INTERACTION OF TRACE METALS WITH PHENOTHIAZINE DRUGDERIVATIVES, III. THEORETICAL PART*

BY DONALD C. BORG AND GEORGE C. COTzIAS

MEDICAL RESEARCH CENTER, BROOKHAVEN NATIONAL LABORATORY, UPTON, L. I., NEW YORK

Communicated by Donald D. Van Slyke, February 23, 1962

In the preceding papers," 2 a chromogenic reaction was reported between deriva-tives related to chlorpromazine and appropriate oxidants, including higher valencestates of metals that exist in cells. Under mild aerobic conditions aqueous man-ganous ions were similarly active, although with lower efficiency.' Several inde-pendent experimental means served to identify the chromatic products as semi-quinone radical ions and to characterize some of the reaction steps leading to theirformation and disappearance.2

In this paper, the data of the preceding ones are correlated with each other andare related to available general biological information. From these correlationssome hypotheses emerge: The molecular properties that underlie the pharmacologi-cal mechanisms common to all phenothiazine drugs are noted and are distinguishedfrom the properties that differentiate them. A series of electron transfer interac-tions is set forth describing the oxidation-reduction sequence of chlorpromazine.Possible mechanisms of biological activity associated with the formation of asemiquinone radical ion are proposed. Metal-phenothiazine reactions are sug-gested as prototypes for the further study of the biochemical roles of trace metalions in vivo.

Relationship of the Present Work to Chemical Structure-Activity Correlationsamong Phenothiazine Drugs.-The N-amino substituted phenothiazine drugs ex-hibit an extremely broad spectrum of pharmacological properties. Well-docu-mented activities in vivo include "tranquilization," antiemesis, antipruritic action,local anesthesia, antibiotic and antihelminthic potency, tuberculostasis, sympa-tholysis, antagonism of serotonin, antihistamine effects, efficacy in acute porphyria,and hypocholesterolemic capacity.3-7 Two features of these many and diverseattributes are striking: (1) All phenothiazine drug congeners share to a greater orlesser extent this range of properties, differing primarily only in their potenciesand in the relative degree of prominence of these various capabilities, when comparedwith one another.4 68 (2) Attempts to unify this pluripotential action spectrumthrough correlation with purely pharmacological properties-such as adrenolysis,cholinolysis, histaminolysis, or sedation-have failed.7' 9

The qualitative similarity of the phenothiazine drug derivatives suggests a com-mon chemically reactive site that is critical to the action of all the congeners as aclass. If chromogenic reactions with metal ions be postulated as an index ofpharmacological action of these drugs, the experiments of Part I (see ref. 10) es-tablish the thiazine nucleus with its thioether linkage as the crucial molecularlocus. Further evidence to implicate the central role of the heterocyclic sulfurwas provided by the failure of chlorpromazine sulfoxide to form a colored reactionproduct with manganese, iron, or cobalt.' In a strikingly parallel fashion, thesulfoxide is equally inert in many metabolic systems that are inhibited by thi-azines,"'-14 and it is inactive as a psychotropic agent in man.'5 Furthermore, a

643

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020

644 BIOCHEMISTRY: BORG AND COTZIAS PROC. N. A. S.

drug which resembles chlorpromazine except for the substitution of an ethylenebridge for the thioether linkage ("imipramine"; (3-dimethylaminopropyl)-di-benzyl amine) also was inactive in vitro in a test system which had responded to allphenothiazines that were tried.14The thesis emerging here is that the shared characteristics of diverse substituted

phenothiazines derive from qualities strongly dependent on the thiazine ring. Inlater discussion, these qualities will be identified with the electron-donor propensi-ties of this group of drugs, an aspect of thiazine chemistry that has been empha-sized recently by several authors.'6-'8 In the present experiments, metal cationscapable of undergoing univalent reduction have served as the electron acceptor.'. 2Of these, manganese, iron, and cobalt are recognized microconstituents of organ-isms; and redox transitions at least of iron between the tri- and divalent states areknown to occur during metabolism."9 Manganese also is thought to become tri-valent at some points in its physiological cycle;20 but in any case, the conditions ofpH and oxygen tension required for the reaction of thiazine drugs with man-ganous ion are within the range of those encountered in the intracellular milieu.These experiments do not prove that phenothiazines interact with metals in

