Proc. Natl. Acad. Sci. USAVol. 90, pp. 7084-7088, August 1993Genetics

Identification and characterization of the Escherichia coli genedsbB, whose product is involved in the formation of disulfidebonds in vivoDOMINIQUE MISSIAKAS*t, COSTA GEORGOPOULOS*t, AND SATISH RAINA*tt*Department of Cellular, Viral and Molecular Biology, University of Utah School of Medicine, Salt Lake City, UT 84132; and tDNpartement de BiochimieMedicale, Centre M6dical Universitaire, 1 rue Michel-Servet, CH-1211 Geneve 4, Switzerland

Communicated by Jonathan Beckwith, March 4, 1993 (received for review December 15, 1992)

ABSTRACT We have identified and characterized theEscherichia coli gene dsbB, whose product is required fordiaulfide bond formation of periplasmic proteins, by using twodifferent approaches: (i) screening of a multicopy plasmidlibrary for clones which protect E. coli from the lethal effectsof dithiothreitol (DTT), and (ii) screening of insertion librariesof E. coli for DTT-sensitive mutants. Mapping and character-ization of mutations conferring a DTT-sensitive phenotype alsoidentified the dsbA, trxA, and trxB genes, whose products areinvolved in different oxidation-reduction pathways. Null mu-tations in dsbB conferred pleiotropic phenotypes such as sen-sitivity to benzylpenicillin and inability to support plaqueformation of filamentous phages, and they were shown toseverely affect disulfide bond oxidation of secreted proteinssuch as OmpA and &3lactamase. These phenotypes resemblethe phenotype of bacteria carrying either a null mutation in thedsbA gene or the double mutation dsbA dsbB. Sequencing andexpression of the dsbB gene revealed that it encodes a 20-kDaprotein predicted to possess an "exchangeable" disulfide bondin -Cys-Val-Leu-Cys-. The dsbB gene maps at 26.5 min on thegenetic map of the E. coil chromosome, and its transcription isdirected from two promoters, neither of which resembles thecanonical Eo7'-recognized promoter.

Relatively little is understood thus far regarding the processof folding and assembly of secreted proteins in prokaryotes.Exported proteins face a harsher environment in the peri-plasm, since this space is more vulnerable to environmentalstresses than is the cytosol (1). It is thus reasonable to expectthat in Escherichia coli there may be periplasmic chaperoneswhose function is to ensure the proper folding and mainte-nance of periplasmic proteins. Another expected class ofproteins is enzymes that can either directly catalyze disulfidebond formation or maintain an appropriate oxidized environ-ment in the periplasm. It is indeed well established that thevast majority of disulfide bond-containing proteins are eithersecreted or integral membrane proteins, since the bacterialcytoplasm is a comparatively reducing environment (2). Thusit is quite important that in E. coli a periplasmic protein,designated DsbA, was recently discovered and shown to beinvolved in disulfide bond formation (3). DsbA has beencharacterized in vivo as well as in vitro (3-5). It has beenshown that in dsbA mutants a variety of secreted proteinslack disulfide bonds (3, 4).

In an attempt to identify genes whose products are in-volved in the stability or assembly of proteins in the peri-plasmic space, we initiated a series of experiments utilizingvarious genetic approaches. In the present study, we tookadvantage of the phenotypes of either resistance or sensitiv-ity to strong reducing agents such as dithiothreitol (DTT) to

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

characterize gene products involved in the oxidation and/orreduction pathways. In one approach, a multicopy E. coligenomic library was used to identify those proteins whoseoverproduction would protect bacteria against the lethalreducing effects of DTT. In a complementary approach,insertional mutants that exhibited hypersensitivity to suble-thal concentrations of DTT were isolated. Both studiesidentified the existence of a gene designated dsbB.§ A nullmutation in dsbB leads to the pleiotropic phenotype, similarto the one exhibited by dsbA null mutants, of deficiency indisulfide bond formation of secreted proteins such as OmpAand ,B-lactamase. These results are consistent with indepen-dent studies (6, 7) also reporting the identification ofthe dsbBgene. The insertional mutational approach for DTT sensitiv-ity also resulted in the isolation of mutations in the knowngenes trxA (encoding thioredoxin), trxB (encoding thiore-doxin reductase), and the recently described dsbA (3, 4, 8).

