Vol. 171, No. 1JOURNAL OF BACTERIOLOGY, Jan. 1989, p. 547-5570021-9193/89/010547-11$02.00/0Copyright C) 1989, American Society for Microbiology

Cloning and Analysis of Genes Involved in Coenzyme B12Biosynthesis in Pseudomonas denitrificansBEATRICE CAMERON,t KATHLEEN BRIGGS,4 SYLVIE PRIDMORE,§

GEORGES BREFORT,t AND JOEL CROUZETt*Gene'tica S.A., 160 Quai de Polangis, 94340 Joinville-le-Pont, France

Received 17 February 1988/Accepted 4 October 1988

Cobalamin synthesis probably requires 20 to 30 different enzymatic steps. Pseudomonas putida andAgrobacterium tumefaciens mutants deficient in cobalamin synthesis (Cob mutants) have been isolated. In P.putida, Cob mutants were identified as being unable to use ethanolamine as a source of nitrogen in the absenceof added cobalamin (deamination of ethanolamine requires coenzyme B12 as a cofactor). In A. tumefaciens, Cobmutants were simply screened for their reduced cobalamin synthesis. A genomic library of Pseudomonasdenitrificans was constructed on a mobilizable wide-host-range vector. Eleven plasmids from this library wereable to complement most of these mutants. By complementation and restriction mapping analysis, four genomicloci of P. denitrificans were found to be responsible for complementation of the Cob mutants. By subcloningfragments from the four genomic loci, we identified at least 14 different genes involved in cobalamin synthesis.

Cobalamins are among the most complex nonpolymericnatural products known (4). The coenzyme B12 biosyntheticpathway contains the following steps (Fig. 1): (i) formationof uroporphyrinogen III, the common intermediate for thesynthesis of hemes, chlorophylls, cobalamin, sirohemes, andF430 (23); (ii) conversion of uroporphyrinogen III into coby-rinic acid; (iii) formation of cobinamide from cobyrinic acid;and (iv) conversion of cobinamide into coenzyme B12 (forreviews on cobalamin biosynthesis, see references 3, 22, 26,41, and 46). Uroporphyrinogen III is synthesized in fourdistinct and well-known enzymatic steps from succinyl co-enzyme A and glycine (3). In contrast, the biosyntheticpathway from uroporphyrinogen III to coenzyme B12 is notclearly understood, since only one enzymatic activity hasbeen partially purified (38) and the precise number of enzy-matic steps is not known. In addition, the intermediatesbetween porphobilinogen and cobyrinic acid are very unsta-ble and sensitive to oxygen (3).

Cloning of genes coding for enzymes involved in adenosyl-cobalamin biosynthesis and the associated regulatory geneswould be a valuable tool in understanding the pathway. Theisolation and genetic characterization of mutants blocked incobalamin biosynthesis in Salmonella typhimurium and Ba-cillus megaterium have been recently described (29, 30, 48).Three bacteria are presently used in industry for vitamin

B12 production: Pseudomonas denitrificans, Propionibacte-rium shermanii, and Propionibacterium freudenreichii (19).We report here a genetic study ofcoenzyme B12 biosynthesisin a P. denitrificans strain of industrial interest (19). We haveisolated mutants ofPseudomonas putida and Agrobacteriumtumefaciens that are deficient in the synthesis of cobalamin(Cob mutants). These bacteria were used to isolate Cobmutants for two reasons: (i) molecular genetic techniques are

* Corresponding author.t Present address: Laboratoire de Gdndtique, Institut de Biotech-

nologie, Centre de Recherche de Vitry, Rh6ne-Poulenc Sante, 94403Vitry sur Seine Cedex, France.

t Present address: Amersham International P.L.C., Amersham,Buckinghamshire HP7 9LL, England.

§ Present address: 52 Habsbarger Strasse, 4310 Rheinfelden,Switzerland.

far more established in P. putida and A. tumefaciens than inP. denitrificans (2, 25), and (ii) these mutants are, like P.denitrificans, gram-negative aerobic rods that synthesizecobalamin. We constructed a genomic library of P. denitri-ficans on a mobilizable wide-host-range vector, and thislibrary was used to complement P. putida and A. tumefa-ciens Cob mutants in order to clone genes from the coba-lamin biosynthetic pathway. Four genomic fragments of P.denitrificans were found to be responsible for complemen-tation of most of the mutants, which were classified into 14complementation groups.

MATERIALS AND METHODSBacterial strains and plasmids. Unless otherwise specified,

chemicals were purchased from Sigma Chemical Co., St.Louis, Mo. Bacterial strains and plasmids used are listed,with their relevant characteristics, in Table 1. Strains relatedto SC510 are used for the industrial production of vitaminB12 (19). Strain SC510, which is characterized by signifi-cantly improved cobalamin productivity, was derived fromstrain MB580, originally defined as a strain of P. denitrifi-cans (Rhone-Poulenc Sante, U.S. patent 3,018,225), afternumerous chemical and physical mutagenesis steps. Al-though the taxonomic validity of the species P. denitrificansis questionable (14), we retain this taxonomic definition forstrains derived from MB580, in accordance with previouspublications (17, 33, 47). Rifampin-resistant isolates wereobtained by spreading fresh culture (5 x 109 cells per ml) onLB agar medium supplemented with 100 mg of rifampin perliter. Resistant colonies were streaked on LB agar mediumcontaining rifampin to obtain single-colony isolates, some ofwhich were checked for parental characteristics. To obtainnalidixic acid-resistant isolates, a similar procedure wasused at a nalidixic acid concentration of 20 mg/liter.Media, bacteriological techniques, and chemicals. For rou-

tine culture, all bacteria were grown in LB medium (36) at37°C for Escherichia coli and at 30°C for P. denitrificans, P.putida, and A. tumefaciens. PS4 medium used for theproduction of cobalamin, derived from the medium of Lagoand Demain (34), consisted of sucrose (30 g/liter), glutamicacid (5.8 g/liter), NZ case (Kraft Inc., Norwich, England) (10g/liter), betaine (10 g/liter), MgSO4 - 7H20 (1.5 g/liter),