vivo. Nonetheless, the present demonstration of specific reactions with biologicaltrace metals under conditions which might prevail in vivo is highly provocative.The work of Yamamoto et al.,2' disclosing an inhibition by chlorpromazine of enzymescontaining ferric iron but not of those containing cupric copper, further supportsthe interpretation that the chromogenic reactions with metal ions may model anessential step in the chemical pharmacology of thiazines.The unification of thiazine activities about the properties attendant to a com-

mon structural nucleus also must be reconciled with the pharmacological differencesknown to exist among the various drug congeners. Similarities between pheno-thiazines are striking while their dissimilarities appear, by comparison, to beof lesser order: for example, unequal clinical potency on a weight basis or a dif-ference in the relative -prominence of their properties in Vivo.4 6,7 Nonetheless,the differences between the extremes of the phenothiazine spectrum are signifi-cant. It is evident that substances of similar chemical character could havemarkedly disparate effects in vivo should they be distributed differently withinthe organs and tissues of intact animals or even within individual cells thereof.7 8Alteration of lipid-water partition coefficient would change the distribution ofa drug at both the tissue and intracellular levels. Hansson and Schmiterlowdemonstrated a great difference in tissue distribution between two phenothia-zine drugs only one of whose side chains provided water solubility at physiologicalpH. There resulted a wide divergency in their biological fates and in their phar-macological actions.22 Furthermore, since the molecular- charge of a thiazinemolecule depends strongly on the extent of protonation of prosthetic groups,metal cation repulsion and water solubility should be sharply sensitive to pH, andthis pH dependency should be a function of the basicity of the nitrogen atomof the nuclear radical group ("RI" in Fig. I-l). This influence of pH was observedin the experiments reported here (Fig. 1-3).

In addition to altered drug absorption and distribution in vivo, changes inthe prosthetic groups could modify the distinctive chemical nature of the thiazinenucleus without changing its fundamental character. If a proclivity for electron

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020

VOL. 48, 1962 BIOCHEMISTRY: BORG AND COTZI.-IS 640

donation is a critical phenothiazine property, as has been implied in this discussion,then ring substitutions that affect the basicity of the reactive [thioether] site shouldalter pharmacological activity quantitatively and might even account for theenhanced potency of many ring-halogenated derivatives. 6 23

Alternative effects of prosthetic groups are also possible. For example, sterichindrance has been reported with regard to the inhibition of amine oxidase bychlorpromazine analogues,23 and the characteristic properties of certain largeprosthetic radicals have been superimposed on the traits of the phenothiazine groupin other cases.4 24

Formulation of the Integrated Reaction Sequence for Bivalent Oxidation of Chlor-promazine.-From the series of reactions already defined,2 it Was concludedthat a red semiquinone radical (Fr) could be formed from the reduced (native)chlorpromazine (R) by the donation of one electron.2 This can be represented bythe half reaction

R Fre+ e9, (1)

where appropriate metal ions or other oxidants can serve as electron acceptors.Clearly, from the conservation of charge implied by equation 1 the semiquinone ofchlorpromazine must be a charged radical ion. However, reducing agents and re-ducing systems may inhibit or reverse semiquinone formation (Table II-1),and the native drug may reform spontaneously from the free radical by dismutation(Figs. II-9, 10). Therefore the reaction given by equation 1 is reversible.A further oxidation to the colorless chlorpromazine sulfoxide (- S=-O) was

revealed by oxidimetric titrations (Fig. II-2, 3, 4) and by stoichiometric considera-tions (Fig. II-9). Ignoring residual charges, this step can be represented by

Fr ==SO. (2a)

