The B form of dihydroorotate dehydrogenase from Lactococcus lactis consists of two different subunits, encoded by the pyrDb and pyrK genes, and contains FMN, FAD, and [FeS] redox centers.

The B form of dihydroorotate dehydrogenase from Lactococcus lactis (DHOdehase B) is encoded by the pyrDb gene. However, recent genetic evidence has revealed that a co-transcribed gene, pyrK, is needed to achieve the proper physiological function of the enzyme. We have purified DHOdehase B from two strains of Escherichia coli, which harbored either the pyrDb gene or both the pyrDb and the pyrK genes of L. lactis on multicopy plasmids. The enzyme encoded by pyrDb alone (herein called the δ-enzyme) was a bright yellow, dimeric protein that contained one molecule of tightly bound FMN per subunit. The δ-enzyme exhibited dihydroorotate dehydrogenase activity with dichloroindophenol, potassium hexacyanoferrate(III), and molecular oxygen as electron acceptors but could not use NAD+. The DHOdehase B purified from the E. coli strain that carried both the pyrDb and pyrK genes on a multicopy plasmid (herein called the δκ-enzyme) was quite different, since it was formed as a complex of equal amounts of the two polypeptides, i.e. two PyrDB and two PyrK subunits. The δκ-enzyme was orange-brown and contained 2 mol of FAD, 2 mol of FMN, and 2 mol of [2Fe-2S] redox clusters per mol of native protein as tightly bound prosthetic groups. The δκ-enzyme was able to use NAD+ as well as dichloroindophenol, potassium hexacyanoferrate(III), and to some extent molecular oxygen as electron acceptors for the conversion of dihydroorotate to orotate, and it was a considerably more efficient catalyst than the purified δ-enzyme. Based on these results and on analysis of published sequences, we propose that the architecture of the δκ-enzyme is representative for the dihydroorotate dehydrogenases from Gram-positive bacteria.

Dihydroorotate dehydrogenase catalyzes the fourth chemical reaction in the biosynthesis of UMP, which is oxidation of 5,6-dihydroorotate to orotate. Genes encoding this enzyme have been cloned and sequenced from a variety of organisms. The milk-fermenting bacterium Lactococcus lactis is the only organism so far known to contain two dihydroorotate dehydrogenases. They have been termed dihydroorotate dehydrogenase A (DHOdehase A) 1 and dihydroorotate dehydrogenase B (DHOdehase B) and are encoded by the pyrDa and pyrDb genes, respectively (Andersen et al., 1994). Both enzymes are able to function in pyrimidine biosynthesis, since both of the genes must be inactivated by mutation in order to impose a pyrimidine requirement on L. lactis and since either of the two genes is able to correct the pyrimidine requirement of a pyrD deletion strain of Escherichia coli (Andersen et al., 1994). The polypeptides encoded by the pyrDa and pyrDb genes both consist of 311 amino acid residues, and the predicted amino acid sequences are 30% identical with each other. However, the sequence of DHOdehase A shows 71% amino acid identity with the sequence of the cytosolic dihydroorotate dehydrogenase from bakers' yeast, while the sequence of DHOdehase B shows 60 -70% amino acid identity with the deduced amino acid sequence of dihydroorotate dehydrogenases from B. subtilis and several other Gram-positive bacteria (Andersen et al., 1994;Nielsen et al., 1996).
We have initiated a study of these two dihydroorotate dehydrogenases from L. lactis, with the aim of comparing their functional and structural properties with the properties of the enzymes from E. coli (Larsen and Jensen, 1985) and other organisms. We began by purifying the two lactococcal enzymes from strains of E. coli that carried either pyrDa or pyrDb cloned on multicopy plasmids. DHOdehase A was a stable enzyme and proved to be a dimeric protein, containing one molecule of FMN per subunit. It was an efficient catalyst that could use dichloroindophenol, potassium hexacyanoferrate(III), or, to a lower extent, molecular oxygen as an acceptor of the reducing equivalents from dihydroorotate (Nielsen et al., 1996). However, the DHOdehase B encoded by the pyrDb gene turned out to be an unstable and inefficient enzyme, although it could be produced in substantial quantities in E. coli. We considered the possibility that this enzyme might require unusual electron acceptors or unusual incubation conditions for optimal function. While we were searching for such assay conditions, it appeared as a result of genetic studies that the activity of DHOdehase B in L. lactis was dependent on the integrity of a neighboring open reading frame, now termed pyrK, which is co-transcribed with pyrDb (Andersen et al., 1996). Therefore, we introduced the pyrK gene into our expression vector, which already carried pyrDb, and purified DHOdehase B from a strain of E. coli harboring this plasmid. The resulting enzyme was dramatically different from the enzyme encoded by the pyrDb gene alone, as it appeared to be a stable stoichiometric complex of PyrDB and PyrK polypeptides. Furthermore, the protein had acquired the ability to use NAD ϩ as a co-substrate for the oxidation of dihydroorotate.
