Nitrate-inducible Formate Dehydrogenase in Escherichia coli K- 12

‘rHE I(:’

Vol . 266, No. 33, Issue of November 25, pp. 22380-22385,1991 Printed in U.S.A.

JOURNAL OF BIOLOGICAL CHEMISTRY

1991 by The American Society for Biochemistry and Molecular Biology,Inc.

Nitrate-inducible Formate Dehydrogenasein Escherichia coliK- 12 I. NUCLEOTIDESEQUENCE

OF T H E fdnGHI OPERONANDEVIDENCETHATOPAL(UGA)ENCODES

SELENOCYSTEINE* (Received for publication, June 11, 1991)

Barbara L. Berg$$, Jing Lis,Johann Heidern, and Valley Stewart$(( From the $Section of Microbiology, Cornell University, Ithuca, New York 14853-8101and the lkhrstuhlf u r Mikrobiologie der Universitat Munchen, 0-8000 Munchen 19, Germany

the active site. The function of the p subunit is unknown, and the y subunit is probably cytochrome b;& (2, 5; reviewed in Ref. 1). The structuralgenes for formate dehydrogenase-N, fdnGHI, are organized in an operon at 32 min on the E. coli genetic map (6). Transcription of the fdnGHZ operon is induced by nitrate during anaerobic growth. Anaerobic induction is mediated by the transcriptional activator protein, FNR, while nitrate induction is mediated by the transcriptional activator protein, NarL (6). Our previous studies identifiedaclone, containingthe fdnGHZ operon,that encodes theformate dehydrogenase-N polypeptides of approximately 110, 32, and 20 kDa (6). The other major selenoprotein in E. coli, formate dehydrogenase-H, is a component of theformate-hydrogen lyase complex (1, 7). Formate dehydrogenase-H is encoded by the fdhF gene at 92 min on theE . coli genetic map. Incorporation of selenocysteine into formate dehydrogenase-H is directed In the absence of oxygen, Escherichia coli uses a variety of by an in-frame opal(UGA) codon (8,9). This cotranslational terminal electron acceptors for respiration, including nitrate.incorporation requires the productsof four genes, selA, selB, Formate, produced from pyruvate during anaerobic growth, selC, and selD. selC encodes tRNA&, which is initially amiserves as a major electron donor for nitrate respiration (renoacylated with serine. Conversion of the seryl residue to a viewed in Ref. 1).Formate oxidation coupled to nitrate res- selenocysteyl residue requires the products of the selA and piration is catalyzed by a major anaerobic respiratory chain, selD genes. selB encodes a translation elongation factor that formate-nitrate oxidoreductase,which consists of formate binds specifically to selenocysteyl-tRNA&. Mutational le(2, 3). Syn- sions in any of the four sel genes eliminate synthesis of all dehydrogenase-N, quinone, and nitrate reductase thesis of formate dehydrogenase-N and nitrate reductase is formate dehydrogenase selenopeptides (reviewed in Ref. 10). induced by anaerobiosis plus nitrate(reviewed in Ref. 1). Here we report the complete nucleotide sequence of the Formate dehydrogenase-N and nitrate reductase are multi-formatedehydrogenase-Nstructural gene operon, fdnGHI. subunit,membrane-bound enzymecomplexes that contain The deduced amino acid sequence of FdnG shares consideramolybdenumcofactor,heme, andnon-heme iron.Addible similarity with those of formate dehydrogenase-H of E. tionally,formatedehydrogenase-N is one of three E. coli coli and the formate dehydrogenase of Methanobacterium proteins that containsselenocysteine (1, 4). Purified formate forrnicicurn. We also provide evidence that an in-frame opal dehydrogenase-N consists of three subunits, a, p, and y, of (UGA) codon in fdnG directs selenocysteine incorporation. 110, 32, and 20 kDa, respectively. The a subunit contains T h e deduced amino acid sequence of fdnH reveals cysteine molybdenum cofactor and selenocysteine and islikely to form clusters typical of iron-sulfur centers. Our data from spectral analysis indicate thatfdnI encodes cytochrome b;$. Finally, * This work was supported by Grant GM36877 from the National we discuss implications of the predicted aminoacid sequence Institute of General Medical Sciences (to V. S), by a grant from the BundesministeriumfurForschungund Technologie (Genzentrum for the membrane topology and bioenergetic function of forMunchen) (to A. Bock), and by a short-term fellowship from the mate dehydrogenase-N.

