1090 H. BOTHE AND B. FALKENBERG

The Reduction of from vinelandii by Pyruvate

H. BOTHE and B. FALKENBERG

Ruhr-Universität Bochum, Abteilung Biologie, Lehrstuhl für Biochemie der Pflanzen, 463 Bochum, Postfach 2148, West-Germany

(Z. Naturforsch. 27 b, 1090—1094 [1972] ; received May 10. 1972)

Flavodoxin, FMN containing , pyruvate phosphoroclastic reaction, nitrogen-fixation

Pyruvate is shown to be a substrate for the reduction of flavodoxin from Azotobacter vinelandii, using the phosphoroclastic system from Clostridium pasteurianum. In this reaction, the reduction of this flavoprotein proceeds beyond the semiquinone state partly to the hydroquinone form. Pyruvate is possibly the substrate for the reduction of flavodoxin in the intact organisms.

The flavoprotein from Azotobacter, first isolated Material and Methods by SHETHNA et al. 1, is chemically well characteriz- ed, but its enzymatic function is still obscure. AR- Lyophilised stock material of Azotobacter vinelandii CCM 289 was a kind gift of the Czechoslovak Collec- 2 NON and coworkers found this flavoprotein to be tion of Microorganisms at Brno. Azotobacter was active in substituting for ferredoxin in the reduc- grown, and ferredoxin, flavodoxin and all the other tion of acetylene by a cell-free nitrogenase prepara- electron carriers used in the experiments were prepar- tion from Azotobacter. We have recently confirmed ed as previously described 3. Clostridium pasteurianum (ATCC 6013) was grown their observation3. In addition, we could demon- in carboys containing 10 liters of sterilized medium strate that this flavoprotein is able to replace ferre- with the following composition: sucrose, 100 g; doxin in the NADP+-reduction by either illuminated extract (Difco), 10 g; calcium carbonate, 0.5 g; spinach chloroplasts or by molecular hydrogen and Na2S204 , 300 mg. The medium was flushed with nitro- hydrogenase from Clostridium with a maximal effi- gen for 30 min before the inoculation by approximately 500 ml bacterial suspension in logarithmic growth. 3 3 ciency of about 40 - 50% . Evidence was presented After 20 — 24 hrs of growth, the bacteria were harvest- that in the flavoprotein from Azotobacter ed and stored under nitrogen at —20 °C prior to use. functions as a substitute for a one electron carrier In preparing the phosphoroclastic enzyme system which shuttles between the fully reduced and the semi- from Clostridium pasteurianum, 20 g of cells were quinone form as do the 4?5. The experi- thawed, suspended in 10 ml of 0.1 M phos- phate buffer pH 7.0, flushed intensively (20 min) 3 ments previously reported led to the conclusion with argon, and anaerobically disrupted in a Ribi cell- that the flavoprotein from Azotobacter is — in its fractionator (Ivan Sorvall Inc., Connecticut) at 30.000 enzymatic functions — closely related to the class psi. The extract was anaerobically centrifuged at of flavodoxins, and, therefore, we suggested that the 38.000 g for 15 min. The supernatant was chromato- graphed on a Whatman DE-52 cellulose column proper name to be used for this flavoprotein is (2.5x5 cm), which was preequilibrated with 0.1 M flavodoxin from Azotobacter. phosphate buffer at pH 7.0. The phosphoroclastic en- zyme system was not absorbed by the column, which is In this communication, we wish to report on ex- in contrast to ferredoxin 6. The latter could be eluted periments in which the reduction of flavodoxin from by a solution containing 0.6 M NaCl in 0.1 M phos- Azotobacter could be achieved by using the phos- phate buffer pH 7.0. The ferredoxin free eluate, which phoroclastic enzyme system from Clostridium pas- contained approximately 5 mg of /ml, was used in the experiments in Fig. 1 and Table II. teurianum. The suggestion will be made that pyru- These experiments were carried out in 8 ml Fern- vate may in fact be the in vivo substrate in the re- bach flasks, sealed with rubber caps, in a water-bath at duction of Azotobacter flavodoxin. 37 °C under continuous shaking. The amount of acetyl phosphate formed in the experiments was determined exactly as described by RAEBURN and RABINOWITZ 7,