For the sake of simplicity we have named the native enzyme encoded only by the pyrDb gene the ␦-enzyme, and the enzyme encoded by pyrDb and pyrK was called the ␦-enzyme.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases, T4 DNA ligase, and Deep Vent (exo-) DNA polymerase were bought from either New England Biolabs or Boehringer Mannheim and used as recommended by the manufacturers. The Sequenase 2.0 kit was from U.S. Biochemical Corp. Diethylaminoethyl-cellulose (DE52) was from Whatman BioSystems Ltd. (Maidstone, United Kingdom), hydroxylapatite (Bio-Gel) was from Bio-Rad, the dye-liganded Matrex Red A was from Amicon, and the Superose 12 column was from Pharmacia (Uppsala, Sweden). Blue dextran-Sepharose was prepared as described previously (Led et al., 1983). Molecular weight marker proteins were bought from Bio-Rad or Pharmacia. Sodium dodecyl sulfate was from BDH (Poole, UK), and acrylamide solutions (Protogel and Sequagel) were from National Diagnostics. Other fine chemicals were either from Merck (Darmstadt, Germany) or Sigma. Radionucleotides were purchased from DuPont NEN.
Construction of Expression Vectors-The expression vectors, pFN2 and pFN4, were constructed by cloning PCR copies of the pyrDb and pyrK genes from L. lactis, present on plasmids pIP51 and pKP6, respectively (Andersen et al., 1996) into the multicopy plasmid pUHE23-2 (obtained from H. Bujard, Heidelberg). This plasmid carries the very strong LacI-repressible P A1/04/03 promoter to drive transcription of cloned genes (Deutschle et al., 1986). For the pyrDb gene, the PCR reaction was directed by two synthetic oligonucleotides (5Ј-CCGGAAT-TCAGGAGAGAAATAATGACTGAA and 5Ј-CGCGGATCCGAATTA-TTTTTTGCCTTCTTTTACT), which were designed to generate an EcoRI and a BamHI site at the start and the end of the resulting PCR fragment. For the pyrK gene, the PCR reaction was directed by two synthetic oligonucleotides (5Ј-CGGGATCCCGTCCGTAAATAAA-AGAATGGA and 5Ј-AACTGCAGAATTAGAATGAAAGCTGTTT), designed to generate a BamHI and a PstI site at the start and the ends of the DNA fragment. The resulting PCR fragments, as well as the vector pUHE23-2, were digested either with EcoRI and BamHI (for pyrDb) or BamHI and PstI (for pyrK), and after removal of the phosphates from the ends of the digested vector with calf intestine alkaline phosphatase, the DNA fragments were ligated together by standard techniques. After transformation of the E. coli strain SA6645 (araD139⌬(ara-leu)7679 galU galK ⌬(lac)174 ⌬pyrD(MluI-BssHII::Km r )[FЈ proAB lacI q Z⌬M15 Tn10]) with the ligation mixture, colonies that were resistant to ampicillin were selected on agar plates. Plasmids were isolated from several independent colonies and tested for the content of cloned PCR fragments, and the structures of the selected plasmids, pFN2 and pFN4, are shown in Fig. 1. The plasmid pFN3 was constructed by transferring the EcoRI-BamHI fragment carrying the pyrDb gene from plasmid pFN2 to pFN4, also cut with EcoRI and BamHI. The nucleotide sequences of the cloned PCR fragments were determined by the technique of Sanger et al. (1977) using the Sequenase 2.0 kit (U.S. Biochemical Corp.) and the PCR primers combined with two universal sequence primers located on either side of the cloning region of pUHE23-2. The sequence of the pyrDb fragment was found to be identical to the published sequence of the pyrDb gene of L. lactis (Andersen et al., 1994), but the sequence of codon number 201 in the pyrK gene of our PCR product (plasmids pFN3 and pFN4) was read as a GCC codon (encoding alanine) instead of the CGC codon (encoding arginine) in the published sequence (Andersen et al., 1996). However, sequencing of the template plasmid pKP6 and inspection of the original gels showed that this difference was due to an error in the published sequence, accession number X74207.