547

548 CAMERON ET AL.

UROPORPHYRINOGEN IIICOCH

COBYRINIC ACID

HEMES

"PI',)

(IV)

2 v

OC

COBINAMIDE

FIG. 1. Cobalamin biosynthetic pathway. R = CH3, Methylcobalamin; R = OH, hydroxocobalamin, R = 5'-deoxyadenosyl, adenosyl-cobalamin; R = CN, cyanocobalamin; X, cobalt ligands (e.g., OH-); CoA, coenzyme A. The cobalamin pathway contains the following steps:(i) formation of uroporphyrinogen III; (ii) conversion of uroporphyrinogen III to cobyrinic acid, which requires eight methylations,decarboxylation of the acetic acid side chain at C-12, loss of C-20, and insertion of cobalt(III) (3, 41, 42); (iii) formation of cobinamide fromcobyrinic acid through six amidations and the addition of aminopropanol (22); and (iv) conversion of cobinamide to coenzyme B12, whichincludes phosphorylation of the aminopropanol residue and addition of GMP from GTP and of the lower base, 5,6-dimethylbenzimidazole,to form cobalamin phosphate, which, after dephosphorylation, gives coenzyme B12 (26).

(NH4)2HP04 (3 g/liter), MnSO4* H20 (0.02 g/liter), ZnSO4 .

7H20 (0.02 gfliter), FeSO4 * 7H20 (0.03 g/liter), MoO3Na-2H20 (0.005 g/liter), CoCl2 - 6H20 (0.12 g/liter), KCI (0.9 g/liter), and 5,6-dimethylbenzimidazole (0.045 g/liter). Cobaltand 5,6-dimethylbenzimidazole are incorporated into coba-lamin, and betaine is known to stimulate vitamin B12 pro-duction (17). Eth medium consisted of glucose (4 g/liter),K2HPO4 (7 g/liter), KH2PO4 (3 g/liter), Na2SO4 (1 g/liter),MgSO4. 7H20 (0.25 g/liter), and ethanolamine (1 ml/liter) asa nitrogen source. Vitamin B12 or cobinamide dicyanide was

added at a concentration of 1 pLg/ml. For E. coli, antibioticswere used at the following concentrations: kanamycin sul-fate, 100 mg/liter; tetracycline hydrochloride, 12 mg/liter;and spectinomycin dihydrochloride, 50 mg/liter. For P.putida and A. tumefaciens, the concentrations used were:kanamycin sulfate, 50 mg/liter; lividomycin sulfate (Rhone-Poulenc Sante), 100 mg/liter; rifampin, 100 mg/liter; andnalidixic acid, 20 mg/liter. P. putida and A. tumefacienscobalamin production was determined by using the followingprocedure: (i) a colony from a fresh reisolate was cultured in

SUCCINYL CoA

CONC

CONH

L2

GLYCINE

COBALAMIN

J. BACTERIOL.

HOOC

COENZYME B12 BIOSYNTHESIS IN P. DENITRIFICANS 549

TABLE 1. Bacterial strains and plasmids used

Bacterial strain or plasmid Relevant genotype, phenotype, or properties Source or reference

StrainE. coliMC1060 A(lacIPOZYA)X74 galU galK strA2 hsdR 9LG90 A(lac-pro)XIII ATCC 37114J53 nal Pro Met 5113-3 metE 11113-3 Cbll Mutant of 113-3 that cannot convert cobinamide dicyanide to This paper

cobalaminP. putidaKT2440 r- m+ 2KT2440 Rif Nalr Spontaneous nalidixic acid- and rifampin-resistant derivative of This paper

KT2440 RifP. denitrificans SC510 High-cobalamin-producing strain obtained from MB580 (U.S. patent Rh6ne-Poulenc Santd

3,018,225) by numerous random-screening mutagenesis cyclesA. tumefaciensC58-C9 Strain C58 cured of its Ti plasmid 25C58-C9 Rifr Nalr Spontaneous nalidixic acid- and rifampin-resistant derivative of C58-C9 This paper

PlasmidpRK2013 ColEl Kmr Tra+ (RK2) 18pRK2073 (pRK2013::Tn7) ColE1 Sp' Tpr Tra+ (RK2) 35pFR10 ColEl multiple-site cloning polylinker 43pFR237 incQ Kmr Mob' F. Richaud, unpublished datapXL59 incQ Kmr Mob' This paperpKT230 incQ Kmr Smr Mob' 2pXL435 incQ Kmr Mob' multicloning site This paperpJB4J1 incP pPHIJ1::Mu::Tn5 5pRK2013::TnS ColEl Kmr Tra+ (RK2) This paper

a 250-ml Erlenmeyer flask containing 25 ml of PS4 medium;(ii) after incubation (4 days for P. putida and 5 days for A.tumefaciens) on a rotary shaker at 30°C, 1 ml of the culturewas cyanurated with 1 ml of a solution of 50% (vol/vol)acetonitrile-75 mM KCN and incubated for 1 h at 56°C;during this treatment, cells were lysed and cobalamins wereconverted to the more stable cyano forms, which could beassayed.

General methods. Genomic digests were obtained by incu-bating 5 jxg of DNA with 20 U of restriction enzyme (NewEngland BioLabs, Inc., Beverly, Mass.) for 2 h. Plasmidsand DNA fragments were labeled by nick translation with[a-32P]dCTP as described elsewhere (40). E. coli strainswere transformed by the standard calcium chloride proce-dure (36). Agarose gel electrophoresis, purification of DNAfragments, preparation of plasmid DNA, and Southern blot-ting were carried out as described previously (6, 36). Ge-nomic DNA was prepared according to the method ofShepard and Polisky (44).

Mutagenesis. E. coli, P. putida, and A. tumefaciens weremutagenized with N-methyl-N'-nitro-N-nitrosoguanidine ac-cording to published procedures (37). TnS transposon muta-genesis was carried out by using either pRK2013::TnS (con-structed in this laboratory by a TnS insertion into pRK2013;data not shown) or pJB4J1, each of which is a Tn5-carryingplasmid. pRK2013::Tn5 and pJB4J1 are self-transmissibleplasmids (5); they do not replicate in bacteria such as P.putida and A. tumefaciens (pRK2013 has only the origin ofreplication of ColEl [18], and pJB4J1 contains bacteriophageMu, which prevents the establishment in non-Enterobac-teriaceae of the carrier plasmid [5]). These plasmids wereintroduced into the bacteria to be mutagenized by mating thedonor organism, HB1O1(pRK2013::TnS) or J53(pJB4J1), andthe recipient organism, KT2440 Rif Nalr or C58-C9 Rif' Nalrand selecting rifampin-, nalidixic acid-, and kanamycin-resistant clones.

