase in absorption at 358 nm0 Linear first-order plots were obtained for the 62 reaction with 0*1 M iodide and used to calculate a second- order rate constant for the reaction (k2 = 9*6 x 10 M s “ 1 in methanol at 25° C)* The amount of triiodide formed

during reaction with the epoxide was determined from the decrease in absorbance at 358 nm observed immediately

after addition of sodium thiosulfate (Equation 5)o

2 S2 0^ + I2 — 1 + 2 I (5)

(Control studies showed that thiosulfate has no affect on

the absorption spectrum of the epoxide or the parent com

pound )B The absorption spectrum of the reaction mixture

after addition of thiosulfate was identical to that ob

served for authentic 3 910-dimethyl-5“deazaisoalloxazine

(Figure 13)0 Formation of the latter was confirmed by

thin layer chromatography (in solvent systems 1-3* as de

scribed in Methods)0 The amount of compound II formed

during the reaction was determined from its absorbance at

397 nm (e = 729300 M“’1 cm"1). The amount of epoxide used

in the reaction was determined from the initial absorbance

of the sample at 326 nm ( e = 16 * 100 cm"1)® Values ob

tained from duplicate reactions were used to calculate the

following stoichiometries for the reactions ^a,5-epoxy-

3 j. lO-dimethyl-5-deazaisoalloxazine (1 *0k ± 0*03)s 1^ (1 *09

± 0*7): 3 910-dimethyl-5-deazaisoalloxazine (1*00)* Simi

lar results were obtained with the epoxide derivative

formed with 10-®ethyl-5“d©©zaisoalloxazine: (1*19 * 0*06s 63

tu o z < m oc o cn CD <

300 400 500 nm

Figure 13. Reaction of 4a,5-epoxy-3,10-dimethyl-5-deazaisoalloxazine with iodide: Curve 1 is the initial spec trum of the epoxide (8.0 x 10“^ M) in 0.80 ml of methanol at 25° C. Iodide was added (0.20 ml of 0.5 M methanolic potassium iodide) to a final concentration of 0.1 M and the reaction was monitored at 358 nm until the formation of triiodide was complete (12 min). Sodium thiosulfate (10 jil of a 20 mM aqueous solution) was then added and the spectrum shown in curve 2 was recorded. The mechanism proposed for the iodide reaction is

shown in Scheme 19®

+ OH R H* I I* Scheme 19

The final equilibrium step in this mechanism lies com

pletely to the right since the 5-hydroxy adduct is not

stable under the reaction conditions (K^ = 6«7 M (43))®

While addition reactions are commonly observed with epox

ides and iodide, a similar oxidation reaction has been

observed with epoxide derivatives substituted with unsat urated groups (66,67) (Scheme 20)®

KI (CHg)sC C-HC— CH-d CtCHsJg * (CHs)sCC-HC=CH-CC(CHs)g o Scheme 20

The results which I obtained for the iodide reactions

are similar to those reported by Chan and Bruice (4 4 ) for

the reaction of iodide with their product formed with 3 910

dimethyl-5-deazaisoalloxazine and HOOH (Table 1)® How

ever, these workers reported that reaction of the same

product with thioxane would also regenerate the parent

compound (k2 = 0„013 M“1 s“1 in methanol at 30° C)«

While these results might be explained in the case of Table 1

Reactivity of the l+a, 5-Epoxides of

10-Methyl-5-deazaisoalloxazine (III), 3,10-Dimethyl-5-deazaisoalloxazine (IV),

and 5-Deazariboflavin (VI) with

Iodide and Thioxane in Methanol at 25° C

k2 (M“ 1 s"1)

Reagent III IV VI

Iodide 8.7 x 10"2 9.6 x 10“2 1.7 x 10~1

1.3 x 10 1 ---

Thioxane N.R. N.R. --- -P* 1.3 x 10 ---

Results of Chan and Bruice at 30° C (l+k)» 66 reaction of thioxane with a 5-hydroperoxy derivative

(Equation 6 (48), where ROH is the labile 5-hydroxy adduct which would decompose to regenerate the parent compound), the data would be difficult to rationalize in terms of thioxane reacting with a 4a * 5-epoxide adduct.

ROOH + S 0 ROH + 0= S 0 (6)

I attempted to repeat the experiment using vacuum-distill ed thioxane (0*1 M) and 4a95-epoxy-3,10-dimethyl-5-deazaisoalloxazine (7.2 x 10”^ M) in methanol at 25° C but found no evidence for reaction during a 4 day incubation period© In other studies I found that an aliquot of the same stock of thioxane would oxidize HOOH (Figures 74 &

7 5). The rate constant observed for this reaction (k2 =

6.7 x 10“5 M"1 s“ 1 at 25° C) is similar to the value pre viously reported under similar conditions (k2 = 6.76 x

7 0 “^ m”1 s“1 (48)). The results indicate that the thiox ane used in these studies cannot account for the failure to observe a reaction with the 4a *5“©Poxy derivative.

C. Kinetic studies of the epoxidation of 3 . 70-dimethyl-

5-deazaisoalloxazine with HOOH in aqueous buffers

As previously discussed, epoxide formation with 5-deazaisoalloxazine and HOOH in organic solvents requires a basic catalyst, suggesting that H00" is the reactive spe cies. To obtain further evidence for this mechanism the Figure 14. Reaction of hydrogen peroxide with thioxane, A reaction mixture containing 8.8 mM HOOH and 0.1 M thioxane was incubated at 25° C in methanol. The HOOH concentration was assayed by adding 10 fil aliquots of the above solution to 1 ml of 0.2 M methanolic potassium io dide. The amount of triiodide formed was measured at 358 nm ( = 259000 M“ 1 cm"1 (59)) and plotted in curve 2. Curve 1 was obtained for a control sample which did not contain thioxane. The data in curve 2 were corrected for the slow decomposition of HOOH and used to generate the first-order plot for the reaction of thioxane with HOOH shown in Figure 15.

Figure 15. The first-order plot for the reaction of thioxane with hydrogen peroxide in methanol at 25° C.

67 ,358 2.0 0.5 J

aqueous solution was examined,, As shown in Figure 16 the

reaction in aqueous solution is not isosbestic, in con

trast to results obtained in organic solvents® Product

formed under the latter conditions was also found to be

unstable in aqueous solutions® The results suggest that

the nonisosbestic course of the epoxidation reaction in

aqueous solutions can be attributed to the fact that the

rate of epoxide decomposition relative to the rate of its

formation is far less favorable as compared with the reaction in organic solvents® In a separate experiment the reaction in aqueous solution was allowed to proceed until the absorbance of the sample had reached a maximum

at 330 nm® The solution was then extracted with chloro form, a solvent where the epoxide is known to be stable®

Thin layer chromatography studies on the chloroform ex tract showed that the product formed in water was identi cal to the product formed in organic solvents® Although the epoxide is unstable in aqueous solution, the rate of its formation can be monitored by the loss of absorbance of the starting compound at 392 nm since neither the epox ide nor its decomposition product(s) absorb at this wave length®

To study the effect of pH on the reaction of HOOH,

3,IO-dimethyl-5-deazaisoalloxazine was selected since studies using IO-methyl-5-deazaisoalloxazine would be W o z < m

300 4 0 0 500 nm

Figure 16. Reaction of 10-methyl-5-deazaisoalloxazine with HOOH. Curve 1 is the initial spectrum of 10-methyl- 5-deazaisoalloxazine (0.1 mM) in 0,1 M sodium pyrophos phate buffer, pH 9.0 at 25° C, Curves 2-4 were recorded 3, 10 and 25 min, respectively, after the addition of HOOH (87 mM). 71 complicated by the ionization at the N(3) position (pKa about 10)* Buffers were selected based on preliminary studies in which the reaction of 3,10-dimethyl-5-deazaisoalloxazine (0*1 mM) with HOOH (82 mM) was monitored at pH 9„0 using various buffers at different concentrations

(0,10-1*0 M, except for sodium pyrophosphate where the highest concentration tested (0*2 M) was limited by the solubility of the buffer)c A constant ionic strength

( U = 1o5) was maintained with sodium chloride* Phosphate9 pyrophosphate and glycine buffers were judged suitable since similar rates were observed with all three buffers

(kQbs = 0*05 ± OoOl min*”1) and were independent of buffer concentration* (Borate buffers inhibited the reaction and were excluded* Carbonate was also excluded because of an unusual effect observed at high buffer concentrations which will be discussed later)* Studies in which the HOOH concentration was varied in 1*0 M glycine buffer at pH