Assays of Dihydroorotate Dehydrogenase Activity-In the standard assay for dihydroorotate dehydrogenase activity, the oxidation of dihydroorotate was coupled to the reduction of the synthetic quinone dichloroindophenol (DCIP). The reduction of 1 mol of DCIP causes a decrease in the absorbance at 600 nm, ⑀ ϭ 20 ϫ 10 3 M Ϫ1 cm Ϫ1 (Karibian, 1978). The spectra were recorded in a Zeiss Specord S10 diode-array photometer. The standard assay mixture contained 0.1 M Tris-HCl, pH 8.0, 5 mM KCN, 1 mM dihydroorotate, and 50 M DCIP. The assay temperature was 37°C. One unit of enzyme activity is defined as the amount of enzyme that produces 1 mol of orotate/min under these conditions. In assays with different electron acceptors, we used the absorption at 295 nm to obtain a quantitative measure of the production of orotate (⑀ ϭ 3.67 ϫ 10 3 M Ϫ1 cm Ϫ1 ).
Growth of Cells for Purification of DHOdehase B-Two forms of DHOdehase B were purified from strain SA6645, transformed either with the expression vector pFN2, which carried the pyrDb gene, or with plasmid pFN3, which carries both the pyrDb and pyrK genes of L. lactis. The cells were grown to stationary phase at 37°C with vigorous aeration in LB broth medium (Miller, 1972) supplemented with 0.1 g/liter ampicillin. The synthesis of DHOdehase was induced by the addition of 0.75 mM isopropyl-␤-D-thiogalactoside when the optical density (A 436 ) of the culture was 1.0. For plasmid pFN2, growth was continued for 24 h until the culture had been stationary for several hours, while the cultures containing pFN3 were harvested 3.5 h after induction because the PyrK polypeptide was slowly degraded upon prolonged incubation in the stationary phase. The cells were harvested by centrifugation for 20 min at 6000 rpm using a GS-3 rotor in a refrigerated Sorvall centrifuge, washed with 0.9% NaCl, and kept frozen at Ϫ20°C.
Purification of DHOdehase B-The buffer used during all steps in the purification was 50 mM sodium phosphate, pH 6.0, containing 10% glycerol, termed Buffer A. Unless otherwise stated, all operations were carried out on a melting ice bath or in a refrigerated room at 4°C. All columns were run with a flow rate of 1 ml/min, and 5-ml fractions were collected.
The ␦-Enzyme-To purify the ␦-enzyme, frozen cell pellets from 2.5liter stationary cultures of SA6645/pFN2 were suspended in 75 ml of ice cold Buffer A and disrupted by ultrasonic treatment using a Branson sonifier for 15 ϫ 0.5 min, interrupted by cooling in an ice bath for 1.5 min between cycles of sonication. Cell debris was removed by centrifugation for 20 min at 12.000 rpm in a Sorvall SS-34 rotor. 1 ⁄10 volume of a 10% solution of streptomycin sulfate was added to the supernatant. After stirring for 15 min, the precipitate was removed by centrifugation as described above. The supernatant was dialyzed for 2 h against 2 liters of 5 mM sodium phosphate containing 10% glycerol (pH 6) and cleared by centrifugation. The supernatant was applied onto a 25-ml column of Matrex Red A (Amicon). After washing the column with 75 ml of Buffer A, the enzyme was eluted with a linear gradient from 0 to 0.80 M NaCl in Buffer A. The fractions with most DHOdehase activity were pooled and dialyzed for 3 h against 1 liter of 5 mM sodium phosphate, pH 6, containing 10% glycerol. The dialyzed sample was applied on a 25-ml column of DE52 (Whatman) equilibrated with Buffer A. After washing the column with 75 ml of Buffer A, the enzyme was eluted with Buffer A containing 0.2 M NaCl. Fractions with the most DHOdehase activity were pooled and dialyzed against 1 liter of Buffer A overnight. Subsequently, the enzyme solution was loaded on a 25-ml column of blue dextran Sepharose. The column was washed with 100 ml of Buffer A and eluted with a 100-ml linear gradient from 0 to 1.0 M NaCl in Buffer A. Fractions with most DHOdehase activity were pooled, concentrated, and dialyzed against Buffer A using a Micro Ultrafiltration system (Amicon). Glycerol was added to 50%, and the enzyme was stored at Ϫ20°C.