Attempts to reverse this reaction directly with reducing agents or by re-equilibra-tion with added reduced chlorpromazine failed.2 Moreover, in oxidimetric titra-tions the addition of nearly two equivalents of electron acceptor produced a gradualdownward drift of recorded potentials (Table I1-2 and Fig. 11-4). This wasshown to be consistent with a relatively slow, irreversible terminal reactionstep.2 Yet, reproducible and characteristic oxidation-reduction titration curveswere obtained with lower concentrations of reactants (Figs. 11-2-4); so, as Conanthas said,25 ... . one can hardly go wrong in accepting the reversibility of thesystem as established." This may be redefined to mean only that somereversible step must be controlling.25 However, if the recorded emf's from thedata cited in Part II are taken to represent "apparent oxidation potentials,"25than an " . . . irreversible transformation [of an oxidation product] (under condi-tions where its rate is slower than that of the initial reversible oxidation) should bealso capable of formulation in terms of 'apparent oxidation potentials.' "25 Polaro-graphic studies of organic substances often have demonstrated redox reactions ofthis type, wherein the initial reaction products are in equilibrium with the sub-strate from which they are formed but are then removed by an irreversible secondaryprocess.26 Accordingly, equation (2) may better be written as

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020

646 BIOCHEMISTRY: BORG AND COTZIAS PROC. N. A. S.

fast slowFr =e+ Ox ==S 0 (2b)

again neglecting to balance electrical charges.For the case of unsubstituted phenothiazine, Craig et al. have proposed a sequence

of reactions to account for the tautomerization of a reversible oxidation product,probably a quininoid phenazothionium ion (III), to the sulfoxide (V) by means of aproton exchange involving a sulfonium base (IV) under weakly acid conditions.27The reductive chlorination of some sulfoxides of thiazine derivatives by refluxingwith concentrated acid also was presumed to involve phenazothionium ions andsulfonium bases as intermediates.28 A similar tautomerization for chlorpromazinecan be written as

R R R R

N N0 CyH

~ N yCl N y C.O

OH °+S

OHe HeIIIa IIIb IV V

or over-all, where the diacid phenazothionium ion (represented by two possibleresonance forms IIIa and IIIb) is taken to be the reversible primary oxidation prod-uct, Ox:

Ox2@D + H20 -0 =S=0 + 2H@. (3b)The internal electrostatic repulsion of the phenazothionium ion (III) might beexpected to force the reaction almost completely in the forward direction, as indi-cated by equation (3.)However, equation (3b) suggests that with sufficient hydrogen ion concentration

the reaction might be reversed. The observed reduction of phenothiazine sulfoxideswith hydrofluoric acid28 or in association with chorination, as noted in the previousparagraph, confirms the possibility of reverting from the sulfoxide to lower oxida-tion states, at least under severe conditions. Subsequent reversal of equation (2)would give again the colored semiquinone radical ion. This hypothetical reactionsequence would explain the reported formation of identical chromophores fromchlorpromazine and from its sulfoxide in the presence of both iron and concentratedacid. 29-31

In addition to the primary sequence of reactions for chlorpromazine, certain"complicating" reactions were observed, such as dimerization of reduced chlor-promazine (I) (Figs. II-5, 7) and dismutation of the semiquinone radical ion (II)(Figs. II-9, 10). Incorporating these reactions, the complete scheme of interactionsrelating to the oxidation of chlorpromazine and its congeners via univalent electrontransfer steps can be summarized by the following:32

M/2 R2 5 9-e -'sE - o2!4 - =S=O + 2 HA

(4)I II III Vt lit A a To~~~+'

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020

VOL. 48, 1962 BIOCHEMISTRY: BORG AND COTZIAS 647

In equation (4), the molecular formulations of III and V have been given before.Native chlorpromazine is

R

,N >,C1

where R is dimethylpropylamine (Fig. 1-1). Several resonance forms of the freeradical ion (Fr'3) may be envisioned:

R R R

aSN0'Cr1 3c12yCl

Ila IIb IIc

However, the consideration of electron spin resonance (ESR) data which followseems to rule out important contributions from the resonance form Ila, whiledelocalization of the unpaired electron into the molecular Tr orbitals of the aromaticthiazine nucleus (IIc) would appear most likely.

