A STUDY OF THE HYDROGENLYASE REACTION WITH SYSTEMS DERIVEDFROM NORMAL AND ANAEROGENIC COLI-AEROGENES BACTERIA'

HOWARD GEST AND HARRY D. PECK, JR.'Department of Microbiology, School of Medicine, Western Reserve University, Cleveland,

Ohio

Received for publication March 24, 1955

Formate is anaerobically decomposed tomolecular hydrogen and carbon dioxide by manycoli-aerogenes bacteria according to the equation:

HCOOH C0 + H2

The mechaDism of this process, i. e., the hydro-genlyase reaction, has been the subject of con-siderable research and speculation over severaldecades. One fundamental aspect of the reactionwhich has received particular attention is thequestion of whether one or more enzymes arerequired for the conversion. It seems reasonablethat definitive and unambiguous evidence relat-ing to this point will necessaily require detailedanalysis of the system in the cell-free state. Anumber of properties of the cell-free system(Gest, 1952; Barkulis and Gest, 1953) and theavailable results of growing and resting cell ex-periments were recently reviewed in detail byone of the present authors (Gest, 1954), who con-cluded that formate is decomposed to H2 andC02 by a multienzyme system containing "formicdehydrogenase" and hydrogenase.The concept that the hydrogenlyase reaction

requires a series of catalysts which may berepresented as:

formicdehydrogenase .... carriers .... hydrogenasehydrogenlyase

was also supported by the early experiments ofOrdal and Halvorson (1939) who investigated thedistribution of formic dehydrogenase, hydro-genase, and hydrogenlyase activities in intactcelLs of normal Escherichia coli and severalanaerogenic variants. The present study ofhydrogen metabolism in ceU-free systm derivedfrom normal E. coli and six anaerogenic coli-

IThese studies were supported by a grant(Contract No. AT(30-1)-1050) from the AtomicEnergy Commission.

' Predoctorate Fellow of the National ScienceFoundation.

aerogenes variants provides additional and moredirect evidence in favor of this interpretation.

MATERIALS AND METHODS

Organims. As in previous related investiga-tions from this laboratory, E. coli strain Crookes(American Type Culture Collection 8739) wasused as a representative normal type. Six variantsincapable of producing H2 from glucose were ob-tained by screening 168 strains of gram negativecoli-aerogenes bacteria isolated from patients atUniversity Hospitals (Cleveland); the six or-ganisms, which were arbitrarily designated asWR 1 through 6, grew abundantly but producedvirtually no gas (Durham fermentation tubes) ina 1 per cent glucose plus nutrient broth medium.A number of significant biochemical charac-teristics of the organisms employed in the presentwork are listed in table 1.The variants were maintained on nutrient

agar slants and showed no evidence of instabilitywith regard to the enzymatic properties examinedduring the course of this investigation.

Resting cel suspenions. For the intact cell ex-periments summarized in table 2, the organismswere grown in stationary culture at 37 C in amedium containing 1 per cent glucose, 1 per centyeast extract, and 1 per cent tryptone. Approxi-mately 17 hours after inoculation, the cells from300 ml of medium (incubated in 500 ml Erlen-meyer flasks) were harvested, washed once with15 ml of deoxygenated water and resuspended in5 ml of water. Depending on the organism, thebacterial density of these suspensions variedfrom 9 to 21 mg dry weight per ml.

CeU-free preparatioms. The bacteria were culti-vated in deep stationary culture (10 liters of me-dium in 12 liter Florence flaks) at 37 C using themedium described above. At approximately 12 to14 hours, the cells were harvested and groundwith alumina A301 according to the procedure ofMcflwain (1948). The ground miture was ex-

326

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

MECHANISM OF THE HYDROGENLYASE REACTION

TABLE 1Some biochemical characteristics of Escherichia coli (Crookes) and the six anaerogenic variants

Organism | Glucose Ferm. Lact Ferm. Vos Methyl Growth on Indole UreaOlEe.nISm 'JUCOSC CtOSC ~Pr ue Red Citrate Formation Hydrolysis

WR.1 ........................ A + + +WR 2........................ A A _ + _ +_WR 3........................ A A + _ _ _ +WR4........................ AA + + +_WR 5....................... A + +WR 6........................ AA + + +E. coli (Crookes)............... AG AG + + -

A = acid produced.G = gas produced.