The purification procedure is summarized in Table I, and an SDS-PAGE analysis of the enzyme product is shown in Fig. 2.
The ␦-Enzyme-In order to purify the ␦-enzyme, frozen cell pellet from a 10-liter culture of SA6645/pFN3 was suspended in 75 ml of Buffer A and disrupted by sonication as described above. Cell debris was removed by centrifugation for 20 min at 12,000 rpm in a Sorvall SS-34 rotor. 1 ⁄10 volume of a 10% solution of streptomycin sulfate was added to the supernatant. After stirring for 15 min, the precipitate was removed by centrifugation as described above. The supernatant was dialyzed for 2 h against 2 liters of 5 mM sodium phosphate containing 10% glycerol (pH 6) and cleared by centrifugation. The supernatant was loaded onto a 100-ml column of DE52. After the loading was completed, the column was first washed with 400 ml of Buffer A and then eluted with a linear gradient (200 ml) from 0 to 0.20 M NaCl in Buffer A followed by a 100-ml linear gradient from 0.20 M NaCl to 1.0 M NaCl. The orange-brown ␦-enzyme appeared from the column with a peak at 0.22 M NaCl, while a surplus of the yellow-green ␦-enzyme peaked at 0.15 M NaCl. Fractions containing the ␦-enzyme were pooled, dialyzed for 2 h against 2 liters of 5 mM sodium phosphate containing 10% glycerol (pH 6), and applied onto a 25-ml column of hydroxylapatite. This column was washed with 100 ml of Buffer A and eluted with a linear gradient over 200 ml from Buffer A to 500 mM sodium phosphate, pH 6.0, containing 10% glycerol. Fractions containing most dihydroorotate dehydrogenase activity (peaking at 0.25 M sodium phosphate) were pooled, dialyzed for 2 h against 2 liters of 5 mM sodium phosphate containing 10% glycerol (pH 6), and loaded onto a 25-ml column of Matrex Red A. After washing the column with 50 ml of Buffer A, the enzyme was eluted with a 200-ml gradient from 0 to 1.0 M NaCl in Buffer A. The activity peaked at 0.15 M NaCl. The active fractions were pooled and dialyzed as described above, and the chromatography on Matrex Red A was repeated. The active fractions were concentrated using a Micro Ultrafiltration system (Amicon), dialyzed exhaustively against 5 mM sodium phosphate, pH 6.0, containing 50% glycerol, and stored at Ϫ20°C.
Determination of Flavin Content-The flavin was released from aliquots of the enzymes by treating with 0.25 M formic acid and analyzed by chromatography on poly(ethyleneimine)-impregnated cellulose thin layer plates together with authentic FMN (R F ϭ 0.35) and FAD (R F ϭ 0.17) as described by Larsen and Jensen (1985). In addition, the flavin was extracted from the ␦-enzyme by treatment with 4% ammonium sulfate in 75% methanol as described by Aleman and Handler (1967). After pelleting the protein part of the enzyme by centrifugation, the spectrum of the supernatant was recorded and compared with the spectra of authentic FMN, FAD, and mixtures of the two flavin compounds, dissolved in 4% ammonium sulfate, 75% methanol.
Determination of Iron Content-Aliquots of the ␦-enzyme (800 l containing 1-12 nmol of enzyme in 5 mM sodium phosphate, pH 6) were mixed with 100 l of 8 M HCl and incubated for 10 min at 0°C. Protein was precipitated by the addition of 100 l of 80% trichloroacetic acid for 10 min, and the solution was cleared by centrifugation. 200 l of 75% ammonium acetate was added to 800 l of the supernatant to adjust the pH to 4.5. Subsequently, 80 l of 10% hydroxylamine hydrochloride and 80 l of 4 mM tripyridyl-s-triazine were added, and the mixtures were incubated for 10 min. The amount of iron was quantified by measuring the absorption at 593 nm as described by Fischer and Price (1964). One nmol of Fe gave an absorption A 593 ϭ 0.015.