10o1 showed that the reaction was first-order with respect to HOOH (Figure 17)®

The effect of pH on the observed rate of the epoxida tion reaction with HOOH is summerized in Table 2. Assum ing the H00“ is the reactive speciess an expression for k 0bS/ [HOOHjj as a function of hydrogen ion concentration can be derived (Equations 7-10)„

kcbs ' k2 l?100^ <7) Ka = fel ^0O"]|iO0H] (8) 72 1.0

0.8

0.6 •c E <0 -Q o 0.4

0.2

0 3 6 9 12 15 18 [HOOH] x I02

Figure 17* Plot of kobs versus hydrogen peroxide concen tration for the reaction of 3»10-dimethyl-5-deazaisoalloxazine (0.1 mM) with HOOH in 1*0 sodium glycine buffer, pH 10.1 at 25° C. Table 2: Effect of pH on the Epoxidation of 3*10-Dimethyl- 5-deazaisoalloxazine with Hydrogen Peroxide,

A cuvette containing 0*98 ml of 0*10 M buffer (/* = 1*5 with NaCl) was mixed with 10 fil of 2 mM 3* 10-dimethyl- 5-deazaisoalloxazine in 0*3 N HC1* The solution was in cubated at 25° C for 3 mins HOOH (l0 /il) was rapidly added using an adder-mixer and the reaction was monitored at 392 nm* In a separate study the pH of the reaction mixtures were found to remain constant throughout the bleaching of the absorption band at 392 nm* The observed rate con stants were obtained by averaging the results from at least three separate runs at each pH value*

73 7k

Table 2

Effect of pH on the Epoxidation of

3,1O-Dimethyl-5-deazaisoalloxazine

with Hydrogen Peroxide

HOOH. /HOOH^ (M"1s~ Buffer Reaction pH c ^obs sodium pyrophosphate 8,29 O.iflf 0.0026 it 8.87 0.18 0.010 glycine 9.39 i i O.Olfif it 9.80 0.11 10.11 0.089 0.23 sodium phosphate lO.ifO o.Zf i ii 10.6 0.51 10.7 0.69 11.5 1.20 11.9 1.70 it 12.0 1.95 " 12.If 2.00 sodium hydroxide (IN) }k 2.10 75

[h o o h ^ = (HOO"] + [HOOH] (9)

kobs/[H°°H^ = k2Ka/( [H+> Ka) (10)

A plot of k0bs/(HOOHk as a function of pH is shown in Fig ure 18® The solid curve in this plot is a theoretical titration curve. This curve was constructed using a value of 2®07 ± 0.007 M " 1 s " 1 for k 2 and a value of 11.10 ± 0o0k for the pKa of HOOH (literature pKa = 11.6 (68))® These values were obtained by a nonlinear least-squares fit (69) of the data to equation 10® The close fit between the experimental points and the theoretical curve indicate that H00” is the reactive species in the aqueous epoxida tion reactions similar to results observed in organic sol vents®

D® The epoxidation of 5-deazaisoalloxazine with HOOH in

carbonate buffer

The spectral course of the reaction of IO-methyl-5-

deazaisoalloxazine with HOOH (87 mM) at low carbonate buffer concentration (0®01 M„ pH 9°0S Figure 19) is simi lar to results obtained with other buffers (e®g® phos phate) at this pH value® The following results indicated that the small amounts of epoxide detected in these re actions reflects comparable rates for epoxide formation and decomposition® The observed rate of epoxide formation in 0.01 M sodium carbonate buffer (kobs = 10®i* x 10"2 min"*1) was obtained from © linear first-order plot (Figure 19)® 76 2.4

0.6

0.0 pH

Figure 18. Plot of the apparent second-order rate con stant as a function of pH for the epoxidation of 3,10- dimethyl -5-deazaisoalloxazine with HOOH in aqueous sol ution. The solid curve is a theoretical best-fit curve as described in the text. 300 400 500 n m

Figure 19. Reaction of IO-methyl-5-deazaisoalloxazine with HOOH in 0.01 M sodium carbonate buffer, pH 9,0 at 25° Ce Curve 1 is the initial spectrum of 10-methyl- 5-deazaisoalloxazine (0*1 mM). Curves 2-Jf were record ed 5, 15 and 31 min, respectively, after the addition of HOOH (87 mM). This value is similar to the rate measured in a separate experiment for the decomposition of 4®15-epoxy-10-methyl-

5-deazaisoalloxazine in the same buffer (kQ^s = 8®9 x 10 min” 1)® The rate of epoxide decomposition is not appre ciably affected in 1®0 M carbonate buffer* pH 9®0 (kQt)S = P «1 8,7 x 10 min )* but the rate of epoxide formation is greatly accelerated (complete within about 2 min) and a

>90% yield of the epoxide is detected before the slower decomposition reaction occurs (Figure 20)® First-order plots for epoxide formation in 1o0 M carbonate are not linear® These plots exhibit a distinct lag phase which can be eliminated by prior incubation of HOOH with the buffer (Figure 21)® The rate observed after the lag is identical with the rate obtained from the linear plots ® 1 (k0bs = 2®1 min )® A relatively slow reaction between

HOOH and C02S resulting in the formation of peroxycarbonic acid (analogous to the reactions observed for C02 with H20 and 0H“ (70))s could account for the observed lag (Scheme

21)®

co2 + h o o h h o Eo o h “0§00H C02 + 00H” Scheme 21

When the peroxycarbonic acid reaches a steady-state con centration the rate would then be identical to that ob served after prior incubation of HOOH with the buffer®

Organic peroxy acids are generally weaker acids than the corresponding carboxylic acid (71)9 suggesting a similar 79

0.9 14 O Z < (Q K O (0 CO < 0.6

0.3

3 0 4 0 030 5 0 0 nm

Figure 20. Reaction of .10-methyl“5-deazaisoalloxazine with HOOH in 1,0 M sodium carbonate buffer, pH 9.0* Curve 1 is the initial spectrum of the 5-deazaisoalloxazine (0,1 mM) at 25° C, Curves 2 and 3 were recorded iamediately after and 5 min after, respectively, the addition of HOOH (8? mM), 80

1.00

0.80

0.60

0.40

0.20

0.10 0.08

0.06

0.04

0.02 30 60 90 120 SECONDS

Figure 21* First-order plots for the reaction of 10-methyl-5-deazaisoalloxazine (8*0 x 10“^ M) with HOOH (71 mM) in 1*0 M sodium carbonate buffer, pH 9*0 at 25° C, Lines 1-3 were obtained for reactions inititated after prior incubation of HOOH with the buffer for 0, 0*5 and 2 min, respectively* 81

relationship for the two ionizable protons in peroxycar-

bonic acid. A tentative mechanism involving aucleophilic

attack by "OCO^H is proposed since this is likely to be

the major species present in solution at pH 9 (Scheme 22),

QwOT § Scheme 22

Epoxide formation would be facilitated, in part, as com

pared with that in the uneatalyzed reaction since HCO" replaces OH" as the leaving group.

E. Kinetic studies of the enoxidation of 5.10-dimethyl-

5-deazaisoalloxazine with m-chloroperoxybenzoic acid

in aqueous buffers

Although several mechanisms have been proposed for

the epoxidation of carbon double bonds by peracids, the

generally accepted view is that of the one-step mechanism proposed by Bartlett (72) (Scheme 23).

Q \ t Q ' ' RCOOH + C=C _ = > RCOH + C— C t \ / \ Scheme 23

Previous studies with a variety of alkenes indicate that the unionized peracid is the reactive species. Unlike the alkaline epoxidation reactions observed with HOOH and

t-butyl hydroperoxide, the peracid cannot be acting as a nucleophile since electron-withdrawing substituents accelerate the reaction while the converse effect is ob served with electron-withdrawing substituents in the al- kene (73)o The latter effect would suggest that 5-deazaisoalloxazine should be relatively unreactive towards

epoxidation with peracid®, Although the kinetics of the reaction of m-chloroperoxybenzoic acid with 5“deazaisoal« loxazine in chloroform are complex (see below), the re sults are at least consistent with a reaction involving

the unionized peracid since the reaction does not require

a basic catalyst® In contrast, preliminary studies indi

cated that the reaction in aqueous solution was consider ably faster than the reaction in chloroform and that the rate of the reaction increased with pH in the range of

pH 5-9o The fact that the reaction is faster in water is not consistent with results obtained for the epoxidation of other alkenes with peracids since optimal conditions

for these reactions involve use of a nonpolar solvent which cannot form hydrogen bonds with the peracid® These preliminary studies suggested that reaction via the un ionized peracid might be suppressed in water and that a different mechanism involving nucleophilic attack of ionized peracid might be applicable® 83

To test this hypothesis the kinetics of the reaction of 3*10-dimethyl«5-deazaisoalloxazine (1 x 10”^ M) with excess m-chloroperoxybenzoic acid was studied at 25° C in

0©10 M sodium pyrophosphate buffers of constant ionic strength at various pH values as summarized in Table 3®