MATERIALS AND METHODSSelection Strategies and Cloning of dsbB. Chromosomal

DNA isolated from E. coli wild-type strain MC4100 wassubjected to partial digestion with Sau3A to produce DNAfragments 2-6 kb in length. These DNA fragments weregel-purified and ligated into the BamHI site of the plSA-based vector pOK12. A library of at least 20,000 independentrecombinant clones was made, and DNA from such a poolwas used to transform various wild-type E. coli strains.Twenty-four plasmid clones whose presence conferred re-sistance to otherwise lethal DTT levels (20 mM) were thusselected. The restriction enzyme digestion pattern ofthese 24plasmids showed that they all carried the same chromosomalinsert. Plasmid clones pDM243 and pDM244, which carry thesame 1.5-kb Sau3A fragment but in opposite orientationswith respect to the T7 promoter of the pOK12 vector, wereretained and shown to contain dsbB. Subcloning ofthe 950-bpKpn I-Sau3A fragment, which contains the minimal dsbBgene, resulted in plasmid pDM353 (Table 1).The overexpression of DsbB was achieved by first ampli-

fying the dsbB gene minimal coding region by the PCRmethod (14), using primers 5'-GGAGCGCCGAATGGATC-CGCGACCGAA-3' and 5'-TATTGCAGGCATATGAATA-TGTTG-3' and cloning in the T7 promoter expression vectorpET-3a (pDM507). The dsbA gene (1.8-kb Sau3A DNAfragment) was cloned, using the genomic library describedabove (pSR1865), by selecting for its ability to complementSR1790 (dsbAIOI::TnlO) and DM547 (dsbA43::TnlO) mutantbacteria and was verified by its ability to recombine withvarious TnlO insertions within the dsbA gene.

Abbreviations: DTT, dithiothreitol; DTNB, 5,5'-dithiobis(2-nitrobenzoic acid); Kanr, kanamycin resistance.*To whom reprint requests should be sent at the t address.§The sequence reported in this paper has been deposited in theGenBank data base (accession no. L03721).

7084

Dow

nloa

ded

by g

uest

on

July

20,

202

1

Proc. Natl. Acad. Sci. USA 90 (1993) 7085

Table 1. E. coli strains and plasmids used in this studyStrain

E. coliCA8000DM215DM391DM547SR688SR1748SR1753SR1789SR1790SR1855CAG18599GW2100RP6866

Relevant genetic markers

Wild type (9)MC4100 rfaH::TnJOCA8000 dsbB36::TnlOCA8000 dsbA43::TnlOCA8000 zbh::TnSCA8000 trxB::TnlOCA8000 dsbB65::TnlOCA8000 trxA::TnlO KanrCA8000 dsbAlOl::TnlO KanrCA8000 dsbAlOl::TnlO Kanr dsbB36::TnlOMG1655 ilv::TnlO KanrAB1157 umuC122::TnS (10)(motA-motB) DEM12-13 his4/pJJ20 carrying

in-frame TnphoA at codon 84 of motB (11)PlasmidspOK12 pl5A-based cloning vector used to construct

genomic librarypET-3a T7 expression vector (12)pHP45 ColEl-based vector conferring resistance to

chloramphenicol and ampicillin (13)pDM243/244 pOK12 carrying 1.5-kb Sau3A dsbB+

fiagmentpDM353 pOK12 carrying 950-bp Kpn I-Sau3A dsbB+

fragment in orientation with T7 promoterof the vector

pDM507 pET-3a carrying 536-bp minimal codingregion of the dsbB+ fragment cloned inNde I-BamHI sites

pSR1865 pOK12 carrying 1.8-kb Sau3A dsbA+fragment

Kanr, kanamycin resistance.

To isolate chromosomal mutations which render E. colisensitive to DTT, pools of approximately 5 x 105 indepen-dent insertional events of either ATnlO (tetracycline resis-tance) or ATnlO (Kanr) (15) were constructed in wild-type E.coli strain CA8000. These bacterial pools were subsequentlyscreened for sensitivity to otherwise nonlethal concentra-tions of DTT (7 mM). Seventy-eight such independentlyisolated mutants were mapped by hybridizing the comple-menting clones to the ordered E. coli genomic library (16) andwere confirmed by linkage to nearby known genetic markers.The assignment of mutations to the dsbA locus was done byusing rfaH: :TnlO from DM215 (Table 1) as the linked marker;the dsbB insertions were mapped by using the umuCl22: :TnS(10) linked marker; the trxB: :TnJO insertions were mapped byconfirming their linkage to the Kanr marker of SR688, whichcarries a TnS insertion located 115 nt downstream of the trxBtermination codon. ilv::TnlO Kanr marker from strain CAG18599 (17) was used to assign the trxA::TnlO insertions.DNA Sequence Analysis. The 1.5-kb Sau3A fragment of

pDM243 (dsbB+) or its derived subclones such as pDM353were sequenced by using the United States BiochemicalSequenase kit. To locate the exact positions of the TnlOinsertions in the dsbB gene, these insertions were firstrecombined onto plasmid pDM243. The junction betweenTnlO and the chromosomal DNA was determined by usingthe synthetic 24-mer oligonucleotide primer 5'-ATTTGAT-CATATGACAAGATGTGT-3', which directs DNA replica-tion away from the ends of the TnlO insertion element. Thesequence ofthe dsbB gene along with the flanking regions hasbeen deposited in GenBank.§RNA Isolation, Northern Blotting, and Mapping of 5' Ter-

mini ofdsbB. Total cellularRNA was isolated by using the hotSDS/phenol extraction procedure (18). Approximately 5 pgofeach RNA sample was analyzed by the Northern technique