tracted with 1.5 ml of water for each gram of cellpaste used and the suspension centrifuged for 10minutes in the Sorvall Model SS-1 centrifuge at12,000 X G. Supernatant fluid from this step wasrecentrifuged for one hour in the high speed headof the International refrigerated centrifuge at20,000 X G in order to remove the bulk of theinsoluble particulate fraction. The extracts pre-pared in this manner (in the cold) were dispensedinto tubes, which were flushed with helium,stoppered, and stored at -20 C.The extract of Clostridium buylicum, which

was kindly supplied by Dr. G. D. Novelli, wasprepared essentially by the procedure describedabove, except that (a) the organism was grownin the medium devised by Wilson et al. (1948),and (b) the alumina ground mixture was ex-tracted with a 0.005 M potassium phosphatebuffer solution (pH 6.8) containing 0.1 per centcysteine.

Enzymatic assays. Hydrogenase and hydro-genlyase activities were determined by measuringthe rate of H2 utilization or formation, respec-tively, in the Warburg apparatus (alkali in centerwell) at 30 C with vessels of approximately 10 mlcapacity containing a final fluid volume of 1.2 ml.Formic dehydrogenase activity was measuredwith several different electron acceptors bymanometric measurement of the rate of CO2evolution. The gas phase employed for estimationof formic dehydrogenase and hydrogenlyaseactivities was helium, whereas 100 per cent H2was used for hydrogenase assay. The other condi-tions of the asays are noted in connection withthe results.

Special reagents. Benzyl viologen was obtainedfrom British Drug Houses, Ltd., Toronto,Canada; methyl viologen from Jacobson Van Den

Berg and Co., 73 Cheapside, London, E.C.2,England, and pyocyanin (chloride) from Hoff-mann-LaRoche Inc., Nutley, N. J., U. S. A. Theviologens are dyes of low redox potential, whichcan be reduced (reversibly) by addition of oneelectron per molecule (Michaelis and Hill, 1933);they differ from most other oxidation-reductiondyes in that they are colorless in the oxidizedform and colored in the reduced form.

RESULTS AND DISCUSSION

Occurrence offormic dehydrogenase, hydrogenaseand hydrogenlyase activities in normal E. coli andthe anaerogenic variants. In addition to the workby Ordal and Halvorson (1939) noted previously,a number of investigators have examined thedistribution of formic dehydrogenase, hydro-genase, and hydrogenlyase activities in cels ofnormal organisms grown under different condi-tions or in such cells treated in some particularmanner. Although cells which display hydro-genlyase activity almost invariably show formicdehydrogenase and hydrogenase activities, re-ports have occasionally appeared which purportthat a reasonably high level of hydrogenlyase hasbeen observed in the absence of formic dehydro-genase or hydrogenase activity. If it is assumedthat the assays employed in these instances forformic dehydrogenase and hydrogenase wouldunquestionably detect and adequately measurethe latter enzymes, then of course it would appearthat hydrogenlyase is not a complex enzyme sys-tem composed, in part, of formic dehydrogenaseand hydrogenase. The validity of the assays istherefore of paramount importance and, in theopinion of the authors, much of the difficulty nowcurrent in discussions of the mechanism of the

3271955]

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

H. GEST AND H. D. PECK, JR.

TABLE 2Distribution offormic dehydrogenase, hydrogenase, and hydrogenlyase activities in Escherichia coli and the

anaerogenic variantsActivites for intact cells are given in terms of IAL/hr/mg dry weight; for extracts, AL/hr/0.25 ml extract

Formic Dehydrogenase Hydro Activity Hydrogenlyase ActivityOrganism (4. ~~~~Activity MAuilized/hr) L (vle/rOrganism (I L C2O evolved/hr) Hi U" r ''Hz evolved/hr)

Cell Extractt Cell* Extractt CeUl Extractt

WR 1.............................. 600 2080 440 1780 54 126WR 2..............................0 0 400 1850 0 0WR 3.............................. 400 1810 0 0t 0 0WR4.............................. 310 0 0WR 5........................... 390 1010 470 1410 0 0WR6.............................. 270 670 0 0 0 0E. coli (Crookes) ....................... 340 450 450 700 190 400

* Intact cell assays: 0.3 ml of suspension in 0.1 M potassium phosphate buffer pH 6.6; 15 pM of sodiumformate; 12 ,si of methylene blue.

t Extract assays: 0.25 ml of extract in 0.083 M potassium phosphate buffer pH 6.1; formate and methyl-ene blue additions as above.

t <50.

hydrogenlyase reaction can be attributed to lackof appreciation of this aspect of the problem.The traditional and arbitrary electron acceptor

used for the assay of formic dehydrogenase andhydrogenase is the dye methylene blue (= MB).The occurrence of these enzymes, as indicated byassay with this dye according to the reactionslisted below, and of hydrogenlyase activity inintact cells of E. coli (Crookes) and the anaero-genic strains is illustrated by the data in table 2.

formic dehydrogenase activity: HCOOH +MB -- C02 + MBH2

hydrogenase activity: Hs + MB * MBH2hydrogenlyase activity: HCOOH * CO + H2

Also shown for comparison are typical activitiesobserved in cell-free preparations.