Determination of Acid-labile Sulfide-Aliquots of enzyme (320 l) were treated with 2.6% Zn(CH 3 COO) 2 and 0.75% NaOH for 2 h; 100 l of 0.1% N,N-dimethyl-p-phenylenediamine, dissolved 5 M HCl, and 40 l of 11.5 mM FeCl 3 in 0.6 M HCl were then added, and the solution was mixed by shaking for 1 min. Finally, 320 l of water was added, and the sample was cleared by centrifugation. The acid-labile sulfide was quantified by measuring A 670 as described by King and Morris (1964). One nmol of S 2Ϫ gave an absorption A 670 ϭ 0.032.
Kinetic Analyses-Saturation curves from kinetic experiments were fitted to the Michaelis-Menten equation using the BIOSOFT program Ultrafit for the Macintosh.

Production of Dihydroorotate Dehydrogenase B-
The expression vectors used for production of DHOdehase B in E. coli are described in Fig. 1. All three plasmids are derivatives of pUHE23-2 and contain the strong LacI-controlled P A1/04/03 promoter to drive transcription of the cloned genes. Plasmid pFN2 contains only the pyrDb gene, which we initially thought would contain all coding information for DHOdehase B of L. lactis (Andersen et al., 1994). Plasmid pFN3 contains both pyrDb and pyrK, but the two genes are inserted in opposite order relative to the order by which they are transcribed from the chromosome of L. lactis. Plasmid pFN4 carries only the pyrK gene. The plasmids, pFN2 and pFN3, were able to complement the pyrimidine requirement of the E. coli strain SA6645, which is deleted for the pyrD gene, but pFN4 was not. In order to use SA6645, transformed with pFN2 or pFN3, for production of DHOdehase, it was important to grow the cultures to a considerable density while the strong P A1/04/03 promoter was kept repressed, since growth terminated approximately one generation after induction of promoter activity by the addition of isopropyl-␤-D-thiogalactoside.
Purification of the ␦-Enzyme, Encoded by the pyrDb Gene on pFN2-The purification procedure for DHOdehase B encoded by pyrDb is described under "Experimental Procedures" and summarized in Table I. The enzyme could be produced in substantial amounts in an electrophoretically homogeneous form (Fig. 2). However, the enzyme was unstable, and the specific activity decreased slightly in the last steps of purification (Table I). In earlier versions of the purification, the fall in specific activity during purification was even more dramatic. The half-life of the ␦-enzyme was about 45 s under assay conditions at 37°C and was 4 min when the assays were performed at 25°C. If a solution of the ␦-enzyme in the purification buffer was left at room temperature overnight, no dihydroorotate dehydrogenase activity remained.
Purification of the ␦-Enzyme, Encoded by the pyrDb and pyrK Genes on pFN3-The purification procedure for the ␦enzyme is described under "Experimental Procedures" and summarized in Table II. The resulting enzyme contained equal amounts of PyrDB and PyrK polypeptides (Fig. 2). These two polypeptides seemed to form a very stable complex with each other, since they have resisted separation over many steps of column chromatography and since they migrated as a single protein during electrophoresis in a nondenaturing agarose gel with mobility very different from the mobility of the ␦-enzyme, which contained only the PyrDB subunits (Fig. 3). The complex FIG. 1. Structure of the expression  vectors pFN2, pFN3, and pFN4. Transcription of the cloned genes is driven by the very strong P A1/04/03 promoter, which is a synthetic derivative of the early A1 promoter of phage T7 containing two binding sites for the lac repressor. Expression of cloned genes is kept repressed by the lacI q repressor until induction with isopropyl-␤-D-thiogalactoside. cat, gene for chloramphenicol acetyltransferase; bla, gene for ␤-lactamase, t o , transcription terminator; pyrDb, gene encoding the PyrDB polypeptide (deduced mass ϭ 33.0 kDa); pyrK, gene encoding the PyrK polypeptide (deduced mass ϭ 28.6 kDa).
␦-enzyme was a very stable protein. The activity could be assayed without problems at 37°C, and approximately 75% of the activity remained when the protein was incubated in the purification buffer for 20 min at 55°C.