using additional FeCl3 to overcome the interference of Requests for reprints should be sent to Dr. H. BOTHE, Ruhr-Universität Bochum, Abteilung Biologie, Lehrstuhl colour development by mercaptoethanol present in the für Biochemie der Pflanzen, D-4630 Bochum, Postf. 2148. reaction mixture. THE REDUCTION OF FLAVODOXIN FROM AZOTOBACTER V1NELANDII BY PYRUVATE 1091

cally added to a solution of oxidized flavodoxin (Table I a). The oxidized form of Azotobacter flavo- doxin may be reduced directly to the hydroquinone form by dithionite via a two electrons transfer, and the may subsequently be generated by the comproportionation reaction oxidized flavodoxin -f fully reduced flavodoxin ^ 2 flavodoxin semiquinone

which was shown to proceed fast5. The stabilization of the radical of flavodoxin mjjmoles flavodoxin added from Azotobacter is only pronounced at neutrality. Fig. 1. Flavodoxin from Azotobacter vinelandii as a At the more alkaline pH-value of 9.0, this flavo- of the phosphoroclastic reaction catalyzed by a ferredoxin-free, doxin can be reduced to the hydroquinone form crude extract from Clostridium pasteurianum. The experi- ment was run at pH 8.0. Besides this, the reaction mixture even by a low excess of dithionite (Table I a). was the same as in Table II. •• —• under aerobic conditions; Other flavodoxin, e. g. the flavodoxin from blue- • — • under anaerobic conditions. green algae, referred to as phytoflavin (ref. 5), are always reduced to the hydroquinone form by a small Results excess of dithionite at all pH-values. In the phosphoroclastic reaction of Clostridium, As pointed out previously8-10, the blue radical pasteurianum, pyruvate is cleaved in a thiamine of flavodoxin from Azotobacter exhibits an unusual pyrophosphate dependent reaction to carbon di- stability towards reduction and reoxidation by air. oxyde, acetyl phosphate and two reducing equi- At neutrality, this flavoprotein cannot be reduced valents, the latter of which reduce the ferredoxin in beyond the semiquinone state in the presence of a Clostridium 11. The reduced ferredoxin can be re- hundred fold excess of dithionite (Table I a). More- oxidized aerobically by oxygen or anaerobically over, the stabilization of the radical is so strong, that through hydrogen evolution which is catalyzed by — when once formed — it cannot be further reduced the hydrogenase enzyme present in Clostridial ex- to the hydroquinone form by the anaerobic addition tracts. Ferredoxin can easily be removed from the of even a very high excess of dithionite 5. A partial clastic enzyme system by treatment with DEAE-cel- reduction beyond the semiquinone state can be lulose. When an extract is so treated, it is possible achieved at pH 7.0, however, when dithionite at a to test the activities of various electron carriers, be- high final concentration of 0.5 molar is anaerobi- sides ferredoxin, in the phosphoroclastic reaction.

Table I. Reduction of flavodoxin from Azotobacter by various substrates. The given numbers are estimated values. In each of the enzymatic essays, equal but limiting amounts of different were used at the comparison of the reduction rates of flavodoxin from Azotobacter and flavodoxin (phytoflavin) from Anacystis.

Substrate M pH Flavodoxin from Azotobacter Flavodoxin from Anacystis 1/2 time of % reduction 1/2 time of % reduction reduction to leucoform reduction to leucoform oxid./semiquin. oxid./semiquin. a) chemically 1. Na2S204; 0.015 7.0 3 min not 5 min 100%

2. Na2S204; 0.5 7.0 sec 20%

3. Na2S204; 0.0025 9.0 2 min 80% 15 min 100% b) enzymatically 4. chloroplasts, light 9.0 45 min 20% 1 min 100% 5. H2, hydrogenase 8.4 15 min 10% 15 min 9% 6. Pyruvate, phosphoroclastic 8.5 5 min 20% 5 sec 100% system 7. NADPH, NADPH-ferredoxin 9.0 15 min 0% 30 sec 0% reductase 1092 II. BOTHE AND B. FALKENBERG