It is of importance to note that the arbitraryasay for hydrogenase is based on utilization ofH2, even though the natural function of thisenzyme in these organisms is undoubtedly con-cerned with formation of H2 (see below).A number of conclusions may be drawn from

the data of table 2. First, it is evident that thepreparations which show hydrogenlyase activityalso show both formic dehydrogenase and hydro-genase (i. e., WR 1 and normal E. coli). This istrue of both intact cells and extracts and it maybe further noted with regard to the activitiesunder consideration that the enzymatic contentof the cell-free extracts is qualitatively identical

with that of the intact cells in all the organismslisted.

It should be emphasized that the data of table2 are to some degree approximate (particularlyvalues greater than 400). This is a consequenceof the fact that relatively large amounts of thepreparations were routinely employed to obviatethe possibility that lack of hydrogenlyase activityin a particular preparation could be due to thedilution effect to which this system is subject andto provide a more critical test for the absence ofany particular activity. In general, the valueslisted are based on data obtained during the first5 to 10 minutes after addition of substrate oracceptor. The endogenous activities displayed bythe different preparations were negligible; theonly definite endogenous activity observed was avery low rate of H2 formation by the intact cellsof E. coli.

E. coli (Crookes) rand variant WR 1. It wassomewhat surprising to find that WR 1 containedan active hydrogenlyase system since this or-ganism does not produce H2 from glucose (orlactose). As indicated, extracts fromWR 1 displayhydrogenlyase activity and in separate tests itwas found that these preparations do not produceH, from pyruvate. These observations suggestthat failure of WR 1 to produce gas from glucoseis probably due to impaired production of formatefrom the intermediate pyruvate, i. e., a "block"

328 [VOL. 70

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

MECHANISM OF THE HYDROGENLYASE REACTION

in the phosphoroclastic cleavage of puruvate toacetate and formate.Comparing the data obtained with WR 1 and

E. coli from a quantitative standpoint it wouldappear that the level of hydrogenlyase activity isnot related to the amount of formic dehydro-genase or hydrogenase activities in the prepara-tions. This does not, of course, necessarily provethat these enzyme activities are unrelated andindependent. It may simply indicate, for example,that the assay of formic dehydrogenase activitywith methylene blue is not an adequate measureof the "formic dehydrogenase" activity requiredfor the coupling with hydrogenase (see below).Variants of the WR 1 type were apparently notobserved in the studies of Ordal and Halvorson(1939).

Variant WR 2. This organism does not displayhydrogenlyase activity and shows no formicdehydrogenase when tested with methylene blueas the electron acceptor. Similarly, when extractswere tested for formic dehydrogenase (under thesame conditions) with benzyl viologen, pyo-cyanin, and ferricyanide as potential electronacceptors, no activity was observed. The sameextracts exhibited readily measurable or highhydrogenase activity with all of the foregoingacceptors (MB included). It thus appears thatabsence of hydrogenlyase activity in this or-ganism may be correlated with absence of the'formic dehydrogenase" component. WR 2 seems'to be equivalent to the variant X described byOrdal and Halvorson, who state that most of thevariants they tested were of this type.

Variants WR 8, 4, and 6. The data in table 2indicate that these three organisms show thesame pattern of enzyme activities, i. e., theypossess formic dehydrogenase but apparentlycontain no hydrogenase, which may be correlatedwith the absence of hydrogenlyase activity. It isof considerable interest that a comparable situa-tion is found in intact cells and cell-free prepara-tions from normal E. coli grown with vigorousaeration, i. e., presenee of a high level of formicdehydrogenase but no hydrogenase or hydro-genlyase (Gest, 1952, 1954). Of this group ofvariants, WR 3 was further examined with respectto hydrogenase activity, particularly becausesome extract preparations from this organismdid display a very slight activity with methyleneblue as the electron acceptor (always less than50 pL H2/hr/0.25 ml). No H2 was consuimed by

the cell-free extracts in the presence of benzylviologen, pyocyanin, or ferricyanide. In con-formity with the tests with methylene blue, how-ever, a very slight hydrogenase activity wasdetectable by a different type of procedure whichis discussed below.3 In any event, when thepresent data for hydrogenase activity of thevarious organisms listed in table 2 are compared,it appears that variants WR 3, 4, and 6 are vir-tually devoid of this enzyme. These strains seemto be equivalent to variant XI of Ordal andHalvorson (1939).