Screening for Cob mutants of A. tumefaciens and P. putida.In the case of P. putida mutants, each single colony obtainedafter mutagenesis was transferred to 100 ,u of 10 mM MgSO4in a 96-well microdilution plate. A drop of each suspensionwas transferred, by using a replicating device, from themicrodilution plate onto a square plate of Eth medium withor without vitamin B12 in order to identify clones that wereauxotrophs for vitamin B12. For A. tumefaciens, each colonyobtained after mutagenesis was individually transferred to100 RI of PS4 medium in a 96-well microdilution plate. Afterincubation and cyanuration, the Cob mutants were detectedby microbiological assay or enzyme-linked immunosorbentassay (ELISA).

Mobilization of plasmid DNA from E. coli to other gram-negative bacteria. Triparental mating (13) was carried out bymixing the recipient strain (P. putida or A. tumefaciens) withHB1O1(pRK2073) and the E. coli strain carrying the mobi-lizable plasmid to be transferred. Clones of the genomiclibrary stored in Hogness freezing medium were mobilizedas described elsewhere (21). The transconjugants were se-lected on LB medium supplemented with rifampin andnalidixic acid for selection of the recipient bacteria and withkanamycin or lividomycin for determination of the presenceof the plasmid. Lividomycin was used for selection ofpXL59derivatives in Tn5 mutants of A. tumefaciens and P. putidasince the enzyme, aminoglycoside-O-phosphotransferase(3')I, encoded by pXL59 (originating from Tn9O3) conferslividomycin resistance, contrary to aminoglycoside-O-phosphotransferase(3')II from TnS (20).Complementation of cobalamin-deficient mutants of P. pu-

tida and A. tumefaciens. After transfer of plasmids from thegenomic library into mutants, we screened for complement-ing plasmids. In the case of P. putida, transconjugants werereplicated on Eth medium with or without vitamin B12; after3 days of incubation, clones that were able to grow on bothmedia were reisolated and their cobalamin production was

VOL. 171, 1989

550 CAMERON ET AL.

studied. In the case of A. tumefaciens, transconjugants wereinoculated 88 by 88 into a 96-well microdilution plate, eachwell containing 100 ,ul of PS4 medium supplemented withkanamycin or lividomycin. After incubation and cyanura-tion, vitamin B12 production was measured by the microbi-ological assay. Cobalamin production by positive clones in a250-ml Erlenmeyer flask was compared with the level ofcobalamin synthesized by the mutant alone or the mutantcontaining pXL59. Mutants were considered complementedfor the Cob mutation when the cobalamin level was compa-rable with that of the parent strain.

Vitamin B12 assays. Three types of assays were alterna-tively used to determine cobalamin concentrations: (i) amicrobiological assay for which the indicator strain is an E.coli vitamin B12 auxotroph (113-3 or 113-3 Cbll); (ii) aradiochemical assay derived from a method previously pub-lished (24) that uses intrinsic factor (IF) or nonintrinsicfactor (NIF) (from porcine gastric mucosa); and (iii) anELISA that involves IF and an anti-B12-KLH goat serum.These assays are sensitive in the range of 10 to 80, 0.02 to0.08, and 1 to 5 ,ug of vitamin B12 per liter, respectively. Thestandard deviation was 5% for the radiochemical assay andaround 15% for the microbiological assay and ELISA. Themicrobiological assay was carried out as described previ-ously (J. Crouzet, these de docteur ingenieur, Institut Na-tional Agronomique de Paris-Grignon, Paris, 1987).

In the radiochemical assay, IF and NIF bind 57Co-labeled,3H-labeled, and nonradioactive cyanocobalamins indiscrim-inately but display different specificities, since vitamin B12and cobinamide bind to NIF with similar affinities whereascobinamide does not bind to IF, which is more specific forthe assay of vitamin B12 and related products (39). To assay400 ,ul of a vitamin B12 sample, 325 pul of a solution of 350mM Tris hydrochloride (pH 10), 75 mM NaCl, and 1.5 mg ofbovine serum albumin per ml, containing 10 nCi of cyano[G-3H]cobalamin (4.4 Ci/mmol; Amersham, United Kingdom)and 15 U of IF or NIF, was added. After incubation for 1 hat 37°C, 250 p.l of a suspension containing 2.5% activatedcharcoal and 0.5% bovine serum albumin was added to thesolution. After a 5-min incubation at room temperature, thecharcoal was removed by centrifugation (the free forms ofcobalamins and cobinamides bind to activated charcoal,whereas IF or NIF complexes do not), thus allowing elimi-nation of the unbound corrinoids; the amount of radioactiv-ity bound to IF or NIF was then measured by counting theradioactivity present in the supernatant. If cyano[57Co]cobalamin (180 to 300 puCi/,ug) was used, the sensitivity ofthe assay could be improved by a factor of 200, thus allowingdetermination of concentrations ranging from 100 to 400 pg/liter.The ELISA was a sandwich-type assay derived from a

previously described procedure (15) and adapted to vitaminB12. The solution to be assayed was added to the IF-coatedplates and incubated in the presence of an anti-B12-KLHgoat serum (Biospecia Ltd., Wembley, United Kingdom). Asheep anti-goat serum conjugated with alkaline phosphatase(Biosys S.A., Compiegne, France) was then added, andalkaline phosphatase activity was assayed as describedelsewhere (16).

Construction of pXL59, pXL435, and plasmids carrying P.deniificans DNA fragments. A vector was constructedwhich (i) has unique restriction sites in the coding region ofa testable genetic marker and (ii) carries a wide-host-rangereplicon that can be mobilized in many gram-negative bac-teria. The replicon is of the IncQ type and contains part ofplasmid pFR237 (a generous gift of F. Richaud) (Fig. 2).