Electron resonance data.-The ESR data presented in Part II (Fig. II-12)revealed the usual hyperfine splitting due to divalent manganese.33 34 In the pres-ence of oxidizing cations the free radical signals were recorded as single peaks withoutevidence of the fine structure that would be anticipated if the free electron were de-localized through the 7r-electron system of the phenothiazine nucleus,33-35 as repre-sented by structure I1c. However, in those experiments paramagnetic ions werepresent in sufficient concentration (0.005 M) to cause line broadening of the free radi-cal signal due to spin-spin interactions,334 thus "smearing out" any hyperfine reso-nance structure. The '-30 gauss width of the ESR signals from radicals in solutionwith metal ions (Figs. It-12, a and b) is consistent with this, while in the amorphoussolid (Fig. II-12c) comparable broadening occurred as a result of the randomorientations of the anisotropic components of the hyperfine interactions.33 Ac-cordingly, with the radical stabilized in strong acid in the absence of metals, a nar-rower ESR pattern with considerable fine structure was observed (Fig. II-13),36thus implying that the delocalized semiquinone electronic structure suggested byIlc is most likely. The asymmetry of the resonance pattern in 11.5 N sulfuric acid-with loss of detail in the tracing with increasing magnetic field (Fig. II-13)-may result from the high viscosity of the H2S04, which so slows the thermal tumblingof the semiquinone molecules that the anisotropic part of the hyperfine interactionsis not quite averaged out during the time of resonance.

In Part II, no ESR signal was obtained from the colorless long-lived excited stateof chlorpromazine initially produced by ultraviolet light and whose ultravioletabsorption was indistinguishable from that of the native drug.2 Although a weakspin signal was elicited from chlorpromazine discolored by prolonged irradiation,2the undegraded ultraviolet product may be a triplet state of the drug, because themost stable excitation states of molecules usually are triplet,37 and ESR signals arenot expected from triplet-excited molecules in solution.33' 34 38 Another experi-mental observation consistent with the ultraviolet product being triplet is its reac-

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020

648 BIOCHEMISTRY: BORG AND COTZIAS PROC. N. A. S.

tions with metal, forming characteristic chromopbores ;2 because triplet states reactreadily with paramagnetic ions.'9 Moreover, semiquinone radicals have beenformed directly from triplet states of thionine, a thiazine derivative.40' 41 Thismay explain why Forrest et al. obtained electron resonance for crystals of derivativesof irradiated chlorpromazine, although they did not examine directly the irradiatedmaterial.42

Speculation on Biological Effects Associated with the Formation of the SemiquinoneRadical.-The significance of the semiquinone form on the effects of phenothiazineson living systems remains moot. No direct correspondence between free radicalformation and biological effect was established by the data of Parts I and II, butthe structure-activity correlations may logically be extended to hypothesize thatat least a part of the pharmacological actions of phenothiazine drugs may stem fromtheir transformation to semiquinone radicals in vivo. Craig and his colleaguesreached a similar conclusion after correlating potentiometric data with antihel-minthic activity of phenothiazine derivatives: namely, that pharmacological efficacyrequired formation of a high proportion of stable semiquinone radical." Theseauthors concluded that the semiquinone per se probably was not the active agent intheir experiments because phenothiazine oxidation potentials were incompatiblewith metabolic processes,43 but this current work has shown oxidation of chlor-promazine congeners by ions of transition-group metals that exist in living material1(Fig. II-3). Further evidence to support the formation of free radicals during thebiological action of chlorpromazine is provided by Yamamoto et al., as noted pre-viously. They showed2' that the drug attacks the iron moiety only of those en-zymes (cytochrome oxidase, catalase) where it exists in the trivalent state or mustundergo redox reaction but not of enzymes containing divalent iron (homogenticase)or copper (ascorbic acid oxidase); and it is only the ferric iron that formed semi-quinone from chlorpromazine in our hands' (Fig. II-3).Mechanisms whereby such a putative formation of semiquinone radicals in vivo

might alter cellular metabolic processes also may be conjectured from the presentdata. The chlorpromazine radical ion is chemically labile and reacts readily withboth [univalent] electron donors (Table II-1) and acceptors (e.g.,; Fig. II-3).Hence, it could act in vivo to shunt the normal electron transport sequence of therespiratory chain of enzymes or to divert electrons into nonphosphorylating path-ways, thus explaining some of the reported effects of chlorpromazine on respiringand phosphorylating biological test systems.7' 11 14, 23, 44 On the other hand, theimpact of semiquinone formation upon cellular function could be exerted indirectlythrough the metal ions that might be involved. For example, the conversion of aphenothiazine drug to its semiquinone via oxidation in situ by a metal' (Fig. II-3),or the further redox transformation of the radical by reducing (Table I-1) or oxidiz-ing (Fig. 11-3) metal cations, could alter the local availability of ions (e. g., Fe+++,or Co+++) that might be rate-limiting cofactors in some essential biochemicalprocesses.