(18). To probe for dsbB message, 100 ng of the 280-bp NsiI-Hpa I DNA fragment (internal to the dsbB gene) wasisolated from pDM353 and 32P-labeled with [a-32P]dATP(3000 Ci/mmol; 1 Ci = 37 GBq) using the nick-translationprocedure (18). To define the transcriptional start sites, -10ng of the oligonucleotide probe 5'-ACTGCTCTGGCACTG-GAACTG-3', which is complementary to nucleotide posi-tions 66-86 of the dsbB sequence, was annealed with 10 ,ugof total cellular RNA. The annealed primer was extended byreverse transcriptase from avian myeloblastosis virus, essen-tially as described (9). The primer extension products wereelectrophoresed on the same gel as the dideoxy sequencingreaction products, using the same primer.

Labeling, Fractionation, and Immunoprecipitation. For la-beling of proteins, cells were grown in M9 minimal medium(18), supplemented with each of the amino acids (exceptmethionine and cysteine) at 20 ug/ml and 0.4% glucose.Bacterial cultures were pulse-labeled with [35S]methionine at100 ,uCi/ml for 45 sec followed by a 1000-fold excess ofunlabeled methionine. Disulfide bond formation in OmpAand 3-lactamase was assessed by the method described byPollitt and Zalkin (19). The anti-OmpA antiserum was a kindgift of S. Mizushima, and anti-t3-lactamase antibody waspurchased from 5 Prime -- 3 Prime, Inc.

RESULTSCloning of dsbB. In an attempt to identify genes whose

products are involved in protein stabilization, particularly inthe periplasm, we selected for tolerance to otherwise lethallevels of DTT, a reducing agent that has been shown toreduce a variety of proteins, including protein disulfideisomerase and thioredoxin. Thus, it is expected that thepresence of high amounts ofDTT will result in the misfoldingof many proteins which require disulfide bond(s) to achievetheir functional native form. Overexpression of either achaperone or a cellular component able to catalyze or toparticipate in disulfide exchange reactions may relieve thecells of DTT's toxic effect. A library of E. coli chromosomalDNA prepared in the p15A-based vector pOK12 was used toselect for clones which enable wild-type E. coli to formcolonies on L agar (18) supplemented with 20 mM DTT, aconcentration at which wild-type bacteria do not normallygrow. Twenty-four large colony formers were thus isolatedand subsequently shown to share an identical 1.5-kb Sau3ADNA fragment (pDM243 and pDM244). Further subcloningexperiments showed that DNA sequences required to conferresistance to DTT were present on a 950-bp Kpn I-Sau3Afragment (pDM353), defining the dsbB gene.Mapping of dsbB. We hybridized the 32P-labeled 1.5-kb

Sau3A fragment derived from plasmid pDM243 (dsbB+) tothe E. coli DNA library in bacteriophage A (16). Hybridiza-tion was observed with bacteriophages A243 (2A3) and A244(11G8), which carry the overlapping region corresponding tokbp 1235-1245 of the E. coli physical map. Further compar-isons of the restriction maps of the dsbB+ subclones con-taining the 950-bp Kpn I-Sau3A fragment with the restrictionmap of E. coli covering this region confirmed that dsbB islocated in the 1238-kbp region, corresponding to the 26.5-minregion ofthe recalibrated E. coli genomic map (20). Since theumuC gene, whose product is required for inducible muta-genesis, also maps in this region, we used plasmids pSE114and pSE116 (10) to verify our results. We found that, whereasboth plasmids carry the umuCD genes, pSE114 also containsthe dsbB gene; thus we were able to place dsbB immediatelyupstream of umuCD. The position of dsbB was also verifiedgenetically by using the umuC122::TnS as a marker forbacteriophage P1 transduction experiments with ourdsbB::TnlO isolates.

Genetics: Missiakas et al.