BE

p / r

c s4ripFR237 ) X P

mb 17 kb I19 kb

Ric

x

H E C.Ss,P,H*C.X,Xb.SC

H Pxc x

p

pXL59 ~~~~pXL435 sripXL59 iO1.6 kcb

mb 17 kb sick

sic~~~~~~~~~~~~~~i

sri a obx

Xb I M 2 M 3 E 4 M

FIG. 2. Physical maps of plasmids pFR237, pKT230, pXL59,and pXL435. B, BamHI; C, CMaI; E, EcoRI; H, HindlIl; P, PstI; S,SstI; Sa, Sall; X, XhoI; Xb, XbaI. Key to symbols: 1, PstI-SstIfragment of RSF 1010, which contains the origin of vegetativereplication (ori) (12), the relaxation nick site (nic), and a determinant(mob) that is essential for plasmid mobilization (2); 2, PstI-BamHIfragment of pACYC177 (2); 3, BamHI-SstI fragment containing theE. coli lactose operon without the promoter, the operator, thetranslation initiation sites, and the first eight nonessential codons oflacZ (9); 4, Sau3AI fragment of P. putida chromosomal DNA.

P-Galactosidase activity was not detected in strainMC1060(pFR237). We inserted Sau3AI fragments of P.putida KT2440 genomic DNA into the BamHI-linearizedvector pFR237 and searched for recombinant plasmids thatallowed expression of P-galactosidase in E. coli MC1060.One of the plasmids was shown to carry a 75-base-pair (bp)fragment that reconstituted the BamHI site on its rightboundary, toward the lacZ gene, and was named pXLS9(Fig. 2). Moreover, it was confirmed that pXL59 could bemobilized with the same efficiency as could pFR237 orpKT230 from E. coli to gram-negative bacteria such as A.tumefaciens and P. putida (data not shown). pXL435 wasderived from pKT230 by substituting the BamHI-SstI frag-ment for the multiple-cloning fragment from pFR10 (43). Thethree wide-host-range vectors (pKT230, pXL435, andpXL59) were chosen for subcloning of P. denitrificansfragments. MC1060 was used as the recipient strain exceptfor the cloning of EcoRI or SstI fragments into pKT230; inthese cases, the streptomycin-sensitive strain LG90 wasused and screened for streptomycin-sensitive recombinants.

Construction of a P. denitriicans genomic DNA library. Agenomic DNA library from strain SC5150 was constructed inpXL59. A partial Sau3AI digest of SC510 chromosomalDNA (200 pug) was fractionated by sucrose gradient centrif-ugation (36). Fractions corresponding to DNA fragmentswith sizes ranging from 10 to 15 kbp were pooled. Thesize-fractionated genomic DNA fragments (100 pug/ml each)were ligated with BamHI-digested pXL59, using T4 DNAligase (New England BioLabs) at a concentration of 300 U/ml. A fraction of the reaction mixture was used to transformcompetent MC1060(pRK2073) cells. Transformed cells were

J. BACTERIOL.

COENZYME B12 BIOSYNTHESIS IN P. DENITRIFICANS 551

plated on LB agar medium containing kanamycin, specti-nomycin, and X-Gal (5-bromo-4-chloro-3-indolyl-,-t-galac-topyranoside). A total of 3,200 white colonies on X-Gal-supplemented medium were picked and individuallyinoculated in a 96-well microdilution plates containing 200 ,udof Hogness freezing medium per well for conservation (21).Analysis of 24 clones showed that the average size of theinserted DNA fragments in pXL59 was around 13 kbp.Assuming that the P. denitrificans genome has a molecularsize of about 5,000 kbp, this library is large enough torepresent any DNA sequence of the genome with a proba-bility greater than 99% (10).

Isolation of an E. coli mutant deficient in the transformationof cobinamide dicyanide into methylcobalamin. It has beenreported that E. coli 113-3, a metE mutant, requires vitaminB12 to grow without an exogenous supply of methionine (11).It has been also reported that E. coli is able to catalyze thetransformation of cobinamide into methylcobalamin (39).Therefore, we searched for a mutant blocked in methylco-balamin biosynthesis from cobinamide dicyanide. After N-methyl-N'-nitro-N-nitrosguanidine mutagenesis, 800 colo-nies of E. coli 113-3 plated on LB medium were replicaplated on M9 medium supplemented with cobinamide dicya-nide. We found one mutant, named 133-3 Cbll, that wascomplemented by vitamin B12 or methionine but not bycobinamide dicyanide or 5,6-dimethylbenzimidazole. E. coli113-3 Cbll is apparently blocked in the synthesis of methyl-cobalamin from cobinamide dicyanide, since in E. colicobinamide dicyanide and vitamin B12 use the same trans-port system (7). This strain was used as an indicator strainfor the microbiological assay of compounds that are formedbetween cobinamide and methylcobalamin. Since all sam-ples assayed for cobalamin content were cyanurated, mutant113-3 Cbll is an indicator strain that should give a morespecific response for the assay of the end products in thebiosynthesis of cobalamin.

RESULTS

Isolation of mutants deficient in cobalamin synthesis. Sincecobalamin synthesis is a biochemical pathway composed ofmany different steps, two mutagenesis strategies were usedto obtain as many different mutants as possible: transposonmutagenesis with TnS, which is known to have a lowtranslocation specificity (32), and N-methyl-N'-nitro-N-nitrosoguanidine mutagenesis. Two gram-negative bacteriathat produce cobalamin under aerobic growth conditionswere mutagenized: P. putida KT2440 (2) and A. tumefaciensC58-C9 (25). These two bacteria synthesize 50 and 500 pug ofcobalamin per liter, respectively, under our culture condi-tions (see Materials and Methods).

It is known that bacterial ethanolamine ammonia-lyase(EC 4.3.1.7), which catalyzes the deamination of ethanol-amine to acetaldehyde and ammonia, requires adenosylco-balamin as a cofactor; this property has been described forenzymes of various bacteria, such as E. coli, Klebsiellaaerogenes, S. typhimurium, and Clostridium sp. (1). Ifethanolamine is the sole source of nitrogen, then the bacte-rium requires adenosylcobalamin to grow; therefore, E. coliand K. aerogenes, which do not produce cobalamin aerobi-cally, grow aerobically on ethanolamine as a nitrogen sourceonly in the presence of added cobalamin (1).