Metal-Phenothiazine Interactions as Models of Trace Metal Reactions that Occurin Vivo.-The value of the metal-thiazine reactions in providing added insight intothe role of metal ions can be illustrated by several examples of peroxidase reactions:In Part II, work of Cavanaugh4' was cited wherein peroxidase could produce froma chlorpromazine-H202 mixture a red chromophore that appeared identical with

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020

VOL. 48, 1962 BIOCHEMISTRY: BORG AND COTZIAS 649

the semiquinone radical described in these communications.1' 2 Accordingly, itcould be concluded from the chlorpromazine reaction alone that peroxidase canfacilitate separate univalent electron transfer steps in the reduction of peroxides,producing free radical forms of the substrates. This conclusion is reasonable, be-cause Yamazaki and Piette were able to show by ESR that the main pathway of theperoxidase reaction (and of ascorbic acid oxidase) involved the formation of freeradicals from the substrates."The metal-thiazine model also leads to a logical explanation of the following com-

plex interactions that are of biochemical interest: Reduced pyridine nucleotides,epinephrine, thyroxine, phenolic estrogens, and ferricytochrome c are oxidizableby H202 and peroxidase; and when they are not the primary substrates themselves,thyroxine and estradiol can stimulate these reactions.47 Divalent manganese andmolecular oxygen can replace peroxide in the oxidations involving peroxidase.47-49The mechanism of the manganese effect in these aerobic reactions has not beenclear, but manganese activation of the enzyme has been suggested,0' 51 and Kentenand Mann proposed that divalent manganese was oxidized enzymatically as theresult of its reduction of primary [phenolic] oxidation products formed by the en-zyme from its substrates.52 In keeping with the preceding paragraph, the for-mation of free radicals of the substrates may be presumed." However, experiencewith manganese-phenothiazine reactions showed that substrate radicals can beproduced by manganic ions that are formed not enzymatically but [transiently]by autoxidation.' In the same way, autoxidized manganese may serve as an elec-tron acceptor in the peroxidase reactions. The role of the enzyme protein thenmight be mainly to potentiate the oxidation via ligand transfer of electrons55involving a substrate-donor: enzyme: metal-acceptor complex. This speculativehypothesis is supported by Mazelis' recent work56 showing nonenzymatic produc-tion of an oxidant by pyridoxal phosphate, Mn++, and any of many amino acidsat neutral pH. There is a close parallel with the manganese-thiazine model,'because no oxidation occurred under anaerobic conditions, no metal could replaceMn++ (except Co++ at 2% relative efficacy), and although the oxidant was notidentified, it was probably univalent since it was detected by using I- as a reduc-tant.56From this example, it is evident that redox transformations may underly a

significant fraction of the roles filled by [trace] metal ions in vivo. In fact, it hasbeen suggested that in at least certain instances the ability of physiological systemsto distinguish sharply between different metals derives from their distinctive redoxproperties.20' 35 Appreciable selectivity may be achieved on this basis,35 as is il-lustrated by the high specificity of the manganous ion for the autoxidation reactionsinvestigated in Part I of this work and for the similar oxidations reported byMazelis.56

In addition to possible service as a model for reactions of metals and free radicals,phenothiazine semiquinone formation may be useful as a biochemical test to deter-mine metal oxidation states or the presence of univalent oxidation potential insitu in chemical samples. Either the characteristic absorbancies (even visible coloras a simple test) (Fig. II-1) or ESR signals of the semiquinone ions can be recorded.The previous discussion of the red color produced by Cavanaugh45 serves as a goodexample of this application. In a similar fashion, the electrochemical potential of

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020

650 BIOCHEMISTRY: BORG AND COTZJAS PROC. N. A. S.

a system may be determinable by observing the formation or quenching of thechromophoric thiazine radicals, thus employing them as indicators57 of oxidation.Summary.-Data concerning the chromophoric semiquinone radical ions pro-

duced from phenothiazine derivatives by certain metal ions and by other univalentoxidants were discussed, and hypotheses were advanced concerning their signifi-cance. The sequence of electron transfer interactions involved in the full oxidationof such thiazine congeners was formulated, taking note of several complicatingreactions previously encountered.