Reactions were initiated by addition of 5 £*1 of a me than- olic solution of excess peracid to a cuvette containing

1 ml of the 5“*

136** ± 44 M” 1 s'”1 for k2 and a value of 7®63 ± 0 o06 for the pKa of m-chloroperoxybenzoic acid (literature pKa =

7®57 (74))® These values were obtained by a nonlinear

1 east-squares fit of the data to equation 109 substitut ing m-chloroperoxybenzoic acid for HOOH© The data show that ionized perbenzoate is the reactive species and that the reaction is likely to proceed via nucleophilic attack of perbenzoate at the 5-position of the 5-deazaisoalloxazine, similar to the mechanism observed with HOOH and to that proposed for attack by peroxycarbonate in the case of the reaction with HOOH in carbonate buffer© Table 3

Effect of pH on the Epoxidation of

3 >10-Dimethyl-5-deazaisoalloxazine

with m-Chloroperoxybenzoic acid

Reaction pH [m-Cl0CO^,I^ [m-C10CO,Hj;} (M“ * s"1)

5.09 2.8 x 10-5 2.0 5.49 " 5.0 5.99 " 11.3 6.50 11 36.9 7.08 5.6 x 10“^ 217. 7.56 1.1 x 10“4 663. 8.03. " 1030. 8.54 " 1210. 8.89 11 1260.

All reactions were conducted in 0.10 M sodium pyrophos phate buffers (/i = 0.944 with NaCl) at 25° C. 85

[m-CI0CC^Gx I04

Figure 22. Effect of the concentration of m-chloroperoxybenzoic acid on the pseudo-first-order rate constant ob served for the reaction with 7O-methyl-5-deazaisoalloxazine (5 in 0.10 H sodium phosphate buffer, pH 7*0 at 25° C. 1.4

O' 0.8 O

o 0.6

» 0,4 ■O o

0.2

0.0

Figure 23. Plot of the apparent second-order rate con stant as a function of pH for the reaction between 3 , 10- dime thyl-5-deazaisoalloxazine and m-chloroperoxybenzoic acid in 0.10 M sodium pyrophosphate buffers of constant ionic strength (pi = 0.944 with NaCl) at 25° C. The solid curve is a theoretical best-fit curve. 87 F 0 Kinetic studies of the epoxidation of 5»10-dimethyl-

5-deazaisoalloxazine with m-chloroperoxybenzoic acid

in chloroform Since the results indicated that the reaction between

5-deazaisoalloxazine and m-chloroperoxybenzoic acid in chloroform probably involves unionized m-chloroperoxybenzoic acid, an attempt was made to evaluate the rate con stant for this reaction for comparison with results obtain ed in aqueous buffers,, Studies in chloroform were conduct ed at 25° C in solutions containing 3 910-dimethyl-5-deazaisoalloxazine (0o1 mM) and excess peracid and were monitor ed at 400 nm. Unlike the rapid reaction observed in aque- our solutions at millimolar concentrations of peracid, studies in chloroform with m-chloroperoxybenzoic acid con centrations lower than 0.1 M were not feasible owing to the extremely slow reaction rates. Even at 0o1 M peracid the reaction was quite slow and © Guggenhein plot was ‘2 m *| used to determine the value for k0^s (3„23 x 10 min )•

At higher peracid concentrations rate constants were deter mined from linear pseudo-first-order plots. The results are summarized in Table 4o As shown in Figure 24, a plot of kQbs versus the peroxybenzoic acid concentration is nonlinear. This suggests that the reaction is not first- order with respect to the peracid, in contrast to results obtained in aqueous solution and with results obtained for the epoxidation of other alkenes with peracids in nonpolar 88

Table 4

Epoxidation of

3,10-Dimethyl-5”deazaiBoalloxazine

with

m-Chloroperoxybenzoic Acid in Chloroform

[m-Cl0CO^H] * kobs

0.10 3.22 x 10“3 0.19 1.77 x 10"2 0.28 4.62 x 10"2 0.39 9.85 x 10“2 0.46 1.76 x 10"1 0.56** 2.75 x 10"1

* The concentration of m-chloroperoxybenzoic acid was determined via reaction with 0.20 M methanolic KI at 25° C by measuring the amount of triiodide formed at 358 run (€ = 25,000 ( 59)).

** Higher concentrations could not be studied owing to the solubility of the peracid in chloroform* 89

2 5

0 0.1 0.2 0.3 0.4 0.5 0.6 [m-CliDCCy-G

Figure 2 4 . Plot of the observed rate constant against the peracid concentration for the reaction of 3»10-dimethyl-5-deazaisoalloxazine (0.1 mM) with m-chloroperoxybenzoic acid in chloroform at 25° C. 90 solvents, A linear plot was obtained by plotting log kolQS versus log [peracid] (Figure 25) suggesting that the re action might be fractional order with respect to the per acid (i0ee> = k [per&cid]11), A reaction order of 2,5 with respect to the {peracid] was obtained from the slope of this plot. The possibility that the data might re flect mired second- and third-order kinetics (Equation 11) was examined by plotting kQbs/ [peracid] Q versus [peracid] Q

(Figure 26)«, But this plot was not linear.

Rate = (dFl| (peracid} k^ [dFl] [peracid] 2 (11)

The results suggest that the epoxidation of 5-deazaisoalloxazine with the peracid in chloroform is more complex than that observed for other alkenes and was not further investigated,

The reaction of 5-deazariboflavin with peroxides was studied in order to obtain a more appropriate model for the reaction of peroxides with enzyme-bound deazaflavin.

Figure 27 shows the spectral course of the reaction of

5-deazaribo flavin with HOOH (87 mM) in acetonitrile con taining 71 fflM triethylamine. The data provide no direct evidence for epoxide formation since the observed reaction consists of the bleaching of both absorption bands of the starting compound9 in contrast to results obtained under similar conditions with 10-methyl-5**deazaisoalloxazine 91

2.5

2.0

1.0

0.5 -

0.2 0.4 0.6 0.8 - log Cm-chloroperoxybenzoic acid3

Figure 25* Plot of -log kQ^s againet -log [peracid] for the reaction of 3»10-dimethyl-5-deazaisoalloxazine (0.1 mM) with m-chloroperoxybenzoic acid in chloroform at 25° C. 92

0 . 5 i------r

'.E 0 . 4 E

r— i 0 . 3 5* o ^ 0.2 0 1 E

0.1

i 0.1 0.2 0.3 0.4 0.5 0.6 [m -C^COgH]

Figure 26. The Figure shows a plot of k0^s/ [peracid] versus [peracid]for the epoxidation of 10-dimethyl- 5-deazaisoalloxazine (0,1 mM) with m-chloroperoxybenzoic acid in chloroform at 25° C.

4 93

UJ o z < CD QC O CD <

300 4 0 0 500 nm

Figure 27. Reaction of 5-deazariboflavin with HOOH. Curve 1 is the initial spectrum of 5-deazariboflavin (O.T mM) in acetonitrile containing HOOH (87 mM) at 25° C, Curves 2-4 were recorded 0.5> 2 and 4 h, re spectively, after adding 71 mM triethylamine. 94 and 3,lO-dimethyl-5-deazaisoalloxazine, However, evidence consistent with epoxide formation is observed during the reaction of the diorthoformate derivative of 5-deazariboflavin (Compound 11) with HOOH (Figure 28)® The results suggest that the ribityl side chain in 5-deazariboflavin interfere with the epoxidation reaction with HOOH®

The spectral course of the reaction observed with m-chloroperoxybenzoic acid and 5-deazariboflavin is simi lar to results obtained for the diorthoformate derivative

(Figures 29 and 30)® Epoxide formation in these reactions is evidenced by the initial increase in absorption at

340 nm accompanied by the bleaching of the absorption band at 400 nm® The secondary decrease in absorption at 340 nm suggests epoxide decomposition® The derivative formed with 5-deazariboflavin was isolated as described in Meth ods® The compound exhibits an absorption spectrum (Xmax

336 nm in methanol) similar to the epoxide derivatives formed with 5-deazaisoalloxazine, and like the latter de rivatives will also decompose in aqueous solution (Figure

31)® Similar to compounds III and IV, 4a,5-epoxy-5-deazariboflavin will react with 0® 1 M iodide (k2 = 0®17 M”*1 s"1 in methanol at 25° C) to regenerate the parent compound as evidenced by the spectral course of the reaction (Fig- gure 32) and by thin layer chromatography studies (in solvent systems 1-3)® 300 4 0 0 500 nm

Figure 28. Reaction of the diorthoformate derivative of 5-deazariboflavin (Compound 11) with HOOH. Curve 1 is the initial spectrum of compound 11 (0.1 mM) in acetonitrile containing HOOH (87 mM) at 25° C. Curves 2-5 were recorded 0.5# 1.5* 3.5 and 6.5 h, respectively, after adding 71 mM triethylamine. W zo