Besides ferredoxin from Clostridium, also benzyl viologen or by ferredoxin from Azotobacter is viologen, ferredoxin from spinach and Azotobacter, higher at pH 7.0 than at pH 8.5 in the experiments. and flavodoxin from Azotobacter are able to cata- Contrary to this, the rate of the pyruvate eleavage lyze the phosphoroclastic reaction by Clostridial ex- catalyzed by Azotobacter flavodoxin is slightly but tracts (Table II). These results are in contrast to consistently higher at pH 8.5 than at neutrality. This result may indicate that the fully reduced form Table II. Different electron carriers as cofactors in the phos- of Azotobacter flavodoxin is involved in the phoroclastic reaction catalyzed by a ferredoxin-free prepara- catalysis of the pyruvate cleavage, since — as point- tion from Clostridium pasteurianum. The complete reaction mixture (final volume 2.0 ml) contained: ferredoxin-free, ed out above — the fully reduced form of this crude enzyme extract. 1.05 mg protein; and the following in flavodoxin is more readily formed upon reduction «moles: phosphate buffer, pH 7.0 or tricine-buffer pH 8.5, respectively, 400; mercaptoethanol, 50; sodiumpyruvate' 4; at alkaline pH-values than at neutrality. coenzyme A, 1; cofactors as indicated and for the experiments The recorded spectra directly indicate that at pH 8.5 K2HP04 , 20 ^moles. The experiment was carried flavodoxin from Azotobacter is partly (~20%) re- out at 34° under aerobic conditions for 60 min. Acetyl phos- phate formation was determined as described by RAEBURN duced beyond the semiquinone state in the pyruvate and RABINOWITZ 7. clastic reaction catalyzed by the Clostridial enzyme system (Table I). The spectral changes observed m/unole acetyl phosphate formed/min • mg Clostridial during the reduction catalyzed by the phosphoro- protein clastic enzyme system are the same as those de- cofactor m//moles pH 8.5 pH 7.0 scribed earlier with other enzymatic reactions (ref.3, Fig. 5). 1. — cofactor 2.7 1.0 2. + benzyl viologen 1000 13.0 37.4 The rate of acetyl phosphate production by Clos- 3. + ferredoxin tridial extracts was examined under aerobic and an- Clostridium 10 23.2 22.0 4. + ferredoxin aerobic conditions and at different concentrations Azotobacter 40 7.5 11.2 of oxidized Azotobacter flavodoxin in the reaction 5. + flavodoxin Azotobacter 40 13.3 9.7 mixture (Fig. 1). The rates obtained under anaero- 6. + flavodoxin Azotobacter 80 25.8 21.5 7. + flavodoxin Azotobacter 40 17.2 17.7 bic conditions were slightly lower than in the pre- and ferredoxin sence of air. Under anaerobic conditions, the reac- Azotobacter 40 tion may be limited by the activity of the hydro- genase, which is present in the crude extracts and a report where the flavodoxin from Azotobacter was which catalyzes the reoxidation of reduced flavo- found to be completely inactive in the catalysis of doxin. But regardless of the gas phase used in the the phosphoroclastic reaction9. The activity now reaction vessels, the saturation concentration for found in this reaction cannot be caused by contami- Azotobacter flavodoxin was about ten times higher nations, since our preparation of Azotobacter flavo- than that for ferredoxin from Clostridium (ref. 6). doxin was judged to be pure for the following two In the experiments reported by others9, the con- reasons: (i) For the absorption maxima the ratios centrations of flavodoxin from Azotobacter may not 274 nm : 452 nm = 5.2 and 274nm:358nm = have been high enough to yield measurable activi- 5.6 are essentially the same as those of the crystal- ties in the phosphoroclastic assay. lized protein reported by HINKSON and BULEN9. (ii) Our preparation migrates as a single band on Discussion Polyacrylamide gel electrophoresis. Furthermore can be seen from Table II, that the Some enzymatic reactions have now been found addition of both ferredoxin and flavodoxin from which are catalyzed by all flavodoxins. These reac- Azotobacter to the reaction mixture resulted only tions comprise the acetylene reduction by the cell- in additive rates of acetyl phosphate production. free nitrogenase complex from Azotobacter, the This clearly indicates, that ferredoxin and flavo- photosynthetic NADP+-reduction by spinach chloro- doxin from Azotobacter are functioning indepen- plasts, hydrogen uptake catalyzed by hydrogenase, dently and not synergistically in the phosphoro- and — as described in this paper — pyruvate clas- clastic reaction. Also it can be seen from Table IT, tic reaction by cell-free extracts from Clostridium that the rate of the reaction catalyzed by benzyl pasteurianum. Judged from the enzymatic