It was proposed that the many common characteristics of all substituted pheno-thiazines derive from the reactive properties strongly dependent on the thioethersites of their thiazine rings. The demonstrated specific reactions with trace metalsunder mild conditions raised the further possibility that these properties may im-ply interaction with metals in vivo. In any case, the hypothesis demands that thechemical pharmacology of all thiazine derivatives be similar, regardless of the siteof biological action. However, the localization of drug molecules, at both the grossand subeellular levels, should depend strongly on the prosthetic radicals, which thusprovide each congener with its distinctive pharmacological character within thebroader pattern of phenothiazine group behavior.

Possible modes of biological reactivity associated with the formation of the semi-quinone radical were advanced: The chemical lability of the free radical might shuntnormal electron transfer or phosphorylation pathways, or the availability of bi-ologically essential trace metal ions might be altered at critical intracellular sites.The well-delineated metal-phenothiazine reaction system also was proposed as auseful model for the more general study of free radical-metal interactions that mayoften be related to the biochemistry of metal ions in the respiratory chain of bi-ological electron transport or to their role as cofactors in enzymic reactions. Itis also possible that formation of the highly colored phenothiazine semiquinoneions might serve as a test of [metal] oxidation states in chemistry or of the presenceof other interacting free radical reactions.

* This work was supported by the U.S. Atomic Energy Commission. Parts of it have beenpresented at the Conference on Biological Aspects of Metal-Binding (University Park, Pa.,September, 1960) and to the Association for Research in Nervous and Mental Disease (New York,N. Y., December, 1960).

1 Borg, D. C., and G. C. Cotzias, these PROCEEDINGS, 48, 617 (1962).2 Ibid., 48, 623 (1962).3 Massie, S. P., Chem. Rev., 54, 797 (1954).4 Friend, D. G., Clin. Pharm. Therap., 1, 5 (1960).6 Ayd, F. J., Jr., Med. Clin. N. Am., 45, 1027 (1961).6 American Medical Association, Council on Drugs, J. Am. Med. Assoc., 177, 245 (1961).7 Cotzias, G. C., and D. C. Borg, Association for Research in Nervous and Mental Disease,

Proceedings, vol. 40 (Baltimore: Williams and Wilkins (in press), 1962).8 Cotzias, G. C., D. C. Borg, E. R. Hughes, A. Bertinchamps, and P. S. Papavasiliou, Rev.

Canad. Biol., 20, 289 (196 1).9 Gyermek, L., I. LizAr and A. Zs. CsAk, Arch. Int. Pharmacodyn. Therap., 107, 62 (1956).

10 For convenience, references to tables and figures of Part Ii and Part II2 of this seriesof papers will employ the abbreviations I and II, respectively; e.g., Fig. II-3 will refer to Fig. 3of Part II.

11 Bernsohn, J., L. Namajuska, and L. S. G. Cochrane, Proc. Soc. Exp. Biol. Med., 92, 201(1956).

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020

VOL. 48, 1962 BIOCHEMISTRY: BORG AND COTZIAS 651

12 Erd6s, E. G., N. Baart, S.; P. Shanor, and F. F. Foldes, Arch. Int. Pharmacodyn. Therap., 117,163 (1958).

13 Moraczewski, A. S., and K. P. Du Bois, Arch. Int. Pharmacodyn. Therap., 120, 201 (1959).14 Kistner, S., Acta Chem. Scand., 14, 1389 (1960).16 Davidson, J. D., L. L. Terry, and A. Sjoerdsma, J. Pharm. Exptl. Therap., 121, 8 (1957).16 Karreman, G., I. Isenberg, and A. Szent-Gyorgyi, Science, 130, 1191 (1959).11 Szent-Gyorgyi, A., Introduction to a Submolecular Biology (New York: Academic Press,