Since P. putida grows on minimal medium with ethanol-amine as the sole source of nitrogen, we hypothesized that inP. putida this pathway of nitrogen assimilation was alsodependent on endogenous adenosylcobalamin synthesis. P.

TABLE 2. Cobalamin biosynthesis-deficient mutantsof P. putidaa

Strain Phenotype Cobalamin Cofncn Mutant class

KT2440 Rif' Nalr 53G547 Cobl 2 6G548 Cob2 0.3 4G549 Cob3 0.2 1G550 Cob4 0.2 5G551 CobS 0.2 4G552 Cob6 0.3 2G553 Cob7 0.1 3G554 Cob8 0.1 1G555 Cob9 0.2 6G556 CoblO 8 2G557 Cobll 0.2 2G558 Cobl2 0.06 5G559 Cobl3 0.1 4G560 Cobl4 0.6 2G561 CoblS 0.01 4G562 Cobl6 0.06 1G563 Cobl7 0.06 3G566 Cobl8 0.04 5G567 Cobl9 0.04 3G568 Cob2O 0.06G570 Cob2l 0.08G571 Cob22 0.8G572 Cob23 3G573 Cob24 0.1G576 Cob25 0.04G577 Cob26 S

a Mutants G568, G570, G571, G572, G573, G576, and G577 were obtainedby N-methyl-N'-nitro-N-nitrosoguanidine mutagenesis. Cobalamin corlcen-trations were determined by IF, as described in Materials and Methods, whenthe strains were cultivated under standard conditions for cobalamin produc-tion. Data are means of two assays. Mutant class refers to the six classesdefined according to the size of the mutant genomie EcoRI DNA fragmenthybridizing with a 32P-labeled Tn5 probe.

putida KT2440 Rif' Nalr was chosen as the recipient in thetransposon-mediated mutagenesis experiment; 6,000 muta-genized colonies were selected at a frequency of 10-5 perrecipient cell as described in Materials and Methods. Afterreplica plating on ethanolamine medium with or withoutcyanocobalamin, mutants that required cyanocobalamin forgrowth were further investigated by confirming their pheno-type and studying cobalamin production in PS4 liquid cul-ture. A total of 19 transposon-mediated mutants were foundto produce reduced levels of cobalamin (Table 2). Thesetransposon-nmediated mutants were also studied by Southernblot analysis on total EcoRI-digested chromosomal DNA(there is no EcoRI restriction site in TnS [31]) by using32P-labeled TnS as a probe. This analysis revealed that foreach mutant a single fragment of genomic DNA hybridizedwith the probe; in each case, therefore, insertion of thetransposon was likely to be unique in the chromosome of thestrain. There was no detectable level of hybridization withchromosomal DNA of the parent strain (data not shown).The Cob mutants were then classified according to theestimated size of the EcoRI fragment hybridizing to 32p_labeled TnS. This classification defined six classes of mu-tants (1 to 6; Table 2), which corresponded to estimatedEcoRI chromosomnal fragments of 17, 14.5, 13, 12, 10.5, and6 kbp, respectively (data not shown).

P. putida mutants were also selected after treatment withN-methyl-N'-nitro-N-nitrosoguanidine as described in Mate-rials and Methods. The screening procedure described abovewas applied to 1,400 mutagenized colonies. Seven selected

VOL. 171, 1989

552 CAMERON ET AL.

TABLE 3. A. tumefaciens Cob mutantsa

Mutant Cobalamin concnStrain Phenotype class (,ug/liter)

C58-C9 Rifr Nair Cob' 500G159 Cobl VI <10G160 Cob2 II <10G161 Cob3 V <10G162 Cob4 IV <10G164 CobS I <10G165 Cob6 II <10G166 Cob7 I <10G168 Cob8 I <10G169 Cob9 VI <10G170 CoblO III <10G171 Cobll V <10G172 Cobl2 I <10G256 Cobl3 IV <10G258 Cobl4 VI <10G260 CoblS III <10G261 Cobl6 I <10G262 Cobl7 III <10G609-617 Cobl8-26 <1obG619-644 Cob27-52 <1obG646, G648 Cob53-54 <1obG1%3-2054 Cob55-146 <jobG2056, G2057 Cobl47-148 <10

a The cobalamin assay was performed with E. coli 113-3 Cbll. Cobalaminconcentrations were not significantly different when tested with E. coli 113-3except for mutants G642 and G2038 to G2043. Mutants G609 to G617, G619 toG644, G646, G648, G1963 to G2054, G2056, and G2057 were obtained bymutagenesis with N-methyl-N'-nitro-N-nitrosoguanidine.

b Most of the mutants produced cobalamin at levels of lower than 10 p.g/liter. Exceptions (with production [in micrograms per liter] indicated inparentheses) were G614(80), G615 (30), G616(30), G617 (30), G640(40), G643(70), G644 (70), G646 (200), G1663 (60), G1965 (120), G1975 (260), G1983 (30),G1991 (260), G1992 (200), G2008 (20), G2010 (150), G2012 (90), G2018 (220),G2026 (40), G2034 (40), G2045 (280), and G2054 (20). Data are means of twoassays. Mutant class refers to the six classes defined according to the size ofthe mutant genomic EcoRI DNA fragment hybridizing with a 32P-labeled Tn5probe.

mutants were found to exhibit the same properties as did theCobl to Cobl9 mutants (Table 2) and to produce a low levelof cobalamin.We noticed that all mutants except G547, G556, G572, and