300 400 500 nm

Figure 29. Reaction of 5-deazariboflavin with m-chloroperoxybenzoic acid. Curve 1 is the initial spectrum of 5-deazariboflavin (0.1 mM) in chloroform at 25° C. Curves 2-4 were recorded 0„5* 7 and 30 min, respectively, after the addition of 0.56 M peracid* Ll ) O z < CD o: o CO CD <

300 4 0 0 500 nm

Figure 30. Reaction of the diorthoformate derivative of 5-deazariboflavin (Compound 11) with m-chloroperoxybenzoic acid. Curve 1 is the initial spectrum of compound IT (0.1 mM) in chloroform at 25° C. Curves 2-i* were recorded 0.5, 7 and 30 min, respectively, after adding 0.56 M peracid. LlJ o < CD CH O CO CD <

0.5

0.0 300 400 500 nm

Figure 3 U Decomposition of i+a,5“epoxy~5-deazariboflavin in aqueous solution. Curves 1-6 were recorded 1, 10, 25, 5 0 , 100 and 200 min, respectively, after adding ^a,5- epoxy-5-deazariboflavin (0,1 mM) to 0,1 M sodium phos phate buffer, pH 7,0 at 25° C, 99 T T

l.€

1.2 -

oUJ z < CO a: o v> 0.8

0 .4

300 4 0 0 500 nm

Figure 32. Reaction of 4a,5-epoxy-5-deazariboflavin with iodide. Curve 1 is the initial spectrum of the epoxide (9»2 x 10-5 in 0.80 ml of methanol at 25° C. Iodide was added (0,20 ml of 0.5 M methanolic KI) to a final concentration of 0.1 M and the reaction was monitored at 358 nm until the formation of triiodide was complete (12 min). SoditHii thiosulfate (10 fil of a 20 mM aqueous solu tion) was added and the spectrum shown in curve 2 recorded. 100

Compared with the NMR spectrum of 5-deazariboflavin, the major differences in the NMR spectrum of its epoxide a r e : a 0 The position of the peak attributable to the hydro

gen at C(5) (8*81 ppm for 5-deazariboflavin in

DMSO-dg) is shifted upfield and appears at 5*13 ppm,

similar to results obtained with compounds III and

IV0 The peak at 5*13 PP® appears as a singlet at

the lower field edge of the multiplet attributed to

the 11 protons of the ribityl side chain (3*57--J!f*93

ppm) and is conspicuously absent in the spectrum of

5-deazariboflavine

b0 Peaks due to the two aromatic protons in 5-deaza

riboflavin appear at 7*83 and 7*93 ppm* ^a,5-Epoxy-

5-deazariboflavin exhibits four peaks in this re

gion (7«359 7*56, 7*60 and 7»83 ppm) with an inter-

gration (2„5 H) higher than that expected,,

The additional peaks in the aromatic region are likely

due to decomposition products since, unlike compounds III

and IV, Zfa,5-epoxy-5-deazariboflavin is not stable in

organic solvents (e0g0, acetonitrile, dimethyl sulfoxide,

methanol and chloroform)a (Decomposition in organic sol

vents is slower than that observed in aqueous solution)*

The instability of ifa85-epoxy“5“deazariboflavin in organic

solvents and in water interfered with attempts to further

purify the compound and also with studies to establish its purity by thin layer chromatography*

H. Relationship of 4a^-epoxide to deazaFAD-X

Reaction of HOOH with 5-deazaFAD bound to oxynitrilase causes an irreversible modification of the coenzyme designated as EdeazaFAD-X (25)® Unlike the 4a95“©poxide derivatives, which exhibit a single absorption band in the visible region and are unstable in water, deazaFAD-X exhibits two maxima (343 and 359 nm) in its visible spec trum, is stable in water and does not react with iodide®

In addition, deazaFAD-X exhibits a characteristic purple fluorescence which distinguishes it from the nonfluores- cent epoxide derivative and from unmodified 5~deazaflavin which emits a blue fluorescence® The results indicate that deazaFAD-X is not an epoxide derivative® However, studies by Pokora and Jorns (45) showed that EdeazaFAD-X could be formed by reacting 5~deazaFAD enzyme with either

HOOH or with m-chloroperoxybenzoic acid, and that both reactions proceeded via an intermediate which exhibited spectral properties similar to my model epoxide compounds®

Reaction of the enzyme intermediate with iodide regener ated unmodified 5-“deazaFAD enzyme, similar to the reaction observed with model epoxide compounds® These results sug gested that the enzyme intermediate was 4a95“epoxy-5"* deazaFAD and that under appropriate conditions 4®95“©poxy-

5-deazaisoalloxazine derivatives might be converted to the corresponding 5-deazaisoalloxazine-X compounds in model studies, Since the results also suggested that deazaFAD-X did not contain an epoxide moiety, I attempted to generate 5-deazaisoalloxszine-X by studying the re action of 4a,5-epoxy-10-methyl- and 4a ,5“*epoxy-3 910-di me thyl-5-deazaisoalloxazine (Compounds III and IV, re spectively) under conditions which might cleave the epox ide ringo In these studies compounds III and IV were in cubated in aqueous solutions with pH values in the range from pH 0 (l N HCl) to pH 9 (0,1 M sodium carbonate),

The slowest decomposition rate was observed at pH 5®

The spectral course observed for these reactions is simi lar to that shown in Figure 33 for the reaction observed at pH 1,45, No evidence was obtained at any pH value tested to suggest formation of 5-deazaisoalloxazine-Xc

However, recent studies by Pokora strongly suggest that at pH >10 compounds III and IV can be converted to a species which exhibits spectral properties similar to deazaFAD-X, My failure to detect this reaction is clearly due to the fact that detailed spectral studies were not conducted at more alkaline pH values0 My kinetic studies on the effect of pH on the epoxidation of compound II were conducted over a pH range (pH 8,29-14) where conversion of IV to 5““deazaIsoalloxazine-X would occur. However, this reaction is slower than the epoxidation reaction under the conditions studied and would not have been de tected during the time required to measure epoxidation 300 400 500 n m

Figure 33* Decomposition of Jfa,5-epoxy-3#10-dimethyl- 5-deazaisoalloxazine (IV) in aqueous solution* Curve 1 is the initial spectrum of the epoxide (9*7 x lu“^ M) in 1 ml of water* An aliquot of 6 N HCl (10 fil) was added to lower the pH to 1*A-5» and curves 2-7 were recorded 10, 20, JfO, 80, 160 and 300 min, respectively, after the addition* Curve 8 was recorded after 18 h at 25° C* 104 ratese Furthermore8 the kinetic studies were monitored only at a single wavelength (392 nm) where neither IV nor 5°deazaiso©llox&zin®“*X exhibit significant absorption®

I 0 Studies with Glycolate Oxidase

Glycolate oxidase is a flavoprotein which catalyzes the oxidation of glycolate by molecular oxygen (Equation

12). “00CCH20H + 02 — "00CCH0 + HOOH (12)

The enzyme contains 2 moles of FMN per molecular weight of 100,000 (56)8 Like other flavoprotein oxidases it forms a red anionic seraiquinone, is reoxidized rapidly by oxygen and forms a covalent adduct with sulfite (23)e In the following sections the reactivity of 5-deazaFMN»recon*» stituted glycolate oxidase (Figure 34) with HOOH and with m-chloroperoxybenzoic acid is discussed in detail and the results compared to similar studies with oxynitrilase and flavodoxine 1 * Reaction of 5-deazaFMN glycolate oxidase with HOOH

As noted in the model studies 5-deazaflavin reacts with HOOH in alkaline buffer to form a 4a,5-epoxy adducte

This reaction is characterized by bleaching of the 390 nm absorption band accompanied by a 4-7 nm bathochromic shift of the band near 320 nmQ In 0.1 M sodium phosphate buffer, pH 7»0, the reaction proceeds very slowly (kQl3S = 5»9 x