Variant WR 5. In this variant both formic de-hydrogenase (as measured with methylene blue)and hydrogenase are present, yet there is nohydrogenlyase activity. A variant of E. coli show-ing this pattern was also described by Ordal andHalvorson (1939; variant XII). A similar enzy-matic picture has been observed in normal E.coli cels and preparations under particular condi-tions. For example, if E. coli is grown with slightor even moderate aeration, both formic dehydro-genase and hydrogenase are present but there isno hydrogenlyase activity (see discussion by Gest,1954). Similarly, extract from anaerobicallygrown cells shows all activities at pH 6, whereasat pH 7.2 or above hydrogenlyase disappearscompletely, though there is active formic dehy-drogenase and hydrogenase present (Gest, 1952).A third example comparable to WR 5 is providedby aged particulate fractions also derived fromcells grown in the absence of oxygen (Gest, 1952).The hydrogenase component in extracts of

WR 5 shows the same range of electron acceptorspecificity as was found with WR 2, i. e., benzylviologen, methylene blue, pyocyanin, andferricyanide all act as acceptors.A number of investigators have viewed the

existence of the WR 5 type of preparation as con-tradictory to the hypothesis that formic dehy-drogenase and hydrogenase are involved in thehydrogenlyase mechanism. Ordal and Halvorson(1939), however, felt that this was not necessarilyso and suggested that a "factor" mediating"electron conduction" between formic dehydro-genase and hydrogenase might be missing fromsuch preparations. This plausible suggestion

3 It would be of interest to test these variants(as well as the normal strain) for their quantita-tive ability to catalyze the exchange reaction be-tween isotopic H2 and water.

3291955]

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

H. GEST AND K D. PECK, JR.

TABLE 3Reconstruction of hydrogenlyase activity by mixing

extracts from the variants

Formic HydrogenlyaseHydrgease Activity ofDehyd'System" Combined Systems"System" H(L Huhr)*

WR 3 WR 2 1000WR 3 WR 5 500WR5 WR2 0WR 6 WR 2 0

* 0.25 ml of each extract present under the usualconditions for hydrogenlyase assay, i.e., 0.083 Mphosphate buffer pH 6.1; 15 im HCOONa; KOHin center well.

finds experimental support in the current study(see below).Reconsruction of hydrogenlyase activity using

mixtures of variant extracts. The observationssummarized in table 2 suggested that WR 2 andWR 3 (or WR 6) might contain complementaryparts of the hydrogenlyase system. This possi-bility was studied by testing for hydrogenlyaseactivity in mixtures of extracts under the condi-tions usually employed for asay (of hydrogen-lyase) with normal E. coli preparations. Typicalresults of such experiments are given in table 3.

It is apparent that a high level of hydrogenlyaseactivity is observed when WR 3 is mixed witheither WR 2 or WR 5, the combination WR 3 +WR 2 being most effective. If either the WR 3 orWR 2 extract is boiled, the combination is inac-tive. Similarly, mixtures of WR 5 or WR 6 withWR 2 are completely inactive. In our interpre-tation of these observations, it seemed most likelythat the "formic dehydrogenase" component ofWR 3 could in fact link very effectively with thehydrogenase component of WR 2, whereas thecoupling with the hydrogenase component ofWR 5, though successful, was less efficient forone reason or another.

Following this line of reasoning it was con-cluded that, since WR 5 and WR 6 failed to linkwith the hydrogenase component of WR 2, theformic dehydrogenase constituents of WR 5 and6 must be "incomplete" as compared with theformic dehydrogenase system present in WR 3.Since the hydrogenase of WR 5 can couple withthe formic dehydrogenase component of WR 3,it also seemed probable that absence of hydro-ganlyase activity in WR 5 alone was due to adefect in the formic dehydrogenase portion of the

system. In experiments designed to determine thenature of this defect a number of coenzymes weretested for ability to activate the hydrogenlyasesystem in WR 5 extracts. None of the compoundstested thus far had any effect; these include di-and triphosphopyridine nucleotides, riboflavin,flavin mononucleotide, flavin adenine dinucleo-tide, folic acid, dihydrofolic acid, tetrahydrofolicacid, anhydrocitrovorum factor and leucovorin.