Molecular Masses and Subunit Composition-The two forms of DHOdehase B were subjected to gel filtration chromatography on a Superose 6 HR 10/30 column (Pharmacia) together with standard marker proteins. Fractions were collected and analyzed by measurements of enzyme activity and by SDS-gel electrophoresis. The ␦-enzyme eluted from the column together with bovine serum albumin, molecular mass 64 -66 kDa. This gel filtration behavior indicated that the protein is a homodimer consisting of two PyrDB subunits, since the molecular mass of the subunit is 33 kDa. On the other hand, the native ␦-enzyme eluted at a position corresponding to a protein with a molecular mass of 130 kDa. This indicated that the ␦-  2. SDS gel electrophoretic analyses of purified DHOdehase B. A, lanes 1 and 6, marker proteins; lanes 2 and 3, purified ␦-enzyme (3 and 12 g); lanes 4 and 5, purified ␦-enzyme (1.5 and 6 g). The protein band below the main protein in lanes 4 and 5 was generated during prolonged storage of the ␦-enzyme. B, the lanes between the two marker lanes (labeled M) contain samples of fractions over the activity peak of the ␦-enzyme as it eluted from the second Matrex Red A column. The 12.5% polyacrylamide gels were prepared as described by Laemmli (1970), run in a Mini Protean System II apparatus (Bio-Rad), and stained with Coomassie Brilliant Blue G-250. a The amount of protein in the samples were determined by the Lowry procedure (Lowry et al., 1951). According to a determination of the content of amino acid in acid hydrolysates of the protein, the Lowry procedure overestimated the absolute protein concentration by 41%. The specific activity of the purified ␦-enzyme is therefore 48 units/mg. enzyme is a tetramer composed of two PyrDb subunits (33 kDa) and two PyrK subunits (29 kDa).
Spectral Properties and Cofactor Content of the Two Enzymes-The purified ␦-enzyme was bright yellow, with a trace of green, and it showed an absorption spectrum typical for an oxidized flavoprotein with absorption maxima at 450 and 375 nm (Fig. 4). The absorbance at 450 nm was 0.29 per mg of protein, determined by the Lowry procedure (Lowry et al., 1951). This value indicated that the ␦-enzyme contains 0.9 mol of flavin/mol of subunit (M r ϭ 33,000), since flavoproteins usually have an absorption coefficient A 450 of about 11 mM Ϫ1 cm Ϫ1 (Untuch-Grau et al., 1982). The flavin was released from the protein by treatment with 0.33 M formic acid and found to co-migrate with authentic FMN by thin layer chromatography, while it migrated twice as fast as FAD.
The complex ␦-enzyme was orange-brown instead of yellow, and the characteristic flavin peaks at 377 and 452 nm in the absorption spectrum (Fig. 4) were superimposed onto a broad range of absorption extending from 300 to beyond 600 nm. Upon treatment with formic acid, both FMN and FAD were released from the enzyme. The two flavin compounds were present in approximately equal amounts as judged from the intensities of the two yellow spots on the chromatogram inspected under UV light. The flavins were also extracted from the enzyme with 4% ammonium sulfate and 75% methanol, leaving a yellow supernatant and a brownish protein pellet after centrifugation. The absorbance of the supernatant at 448 nm indicated that one mol of native ␦-enzyme contains 4.1 Ϯ 0.1 mol of flavin. The ratio between the absorption at 268 nm and the absorption at 448 nm was compared with the similar ratios of absorbances of mixtures of authentic FMN and FAD. This analysis revealed that the enzyme contains approximately 40% FMN and 60% FAD. Based on these results, we propose that the ␦-enzyme contains 2 mol of FAD and 2 mol of FMN per mol of native tetrameric enzyme.
The absorption spectrum in Fig. 4, as well as the color of the protein, indicated that the enzyme contains iron as well as the flavins, and since it also developed a characteristic smell of sulfide when it was treated with sulfuric acid, we suspected that the protein contained iron-sulfur redox centers. The iron content was quantified by the method of Fischer and Price (1964) and a value of 3.6 Ϯ 0.3 mol of Fe/mol of native ␦enzyme was found. Furthermore, we found 3.2 Ϯ 0.3 mol of acid-labile sulfide/mol of enzyme by using the method of King and Morris (1964). These iron and sulfur analyses were performed more than six times using two different preparations of the enzyme, and since the method of King and Morris (1964) usually underestimates the true content of sulfur, the data suggest strongly that one mol of native ␦-enzyme contains two mol of [2Fe-2S] redox centers. The iron-sulfur clusters are likely to be bound to the sulfhydryl-rich stretches of amino acid residues near the carboxyl termini of the PyrK subunits (Fig. 5).