The fdnGHI operon of Escherichia coli encodes nitrate-inducible formate dehydrogenase. We report here the entire nucleotide sequence of fdnGHI. The sequence contains three open reading frames of sizes appropriate to encode the three subunits of formate dehydrogenase-N. fdnG contains an in-frame UGA codon that specifies selenocysteine incorporation, and the predicted amino acid sequence of FdnG shows similarity to two other bacterial formate dehydrogenase enzymes. FdnH contains 4 cysteine clusters typical of those found in iron-sulfur proteins. FdnG also contains a cysteine cluster. Evidence from sequence and spectral analyses suggest that FdnI encodes cytochrome bKg& Implications for the membrane topology of formate dehydrogenase-N and its mechanism of proton translocation are discussed.

International Human Frontier Science Program Organization (to B.

L. B.). The costs of publication of this article were defrayed in part by the paymentof page charges. This article must therefore hereby be marked “oduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in this paper has been submitted t o the GenRank’rM/EMBL DataBank with accession number(s) M7,5029. §Present address:Dept. of Genetics,Harvard Medical School, Boston, MA 02115. I( To whom correspondence should be addressed Sectionof Microbiology, Wing Hall, Cornell University, 1t.haca NY 14853-8101.

EXPERIMENTALPROCEDURES

Bacterial Strains andPlasmids-E. coli K-12 strains and plasmids used in this study are listed in TableI. Standard methods were used for genetic and molecular biological manipulations (11). Restriction enzymesandT4DNA ligase were puchased from New England Biolabs (Berverly, MA). T4 DNA polymerase was from U. S. Biochemical Corp. (Cleveland, OH). SequencingStrategy-DNAsequences were determined directly from double-stranded templates by the dideoxynucleotide chain termination method (12) with modified T 7 DNA polymerase (13) and

22380

E. coli fdnGHI Operon

22381

TABLE I Strains and plasmids Genotype

Strain"

S

E. coli K-12 strains

)Ul69

VJS691 fdnIl07:MudJ VJS1222 VJS1661 VJS1938 VJS1945 VJS1948 VJS1949 VJS3171

but

As VJS691 fdnIl07:MudJ but As VJS691 but AS VJS691 but AS VJS691 Asbut VJS691 Asbut VJS691 but As VJS691

Plasmids pVJSlOl pGEM7Zf(+) pVJS525 pRS414 a

narI204::TnlO 9(fdnG-lQCZ)ucA @(fdnG-lacZ)ucA @(fdnG-hZ)ucAselAB::TnlO 9(fdnG-lacZ)ucA selAB::TnlO narI204::TnlO

X X X X

Ap', fdnGHI in PstI site of pHG329 Ap', 910 T7 promoter Ap', EcoRI-BamHI fragment pGEM7Zf(+) of in fdnG Ap', 'lacZ lacY' lacA+

23 6 This work This work This work This work This work This work 6 Promega Corp. This work 21

All strains are F- and X- except as noted.