1960).18 Orloff, M. K., and D. D. Fitts, Biochim. Biophys. Acta, 47, 596 (1961).19 Underwood, E. J., Trace Elements in Human and Animal Ni trition (New York: Academic

Press, 1956).20 Cotzias, G. C., Fed. Proc., 20, Suppl. No. 10, 98 (1961).21 Yamamoto, L., N. Adachi, Y. Kurogochi, and A. Tsujimoto, Jap. J. Pharm., 10, 38 (1960).22 Hansson, E., and C. G. Schmiterlow, Arch. Int. Pharmacodyn. Therap., 131, 309 (1961).23 Nakajima, H., J. Biochem., 46, 559 (1959).24 Long, J. P., A. M. Lands, and B. L. Zenitz, J. Pharm. Exptl. Therap., 119, 479 (1957).25 Conant, J. B., Chem. Rev., 3, 1 (1926).26 Miller, 0. H., Ann. N. Y. Acad. Sci., 40,91 (1940).27 Craig, J. C., M. E. Tate, F. W. Donovan, and W. P. Rogers, J. Med. Pharm. Chem., 2, 669

(1960).28 Schmalz, A. C., and A. Burger, J. Am. Chem. Soc., 76, 5455 (1954).29 Wechsler, M. B., and I. S. Forrest, J. Neurochem., 4, 366 (1959).30 Fels, I. G., and M. Kaufman, Nature, 183, 1392 (1959).31 Forrest, I. S., and F. M. Forrest, Clin. Chem., 6, 11 (1960).32 Equation (4) can apply to the interaction of metal cations with positively charged forms

of soluble phenothiazines,l because many examples are known wherein electron transfer reactionsbetween highly charged ions of the same sign are very fast (Halpern, J., Quart. Rev., 15, 207 (1961)).

33 Ingram, D. J. E., Free Radicals as Studied by Electron Spin Resonance (London: Butter-worths Scientific Publications, 1958).

34 Whiffen, D. H., Quart. Rev. (London), 12, 250 (1958).3 Borg, D. C., Fed. Proc., 20, Suppl. No. 10, 104 (1961).36 Similar observations have been made by L. H, Piette (personal communication). Billon,

Cauquis, and Combrisson also reported recently the presence of hyperfine structure in the ESRsignals of electrochemically formed radical ions from unsubstituted phenothiazine and from tworing-methylated derivatives (C.R. Acad. Sci. (Paris), 253, 1593 (1961).

37 Reid, C., Quart. Rev. (London), 12, 205 (1958).38 Hutchison, C. A., Jr., and B. W. Mangum, J. Chem. Phys., 29, 952 (1958).39 Porter, G., and M. R. Wright, Farad. Soc. Disc., No. 27, 18 (1959).40 Ainsworth, S., J. Phya. Chem., 64, 715 (1960).41 Hatchard, C. G., and C. A. Parker, Trans. Farad. Soc., 57, 1093 (1961).42 Forrest, I. S., F. M. Forrest, and M. Berger, Biochim. Biophys. Acta, 29, 441 (1958).43 Craig, J. C., M. E. Tate, G. P. Warwick, and W. R. Rogers, J. Med. Pharm. Chem., 2, 659

(1960).44 Desci, L., Psychopharm., 2, 224 (1961).45 Cavanaugh, D. J., Science, 125, 1040 (1957).46 Yamazaki, L., and L. H. Piette, Biochim. Biophys. Acta, 50, 62 (1961).47 Klebanoff, S. J., J. Biol. Chem., 234, 2480 (1959).48 Akazawa, T., and E. E. Conn, J. Biol. Chem., 232, 403 (1958).49 Klebanoff, S. J., Biochim. Biophys. Acta, 48, 93 (1961).50 Chance, B., J. Biol. Chem., 197, 577 (1952).61 Mudd, J. B., and R. H. Burris, J. Biol. Chem., 234, 3281 (1959).62 Kenten, R. H., and P. J. G. Mann, Biochem. J., 46, 67 (1950).63 Ibid., 53, 498 (1953).64 The presence of free radical intermediates in the peroxidase oxidations of pyridine nucleotides

n &Iso suggested by the synergistic effect of sulfite on the enzymic reactions.49 Radical formation

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020

652 BIOCHEMISTRY: EISENSTADT, KAMEYAMA, AND NOVELLI PRoc. N. A. S.

also was inferred in the similar case of oxidations of pyridine nucleotides by ceruloplasmin (Walaas,E., and 0. Walaas, Arch. Biochem. Biophys, 95, 151 (1961)).