proper- THE REDUCTION OF FLAVODOXIN FROM AZOTOBACTER V1NELAND1I BY PYRUVATE 1093 ties, the flavoprotein from Azotobacter isolated first Contrary to the photosynthetic NADP+-reduction, by SHETHNA 1 belongs to the class of flavodoxins. the rate of the phosphoroclastic reaction, catalyzed However, there are certainly chemical differences by Azotobacter flavodoxin, is not dependent on oxy- between the flavodoxin isolated from Azotobacter gen exclusion in the experiments. This result is not and the rest of the flavodoxins so far described surprising. In the phosphoroclastic reaction, the (ref.3). The most striking feature fo Azotobacter electrons are accepted by the oxidized (or the radi- flavodoxin is the unusual stability of the radical, cal) form, and the oxidized form is reduced to the which is probably not so pronounced in all the hydroquinone form which is the only oxygen-sensi- other flavodoxins. The reason for the stabilization tive species. of the semiquinone form of flavodoxin from Azoto- ARNON and coworkers 13 were the first to report bacter is not fully understood (compare discussion on cell-free nitrogen fixation activity catalyzed by in ref. 10). The stabilization of the radical is not flavodoxin (or ferredoxin) from Azotobacter, using only seen with the treatment by dithionite, but also NADPH and ferredoxin-NADP+ reductase from spi- in all enzymatic reductions. This finding is sur- nach as electron donor system. Their reported cell- prising, since the redoxpotentials of at least some free activities in the presence of flavodoxin from electron donors (pyruvate, illuminated chloroplasts) Azotobacter, however, were low. Such low activities should be sufficiently electronegative to allow the should be expected, since a reduction of Azotobacter reduction to the fully reduced form completely to flavodoxin to the hydroquinone form by NADPH hundred per cent. is thermodynamical extremely unfavourable. Pyru- The photosynthetic NADP+-reduction by illumi- vate would be a much better electron donor than nated spinach chloroplasts, using flavodoxin from NADPH in the reduction of flavodoxin from Azoto- 14,15 Azotobacter as a substitute for ferredoxin, was bacter. Recently, VEEGER'S group could demon- found to be oxygen sensitive 3. From this finding we strate a pyruvate clastic reaction in Azotobacter. suggested that the fully reduced form of Azotobacter Until now, however, these authors tested only violo- flavodoxin — as the only oxygen sensitive species — gens, but did not try ferredoxin or flavodoxin from is involved in the catalysis of the photosynthetic Azotobacter in this pyruvate clastic reaction and NADP+-reduction. Under aerobic conditions, the also did not connect this reaction with nitrogen- fully reduced form of Azotobacter flavodoxin is fixation. Although all experiments evidence is lack- immediately reoxidized by oxygen; however, under ing, it is nevertheless tempting to speculate a scheme oxygen exclusion, it is able to transfer the electrons for nitrogen-fixation in Azotobacter which comprises to NADP+ in an enzymatic reaction catalyzed by a sequence pyruvate —> flavodoxin (or ferredoxin) ferredoxin-NADP+-reductase. Also in the phosphoro- —> nitrogenase complex. This scheme may be too clastic reaction Azotobacter flavodoxin is reduced simple with Azotobacter but would be similar to the to the hydroquinone form. This is concluded from "classical one" for nitrogen-fixation described for the observation, that the phosphoroclastic reaction Clostridium A carrier with a more negative re- can be conducted under anaerobic conditions, where doxpotential than ferredoxin (Eq ~ — 400 mV) was the reduced flavodoxin can only be reoxidized demanded as direct electron donor for the nitro- 16 through hydrogen evolution catalyzed by the hydro- genase complex ; such a candidate could be the genase enzyme present in Clostridial extracts. Only Azotobacter flavodoxin. Experiments to test the role the couple hydroquinone/semiquinone form with a of flavodoxin and ferredoxin in nitrogen-fixation redoxpotential of — 510 mV 12, but not that one of by cell-free systems from Azotobacter are now in progress. the semiquinone/oxidized form with E0 = + 50 mV (ref. 12) is electronegative enough to transfer the electrons to the hydrogenase. Also the pH- dependence of the phosphoroclastic reaction catalyz- We would like to thank the Deutsche Forschungs- ed by Azotobacter flavodoxin indicates the partici- gemeinschaft for financial support in this work. The excellent technical assistance of Mrs. G. PSCZOLLA is pation of the fully reduced form in this reaction. gratefully acknowledged.