Dow

nloa

ded

by g

uest

on

July

20,

202

1

Proc. Natl. Acad. Sci. USA 90 (1993)

Sequencing of dsbB and Identification of Its Product. Thenucleotide sequence determination ofthe 1.5-kb Sau3ADNAfragment revealed a 534-nt open reading frame (ORF). Thissequence encodes a predicted 178 amino acid residue poly-peptide of 20,329 Da, with a CXXC motif located at position43-46 (Fig. 1). The predicted sequence of the DsbB proteinalso shows at least three potential transmembrane domains(Fig. 1), indicating a potential cytoplasmic membrane loca-tion. Upstream of the dsbB ORF we found a truncated ORFthat corresponds to the end of the nhaB gene, encoding theNa+/H+ antiporter (21). The recently reported nhaB se-quence includes part of the dsbB ORF (21). However, thissequence differs from ours by an extra base at position 794,leading to a frameshift in the dsbB sequence, and at position479 (G vs C), leading to amino acid change from glycine toalanine (amino acid residue 52) of the DsbB protein (Fig. 1).The positions of two of the six dsbB: :TnJO insertions, able

to recombine with the dsbB+-carrying plasmid pDM243,were determined by sequencing the DNA junction betweenTnJO and the dsbB gene, using primer reading out of the TnJOends. The sites of the mini-Tet::TnJO insertions in DM391and SR1753 were found to be within the deduced dsbB geneat nucleotide positions 36 and 65, respectively, from theputative ATG initiation codon (Fig. 1).To identify the dsbB gene product we used [35S]methionine

to label BL21(DE3) cells carrying either plasmid pDM353 orpDM507, which contain the dsbB gene under the exclusive T7promoter expression system (12). When the T7 promoter wasinduced, a simultaneous induction ofan -20-kDa protein wasobserved, consistent with the predicted size ofDsbB protein(Fig. 2).

Phenotypic Characterization and Mapping of DTT-SensitiveMutants. Seventy-eight DTT-sensitive TnlO (Kanr) or TnlO(Tetr) insertions were isolated as described in Materials andMethods. Of these, six mutations were shown to map in thedsbB gene and two in the dsbA gene. Among the other genesidentified in this screen were the trxA (SR1789) and trxB(SR1748) genes (Table 1). The trxA and trxB genes have beenshown to encode thioredoxin and thioredoxin reductase,respectively (8). The majority of the DTT-sensitive mutantswere found to share the following phenotypes: (i) hypersen-sitivity to benzylpenicillin, especially the dsbB and dsbA nullmutant bacteria, with dsbB being more sensitive (15 ,g/ml);(ii) reduced overall expression of alkaline phosphatase [thisphenotype was assayed by first transducing these mutations,by using bacteriophage P1, into strain RP6866, which con-

I..0

44*- 20 kDa

FIG. 2. Identification of theDsbB protein. Cultures of strainscarrying vector pOK12 alone orplasmid pDM353, in which thedsbB gene is inserted under the T7RNA polymerase promoter, weregrown at 37°C in M9 minimummedium. Expression of the T7RNA polymerase was induced byaddition of isopropyl J3D-thioga-lactopyranoside (IPTG; 0.5 mM).After a 30-min incubation, cellswere treated with rifampicin (200,tg/ml) for another 15 min andthen labeled with [35S]methionine(50 ,uCi/ml) for 5 min. Lysateswere electrophoresed on an SDS/12.5% polyacrylamide gel.

tains plasmid pJJ20 carrying an in-frame TnphoA fusion atcodon 84 of the motB gene (11)]; (iii) inability to supportplaque formation of filamentous bacteriophages such as M13or fl. These pleiotropic phenotypes suggested a generalizeddefect in the maturation of translocated proteins. Since theDsbB protein is predicted to lie in the membrane and topossess a disulfide bond active site, we tested whether thesedefects could be due to the accumulation of reduced forms ofperiplasmic proteins. Hence, we performed a quantitativeanalysis of periplasmic fractions prepared by osmotic shock(22) from DTT-sensitive mutant bacteria; the analysis usedEllman's reagent, 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB)(23). Interestingly, the incubation of the periplasmic fractionwith an excess of DTNB (0.8 mM) resulted in a significantincrease in absorbance at 412 nm due to formation of the5-thio-2-nitrobenzoate ion, as compared with the absorbanceobserved with the extracts prepared from wild-type bacteria(Table 2).