containing the UGA to UCA change (described below) were cloned into thegene (translational) fusionvector pRS414 (21). The resulting I constructs werecrossed ontobacteriophage X RS45 (21) to form " " t . t " " 9(fdnG-lacZ)"cA[9(fdnG-lacZ)121(Hyb)]or 9(fdnG-lacz)ucA FIG. 1. Sequencing strategy for fdnGHI. Dideoxy chain-ter- [9(fdnG-lacZ)l22(Hyb)]. These phagewere transduced into VJS691, and single copy lysogens were identified by genetic tests as described minationsequencing was doneas described under"Experimental Procedures." Oligonucleotides complementary to the ends of MudJ, (42). The selAB::TnlO allele was transduced into VJS691 lysogens MudK, Tn5, orregions of fdnCHI were used to prime the sequencing via bacteriophage P1 kc. Site-specific Mutagenesis-The fdnG codon specifying selenocysreactions. The arrowsindicatethedirectionandextent of each teine (UGA) at amino acid position 196 was changedto serine (UCA) sequencing run. using oligonucleotide-directed site-specificmutagenesis (22).The mutagenic oligonucleotide used was (5'-GCGTCTsACACGGAC(d"]dATP labeling (14). DNA for sequencing was prepared by the CAACG-3'). The mutagenized template was sequenced to confirm method of Kraft et al. (15). the mutational alteration and to ensure that no spurious changes Most of the fdnGHI DNAsequence was determined by sequencing were introduced. outward from the ends of three different types of transposon inserCulture Media and Conditions-For fl-galactosidase assays, anaertions located throughouttheoperon.MudJ(MudI1734),MudK obic cultures were grown in MOPS2-buffered medium as described (MudII1734) (16), andT n 5 (17) insertions were isolated in plasmids (23).Cultures for cytochromescans were grownanaerobically in containing fdnCHZ (6).' For sequencing from Mu insertions, primers TYGNmediumas described (6). NaNOs(40 mM) wasadded as were synthesized that were complementary to the right end (5'-TTT indicated. Aerobic cultures for selenium labeling experiments (deTCGCATTTATCGTG-3') and the left end (5"TTTTTCGTACTT scribed below) were grown in: tryptone, l % ; glycerol, 0.5%; potassium CAAG-3) of the elements. We were unable to prime DNA synthesis phosphate, pH 6.8, 100 mM; MgS04, 1 mM; CaC12,0.1 mM; Na2Mo04, from the right end of the MudJ transposon because it contains a 10 WM;andtraceelements(24). Anaerobic cultures for selenium rearrangementwithstrong secondary structure (181.' The MudK labeling were grown in the same medium except that glucose, 0.4%, element does not contain this rearrangement. Because the ends of was substituted for glycerol. T n 5 are identical, subclones were constructed in pUC18 and pUC19 I n Viuo Labeling of Selenopolypeptides-Incorporation of [7sSe] (19) that contained DNA from the Hind111 site within the ends of selenite into anaerobic cultures grown with nitrate was carried out Tn5 to a site outside of fdnGHI. The resulting subclones thus con- according to the method of Cox et al. (25). Na:5Se0:3 was added to tained only oneend of Tn5 adjacent to fdn DNA. Theprimer cultures a t a concentration of 0.33 wM. Cell extracts were electrophocomplementarytotheends of Tn5 was(5"GTTCAGGACGCTresed on 7.5% Laemmli gels (26), dried, andexposed to x-ray film. ACTTG-3'). Gaps in the sequence were filled in by synthesizing 17@-GalactosidaseAssay-6-Galactosidase assays were performed on mer oligonucleotide primers complementary to internal sequences. permeabilized cells essentially asdescribed (23). The complete sequences of both strands of the fdnGHI operonwere Cytochrome Spectral Analysis-Cell suspensions were prepared by determined,except for one 40-base gaponthetopstrandfrom growing anaerobic cultures to late exponential phase, harvesting by nucleotide 1825 to 1864. The sequencing strategy is shown inFig. 1. centrifugation, and resuspending in Z buffer (27) to a concentration Oligonucleotides were synthesized at theOligonucleotide Synthesis of about 2 mg of protein/ml. An Aminco DW-2 scanning spectrophoFacility of the Cornell University Biotechnology Program. T 7 DNA tometer was used to obtain absorption spectra of cell suspensions. polymerase (SequenaseTM2.0) and sequencing reagentswere from U. Samples were reduced with excess dithioniteandreadagainst a S. Biochemical Corp. (Cleveland, OH), and [a-"SIdATP was from reference sample oxidized with nitrate (29). Amersham Corp. DNA sequences were stored and analyzed with the DNA Inspector IIe program (Textco, West Lebanon, NH). HydroRESULTS phobicity was estimated by the method of Hopp andWoods (20) with a n averaging length of 19 residues. Alignments of FdnG and FdnH Features of the DNA Sequence-The nucleotide sequence with other protein sequences were performed by visual inspection. of fdnGHI is shown in Fig. 2.We identified three open reading Construction of NfdnC-lacZ) Gene Fusions-Translational fusions of fdnG to the lacZ gene of E. coli were constructed in several steps. frames that are consistent with previous conclusions regardUsing standard molecular manipulations, an EcoRI site was intro- ing the size, order, and orientation of the three genes (2,5, duced approximately 2 kilobases upstream of the fdnGcoding region 6). Each of the three open reading frames is preceded by a in pVJS101. The resulting plasmid was digested with BamHI, which credible Shine-Dalgarno sequence with appropriate spacing cuts in the polylinker of pVJSlOl and also cuts fdnG at nucleotide (28). postion 2826. The BamHI fragment containing the promoter proximal piece of fdnG was ligatedinto BamHI-digested pGEM7Zf(+) to create The fdnG open reading frame begins with an AUG at pVJS525. The EcoRI-BamHI fragment from this plasmid (containing position 451 and terminates with UAA (ochre) at position thepromoterproximalfragment of fdnG) and a similarplasmid The abbreviation used is: MOPS, 4-morpholinopropanesulfonic ' B. Berg and V. Stewart, unpublished data. acid.