Preliminary studies of several properties of thereconstructed hydrogenlyase (WR 3 + WR 2)have been made in order to determine whetheror not this hydrogenlyase system is comparablewith the "normal" enzyme complex. It wasfound that a pronounced dilution effect on hydro-genlyase activity could be demonstrated with thereconstructed system just as with E. coli extracts.Further, in both systems hydrogenlyase activityis severely inhibited by metal complexing agentssuch as a,a'-dipyridyl and the inhibition isalmost completely relieved by addition of Fe++(see Gest, 1952, for discussion of properties of thenormal extract). These observations, whichshould be amplified, suggest that the two typesof systems are in fact quite similar.The solubility of the components required for

the WR 3 + WR 2 reconstruction of hydro-genlyase was studied by assaying mixtures of theparticulate and soluble fractions obtained fromthese two crude extracts by centrifugation in theSpinco preparative centrifuge at 144,000 X G for1.5 hours. The formic dehydrogenase componentrequired was exclusively in the soluble fractionof WR 3, whereas the effective hydrogenase com-ponent of WR 2 was found to be mainly in theinsoluble fraction (some hydrogenlyase activitywas observed when a relatively large amount ofthe soluble fraction of WR 2 was used). Thiscoupling between a soluble formic dehydrogenaseand an insoluble hydrogenase system is reminis-cent of the synergistic effects (on hydrogenlyaseactivity) observed when soluble and hydro-genase-rich particulate fractions derived fromanaerobically grown E. coli are mixed (Swimand Gest, 1954).The successful reconstructions described in

table 3 indicate that the absence of hydrogenlyaseactivity in variants WR 2, 3, (4), 5 and 6 is notdue to the presence, in these organisms, of "in-hibitors" of the hydrogenlyase complex. Thepossibility that "inhibitors" were present wasfurther ruled out by determining the effects of

3.30 [VOL. 70

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

1MECHANISM OF THE HYDROGENLYASE REACTION

TABLE 4Formic dehydrogenase activity of WR 3, WR 5, and

WR 6 with several electron acceptorsFormic Dehydrogenase Activity (,,L COs

evolved/hr) with:Preparation

Methyl Benzyl Methyleneviologen viologen blue

WR 3 550 1500 1810WR 5 28 41 970WR 6 16 48 540

Assay conditions: 0.25 ml of extract in 0.083 Mphosphate buffer pH 6.1; HCOONa, 15 ,uM;viologen dyes, 20 i&m; methylene blue, 10 pAm. Therates were calculated from data obtained withinthe first 15 minutes after addition of the electronacceptor (and formate).

extracts from these variants on the hydrogenlyaseactivity of preparations from WR 1 and E. coli.In all instances there was either no effect orstimulation of the activity. The latter was par-ticularly noticeable in the tests with WR 3, e. g.,addition of 0.25 ml of WR 3 extract acceleratedthe hydrogenlyase activity of an equal volume ofWR 1 extract by 7-fold and that of E. coli by2-fold. This observation suggests that the elec-tron donating portion of the system is limitingin theWR 1 and E. coli preparations, particularlyin the former.A comparison of the formic dehydrogenase com-

ponents. Marked differences were observed withrespect to ability of the various preparations toreduce the low redox potential dyes benzyl andmethyl viologens upon addition of formate. Thesedifferences are illustrated by the data in table 4.From table 4 it is evident that the viologen dyes

(and methylene blue) are effective electron ac-ceptors for the formic dehydrogenase systempresent in WR 3. On the other hand, WR 5 andWR 6 show good activity with methylene bluebut display negligible activity with the viologens.This was true not only of extracts but also forintact cells as tested with WR 3 and WR 5. Inseparate tests it was found that the viologens arealso very rapidly reduced by extracts of WR 1and E. coli upon addition of formate.4 It is strik-ing that preparations which contain an activehydrogenlyase system (WR 1 and E. coli) or

4 Quantitative data are not given for WR 1 andE. coli because under certain conditions the violo-gens stimulate the rate of CO (and H2) evolutionfrom formate in such extracts by the mechanismdescribed in the next section.

which can furnish the formic dehydrogenase por-tion required for coupling with the hydrogenaseof WR 2 are able to reduce the viologen dyesrapidly. On the other hand, preparations offormic dehydrogenase which cannot couple withthe hydrogenase of WR 2, viz. WR 5 and WR 6,show a greatly impaired ability to reduce theseparticular electron acceptors.