Oxidized DTT Can Suppress Phenotypic Defects Exhibited bydsbB Mutant Bacteria. We further confirmed that the pheno-types observed with dsbB null mutant bacteria were due todefects in disulfide bond formation by testing the ability ofoxidized DTT to revert phenotypes such as hypersensitivity tobenzylpenicillin. It has been shown that the penicillin-bindingprotein 4 (PBP4) has two disulfide bridges, which are predictedto be essential for its enzymatic activity (deacylation ofbenzylpenicillin) (24). Thus, we reasoned that hypersensitivityofdsbB mutant bacteria to benzylpenicillin could be due to the

-351 CTGCGTTCCTGTTC-TGCTGACCTCTGCACTCGCGCCATrGATTCGCCTCTCTTATGGCCGCATGGTGTGGA

P1_1Q 11

7 3 TGGCCCTGCCTTACACCCTCGTCCTGACACTCGTCCGCTTGCTCTGCGTCGAGITTACGCTTGCCCCTGTAA14 5 CCGAATGGTTTATGCAAATGGGCTGGATAGCAACGCTrTGATAACAACTTACCGGGCATATAAATGCCCGGT

P2-35 -10Q

217 TTGCCTTTTCGCCCGATAATTGTCCAATTGCGTCCTANmTTGCTGCCTCCTGGTGGCGCAGCGAATGAATT FIG. 1. Nucleotide sequencev of the dsbB gene and its flanking

289 GGTTTAAACTGCGCACTCTATGCATATGCAGGGAAATGATTATG?rGCGA¶NTTTGAACCAATGTTCACAA region. The various -10 and -35V M I M L R F L N Q C S Q 12 transcription regulation regions

361 GGCCGGGGCGCGGGCGTTGATGGCTGGAAC TGACGGCGCT GGTTCCAGCAT are overlined. The two transcrip-GORG A WLTTLMA F TA L A L E L TA L W F O H 36

toa tr ie ladP r4 3 3 GTGATGTTACTGAAACCTrGCGTGCTCTGTA¶NAT7GAACGCTGCGCGTTATTCGGCGTTrCTGGGTGrGCG tional start sites P1 and P2 areV M L L K P C V L C I Y E R C A L F G V L G A A 60 shownby4X andthetwoinser-

505 CTGATrGGCGCGATCGCCCCGAAAACTCCGCTGCGTTATGTAGCGATGGTTATCTGGTTGATATGGTTC tional sites of TnlO are indicatedL I G A I A P K T P L R Y V A M V I W L Y S A F 84 by v. The potential transmem-

577 CGCGGTGTGCAGTTAACTTACGAGCACACCATGCTTCAGCTCTATCCTTCGCCG'-TGCCACCTGTGATTTT brane domains are underlined.R G V Q L T Y E H T M L Q L Y P S P F A T C D F 108 The potential active site, CVLC

649 ATGGTTCGTTTCCCGGAATGGCTGCCGCTGGATAAGTGGGTGCCGCAAGTGITGTCGCCTCTGGCGATTGC (residues 43-46), is shown in boldMV R F PEW LPLDKWV P V F V ASG D C 132 characters. The complete se-

721 GCCGAGCGTCAGTGG¶FFGTTGATCCCGGCGTGTTTTTCGCTTACC'1Gqec fte15k a3 NA E R Q W D F L G L E M P Q W L L G I F I A Y L. 156 quagenotheinludin-thbsaqueAnce793 ATTGTCGCAGTrGCGGTGrG'ATTTCCCAGCCGTTTAAAGCGAAAAAACGTGATCTGTTCGGTCGCTAATCC fragment, including the sequences

I V A V L V V I S Q P F K A K K R D L F G R 178 upstream covering the nahB gene,865 ATTCGGCGCTCCTGCGGGAGCGC-Ti'-r-CTGCCGCTATATTTATTGACCCTCAGTAAATCAGAACTTCG has been deposited in GenBank937 CG TGTATAACGTGGTAAAGCATGCTCGCTrCATCIGCATTGTGCTGTCC (LO3721).

7086 Genetics: Missiakas et al.

Dow

nloa

ded

by g

uest

on

July

20,

202

1

Proc. Natl. Acad. Sci. USA 90 (1993) 7087

Table 2. Accumulation of proteins with reduced cysteines asjudged by reaction with Ellman's reagent

Genotype A412

Wild type 0.081dsbAlOl::TnlO 0.331dsbB36::TnlO 0.305

Osmotic shock fluids (22) were prepared from the three isogenicstrains harvested at OD595 = 0.4 in 50mM sodium phosphate buffer,pH 7.5, containing 1 mM EDTA and 1 mM phenylmethanesulfonylfluoride. DTNB prepared in the same buffer was added to eachsample to a final concentration of 0.8 mM. The reaction at equilib-rium was assessed by measuring the absorbance at 412 nm.

lack of disulfide bridges in PBP4. In agreement with thisassumption, we observed that the addition of submillimolaramounts of oxidized DTT (0.2 mM) to the growth mediumrestored the colony-forming ability of all six dsbB::TnlOmutant bacteria to the wild-type level on L-agar plates sup-plemented with benzylpenicillin (50 ,g/ml). A similar pheno-typic suppression by supplementation with oxidized DTT wasobtained with bacteria carrying either the dsbAlOl::TnlO orthe dsbA43::TnlO mutation. Taken together, these resultssuggest an important role ofthe DsbB protein in disulfide bondformation of a variety of proteins such as PBP4.