10-5 at 25° C) in 88 mM H00H„ In contrast, when

3 mM HOOH is added to 2,1 x 10“^ M 5-deazaFMN glycolate 105 I5K

IOK

300 400 500 nm

Figure 3*f* Denaturatiou of 5-deazaFMN-reconstituted gly colate oxidase in 2% sodium dodecyl sulfate. Curve 1 is the initial spectrum of EdeazaFMN in 0.1 M sodium phos phate buffer, pH 7.0, containing 0.3 mM EDTA. Curve 2 was recorded 1 min after the addition of an equal volume of k% sodium dodecyl sulfate in the same buffer. The scale was adjusted to show the relative extinction co efficients. oxidase in 0,1 M sodium phosphate buffer, pH 7*0, contain

ing 0*3 mM EDTA, bleaching of the 400 nm absorption band

occurs rapidly (kQbs = 2*4 min"1 at 15° C), leveling off

temporarily at about 26% of the initial absorption* This

reaction is accompanied by a 6 nm bathochromic shift of

the 334 nm band (Figure 35s curve 2)* Studies at various

HOOH concentrations show that this reaction is first-order

with respect to HOOH* This initial reaction is followed

by the formation of a second intermediate as evidenced by

a much slower increase in absorption in the 360-410 nm

region which reaches a maximum after about 1 h (Figure 35»

curve 3)o A small amount of product is also present at

this time* Product formation continues, as monitored by

a decrease in absorption in the 360-410 nm region accom

panied by an increase in absorption at 340 and 355 nm

(Figure 35» curves 4-7)» After about 14 h no further

spectral changes occur0 This combination of HOOH concen

tration (3 mM) and temperature (15° C) offered the best

spectral resolution of the intermediates and was used in

further studies of the reaction*

2* Reaction of 5-deazaFMN glycolate oxidase with m-chloro-

peroxybenzoic acid

When EdeazaFMN was reacted with a 2-fold excess of m-chloroperoxybenzoic acid in the same buffer at 25° C,

the spectral course was similar to that observed for the

reaction with HOOH at the same temperature* However, the 300 400 500 600 Wovalangth (nm)

0 3

Z 02

3 0 0 5 0 0 6 0 0

Figure 35# Reaction of 5-deazaFMN bound to glycolate oxidase with HOOH, EdeazaFMN (2,1 x 10“^ M) in 0,1 M sodium phosphate, pH 7,0, containing 0,3 mM EDTA at 15° C (curve l) was mixed with 3 mM HOOH, Curves 2-6 were re corded 1,3 3 , 5 9, 123, 2 M and if8 l min, respectively, after the addition. Curve 7 was recorded after 23 h. The final phase of the reaction is shown separately for clarity. 108 first intermediate formed too quickly to measure its rate by monitoring A ^ 0 o This faster reaction of 5~deazaflavin with peracid relative to HOOH is in accord with the re sults obtained in the model studies where the second- order rate constant for the epoxidation of 3 ,10-dimethyl-

5 -deazaisoalloxazine with m-chloroperoxybenzoate was about

660 times greater than the rate constant obtained for the reaction with HOOH in aqueous buffer at 25° Cc

3® Characterizing the first intermediate

The rapid formation of the first intermediate with either HOOH (Figure 35» curve 2) or with m-chloroperoxybenzoic acid relative to its slower conversion to the second intermediate permitted studies with the first inter mediate, The reactivity of intermediate I with iodide was measured by adding 5 ^ 1 of 5 M potassium iodide to 0 ,2 5 ml of the reaction mixture when the absorbance at 400 nm had reached a minimum (about 2 min), m-Chloroperoxybenzoic acid was found not to react with iodide under these con ditions, but for intermediate I formed with HOOH the re action mixture was treated with catalase (325 Sigma units/ ml) 30 s prior to the addition of iodide in order to re move excess H00Ho Iodide (0ol M) added to enzyme-bound intermediate I resulted in the regeneration of starting compound (Figure 36 )e Unlike the oxidation of iodide by the 4 a,5-epoxy derivatives in the model studies, the re action in the enzyme system occurs without the apparent 109

0.100

4 0 0 0.075

S> 0.0 5 0

0.025

3 2 300 4 0 0 500 minutes nm

Figure 36, Reaction of iodide with the first intermediate formed with HOOH and 5-deazaFMN bound to glycolate oxidase* EdeazaFMN (1,7 x 10"^ M) in 0,1 M sodium phosphate buffer, pH 7.0, containing 0*3 mM EDTA at 25° C (curve 1) was mixed with 1 mM HOOH for 3*5 min. Formation of intermed iate I was monitored by the bleaching of A^0 0 (left). Excess HOOH was removed by catalase (325 Sigma units/ml) and 0,1 M KI was added 30 s later to regenerate the start ing compound (curve 2 ), n o formation of triiodide® Control studies indicated that the latter was due to a rapid reaction of triiodide with the protein moiety (Figure 37)o Treating intermediate I with catalase alone did not regenerate the starting com pound 9 similar to that observed in the model studies0

This suggests that formation of intermediate I is also not a reversible reaction,, In the presence of catalase conversion of intermediate I to intermediate II and then to product proceeds at rates identical to the control re actions

Further evidence that the initial intermediate is an epoxide derivative was found by examining the properties of protein-free intermediate I at neutral pHe The inter mediate was released from the protein by adding an equal volume of 4 % sodium dodecyl sulfate in buffer to the re action mixtures Sodium dodecyl sulfate causes rapid denaturation of the protein (which remains soluble) accom panied by the release of the modified coenzyme® Unlike enzyme-bound intermediate I9 the free intermediate does not form product9 but rather decays with a loss of absorp tion (Figure 38) similar to that observed for the model epoxide compounds at neutral pH® The residual absorption observed after decomposition of free intermediate I is attributed to the presence of 5"deazaFMN in the enzyme preparation which does not react with peroxide or peracid®

The initial spectrum of the protein-free intermediate m 0 , 1 0 0

0.075

Ld O Z < “ 0.050 O in CD < 0 .0 2 5

0 . 0 0 0 300 4 0 0 500 nm

Figure 37. Reaction of 5-deazaFMN bound to glycolate oxidase with triiodide. Curve 1 is the initial spectrum of the enzyme (6 .8 x 10“^ M) in 0.1 M sodium phosphate buffer, pH 7.0* containing 0.3 mM EDTA at 25° C. Curve 2 was recorded 1 min after adding 0.1 M KI and iodine ( 1 .0 x 10-5 yjjg latter causes only a small increase in absorbance at 3 5 8 nm, in contrast to the absorbance increase (0.25 OB units) expected if the enzyme did not react with triiodide. 112 0.10

(_!

300 4 0 0 5 00 6 0 0 Wavelength (nm)

Figure 38. Effect of sodium dodecyl sulfate on the first intermediate formed with 5-deazaFMN bound to glycolate oxidase and HOOH. EdeazaFMN in 0.1 M sodium phosphate buffer, pH 7*0, containing 0.3 mM EDTA at 25° C (curve l) was reacted with 1 mM HOOH for 3 min. Sodium dodecyl sulfate (2%) was added and the spectrum shown in curve 2 was recorded. Curves 3-5 were recorded 30, 60 and 300 min, respectively, after adding the denaturant. The spectrum of the protein-free intermediate (curve 2 ) was corrected for the presence of unmodified 5-deazaFMN to yield the difference spectrum in curve A. Curve B is the spectrum of 4 a,5-epoxy-5-deazariboflavin. 113

(curve 2 ) was corrected for the contribution of unmodified

5 -deazaFMN (curve 5) to yield a difference spectrum (Fig ure 38) similar to the absolute spectrum of 4 a,5-epoxy-

5-deazariboflavin0

4 ® Characterizing the second intermediate

As previously discussed, conversion of the epoxide intermediate to the second intermediate is evident by an increase in absorption in the 3 6 0 -4 1 0 nm region with the appearance of a new absorption band at 375 a® (Figure 35* curve 3)o Intermediate It does not accumulate in stoichio metric amounts owing to a competing conversion to product,,

This conversion is readily apparent during the subsequent course of the reaction (Figure 35, curves 4-7), which re sults in the formation of additional product accompanied by the loss of intermediate II, as judged by the decrease in absorbance in the 360-410 nm region® The observed quantitative conversion of intermediate II to product in dicates that this step is irreversible®

Evidence regarding the possibility of a reversible reaction between the epoxide intermediate and intermediate

II was sought by examining the effect of iodide on inter mediate II® This intermediate might not react directly with iodide, but if it were in equilibrium with intermed iate I, iodide addition should result in the regeneration of starting compound from both intermediate I and intermed iate II present in the reaction mixture® In these studies catalase (325 Sigma units/ml) was added after allowing the reaction to proceed until intermediate II had reached a maximum concentration (about 1 h as judged by A*1-00), fol lowed 30 s later by the addition of 0,1 M iodide,. The subsequent course of the reaction was similar to that ob served for a control reaction, except for a 2 0 % lower yield of the final product, which was formed at about one- third the rate observed for the control reaction,, The lower product yield can be attributed to the reaction of iodide with residual intermediate I, as evidenced by an increase in absorbance at 4 00 nm immediately following the addition of iodide,. Kinetic studies (presented later) confirm that about 17% of the reactive enzyme is present as intermediate I at this point in the reaction,, The slower rate of conversion of intermediate II to product relative to the control reaction may reflect inhibition by iodide since glycolate oxidase is known to bind vari ous anions at its active site (14)® The results show that iodide does not react with intermediate II nor pre vent its conversion to product,, This indicates that con version of the epoxide intermediate to intermediate II is an irreversible reaction likely to involve modification of the epoxide moiety in the first intermediate®

In order to investigate what role the protein moiety plays in the conversion of intermediate II to product the enzyme was denatured with sodium dodecyl sulfate when 115 intermediate II had reached a maximum concentration. Re lease of intermediate II into solution at neutral pH re sults in its immediate conversion to product3 as evidenced by the loss of absorbance at 3 7 5 nm and the increase in absorbance at 340 and 355 nm. A slower secondary decrease in absorbance in the latter region was observed and is attributable to the decomposition of residual amounts of intermediate I8 as previously described (Figure 38).