"-"*t""-

1UnG

I

fanH

1

fam

]

22382

E. coli fdnGHI Operon 3499. The fdnG sequencealso containsanin-frame UGA (opal) codon at nucleotideposition 1036. The fdnH open readingframe beginswith an AUG at position 3513 and terminates with UAA (ochre) at position 4395. We identified two possible initiation AUG codons for the fdnZ open reading frame at positions4372 and 4390; both are located within the codingsequence of f d n H . We favor the 4390 AUG as the actualtranslationinitiationsite becausealongeroverlap between two codingsequences is quite unusual. The fdnZ open reading frame terminates with UAA (ochre) at position 4930. The calculated molecular masses of the fdnG, fdnH, and f d n I open reading frames are113,219, 32,162, and 20,922 Da, respectively. These values agree well with previous data that estimated themolecular masses of the a, p, and y subunits to be approximately 110, 32, and 20 kDa, respectively (2, 6). Spectral Analysis of Cytochrome b[&-Formate dehydrogenase-N and nitrate reductase are each associated with a unique cytochrome b that absorbs at556 nm (1). The nitrate reductase-associated cytochromeis encoded by narZ (29). We wished toobtain more direct evidence that fdnZ actually encodes cytochrome b&. To do this, we compared the spectral patterns of a wild-type strain, anfdnZ mutant, a narZ mutant, and an fdnZ narZ double mutant. If fdnZ does encode cytochrome bL$, we would expect each of the single mutants to contain approximately half of the wild-type amount of dithionite-reduceable cytochrome at 556 nm. The results of this analysis are shown in Table 11. As predicted, insertions in fdnZ and narl each abolished approximately half of the dithionite-reducible cytochrome absorbing at 556 nm. Analysis of the In-frame UGA Codon in fdnG-fdnG contains an in-frame UGA codon at the position corresponding to aminoacid residue 196. We usedtwo independent methods to examine whether the UGA at position 196 of fdnG encodes a selenocysteine residue. First, we constructed two translational fusions of fdnG to the coding sequence of lacZ, each with a fusion point at the BamHI site (downstream of the UGA) at nucleotide position 2405. Both fusions are carried on X bacteriophages and were integrated into the chromosome in single copy at the X attachment site. Onefusion contained wild-type fdnG DNA and thus retained the UGA codon; the other fusion contained a change of the UGA codon to UCA, which specifies serine. We then examined the expression of these fusions in a wild-type strain andin strains unable to incorporate selenocysteine into proteins. Strains can be prevented from incorporating selenocysteine into proteins by either depleting the medium of selenium or by introducing a mutation in any one of the four sel genes (9, 30). Expression of the UGA-containing fusion was blocked when present ina selAB::TnlO strain background and was severely reduced when cellswere grown in seleniumdeficient medium (Table 111). The low level of activity produced by the fusion when assayed in selenium-deficient medium was probably due to trace amounts of selenium present intheculture medium. Expression o f the UCA (serine)TABLE I1 Cytochrome bss, content in fdnI and narI mutants Strain

I

Cytochrome Genotype

bsi,"

0.017h VJS691 fdnl' narl' 0.010 VJS1222 fdnZ:MudJ narl' 0.010 VJS3171 fdnI+ narI:TnlO ' . . r : F ' C I D R " h i G . . P I C ". E * C P. cated in the cytoplasm. The single putative transmembrane FIG. 5. Sequence similarities between FdnH, NarH (E. coli helix we observed in the FdnHsequence probably serves only nitrate reductase), and DmsB (E. coli dimethyl sulfoxide re- as an anchor to the membrane and does not seem likely to ductase). Residues identical or analagous to FdnG amino acids are in boldface. Dots indicate cysteine residues contributing to potential contribute to formation of a transmembrane channel. Thus, iron-sulfur-bindingsites.Sequences were fromNarG(37), DmsA it is unclear how proton extrusioncould be coupled to formate (38). oxidation. Further work is required to determine thebioeneri:.,*

":,!