Disulfide Bond Formation of OmpA and -Lactamase IsAltered in a dsbB Null Mutant. More direct evidence that DsbBcan facilitate disulfide bond formation in vivo was sought byusing two different secreted proteins as substrates, OmpA and,B3lactamase. Proteins were labeled by pulse-chase and thendirectly treated with iodoacetamide (modifying cysteine resi-dues) or pretreated with DTT in the control experiments, asdescribed by Pollitt and Zalkin (19). The different forms ofimmunoprecipitated OmpA and 1-lactamase were separatedby SDS/PAGE. Fig. 3 shows the pattern of migration ofreduced and oxidized proteins, corresponding respectively toretarded and faster-migrating species. Immediately after thechase (time 0 min), -lactamase (Fig. 3A) or OmpA (Fig. 3B)was found to be mostly oxidized in wild-type bacterial back-grounds. In contrast, these proteins were found primarily in areduced state in the null mutant strain dsbB- (DM391) (Fig. 3)and remained so, even after 10 min of chase. Similar resultswere obtained with bacteria carrying null mutations either indsbA (SR1790) or in both dsbA and dsbB (SR1855).

10min0 min 10 min +DTT

+ + +

+t~~~~~~cx+Nwm +

Aredox

Bredz*.t.. 4..1M. :j0M.t40, *.

ox -~

FIG. 3. dsbB null mutants are defective in disulfide bond forma-tion. Bacterial cultures CA8000 (dsbA+ dsbB+), SR1790 (dsbA01::TnlO), DM391 (dsbB36::TnlO), and SR1855 (dsbA- dsbB-) weregrown in M9 medium and either labeled with [35S]methionine for 45sec or labeled for 45 sec followed by a 10-min chase with excessunlabeled methionine. Cells were lysed and immunoprecipitated withantisera to either p-lactamase (A) or OmpA (B). Autoradiograms ofgel electrophoresis performed under nonreducing conditions areshown. The positions of oxidized (ox) and reduced (red) forms areindicated by arrows. To assay for /3-lactamase, the four strains werefirst transformed with pHP45 plasmid, which carries resistances tochloramphenicol and ampicillin, and then selected only for resistanceto chloramphenicol.

u0C)4

U0C)

m FIG. 4. Northern analysis ofdsbB tran-scripts. RNA was extracted from dsbB+bacteria grown at 30°C or shifted to 42°C.Approximately 5 Lg of total RNA was

< analyzed by the Northern blot technique,using as a probe the 32P-labeled 250-bp NsiI-Hpa I DNA fragment (100 ng), whichcontains the amino end of the dsbB codingsequence.

Transcriptional Regulation of dsbB. Northern analysis ofdsbB transcripts showed that the gene is transcribed as amonocistronic message of -650 nt. As seen in Fig. 4, thedsbB mRNA levels did not substantially change after a shiftup in temperature from 300 to 42°C. Thus the dsbB gene doesnot appear to be regulated by heat shock.Primer extension analysis was used to identify the 5'

termini of dsbB-specific mRNA species. It was found thatdsbB transcription is initiated from two sites, designated asP1 and P2, located 237 and 40 nt upstream of the putativeATG initiation codon (Fig. 5). Interestingly, the P1 transcrip-tional start site is located within the coding region ofthe nhaBgene. The -10 and -35 regions corresponding to the twotranscriptional start sites do not resemble canonical Ecr70-transcribed promoters or those recognized by RNA polymer-ase coupled with any other known o factors. Comparison ofefficiency ofpromoter usage in different genetic backgroundsclearly shows that the more distal promoter P1 is morefrequently utilized in wild-type bacteria, as judged by therelative amounts of transcripts. In contrast, transcripts ini-tiated from the P2 start site were found to be more abundantwhen RNA was analyzed from isogenic bacterial strainscarrying a null mutation in the rpoH (25), katF (26), or lrp (27)gene, encoding o.32, CyS, and leucine response regulatoryproteins, respectively. The efficiency of the P2 promoterusage in rpoHA (oa32) strains is higher than that found for thewild type or katF or lrp mutant bacteria (Fig. 5).

DISCUSSIONWe used two complementary approaches which resulted inthe identification of an E. coli gene, dsbB, whose product isinvolved in disulfide bond formation of periplasmic proteins.