(The final product is stable free in solution at neutral pH). The results indicate that the protein moiety is re quired for conversion of intermediate I to intermediate IIS but it appears unnecessary and actually inhibits the con version of intermediate II to the final product. This is evidenced by the fact that the reaction with protein-bound intermediate II is much slower than that observed with protein-free intermediate II. Studies by Pokora and Jorns

(4 5 ) show that under alkaline conditions the model epoxide compounds are converted to a product which exhibits spec tral properties similar to the product formed with glycolate oxidase and oxynitrilase. The model reaction pro ceeds via two intermediatess A and B (Scheme 24)®

ch 3

A - NHR H

a B C

Scheme 24 116

Formation of intermediate A is the only step in the model reaction which requires alkaline conditions. Intermediate

B exhibits spectral properties similar to intermediate II in the glycolate oxidase reaction, and like the latter, is rapidly converted to product at neutral pH.

5. Characterization of the product

The spectrum of the product formed with glycolate oxi dase is very similar to the spectrum observed for Edeaza-

FAD-X formed with oxynitrilase. Both have characteristic dual absorption bands (340 and 355 nm for EdeazaFMN-X;

347 and 362 nm for EdeazaFAD-X) and are formed by reacting the appropriate 5-deazaflavin-reconstituted enzymes with either HOOH or m-chloroperoxybenzoic acid. Like Edeaza

FAD-X, EdeazaFMN-X is unaffected by catalase and iodide, is stable when released from the enzyme with sodium do decyl sulfate, and emits a purple fluorescence when viewed under ultraviolet light. Further evidence that deazaFMN-X from glycolate oxidase is similar to deazaFAD-X from oxy nitrilase came from thin layer chromatography studies by

Pokora and Jorns (45)® 5“DeazaFMN-X from glycolate oxi dase was shown to comigrate with deazaFMN-X formed by phosphodiesterase treatment of deazaFAD-X from oxynitril ase.

6 . Kinetics of the reaction of EdeazaFMN with HOOH

The individual rate constants in the pathway from

EdeazaFMN to EdeazaFMN-X (Equation 13) were determined by 117 monitoring the reaction at 400 nm and 15° C.

k k k EdeazaFMN + HOOH — I — II EdeazaFMN-X (13)

The initial rapid absorbance change at 400 nm was used to determine k^® The slower absorbance changes at 400 nm were used to determine k2 and k^« The following observa tions indicate that the product, EdeazaFMN-X, and the first intermediate exhibit negligible absorbance at 400 nm which would therefore correspond to an isosbestic point suitable for determining k2 and k^® The residual absorption at 4 0 0 nm observed at the end of the reaction is attributed to unmodified EdeazaFMN as evidenced by thin layer chromatography studies and by the fact that purified deazaFMN-X exhibits negligible absorbance at 400 nm. The presence of unmodified EdeazaFMN can account for nearly all of the absorbance observed for the first inter mediate at 400 nm since the latter is only slightly high er than the value observed at the end of the reaction®

Similar results regarding the presence of unreactive

5-deazaFMN in glycolate oxidase preparations have been obtained in previous studies involving substrate reduc tion or reaction with other nucleophiles (14)»

Intermediate I is formed rapidly ( (HOOH] £ l mM) in comparison with subsequent steps in the reaction and can be considered to be kinetically uncoupled from the latter.

The rate constant for the formation of intermediate I was 118 obtained directly from a first-order plot of the initial rapid decrease in absorbance at JfOO nm (kj = 8 0 0 M ^ m i n ” 1)

(Figure 39), Studies at 1 mM and 3 mM HOOH concentrations show that this reaction is first-order with respect to

H00H9 whereas the subsequent conversion of the epoxide to EdeazaFMN-X proceeds via a HOOH-independent pathway, A value for k^ (5 ,8 x 10“^ min” 1) was obtained from the linear region of a first-order plot of the slower absor bance changes at ifOO nm (Figure ifO) 0 The initial portion of this plot was corrected for the contribution from k^

(as described by Fersht (75)) to obtain a value for k2

(3,0 x 10 OBlP min 09 1 ) (Figure 40), Similar rate constants

(k2 = (3 , 0 * 0 ,1 ) x 10"2 min"*1; k^ = (5 *6 * 0 ,2 ) x 10”^ min” 1) were obtained when the data were analyzed by a non linear least squares fit to equation 16 (76),

k k i ii EdeazaFMN-X (H+)

- k3&l] = K{)k2 e”k2 t - k^l] (15)

Integrating r-j-j 1^)k2 (e“k2t - e k3^.)_ ( 16 \ gives * J (k3 - k2 )

where |jl] is proportional to A A1*®® and [i^ was computed as 0 ,1 5 8 ± 0 ,0 0 2 ,

The composition of the reaction mixture at any time after the initial formation of intermediate I could now be cal culated using the following kinetic expressions from

Frost and Pearson (77)s 1 T9

0.20

o.to •

0.15 *• 0.10 O jM

u OjM

cn 0 .1 0

10 10 40 90

0 .0 5

MINUTES

Figure 39. Monitoring of during the initial reaction of 5-deazaFMN bound to glycolate oxidase <2.1 x 10“^ M) in 0.1 M sodium phosphate buffer, pH 7«0, containing 0.3 mM EDTA at 15° C with 3 mM HOOH. The inset shows the first- order plot of the bleaching of A1*®® following the addition of HOOH, where Af is the minimum value for A^®^ observed at about 2 min. 120

a to o 0.2000 •

0 .1 0 0

0.1000-

£ 0.0500 -• 20 >0 90 TO

0 .0 2 9 0 -

0.0)25

100 200 3 0 0 4 0 0 soc MINUTES

Figure 40. Semi-logrithmic plot of between 5-480 min after the addition of 3 mM HOOH to 5-deazaEMN bound to glycolate oxidase (2.1 x 10”^ M) in 0.1 M sodium phos phate buffer* pH 7*0, containing 0.3 mM EDTA at 15° C. The inset: shows the semi-logrithmic plot of A, the differ ence between A A 1*®® and the value extrapolated back from the linear portion of the plot. 121

(17) 0 = (e“T - e“KT)/( K - 1) (18)

y= 1 + (e “ HT - #fe“T )/( tc - 1 ) (19)

max (20 ) (21) where a = [I] / [EdFMtg Total (2 2 )

ft. [IIJ / [EdFMN^ Total (23)

y = [EdFMN-X]/[EdFMNj Tota]_ (2 if)

r = k£t (25) K = k^/k2 (26)

[i] is the concentration of intermediate I,

(jfj is the concentration of intermediate II,

[EdFMN-X] is the concentration of the final

product, EdeazaFMN-X, and [EdFMN^J

the total concentration of reactive enzyme.

7. Theoretical absorption spectra

The total concentration of enzyme in the reaction shown in Figure 35 was obtained from the initial absorb ance of the sample (curve 1) at /*00 nm using the molar extinction coefficient for 5-deazaFMN bound to glycolate oxidase (Equation 27).