G577 produced 50 times less cobalamin than did the wild-type strain as determined by the radiochemical assay.A. tumefaciens C58-C9 Rif' Nalr was also mutagenized to

obtain Cob mutants. Unfortunately, we have not been ableto determine a procedure for screening vitamin B12-negativemutants of A. tumefaciens, as has been done for P. putidaand for S. typhimurium and B. megaterium (29, 48). There-fore, each mutagenized colony was screened for cobalaminproduction (see Materials and Methods). Mutants deficientin adenosylcobalamin synthesis were kept on the basis of atleast a 20-fold-reduced production compared with that of thewild-type strain. Transposon-mediated mutagenesis in A.tumefaciens was performed as described in Materials andMethods. Each colony was studied for ability to producecyanocobalamin as determined by ELISA. A total of 17clones from 5,000 mutants were deficient in production ofcobalamin. The mutants were analyzed as described above.Six classes, related to the estimated size of the EcoRIchromosomal fragment of the mutants hybridizing with the32P-labeled TnS probe, were found. Fragments of 17, 15.5,14.5, 13.5, 8.5, and 7.5 kbp corresponded to classes I, II, III,IV, V, and VI, respectively (data not shown). Characteris-tics of and cobalamin production by these mutants are

shown in Table 3. In addition, 17,000 colonies mutagenizedby N-methyl-N'-nitro-N-nitrosoguanidine were screened forthe absence of cobalamin production, and 131 cobalamin-negative mutants were obtained. Cob mutants were detectedby ELISA for the first 39 mutants; the other mutants wereanalyzed by the microbiological assay, using E. coli 133-3Cbll. Except for 22 mutants, cobalamin production as de-termined by the microbiological assay was at least 50 timeslower than the parental level (Table 3).

Since the Cob mutants were obtained in two obligateaerobes, it is probable that they are not blocked in the part ofthe cobalamin synthetic pathway common with heme bio-synthesis; otherwise, they should have been somehow defi-cient in heme synthesis and characterized by an impairedrespiration mechanism leading to a lethal phenotype. Forinstance, Rhizobium meliloti mutants blocked in the firstenzymatic step of the heme pathway (corresponding tob-aminolevulinic acid synthetase, which catalyzes the con-densation of succinyl coenzyme A and glycine into 8-aminolevulinic acid) require b-aminolevulinic acid forgrowth (35). Since 5,6-dimethylbeizimidazole is a compo-nent of PS4 medium, it is likely that all of the mutants thathave been obtained are not blocked in 5,6-dimethylbenzimi-dazole synthesis.

In S. typhimurium, cysG mutants have been described;these mutants, which do not synthesize sirohemes, are alsoblocked in vitamin B12 synthesis (29). We investigatedwhether such mutants were isolated in our study. Since theA. tumefaciens mutants were first plated on LB medium andtested for cobalamin production in PS4 medium (containingNZ case), cysteine-requiring mutants would not have beeneliminated. This procedure should therefore have allowed usto isolate mutants blocked in sirohemes and vitamin B12synthesis if such mutants could be found in A. tumefaciens.However, among all of the A. tumefaciens mutants studied,none were cysteine auxotrophs; this finding indicated anabsence of these mutants or a different genetic organizationthan exists in S. typhimurium (30). Since P. putida Cobmutants were screened on Eth medium without cysteine, it isprobable that mutants defective in sirohemes and vitaminB12 synthesis were not isolated.Complementation of Cob mutants of P. puida and A.

tumefaciens by a genomic library of P. denitficans. Eachclone of the genomic library of P. denitrificans was mobi-lized individually into Cob mutants of P. putida as describedin Materials and Methods. The resulting transconjugantswere tested for ability to grow on Eth medium. Clones thatwere no longer vitamin B12 auxotrophs were identified. Itwas verified that they were also complemented for coba-lamin synthesis in 25-ml PS4 medium cultures. Because theplasmids were introduced individually, by replica mating, itwas possible to identify plasmids that complemented thesame mutants. Plasmids pXL151, pXL152, pXL156 topXL158, pXL160, and pXL161 were shown to complement16 P. putida Cob mutants (Table 4). Among the other 10mutants, 9 (G547, G550, G553, G555, G563, G566, G567,G568, and G571) were not complemented by the genomiclibrary of P. denitrificans and 1, G576, was partially com-plemented (5% of the wild-type level) by pXL159.The genomic library was also used to complement Cob

mutants of A. tumefaciens. Cobalamin synthesis by eachtransconjugant was studied as described in Materials andMethods. Mutants were defined as complemented whentransconjugants produced the wild-type level of cobalamin.pXL151 through 154, pXL156 to 161, and pXLS19 comple-mented A. tumefaciens Cob mutants (Table 4). pXL151,

J. BACTERIOL.

COENZYME B12 BIOSYNTHIESIS IN P. DENITRIFICANS 553

TABLE 4. Cob mutants of P. putida and A. tumefaciens and the corresponding plasmids from the genomic libraryof P. denitrificans that allow complementationa

Group Plasmids that complement Complemented Cob mutantsGroup ~Cobmutants.vP. putida A. tumefaciens

A pXL151 G558,b G570 G164/' G261,/ G609, G610, G638, G1985G1998,G2056, G2057

pXL151, pXL152 G548,b G551,b G552,b G557,b G166,b G611-617, G619, G620, G1963-1984G559/b G560/b G556,b G561b

B pXL153, pXL154 G159,b G169,/ G258,b G2035pXL154 G161,/ G171,b G2034, G2037

C pXL156 G549, .c G562,b G573c G170,b G260,' G262b, G622-630, G642,cG2021-G2033, G2036, G2043C

pXL156, pXL157 G632, G633, G2018-G2020pXL157 G631, G2005-G2017pXL157, pXL158, pXL160, pXL161 G554/' G572, G577 G160/' G165,b G634, G644, G1999-02004,

G2054pXL157-pXL161 G162,b G256,b G635-637, 6639, 6640, 6643,

G646, G648

D pXL519 G2038,c G2039,c G20WO,c G2041,c G2042Ca Uncomplemented mutants for P. putida: G547, G550, G553, G555, G563, G566, G567, G568, and G571 (G576 is partially complemented by pXL159).