In view of the fact that hydrogenase acts inmany respects as a classical hydrogen electrode,it seems reasonable to assume that the electronsor hydrogen atoms derived from formate "acti-vation" are transported to the hydrogenase sys-tem through a specific intermediate carrier(s) oflow redox potential. Further, we a#sume thatability to reduce artificial "one electron" dyes oflow redox potential such as benzyl viologen(Eo' = -0.359) and methyl viologen (Eo' =

-0.446) depends on a factor xi, probably anelectron carrier, which is also an essential part ofthe formic dehydrogenase component required forhydrogenlyase activity. The apparent distribu-tion of the factor xi, postulated on this basis, inthe various preparations tested is consistent withthe foregoing view.

Extracts of WR 3 show high levels of formicdehydrogenase activity when tested in the usualmanometric procedure not only with the viologensand methylene blue (Eo' = +0.011), but alsowith pyocyanin (Eo' = -0.034) and ferricyanide(Eo' - +0.36) acting as electron acceptors. Al-though extracts of WR 5 display good formicdehydrogenase activity with the latter three ac-ceptors, activity with the viologens is poor, asalready noted. These observations suggest thatthe defect in WR 5, or in preparations from E.coli which resemble WR S, is the virtual absenceof the factor xi closely associated with "formicdehydrogenase" and, secondly, that the measure-ment of formic dehydrogenase using methyleneblue or other high redox potential acceptors is notin itself an adequate assay of the formic dehydro-genase essential for hydrogenlyase activity.The foregoing concept of the reconstructed

hydrogenlyase complex may be convenientlyrepresented as follows:

methylene blueI

formate-activatingenzyme system .... xi.... carrier(s) .... hydrogenase

WR 3 2WR2viologens

3311955]

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

H. GEST AND H. D. PEC0TR JR. [VOL. 70

A model hydrogenlyase 8ystem in wY*h methylviologen is an intermediate carrier. Extracts fromcertain strictly anaerobic microorganisms such asClostridium bntylicum and Micrococcus lactilyticusdo not possess formic dehydrogenase or hydro-genlyase activities, but do contain a hydrogenasewhich is presumed to be essential for productionof H2 from pyruvate. It was anticipated that thehydrogenase in such preparations could couplewith the formic dehydrogenase component ofWR 3 to provide a complex with hydrogenlyaseactivity. Such combinations, however, did notshow significant hydrogenlyase activity, eventhough the clostridial or micrococcus extractswere capable of forming H2 at a rapid rate frompyruvate. This result suggested that a specificcarrier (designated as x2) required for electrontransport between the formic dehydrogenase ofWR 3 and hydrogenase is present in WR 2, butis absent from the extracts of Cl. butylicum andM. lactilyticw. A corollary to the foregoing is that

180

* WR 3 + VIOLOGEN +C. bufylicum140 o WR 3 + VIOLOGEN

w3 120 /0

J100X

0) 80

I--:i60/0

540/

20

0 5 10 15 20 25 30 35MINUTES

Figure 1. A model hydrogenlyase complex withthe composition: formic dehydrogenase (from var-iant WR 3).... methyl viologen.... clostridialhydrogenase. The vessels contained 0.25 ml ofWR 3 extract and 1 pM of methyl viologen in 0.083M phosphate pH 6.1 (final volume, 1.2 ml). Atzero time, 15 upm of sodium formate and, whereindicated, 0.2 ml of Clostridium butylicum extractwere added; KOH in center well; gas phase, he-lium.

HCOOH1 2MV H2

formic deb drogenase clostridial hydrogenaseof

C02 2MVHthe carrier x2 is evidently not required for electrontransport to hydrogenase in the phosphoroclasticreaction catalyzed by the latter organisms (i. e.,pyruvate + H3P04 -> acetyl phosphate +C02 + H2).The carrier postulated (x2) apparently can be

effectively replaced by catalytic amounts of adye of low redox potential, such as methyl violo-gen. This effect of the dye is illustrated in figure 1.The clostridial extract alone does not produce

H2 from formate either in the absence or presenceof methyl viologen. Molecular hydrogen is pro-duced at a low rate by WR 3 alone upon additionof the dye; this is apparently due to couplingwith the very small amount of hydrogenase pres-ent in such preparations. The rapid evolution ofH2 observed when methyl viologen (MV) isadded to the combination of the two extractsreflects the ability of this dye5 to transportelectrons between the formic dehydrogenase com-ponent of WR 3 and the hydrogenase of C.butylicuZm.