[EdeazaFMN] = a^0 0 /i2 ,1 0 0 = 2 .1 0 x 10“ 5 M (27) Total

The concentration of all coenzyme forms discussed below will be expressed relative to the total enzyme 122 concentration,. As discussed in the preceding section, the composition of the reaction mixture at any time can be calculated® Based on these calculations it should be pos sible to derive theoretical spectra for intermediates I and II® However, in order to do this it is necessary to correct the data in Figure 35 for the presence of unre active deazaFMN in the glycolate oxidase preparation®

The latter was estimated based on the final absorbance ob served at 400 nm (23 h, curve 7 in Figure 35) and corres ponds to 17% of the total enzyme-bound deazaflavin®

Assuming unreactive and reactive enzyme exhibit identical spectra, the spectrum due to reactive enzyme can be de termined as shown in Equation 28®

AEdFMN = C.j(l - 0o17) = 0®83 C 1 (28) r where spectrum of reactive

EdeazaFMN and is the absorbance along curve

1 in Figure 35®

The final spectrum observed for the reaction (curve

7, Figure 35) is attributed to a mixture of EdeazaFMN-X and unreactive EdeazaFMN® Ely subtracting the contribu tion made by unreactive EdeazaFMN, the theoretical spec trum of EdeazaFMN-X at 83% of the total enzyme concentra tion is obtained (Equation 29)®

AEdFMN-X “ C7 “ Oo17C1 (29) where a j& f m n „x tlie theoretical spectrum of EdeazaFMN-X, and and are the absorbances

along curves 1 and ?, respectively, in Figure 35®

The theoretical spectra of intermediates I and II were calculated using spectra recorded near the moment of maximum concentration of each intermediate, namely, at

1«33 min and 59 min, respectively (Figure 35j curves 2 and

3)o From Figure 35 it is seen that neither intermediate absorbs above 450 nmQ Therefore, the region 300-450 nm was considered in calculating the theoretical spectra for these intermediates® Because the spectrum changes rapidly in this region during the formation of intermediate I (the scanning speed of the spectrophotometer recorder is limit ed at 100 nm/min) © question arises concerning spectral changes during the time required to record curve 2e Since the latter spectrum was initiated at 500 nm 1®33 fflin after the addition of HOOH, the spectrum between 450-300 nm re presents the spectral course of the reaction between 2.83-

4®33 min® Equations analogous to Equations 22-26 can be used to describe the composition of the mixture during the early phase of the reaction simply by redefining terms as given in Equations 30=34®

^fmN^ / ^EdFMN^ (30) r _ r Total 0’= M/GWmn] Total (31) r'- M / te,™,]fdFMN _ . (32) r Total T = kjt (33) 124

K = k2/k1 (34)

The composition of the reaction mixture at the beginning and at the end of the recording of curve 2 can be calcu lated (Table 5, Parts A and B, respectively* Part C is the composition of the reaction mixture calculated for the time when intermediate I is expected to be at a max imum concentration)*

Table 5

t (min) t * a* y*

A* 2.83 6*80 0.001 0.929 0.070 B. 4.33 10*40 0.000 0.889 0.111 C. 1.85 4*44 0*012 0.946 0*042

The above results show that the relative concentrations of 9 9 intermediates I (fi ) and II (y ) change inversely by about

4 % during the recording of curve 2* This introduces a small but tolerable error in the determination of their theoretical spectra* Since 0* and y* are expressed as fractions of reactive EdeazaFMN, which constitutes 83% of the total amount of enzyme, curve 2 represents, on the average, 0.83(0.929 + 0.889)/2 = 0*754 intermediate I (35) 0.83(0.070 + 0.111)/2 = 0.076 intermediate II (36) plus 0.17 unreactive EdeazaFMN with each coenzyme

species expressed as a fraction of the total con

centration of enzyme* 125 Before calculating the composition of the reaction mixture represented by curve 3 in Figure 35, tQ must be redefined since Equation 22 implies that at tQ 1, As shown in Table 5 o? is close to 1,0 (0,9^6) after 1,85 min of reaction® As an approximation the new zero time (t£) is set equal to

t£ = tQ - 1,85 min (37)

Consequently, for curve 3 in Figure 35> initiated at t® =

57® 15 min, the spectrum between if50-300 nm represents the spectral course of the reaction between 57.65-59®15 min®

From Equations 17-26 the composition of the reaction mix ture during this interval can be calculated (Table 6,

Parts A and B® Part C is the composition of the reaction mixture calculated for the time when intermediate II is expected to be at a maximum concentration)®

Table 6

t® (min) r a fi Y

A. 57.65 1.73 0,177 0,667 0,156 B, 59.15 1.77 0,170 0,669 0,161 C„ 67.91 2®0if 0,130 0„67*f 0,196

Since a, /?, and Y are expressed as fractions of reactive

EdeazaFMN, which constitutes 83% of the total amount of enzyme, curve 3 represents, on the average,

0,83(0®177 + 0,170)/2 = OolVf intermediate I (38)

0,83(0,667 * 0®669)/2 = 0,554- intermediate II (39) 126

0.83(0,156 + 0.161 )/2 = 0.132 EdFMN-X (lf0)

plus 0.17 unreactive EdeazaFMN, with each coenzyme

species expressed as a fraction of the total concen

tration of enzyme.

The results in Equations 35# 36, 38-40 can besummarized as shown in equations 41 and l+20

C2 = 0.754 Aj + 0.076 An + 0.17 C1 (41)

= 0.144 Aj + 0.554 Ajj + 0.132 Ag^pj^^ + 0.17 (42)

where C^, C2 and are the absorbances along curves

1, 2 and 3, respectively, of Figure 35s and Aj, Ajj

and AgdpM1j_x are the contributions from intermediates

I and II, and EdeazaFMN-X, respectively. Aj, A jj and

AEdFMN-X rePresen^ theoretical spectra of these species at a concentration equal to 100% of the total

enzyme (2.1 x 10*’^ M) in the reaction.

Equation 43 is obtained by substituting Equation 29 into

Equation 42,

C5 = 0.144 Aj + 0.554 An + 0.143 C t + 0.159 C? (43)

Rearranging Equations 41 and 43 yields Equations 44 and

45,

Aj = 1.326 C2 - 0.101 Ajj - 0.17 C1 (44)

A j j = (C3 - 0.144 A j - 0.143 - 0.159 C? )/0.554 (45) 127

Substituting Equation 45 into Equation 44 gives,

Aj = -0.202 C1 + 1„362 C£ - 0.187 C3 + 0.030 C? (46)

Substituting Equation 46 into Equation 45 givess

Ajj. = -0 .2 0 6 Cj - 0.354 C2 + 1.854 - 0.294 C? (4 7 )

By substituting the absorbances in the 450-300 nm region of curves 18 2, 3 and 7 in Figure 35 into Equations 46 and 4 7$ the theoretical spectra for intermediates I and

II9 respectively9 may be drawn (Figure 41)© The spectrum of EdeazaFMN-X9 corrected for the contribution of unre active EdeazaFMN (Equation 29) is also shown in Figure 4 I0

The extinction coefficient calculated for intermediate I at its absorption maximum ( e ^ ^ = 1 2 ,5 0 0 ) is somewhat lower to the value obtained for free 4 & s5°epoxy-5°“deazariboflavin in methanol (^ 3 3 5 = 168500). The value cal culated for enzyme-bound deazaFMN-X (e^q - 19*700) is also lower than the value determined for free deazaFMN-X

(£ 3 4 0 = 24 9 800) .

J® Studies with Flavodoxin Flavodoxins are small flavoproteine (MW 149 500-23,000) which function solely to mediate electron transfer between the prosthetic groups of other proteins (78). They con tain one equivalent of FMN and in many respects resemble the larger flavoprotein dehydrogenases. Both classes of proteins form a blue neutral semlquinone rather than the 20K

I5K

G »0K

5 K

400 450 350900 nm

Figure 41, Theoretical spectra of the coenzyme species formed during the reaction of 5-deazaFMN bound to glyco late oxidase and 3 mM HOOH in 0.1 M sodium phosphate buf fer, pH 7,0, containing 0,3 mM EDTA at 15° C, Curve 1 is the spectrum of the starting compound, EdeazaFMN: A max ( O , 334 (10,900), 402 (12,100), Curve 2 is the spectrum of E-intermediate I: 345 (12,500), Curve 3 is the spectrum of E-intermediate II: 355 (13*200), 375 (13*100), Curve 4 is the spectrum of EdeazaFMN-X: 340 (19,700), 355 (19*400), 129 red anion, they are unreactive towards nucleophilic attack

by sulfite at N-5, and the reduced enzymes are only slowly

reoxidized by oxygen® The converse is generally true

about another class of flavoprotein, the oxidases, and

with oxynitrilase, which exhibits many properties in com

mon with the oxidases® As described in the preceeding

section 5-deazaFMN bound to glycolate oxidase was found

to react with HOOH or m-chloroperoxybenzoic acid to form

EdeazaFMN-X via an epoxide intermediate, similar to that

observed with oxynitrilase® When similar experiments were

conducted with apoflavodoxin reconstituted with 5-deazaFMN

the results differed from those obtained with glycolate

oxidase® These experiments are discussed below®

The absorption spectrum of 5-deazaFMN flavodoxin is

similar to 5-deazaflavin bound to other flavoproteins

(Figure if2)® When ko3 mM HOOH was added to the enzyme

(1 ®65 x 10-5 in Ool M sodium pyrophosphate buffer, pH

7«5j containing 0®3 mM EDTA no change occurred in the ab

sorption spectrum after 10 min at 25° C® Increasing the

HOOH concentration to W7 mM and 0®21 M similarly had no

effect on the spectrum®

Addition of a stoichiometric amount of m-chloroperoxy-

benzoic acid to 1®31 x 10“^ M of enzyme produced no change

in the absorption spectrum after 20 min at 25° C in the

same buffer® Likewise, no reaction occurred at 7-fold and if 5-fold concentrations of the peracid® 130