Uncomplemented mutants for A. tumefaciens: G168, G172, G621, G641, and G2044 to G2053.b Cob mutants obtained by insertion of transposon TnS.c Mutants positively assayed with E. coli 133-3.

pXL152, pXL156 to pXL158, and pXL161 were thereforeable to complement Cob mutants of both A. tumefaciens andP. putida. The complementation data allowed us to classifythe plasmids into four complementation groups, designatedA, B, C, and D. Plasmids in each group complemented aspecific set of mutants not complemented by plasmids fromthe other groups (Table 4). Group A consisted of pXL151and pXL152, group B contained pXL153 and pXL154, groupC included pXL156 to pXL161, and group D contained onlypXL519.The restriction maps of the inserts carried by these various

plasmids were established (Fig. 3). The maps are in agree-ment with complementation groups, since overlapping in-serts always correspond to plasmids belonging to the samecomplementation group. Any physical linkage deduced fromthe restriction pattern was checked by Southern blot analy-sis. For instance, the 2.6-kbp EcoRI-SstI fragment ofpXL156, which overlaps with pXL1574 was purified andradiolabeled with 32p; hybridization with EcoRI-SstI digestsof pXL156 and pXL157 showed that the same fragment waspresent on both pXL156 and pXL157 (data not shown). Thesizes of cloned DNA corresponding to groups A to D were15.5, 25.5, 29, and 8 kbp, respectively. No correlation couldbe found among the restriction maps of the four groups (Fig.3), which indicated that there is no overlap between the fourdifferent fragments. Therefore, at least four genomic loci ofP. denitrificans are involved in cobalamin synthesis and cancomplement Cob mutants of P. putida and A. tumefaciens.We do not yet know whether those loci are tightly clusteredor scattered on the chromosome.

Genetic organization of the Cob loci. A more precisecomplementation map of the Cob loci was established to gaindeeper insight into the genetic organization of the pathway.Smaller fragments representative of the four genomic loci, inthe range of 2 to 4 kbp, were subcloned into pKT230,pXL435, or pXL59, depending on the available restrictionsites (Fig. 2). We thus identified 14 different plasmids thatspecifically complemented most of the mutants investigated

(Table 5): pXL220, pXL227, pXL239, pXL436, pXL444,pXL452, pXL556, pXL617, pXL622, pXL678, pXL684,pXL698, pXL735, and pXL837 (Fig. 3). All of these plasmidscomplemented mutants that were not complemented by theothers. All of the plasmids except pXL220 complemented A.tumefaciens Cob mutants.Complementation data with subclones are in agreement

with the classification of P. putida and A. tumefaciens TnSmutants. For instance, a subclone complemented TnS mu-tants of P. putida or A. tumefaciens belonging to the samehybridization class (i.e., pXL684 complemented the three P.putida mutants of class 2; Table 5); this finding suggests thatin complemented Tn5 mutants the transposon has insertedinto Cob genes. Complemented mutants of a hybridizationclass were all complemented by the same subclone (seeexample cited above) or by subclones of the same cluster(pXL617, pXL622, and pXL239 complemented the P. putidamutants G562, G549, and G554, respectively, which consti-tute class 1; Table 5). This result may indicate that somehybridization classes represent a genomic EcoRI fragmentthat carries more than one gene involved in coenzyme B12synthesis. In addition, uncomplemented P. putida TnS mu-tants belonged to the same hybridization classes (classes 3and 6).Although some inserts of these 14 plasmids overlap, we

can deduce that each insert carries at least one different geneinvolved in cobalamin biosynthesis, since each plasmiddefines a specific complementation group (Table 5). A totalof at least 14 P. denitrificans genes involved in the transfor-mation of uroporphyrinogen III to coenzyme B12 have thusbeen cloned.Complementation by pXL189, pXL190, pXL191, or

pXL300 (Fig. 3) was not specific because several mutantswere complemented by these plasmids and not by theirsubclones (Table 5). For instance, pXL190 complementedG162, which was not complemented by pXL239 or pXL556(Fig. 3). pXL189, pXL190, pXL191, and pXL300 shouldcarry additional genes that are not revealed by the comple-

VOL. 171, 1989

554 CAMERON ET AL.

GROUP A H E H Be

pXL 151pXL 152

1 kb

so x a 8 8 EI I I I I - -

pXL444 pXL67S pXL4528 S

pXLSS8Sg SB9

IpXL436

GROUP BE

pXLSOO

IXES Hl~~~~IS H E E aXsE XHB BH H B HS E XX ES X

I I I I I 1111 -1 I I I11 1 1 I 11 I I I I1pXL1 53

pXL 154

pXL735 E x

pXLS37GROUP C

H _S ESSS8 XSB X SB E B SIX H H E SH H Se E X XI1II- I I I II II- I I I I

pXLI 56pXL157pXLI58

pXLI 59

pAL J VU

pXL6IS

S SI-pXL61 7

S P

pXL622

GROUP D

E S C EcoRVI I I

pXL227 pXL556H S.u

I ISau C pXL239a I E BpXL189 I I

pXL220C SauII

pXLl90C

pXL1 91

S S BE S

pXL519

Bg Bg

pXL6SS

FIG. 3. Physical map of inserts of the 11 plasmids from the genomic library of P. denitrificans that complement Cob mutants of P. putidaand A. tumefaciens. Shown are inserts of each plasmid as well as inserts of the 14 subclones and plasmids pXL300, pXL189, pXL190, andpXL191. The 11 plasmids and their subclones were introduced into mutants by triparental mating; the transconjugants were studied forcobalamin synthesis in a 25-ml PS4 culture. The mutants complemented by these are listed in Tables 4 and 5. B, BamHI; Bg, BgIII; C, ClaI;E, EcoRI; H, HindlIl; P, PstI; Sa, Sall; Sau, Sau3AI; S, SstI; X, XhoI.