Catalytic Properties and Specificity-Both forms of DHOdehase B displayed optimal activity around pH 8 when assayed with DCIP as electron acceptor and dihydroorotate as substrate (Fig. 6). The specificity of the enzyme reactions at pH 8 is shown in Table III. It appears that both forms of DHOdehase B could use dichloroindophenol, potassium hexacyanoferrate(III), and to a lower extent also molecular oxygen as electron acceptors for the conversion of dihydroorotate to orotate, whereas only the ␦-enzyme was able to use NAD ϩ . When assayed under standard conditions at pH 8, using 50 M DCIP as an electron acceptor, the apparent K m for orotate was 28 Ϯ 2 M for the ␦-enzyme and 949 Ϯ 48 M for the ␦-enzyme, while the apparent V max was 3-fold higher for the ␦-enzyme Plus and minus signs indicate the electrodes. The gel was fixed with 10% acetic acid, dried, and stained with Coomassie Brilliant Blue G-250 than for the ␦-enzyme, indicating that the ␦-enzyme is a considerably more efficient catalyst than the ␦-enzyme. The apparent K m of the ␦-enzyme for NAD ϩ was 111 Ϯ 12 M when assayed with 1 mM dihydroorotate as co-substrate. Neither of the enzymes was able to use dihydrouracil as a substrate with DCIP as electron acceptor.
The ␦-enzyme catalyzed efficient conversion of orotate to dihydroorotate at the expense of the reducing equivalents of [NADH ϩ H ϩ ] at pH 6.5, but no formation of dihydroorotate could be detected at pH 8, probably because this high pH favors formation of [NADH ϩ H ϩ ]. No conversion of uracil to dihydrouracil could be detected at pH 6.5 using [NADH ϩ H ϩ ] as co-substrate, showing again that DHOdehase B of L. lactis does not belong to the class of dihydropyrimidine dehydrogenases, which are able to reduce all natural pyrimidine bases.

DISCUSSION
The results presented in this paper document unambiguously that the polypeptides encoded by the pyrDb and pyrK genes of L. lactis form a protein complex, herein termed the ␦-enzyme, since they resisted separation by chromatography on several types of columns and since they migrated as a single protein species during electrophoresis in a nondenaturing gel.
The ␦-enzyme is able to use dichloroindophenol, potassium hexacyanoferrate(III), NAD ϩ , and, to some extent, molecular oxygen as acceptors of the reducing equivalents from dihydroorotate. We propose that the complex ␦-enzyme is the physiological form of DHOdehase B in L. lactis, since disruption of the pyrK gene in mutants lacking DHOdehase A (encoded by pyrDa) gave rise to a 4-fold lower growth rate in the absence of pyrimidines and made it impossible to assay DHOdehase B activity in extracts of L. lactis with the use of NAD ϩ as electron acceptor (Andersen et al., 1996). It seems likely that the archi-tecture of the ␦-enzyme is the prototype of the dihydroorotate dehydrogenases from Gram-positive bacteria, since the PyrDB protein shows very high sequence similarity to all known PyrD proteins of Gram-positive bacteria (i.e. Bacillus subtilis, Bacillus caldolyticus, and Enterococcus faecalis (Nielsen et al., 1996)) and since an orf with very high similarity to pyrK is found immediately upstream of the pyrD gene in these bacteria (Andersen et al., 1996;Ghim et al., 1994;Li et al., 1995;Quinn et al., 1991). Moreover, it was shown in the case of B. subtilis that disruption of this orf, now termed pyrDII, imposes a partial pyrimidine requirement on the bacterium and strongly lowers the activity of dihydroorotate dehydrogenase in cell-free extracts (Kahler and Switzer, 1996).