0 . 2 0

0.I5

LU o z < ffi £K 0.I0 O (0 CD <

0 .05

0.00 300 4 0 0 500 nm

Figure 42. Spectrum of 5-deazaFMN bound to flavodoxin (1.6 x 10-5 m ) in o.l M sodium pyrophosphate buffer, pH 7.5» containing 0.3 mM EDTA. No evidence for the formation of covalent adducts with 5-deazaFMN flavodoxin was obtained in studies at 25°

C with 0*2 M sodium bisulfite in 0*1 M sodium phosphate buffer, pH 7,0, containing 0,3 mM EDTA or with 0,2 M KCN in the same buffer at pH 7,6, Unlike glycolate oxidase and oxynitrilase, the results show that 5-deazaFMN bound to flavodoxin is not susceptible towards nucleophilic attack at position 59 similar to results obtained with native flavodoxin. This can account for the failure of

5-deazaFMN flavodoxin to react with peroxides. The re action of peroxide with free 5-deazaflavin or with 5-deazaflavin bound to glycolate oxidase or oxynitrilase pro ceeds via an epoxide intermediate whose formation is like ly to involve initially nucleophilic attack of R00“ at the

5-position of the coenzyme. While additional enzymes need to be tested, the results suggest that reactions similar to those observed with glycolate oxidase and oxynitrilase will occur only with 5-deazaflavin bound to flavoprotein oxidases. LIST OF REFERENCES

2© Ulrich# I* D©# Vogels# Go D s and Wolfe* R« So (1978)* Biochemistry# 178 4583-4593© 3o Elrich. Lo Do* Vogels. G 0 Do and Wolf©* Ro S« (1979)* Jo Bacterio!© # 140# 20-27©

6© Brustlein# M e and Bruice# To C© (1972)# Jo Am© Cheat© Soco# 94# 6548-6549© 7o Jorns# M e S© and Hersh# L© B© (1974)# ibid©# 96# 4012- ItOI^

9o Averill. B© # Schonbrunn# A0 # Abales# Rn # Weinstock# Le # Cheng. Co# Fisher. J©* Spencer. R© and Walsh# C© (1975)# J® Biol© Chest© * 250# 1603-1605-

11© Hersh. I© B© and Jorns# M© So (1975)# ibid©# 250# 8728-8734© 12© Jorns# M© S© and Hersh# L© B© (1976) in "Flavins and Flavoproteins"# (Ta P© Singer. ®d©)# Elsevier Publish ing Company# Amsterdam# ppQ 362-371© 13© Jorns# M© S© and Hersh# L© B 0 (1976)# ibid©# pp© 372- 380© 14© Jorns# M© S e and Hersh# L© B© (1976)# J© Biol© Chem©* 251# 4872-4881© 132 133

17® Crosarti©® T© H© and Waleh® C© fc (1976)® J® H o i © Ch©So® 251® 329-333® 18® Shinkai® S© ©ad Braie©® T© C© (1973)® J® An© Ohea© Soe© ® 95® 7526-7528©

2 0 o Fisher® H® F 0® Conn® E© E©* ¥©nn©sland® B© and West- heisaer® F© H© (1953)® J® Biol© Oh@a©® 202® 687-697® 21 © Walsh® C0® Fisher® Jo® Spencer® R e® Graham® D© W 0® Ashton® Wo To® Brown® J© E 00 feown0 R© D© and Rogers® E 0 Fo (1978)® Biochemistry® 17® 1942-1951©

22© Bacher® W G and Pfeil® E 0 (1966)® feocheia0 Z© ® 346® 301-321© 23© Massey® V 0® Muller® F 0® Foldberg® R 0® Schuaan® M 0® Sullivan® r© A 0® Howell® L© G©® ffeyhew® S 0 G0« Matthews® R© G© and Foust® G© P0 (1969)® J© mol© Chea© ® 244® 3999-4006© 24© Swoboda® B© E® P© and Massey® V* (1966)® ibid0® 241® 3409-3416® 25© Jorns® M© So (1979)® Ibid©® 254® 12145-121520

27® Massey® V 8® Palmer© G a and fellow® D0 (1973) in "Oxidases and Related Redox Systems"® (T© E 0 King® Ho S© Mason and M 0 Morrison® eds©)® university Park Press® Baltimore® Vol© 1® pp« 25-49®

2 8 0 Massey® V©® Palmer® G© and Ballow® D0 (1971) in "Flavins and Flavoproteins"® (Ha Kamin® ©d®)® Univer sity Park Press® Baltimore® pp© 349-361©

30o Muller® Fo® Jarbandhan® T 0® Gast® R 0 ©nd Grande© H© J» (1975)® in "Reactivity of Flavins" 0 (Ke Yagi® ®d0 )® University of Tokyo Press® Tokyo® pp© 51-70© 134

31. Muller, F., Grande, H. J. and Jarbandhan, T. (1976), in "Flavins and Flavoproteins”, (T. P. Singer, ed.), Elsevier Publishing Company, Amsterdam, pp. 38-50®

34. Massey, ¥© and Hemmerich, P. (1975) in "The Enzymes", (P. Boyer, @d.), Academic Press, New fork, Vol® 12, pp. 191-252®

35® Strickland, S. and Massey, V. (1973), J® Biol. Chem., 248, 2953-2962.

36. Specter, T. and Massey, V. (1972), ibid., 247, 7123- 7127® 37® Presswood, P. and Kamin, Hs (1976) in "Flavine and Flavoproteins", (T. P.-S±ngBFe @d.), Elsevier Publish ing Company, Amsterdam, pp0 145-154®

38. Hastings, Jo 1® and Gibson, Q® H. (1963), J® Biol® Chem®, 238, 2537-2554® 39® Gibson, Q. He, Hastings, J. W., Weber, G., Duane, W. and Mass®, J® (1966), in "Flavins ©ad Flavoproteins", (E© Co Slater, ©d.), ELeevier Publishing Company, Amsterdam, pp. 341-359® 40. Watanabe, T. and Nakamure, T. (1976), J© Biochem., 79, 489-495® 41. Mager, H. 1. X. (1977), Tetrahedron, 33, 981-989®

43® Chan, R® L® and Bruice, T 0 C. (1977), J® Am. Chem® Soc®, 99, 6721-6730.

44. Chan, B. L. and Bruice, T. C. (1978), ibid., 100, 7375-7382.

45® Pokora, A. and Jorns, M. S. Unpublished results.

46. Schwarts, No N. and Blumberg®, J® H. (1964), J® Org. Chem., 29, 1976-1979® 135

5 4 © Karrer© P© and Meerweln© H© P© (1935)® Helv© Chem® Acta® 1 8© 1130-1134© 55o Cresswell. H® M® and Wood® H® C® S® (i960)® J® Chem© See©® 4768-4775©

58® Koltholf® 2© M® and Sandell© E© B© (1943) in "Textbook of Quantitative Inorganic Analysis'98© The Macmillan Com pany © Novr York© p© 6 3 0 ®

66 * Charpentier-Morizo9 M®9 Laszio* P® and Mayer* M® (1966)8 Bull® Soc® Chi®® Fr. „ 2264-2269° 67* Dullaghan* M® E® and Nord* F® F® (1953)» Mikroehim® Aeta9 17*21,

68* Everett* A® J, and KLakoff, 6® J® (1953)© Trass® Fara day Soc,8 49 0 410* © 69© Savington* P® R. (1969) in "Bata deduction and Error Analysis for the Physical Sei@nc@swg McGraw-Hill Book Company* Mew Y@rk0 pp® 219*222®

70* Cottong F® A© and Wilkinsong G® (1967) in "Advanced Inorganic Ch©£&stryc,0 Xmt©rsei©ae© Publishers* M@w York© pp® 308-310®

71* Curci* R. and Edwards* J® 0® (1970) in "Organic P@r- o&ided51* (D® Swern® ©d, )g fil©y-Int©rseiene©g Hew York8 Vol 1* pp, 2Q5-207©

76 o Bevington0 P® R, (1969) in "Data Reduction and Error Analysis for th© Physical Sciences% McGraw-Hill Book Company g, Mew York9 pp, 2 4 0 - 2 4 2 ®

77® Frost9 A, A® and Pearson* R® G® (1961) in "Kinetics and Mechanisms"* 2nd® ©d®0 John Wiley and Sons* Ine*f Hew York* pp® Ib6-169,

78* ftoyhew* S® G® and Ludwig* M® L® (1975) to "The Enzymes" (P® Boyer* ©d®)« 3rd, ©d®* Academic Press* Hew York* Vol. 12* pp. 57-118.