C

C T A G-¢ .ACCTTAC

- Pi *ACCCTC

AATG

P2 *AATTGGTTT

FIG. 5. Mapping of 5' ter-mini of dsbB transcripts: Primerextension reactions of total cel-lular RNA hybridized to a 32pend-labeled DNA oligonucleo-tide probe complementary to thesequence from nucleotide 64 tonucleotide 86 of the dsbB codingregion. RNA was extracted fromdsbB+ bacteria grown at 30°C,from wild type or from isogenicstrains carrying null mutations inone ofthe genes rpoH, katF, andlrp. Lanes labeled G, A, T, andC correspond to the dideoxy se-quencing reactions carried out,using the same oligonucleotideas primer. The asterisk and thearrow indicate the P1 and P2transcriptional start sites, re-spectively.

Genetics: Missiakas et al.

Dow

nloa

ded

by g

uest

on

July

20,

202

1

Proc. Natl. Acad. Sci. USA 90 (1993)

The first approach was to select for genes whose presence ona multicopy plasmid enabled E. coli to survive the otherwiselethal effect of DTT (20 mM). All 24 clones isolated in thisscreen carried the dsbB+ gene (although in reconstructionexperiments we showed that the dsbA gene on the samemulticopy plasmid enabled E. coli to form colonies, althoughsmaller, in the presence of DTT). The second approach wasto screen for insertional mutations which result in increasedsensitivity to 7 mM DTT, a concentration that wild-type cellscan tolerate. One ofthe major targets of insertional mutationswas the dsbB gene. We found that inactivation of the trxA andtrxB genes (8) and the dsbA gene (3, 4) also resulted inhypersensitivity to DTT. Thus, it appears that mutations ingenes whose products are involved in maintaining an oxida-tion-reduction balance, either in the periplasm or in thecytosol, lead to DTT sensitivity.

In this work, we have presented different lines of evidencewhich demonstrate that in vivo there is a strong requirementfor DsbA, DsbB, or both, to oxidize disulfide bonds of varietyof proteins. Consequently, null mutations in the dsbB genewere shown to confer a highly pleiotropic phenotype similarto that observed in dsbA mutants, such as (i) inability tosupport plaque-forming ability of filamentous bacteriophagessuch as M13 [such a phenotype may be due to lack of F pili,as has been reported for dsbA mutants (3)]; (ii) reducedexpression of periplasmic secreted proteins such as alkalinephosphatase; (iii) accumulation of reduced forms of thesecreted proteins OmpA and l3-lactamase in the periplasm, asjudged by direct comparison of amounts of oxidized versusreduced forms of these proteins; (iv) sensitivity to drugs suchas benzylpenicillin whose target is penicillin-binding protein4, which has two disulfide bridges as part of its active site; and(v) overall accumulation of proteins with reduced cysteines inthe periplasm as assayed with DTNB. The fact that thesephenotypic defects could be overcome by supplementationwith submillimolar amounts of oxidized DTT clearly showedthat DsbB is involved in the oxidation of disulfide bonds ofsubstrate proteins.

Since null mutations in either dsbA or dsbB alone or in bothdsbA and dsbB exhibit identical phenotypes, most likelyDsbA and DsbB are part of the same periplasmic pathway ofdisulfide oxidation-reduction. DsbA has been shown to be adisulfide oxidase capable of oxidizing substrate proteins,such as alkaline phosphatase and ribonuclease A (5), orreducing oxidized insulin (3); hence, it is likely that the roleof DsbB is to oxidize reduced forms of DsbA, thus allowingrecycling of active DsbA. It is also possible that DsbB is ableto directly oxidize other periplasmic proteins in a substrate-specific or nonspecific manner. In this case, DsbA would besimply another substrate of DsbB. The independent studyfrom Bardwell et al. (6) favors the model where DsbBpreferentially oxidizes DsbA. Examination of the DsbB se-quence leads to the prediction that it is an integral membraneprotein with a CXXC motif in the periplasmic space, whichis consistent with such a disulfide oxidase role for the protein.The putative membrane localization of DsbB raises thepossibility of its coupling to a membrane electron transportsystem, thus enabling its reoxidation and recycling. If DsbAand DsbB proteins are indeed part of the same pathway, it isvery likely that genes homologous to dsbB will be found inother organisms, as is the case for the dsbA gene (28-30).An interesting point concerns the type of transcriptional

regulation to which the dsbB gene is subjected. Although twopromoters were found, neither was related to any knownconsensus promoter sequence. The P2 promoter was acti-vated in the absence ofgene products, such as o.32, as, or the

leucine-responsive regulatory protein, involved in the tran-scription ofgroups ofgenes under special growth conditions.Perhaps, the P2 promoter is responsible for ensuring contin-uous transcription of the dsbB gene under a variety ofstresses since, under the same conditions, transcription fromthe P1 promoter is drastically diminished.