Several observations support the conclusionthat methyl viologen functions as an inter-mediate carrier in this manner. First, it may benoted that the rate of H2 formation by the modelsystem is proportional to methyl viologen con-centration over a rather large range (0.000025 to0.00025 M). When the dye (and formate) isadded toWR 3 alone, the blue color characteristicof the reduced form appears rapidly and remainsintense. Under these conditions, the dye tends toremain in the reduced form because the latter isreoxidized at a comparatively negligible rate bythe very small amount of hydrogenase present.On the other hand, in the complete system:WR 3 .... MV.... clostridial hydrogenase, thesteady state concentration of reduced dye is

6 Benzyl viologen also can act as an intermedi-ary carrier linking formic dehydrogenase and hy-drogenase, but not as effectively as methyl violo-gen. This may be related to the fact that the Eoof benzyl viologen is somewhat higher than thatof methyl viologen.

332

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

MECHANISM OF THE HYDROGENLYASE REACTION

smaller, owing to relatively rapid reoxidation bythe "excess" of hydrogenase. This is shown by thepale blue color in both this type of model system,as well as in preparations with inherent hydro-genlyase activity (i. e., in WR 1 and E. coliextracts). The color disappears entirely when theformate is completely decomposed.

In a previous study, it was reported thatfreshly made particulate preparations fromanaerobically grown B. coli show hydrogenlyaseactivity after an induction period and that thislag phase is abolished by addition of catalyticquantities of the viologen dyes (Gest, 1952). Asin the current experiments, the effect of the dyewas attributed to its ability to couple the formicdehydrogenase and hydrogenase componentspresent in the particles.The use of reduced methyl viologen, in sub-

strate concentrations, as an electron donor forhydrogenase will be described in a later paper.Metabolim of pyruvate by extracts from the

variants. Extracts from variants WR 1, 2, 3, and5 did not produce H2 from pyruvate under theconditions usually employed for assay of hydro-genlyase activity (i. e., 0.25 ml of extract at pH6.1). Of these preparations, the only one whichutilized pyruvate at an appreciable rate wasWR 3; in agreement with the results noted ingrowing cultures (see table 1), acetoin was formedas a major product. Although each extract byitself was inactive, a combination of WR 3 +WR 5 extracts produced H2 from pyruvate at arapid rate. This observation is consistent withthe results summarized in table 3 and constitutesfurther evidence for the view that formate is anintermediate in production of H2 and CO2 frompyruvate by normal coli-aerogenes bacteria.Comments on other approaches to investigation

of the single vs. multienzyme qution. It has beenreported that aerobically grown cells of Aero-bacter aerogenes show hydrogenlyase activity, of arather low order, 30-40 minutes after exposureto hydrolyzed yeast extract and that such cellsstill do not manifest hydrogenase activity withmethylene blue (Lichstein and Boyd, 1953).Since inability to reduce methylene blue is notnecessarily a valid criterion for the absence ofhydrogenase (Gest, 1954), this observation cannotbe accepted as critical evidence in support of theidea that hydrogenlyase activity is due to asingle distinct enzyme. In any event, using thesame strain of Aerobacter (kindly supplied by Dr.

Lichstein), we also found that hydrolyzed yeastextract stimulated the appearance of hydro-genlyase activity, but in contrast to their findingsthese cells proved to have a readily measurablehydrogenase activity as tested by several pro-cedures.Grunberg-Manago et al. (1951) also believe

that hydrogenlyase is a single enzyme not relatedto formic dehydrogenase or hydrogenase, on thebasis of inhibitor experiments using hypophos-phite and intact cells of E. coli. This compoundwas reported to inhibit formic dehydrogenaseactivity (measured by observation of methyleneblue reduction in a Thunberg tube procedure)completely but to have no effect on hydrogenlyaseactivity. Thus far we have been unable to obtaina differential result of this kind using the cell-freesystem from E. coli. In a typical experiment with0.25 ml of extract and 0.0025 M KH2PO2, theformic dehydrogenase activity with methyleneblue, measured manometrically, was inhibitedabout 57 per cent, whereas the hydrogenlyaseactivity was depressed by some 30 per cent. Thisobservation and other related considerationsdiscussed by Gest (1954) indicate that the effectsof hypophosphite on these activities are equivocaland require further investigation.