55 Halpern, J., Quart. Rev. (London), 15, 207 (1961).66 Mazelis, M., Arch. Biochem. Biophys., 93, 306 (1961).57 Clark, W. M., Oxidation-Reduction Potentials of Organic Systems (Baltimore: Williams and

Wilkins, 1960).

21 REQUIREMENT FOR GENE-SPECIFIC DEOXYRIBONUCLEICACID FOR THE CELL-FREE SYNTHESIS OF f-GALACTOSIDASE

BY J. M. EISENSTADT,* TADANORI KAMEYAMA,t AND G. DAVID NOVELLI

BIOLOGY DIVISION, OAK RIDGE NATIONAL LABORATORY, OAK RIDGE, TENNESSEE

Communicated by Alexander Hollaender, February 19, 1962

The following article (pp. 659-666) describes the preparation and some propertiesof a cell-free system from E. coli that catalyzes the de novo synthesis ofinduced fl-galactosidase.1 The system requires the presence of a'particle and super-natant fraction obtained by centrifugation at 105,000 X g, inducer, an energysource, amino acids, and the nucleoside di- and triphosphates. The cell-free systemalso catalyzes the incorporation of C14 amino acids into protein. The rate of in-crease in enzyme activity is proportional to the rate of C14 labeling of protein thatis precipitable by anti-sera to fl-galactosidase. The incorporation of amino acidsas well as enzyme synthesis is inhibited by treatment with RNase and by the pres-ence of chloramphenicol. One of the surprising observations made was that thesynthesis of #-galactosidase as well as the incorporation of C'4-leucine into proteinwas sensitive to DNase. The present communication represents an extension ofthese observations and presents evidence for the requirement for gene-specificDNA in the cell-free system. A preliminary account of this work has appeared.2

Materials and Methods.-Cell-free extracts from pre-induced cells of E. culi were prepared asdescribed' except that the buffer used was 0.01 M Tris-HCl pH 7.5, 0.01 M Mg acetate, and 0.001AlM j-mercaptoethanol. The reaction mixture for synthesis of ,B-galactosidase was as describedexcept that glutathione (5 Mimoles/ml) was included. Other methods were as described inref. 1. The reaction mixture for the synthesis of t# -galactosidase was composed of the followingin jtmoles/ml: Tris buffer, pH 7.5, 100; Mg acetate, 4; MnCl2, 2; ATP, 10; methyl-,3-D-thio-galactopyranoside (TMG), 5; PEP, 10; PEP kinase, 100 Mg; UTP, GTP, CTP, UDP, GDP,CDP, 0.03 each; GSH, 5; L-amino acids, 50 jug, representing a mixture which reflects the aminoacid composition of ,3-galactosidase described by Wallenfels and Arens.'

Conditions for in vivo inhibition experiments: E. coli was grown overnight in the syntheticmedium with 0.5% glycerol. The cells were diluted tenfold in fresh medium and distributed inculture tubes equipped with aeration tubes and incubated at 370C. Early log phase cells wereX irradiated, with aeration at room temperature, using a G.E. Maxitron X-ray machine operatedat 250 kvp and 30 ma with 3 mm of added aluminium filtration (hvl, 0.34 mm of Cu). The sampleswere mounted in a Lucite frame 15 cm from the target and received an X-ray dose of 2,100 r/min.Each experimental dose point consisted of two such tubes, one which was X irradiated and acontrol that was shielded from the X-rays. Immediately after irradiation TMG was added to aconcentration of 5 X 10( -- A! and incubation continued. One ml samples were withdrawn to pre-chilled test tubes and kept on ice for later assay. To these tubes 0.1 ml of toluene and 10 /Ag ofsodium desoxycholate were added. The tubes were incubated at 370C for 10 min with frequentshaking and assayed for enzyme formation and total protein. Cell-free preparations were also

Dow

nloa

ded

by g

uest

on

Mar

ch 2

6, 2

020