We thank James Bardwell and Jon Beckwith for sharing theirresults prior to publication and S. Mizushima for kindly providingOmpA antibodies. This work was supported by a European Molec-ular Biology Organization Fellowship to D.M. and by Grant FN31-31129-91 from the Fond National Scientifique Suisse and GrantAI21039 from the National Institutes of Health to C.G.

1. Johnson, K., Murphy, C. K. & Beckwith, J. (1992) Curr. Opin.Biotechnol. 3, 481-485.

2. Creighton, T. E. (1984) Proteins: Structures and MolecularProperties (Freeman, New York), pp. 311-313.

3. Bardwell, J. C. A., McGovern, K. & Beckwith, J. (1991) Cell67, 581-589.

4. Kamitani, S., Akiyama, Y. & Ito, K. (1992) EMBOJ. 11,57-62.5. Akiyama, Y., Kamitani, S., Kusukawa, N. & Ito, K. (1992) J.

Biol. Chem. 267, 22440-22445.6. Bardwell, J. C. A., Lee, J.-O., Jander, G., Martin, N., Belin,

D. & Beckwith, J. (1993) Proc. Natl. Acad. Sci. USA 90,1038-1042.

7. Dailey, F. E. & Berg, H. C. (1993) Proc. Natl. Acad. Sci. USA90, 1043-1047.

8. Russel, M. & Model, P. (1986) J. Biol. Chem. 261,14997-15005.9. Raina, S. & Georgopoulos, C. (1990) J. Bacteriol. 172, 3417-

3426.10. Elledge, S. & Walker, G. C. (1983) J. Mol. Biol. 164, 175-192.11. Chun, S. Y. & Parkinson, J. S. (1988) Science 239, 276-278.12. Studier, F. W. & Moffatt, B. A. (1986) J. Mol. Biol. 189,

113-130.13. Fellay, R., Frey, J. & Krisch, H. (1987) Gene 52, 147-154.14. Innis, M. A., Gelfand, D. H., Sninsky, J. J. & White, T. J.

(1990) PCR Protocols: A Guide to Methods and Applications(Academic, New York), pp. 13-20.

15. Way, J. C., Davis, M. A., Morisato, D., Roberts, D. E. &Kleckner, N. (1984) Gene 32, 369-379.

16. Kohara, Y., Akiyama, K. & Isono, K. (1987) Cell 50,495-508.17. Singer, M., Baker, T. A., Schnitzler, G., Deischel, S. M.,

Goel, M., Dove, W., Jaacks, K. J., Grossman, A. D., Erick-son, J. W. & Gross, C. A. (1989) Microbiol. Rev. 53, 1-24.

18. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) MolecularCloning: A Laboratory Manual (Cold Spring Harbor Lab.Press, Plainview, NY).

19. Pollitt, S. & Zalkin, H. (1983) J. Bacteriol. 153, 27-32.20. Rudd, K. E. (1992) in A Short Course in Bacterial Genetics: A

Laboratory Manual and Handbook for Escherichia coli andRelated Bacteria, ed. Miller, J. (Cold Spring Harbor Lab.Press, Plainview, NY), pp. 2.3-2.43.

21. Pinner, E., Padan, E. & Schuldiner, S. (1992) J. Biol. Chem.267, 11064-11068.

22. Neu, H. C. & Heppel, L. A. (1965) J. Biol. Chem. 240, 3685-3692.

23. Riddles, P. W., Blakeley, R. L. & Zerner, B. (1979) Anal.Biochem. 94, 75-81.

24. Mottl, H., Nieland, P., de Kort, G., Wierenga, J. J. & Keck, W.(1992) J. Bacteriol. 174, 3261-3269.

25. Zhou, Y.-N., Kusukawa, N., Erickson, J. W., Gross, C. A. &Yura, T. (1988) J. Bacteriol. 170, 3640-3649.

26. Lange, R. & Hengge-Arnois, R. (1991) J. Bacteriol. 173,4474-4481.

27. Platko, J. V., Willins, D. A. & Calvo, J. M. (1990) J. Bacteriol.172, 4563-4570.

28. Peek, J. A. & Taylor, R. K. (1992) Proc. Natl. Acad. Sci. USA89, 6210-6214.

29. Yu, J., Webb, H. & Hirst, T. R. (1992) Mol. Microbiol. 6,1949-1958.

30. Tomb, J.-F. (1992) Proc. Natl. Acad. Sci. USA 89, 10252-10256.

7088 Genetics: Missiakas et al.

Dow

nloa

ded

by g

uest

on

July

20,

202

1