1 Y. I. SHETHNA, P. W. WILSON, and H. BEINERT, Biochim. 2 J. R. BENEMANN, D. C. YOCH, R. C. VALENTINE, and D. I. biophysica Acta [Amsterdam] 113,225 [1966]. ARNON, Proc. nat. Acad. Sei. U.S.A. 64, 1079 [1969]. 1094 K. D. WATENPAUGH ET AL.

B. VAN LIN and H. BOTHE, Ardi. Mikrobiol. 82, 155 [1972]. 10 G. TOLLIN and D. E. EDMONDSON, in: Flavins and Flavo- S. G. MAYHEW and V. MASSEY, J. biol. Chemistry 244, 803 . III. Internat. Symposium, p. 153, H. KAMIN, ed. [1966]. Baltimore; University Park Press 1971. H. BOTHE, P. HEMMERICH, and H. SUND, in: Flavins and 11 L. E. MORTENSON, R. C. VALENTINE, and J. E. CARNAHAN, . III. Internat. Symposium, p. 211, H. KAMIN, J. biol. Chemistry 238, 794 [1963]. ed., Baltimore; University Park Press 1971. 12 G. TOLLIN, communicated at the International Workshop L. E. MORTENSEN, R. C. VALENTINE, and J. E. CARNAHAN, on Flavins and Flavoproteins, helt at Konstanz, Germany, Biochim. biophysic. Res. Comm. 7, 448 [1962]. in March 1972. S. RAEBURN and J. C. RABINOWITZ, Archives Biochim. Bio- 13 J. R. BENEMANN, D. C. YOCH, R. C. VALENTINE, and D. I. phys. 146,21 [1971]. ARNON, Biochim. biophysica Acta [Amsterdam] 226, 305 Y. I. SHETHNA, H. BEINERT, and P. HEMMERICH, unpub- [1971]. lished data, cited by P. HEMMERICH, C. VEEGER, and H. S. 14 T.W. BRESTERS, J. KRUL, P. C. SCHEEPENS, and C. VEEGER, C. WOOD, Angew. Chemie, int. Edit. 4, 680 [1965]. FEBS Letters 22, 305 [1972]. J. W. HINKSON and W. A. BULEN, J. biol. Chemistry 242, 15 H. HAAKER, T. W. BRESTERS, and C. VEEGERS, FEBS Let- 3345 [1967]. ters, in press. 18 R. W. F. HARDY and R. C. BURNS, Ann. Rev. Biochem. 37. 331 [1968].

Flavodoxin from the Sulfate Reducing Bacterium Desulfovibrio vulgaris. Its Structure at 2.5 Ä Resolution

K. D. WATENPAUGH, L. C. SIEKER, and L. H. JENSEN

Department of Biological Structure, University of Washington, Seattle, U.S.A.

and J. LE GALL and M. DUBOURDIEU

Department of Biochemistry, University of Georgia, Athens, U.S.A.

L. D. B., C. N. R. S., Marseille, France

(Z. Naturforsch. 27 b, 1094—1095 [1972]; received May 10, 1972)

Protein structure, X-ray diffraction, Flavodoxin, Desulfovibrio vulgaris Crystals of oxidized Desulfovibrio vulgaris flavodoxin have been studied by X-ray diffraction techniques. The structure has been determined at 2.5 A resolution. The molecule consists of a pleated sheet core with two sections of helix on each side of the sheet. The FMN group is distinguishable: it is mostly burried but 7- and 8-Methyl groups appear to be at the surface of the protein.

We report here the data obtained from a 2.5 Ä resolution electron density map of D. vulgaris flavo- doxin. The electron density map was relatively easy to interpret with only a few minor breaks in the main chain and no crossovers between neighboring molecules. The molecules of flavodoxin are approxi- mately oblate spheroids of dimensions 25 Ä X 40 Ä x 40 Ä, packed in the crystal lattice as shown in Fig. 1. There are large channels of solvent be- tween molecules alternately parallel to the x-axis and the y-axis. The molecule essentially consists of a central portion of five sections of chain in a parallel pleated sheet configuration with two sec- tions of helix on each side of the pleated sheet as Fig. 1. The conformation of D. vulgaris flavodoxin peptidic chain including the position of FMN. shown in Fig. 1. Approximately 1/3 of the residues are involved in the pleated sheet, 1/3 in the helices The sequence of this flavodoxin has not been and 1/3 in extended chain or other configurations. completed yet and we cannot identify most residues with any degree of certainty. It does appear that Requests for reprints should be sento to K. D. WATEN- none of the cysteine are involved in disulfide bridges PAUGH, Department of Biological Structures, University of Washington, Seattle, U.S.A. or in binding to the FMN ().