Several investigators have assumed that anappreciable difference in Michaelis constant (Km)in the formic dehydrogenase (methylene blue)and hydrogenlyase activities observed withintact cells is critical evidence against the viewthat formic dehydrogenase is a component of thehydrogenlyase complex (e. g., Wolf et al., 1954).This assumption was recently examined byCrewther (1953), who concludes from theoreticaJconsiderations that appreciably different Kmvalues might be expected for a dehydrogenase,depending both on the nature and concentrationof the electron acceptor used. This conclusion wassubstantiated experimentally by Crewther instudies with the formic dehydrogenase of intactE. coli. Crewther's investigation indicates thatthe observed differences in Km cannot be usedas unambiguous evidence to support the argu-ment that hydrogenlyase activity can be at-tributed to a single enzyme.

ACKNOWLEDGMENTS

The authors wish to thank Dr. R. F. Parkerfor the cultures from which the variants wereselected; Mr. Uros Roessmann, who isolated the

3331955]

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

H. GEST AND H. D. PECK, JR.

variants during work carried out in partial ful-fillment of the requirements of the ProjectTeaching Program of the Medical Curriculum atWestern Reserve University School of Medicine;and Miss Marion A. Koser for valuable technicalassistance. We are also indebted to Professor L.0. Krampitz for stimulating discussions arisingfrom his active interest in these studies.

SUMRY

The mechanism of the formic hydrogenlyasereaction has been investigated, using intact cellsand cell-free preparations derived from one nor-mal and six non-gas producing strains of coliformbacteria. The present results indicate that thereaction is catalyzed by a multienzyme chaincontaining at least two enzymes, formic dehydro-genae and hydrogenase, and two intermediatefactors, designated as xi and x2. Two types ofmodel hydrogenlyase complexes are described.An active complex results from the combinationof extracts from two non-gas producing variantstrains, "formate-activating system",.... xi beingsupplied by one extract and the remainder of thesystem (....x2....hydrogenase) by the other.In the second type of model hydrogenlyase com-plex, hydrogenase is supplied by extract fromCletridium butylicum or Micrococcus lactilyticueand x2 is replaced by the dye methyl viologen,which acts as an efficient intermediary electroncarrier.

REFERENCESBARKLUS, S. S., AND GEST, H. 1953 The role

of Cs-derivatives in the decomposition offormate. Biochim. et Biophys. Acta, 11,306-3.

CREWTHE, W. G. 1953 Studies on Aerobacillus

polymyza. V. The significance of the Michaelisconstants of the hydrogenlyase systems andrelated enzymes. Australian J. Biol. Sci., 6,198-204.

GE8T, H. 1952 Molecular hydrogen: oxidationand formation in cell-free systems. Phospho-rus metabolism, pp. 522-43, Vol. II. Edited byW. D. McElroy and B. Glass. The JohnsHopkins Press, Baltimore, Md.

GEsT, H. 1954 Oxidation and evolution of mo-lecular hydrogen by microorganisms. Bac-teriol. Revs., 18, 43-73.

GRUNBERG-MANAGO, M., SZULMAJBTER, J., ANDPROUVOsT, A. 1951 Hydrogenlyase, formi-co-ddshydrog6nase et hydrogenase chez Esch-erichia coli. Compt. rend., 233, 1690-1692.

LIcHSTEIN, H. C., AND BomD, R. B. 1953 Theformic hydrogenlyase system of Aerobacteraerogene8. J. Bacteriol. 65, 617-618.

MCILWIN, H. 1948 Preparation of cell-freebacterial extracts with powdered alumina.J. Gen. Microbiol., 2, 288-291.

MICHABLIS, L., AND HILL, E. S. 1933 The vi-ologen indicators. J. Gen. Physiol., 16, 859-873.

OuAL, E. J., AND HALvoIiON, H. 0. 1939 Acomparison of hydrogen production from sug-ars and formic acid by normal and variantstrains of Escherichia coli. J. Bacteriol., 38,199-220.

SWIM, H. E., AND GnsT, H. 1954 Synergisticeffects between soluble and particulate prepa-rations from E8cherichia coli. J. Bacteriol.,68, 755-756.

WILSON, J., KRAMPITz, L. O., AND WERKMAN, C.H. 1948 Reversibility of a phosphoroclasticreaction. Biochem. J. (London) 42, 598-600.

WOLF, J., STICKLAND, L. H., AND GORDON, J.1954 Enzymes concerned with gas formationby some coliform bacteria. J. Gen. Micro.biol., 11, 17-26.

334 (VOL. 70

on May 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from