The behavior of the ␦-enzyme during SDS-gel electrophoresis and gel filtration indicated strongly that it consists of two PyrDB and two PyrK subunits. Spectrophotometric and chromatographic determinations showed that the native enzyme contains 2 mol of FAD and 2 mol of FMN. In addition, the iron and sulfide analyses indicated strongly that the tetrameric ␦-enzyme contains 2 mol of [2Fe-2S] redox centers. The ability to bind FMN must reside in the PyrDB subunit, while the ability to bind FAD and the [2Fe-2S] clusters as prosthetic groups seems to be linked to the PyrK subunit in the complex, since the protein encoded by the pyrDb gene alone, i.e. the ␦-enzyme, is a functional dimeric dihydroorotate dehydrogenase that only contains FMN. Furthermore, the capability to use NAD ϩ as an electron acceptor is linked to the presence of the PyrK subunits in the complex, since the ␦-enzyme is unable to function with NAD ϩ as a substrate.
The ␦-enzyme resembles several other dihydroorotate dehydrogenases, e.g. dihydroorotate dehydrogenase A from L. lactis (Nielsen et al., 1996), and the enzymes from E. coli (Larsen and Jensen, 1985) and the two protozoans, Crithidia fasciculata and Trypanosoma bruceri (Pascal et al., 1983). These enzymes are all dimeric dihydroorotate dehydrogenases with one FMN per subunit, and they are unable to use NAD ϩ as an electron acceptor. The dihydroorotate dehydrogenases purified from bovine liver mitochondria and the mitochondria of Neurospora crassa were also found to contain FMN (Hines and Johnson, 1989;Hines et al., 1986;Miller, 1975;Miller and Adams, 1971), whereas flavin could not be detected in significant quantities in FIG. 5. Clustering of cysteinyl residues in the PyrK homologues of four Gram-positive bacteria. Ll, L. lactis (Andersen et al. 1996); Bc, B. caldolyticus (Ghim et al. 1994); Bs, B. subtilis (Kahler and Switzer, 1996;Quinn et al., 1991); Ef, E. faecalis (Li et al., 1995). The first residue shown is residue 220 in the PyrK protein of L. lactis. The ends of the sequences represent in all cases the C-terminal ends of the encoded proteins.
FIG. 6. The pH dependence of the activity of the ␦-enzyme (E) and the ␦-enzyme (q). The activities were determined at 37°C in a buffer consisting of Tris (50 mM), NaH 2 P0 4 (50 mM) adjusted to the indicated pH by means of either HCl or NaOH. The substrate concentrations were 1 mM dihydroorotate and 50 M DCIP. The reduction of DCIP was monitored by the absorption at 600 nm.

TABLE III
Specificity for electron acceptors by the two forms of DHOdehase B All reactions contained 1.0 mM of dihydroorotate as substrate. They were carried out under standard assay conditions except that different electron acceptors were used. The reactions were quantified by monitoring the change in absorption at 295 nm stemming from the conversion of dihydroorotate to orotate. the enzymes purified from Plasmodium berghei (Krungkrai et al., 1991) and rat liver (Forman and Kennedy, 1978), and in a recombinant truncated version of the human enzyme (Copeland et al., 1995).
Hitherto, the ability to use NAD ϩ as electron acceptor seemed to be a unique property of the dihydroorotate dehydrogenase from Zymobacterium oroticum, which was discovered by Lieberman and Kornberg (1953). Because this enzyme was made in large amounts when the bacterium was grown with orotate as the sole carbon source and because subsequently discovered dihydroorotate dehydrogenases were unable to use NAD ϩ , the dihydroorotate dehydrogenase of Z. oroticum was generally regarded as being an atypic, catabolic enzyme. To the best of our knowledge, however, it was never shown that this enzyme does not also participate in pyrimidine nucleotide biosynthesis. Z. oroticum is a Gram-positive bacterium (now called Clostridium oroticum), and it is likely that its dihydroorotate dehydrogenase is homologous to the dihydroorotate dehydrogenases of other Gram-positive bacteria (i.e. formed like the ␦-enzyme described herein). The reaction kinetics of dihydroorotate dehydrogenase of Z. oroticum were studied intensively 30 years ago, and the enzyme was shown to contain two FAD, two FMN, and 4 g-atoms of iron, bound in acid-labile iron-sulfur clusters, in each native enzyme molecule, M r ϭ 120,000 (Aleman and Handler, 1967;Miller and Massey, 1965). However, no studies were made on the protein moiety of the enzyme, and the amino acid sequence is unknown.
Currently, we are studying how the FMN redox centers on the PyrDB subunit of the ␦-enzyme of L. lactis interact with the FAD groups and iron-sulfur clusters on the PyrK subunits. The presence of the different types of redox centers on different subunits may facilitate these studies.