J. BACTERIOL.

Sou

II

COENZYME B12 BIOSYNTHESIS IN P. DENITRIFICANS 555

TABLE 5. Relevant Cob mutants of A. tumefaciens and P. putida and the subclones that complement them

Subclones that Complemented mutantsaGroup complement mutants P. putida A. tumefaciens

A pXL444 G612pXL678 G551 (4), G559 (4), G561 (4) G614, G616, G617pXL452 G558 (5), G570 G609, G610pXL684 G552 (2), G556 (2), G557 (2), G560 (2) G166 (1), G620pXL436 G548 (4) G615pXL300b G164 (1), G261 (1), G613, G611, G619, G638

B pXL735 G159 (VI), G169 (VI), G258 (VI), G2035pXL837 G161 (V), G171 (V), G2034, G2037

C pXL617 G562 (1) G170 (III), G260 (III), G262 (III), G624, G629pXL622 G549 (1), G573 G642,c G2043CpXL227 G632, G633pXL556 G637, G648pXL239 G554 (1), G577 G643pXL220 G572pXL189d G631pXL190e G162 (IV), G256 (IV), G635, G636, G639pXL191f G160 (II), G165 (II), G634

D pXL698 G2038,C G2039,c G2040,c G2041,c, G2042Ca Arabic and roman numerals in parentheses indicate genetic classes to which the TnS mutants belong.b Complements all mutants complemented by pXL452, pXL678, pXL684.c Mutants positively assayed with E. coli 113-3.d Complements all mutants complemented by pXL227.e Complements all mutants complemented by pXL556 or pXL239.f Complements all mutants complemented by pXL556, pXL239, pXL220, or pXL190.

mentation data obtained with the 14 subclones mentionedabove.

Analysis of mutants blocked in the conversion of cobinamideinto cobalamin. Among the various Cob mutants isolated,some exhibited the same levels of cobalamin production asdid the wild type when assayed with E. coli 113-3 as theindicator strain (Table 5); in contrast, when production ofcobalamin was assayed by using E. coli 113-3 Cbll (which isblocked in the conversion of dicyanocobinamide into coba-lamin), synthesis of cobalamin by these mutants was at least50 times lower than that of the wild type. Among these Cobmutants were two mutants of P. putida (G549 and G573) andseven mutants of A. tumefaciens (G642 and G2038 toG2043). In addition, cobalamin production was comparablewith that of the wild type when assayed with NIF but wassignificantly reduced when assayed with IF (data notshown). These results suggest that such mutants are blockedin the conversion of cobinamide to cobalamin. They werecomplemented by two subclones, pXL622 and pXL698,defining two classes of mutants (Table 5 and Fig. 3). pXL622and pXL698 should each carry at least one gene whoseproduct is involved in the conversion of cobinamide tocobalamin. Among the 14 genes cloned, two are thus impli-cated in the last part of the cobalamin biosynthetic pathwayand are carried by plasmids pXL622 and pXL698.

DISCUSSION

The selection of mutants unable to grow on ethanolamineas the sole source of nitrogen has enabled us to isolate P.putida mutants deficient in cobalamin biosynthesis. Thisresult supports the idea that in P. putida, ethanolamineammonia-lyase requires adenosylcobalamin to be active.Similar results have been obtained with B. megaterium (48),from which cobalamin biosynthesis-deficient mutants were

isolated by screening for mutants auxotrophic for vitaminB12 in the presence of ethanolamine as the sole source ofnitrogen. It is likely that this procedure can be generalized tomany cobalamin-producing bacteria that are able to metab-olize ethanolamine. However, attempts to find screeningconditions for the detection of A. tumefaciens Cob mutantshave failed, and most A. tumefaciens Cob mutants describedin this report grow on Eth medium. A similar approach,based on the vitamin B12 requirement in anaerobiosis for thegrowth of an S. typhimurium metE mutant, has been de-scribed to screen Cob mutants (29). This type of screeningcan probably be used for bacteria such as R. meliloti, thegrowth of which depends on the presence of vitamin B12 inthe medium when cobalt ions are absent (27, 28). In thiscase, the vitamin B12-dependent enzymes involved are aribonucleotide reductase (28) and a methionine synthetase(27). R. meliloti Cob mutants could probably be detected bythe absence of growth on a minimal medium and correctionof this defect by vitamin B12. Such a screening procedure didnot allow the isolation of Cob mutants in A. tumefaciens.At least 14 genes involved in cobalamin synthesis in P.

denitrificans have been cloned; 12 are implicated in theconversion of uroporphyrinogen III to cobinamide, and 2 areimplicated in the conversion of cobinamide to coenzymeB12. These genes have been identified by heterologouscomplementation of P. denitrificans or A. tumefaciens Cobmutants with amplified P. denitrificans DNA. We do notknow whether the absence of complementation for 10 of 26P. denitrificans Cob mutants and for 15 of 148 A. tumefa-ciens Cob mutants is due to the lack of the desired fragmentsin the genomic library or to problems related to either theabsence of heterologous gene expression or differences ingenetic organization between P. denitrificans and A. tume-faciens or P. putida. Another possibility is that at least someenzymes involved in such a metabolic pathway are arranged

VOL. 171, 1989

556 CAMERON ET AL.

in multienzymatic complexes (45), which would make it lessprobable that mutants in the corresponding genes would becomplemented by heterologous DNA. On the basis of thepercentage of complemented mutants with P. denitrificansamplified DNA, it appears that complementation of A.tumefaciens mutants was more successful than was that ofP. putida mutants.

It can be deduced from several publications (3, 22, 26, 41,42, 46) that more than 20 intermediates should be required inthe pathway from uroporphyrinogen III to cobalamin. How-ever, the number of enzymes involved in this pathway couldbe smaller, since (i) during the building of the corrin macro-cycle, one could carry out more than one methylation whenmethyl groups are transferred at equivalent positions (42) (itseems, for instance, that one enzyme in Propionibacteriumshermanii catalyzes the transfer of two methyl groups touroporphyrinogen III [38]); and (ii) during the conversion ofcobyrinic acid to cobinamide, the six amidations of thepropionic and acetic groups in the corrin macrocycle maynot require six different enzymes. Analyses of mutants andtheir products could establish this point. Moreover, thenumber of cloned genes is likely to be underestimatedbecause more than one gene could be present on eachsubclone defining a complementation group. Although wecannot yet determine whether all of the cloned genes arestructural or whether some of them play only a regulatoryrole, our results illustrate the genetic and biochemical com-plexity of this pathway.The cloned genes are grouped into four genomic regions,

and we do not yet know whether those Cob loci are clusteredor located at distant positions on the chromosome of P.denitrificans. Clustering of Cob genes is not peculiar to P.denitrificans, having already been reported for B. megate-rium and S. typhimurium (8, 29). In B. megaterium, all of theCob mutations found are linked by cotransduction (48); in S.typhimurium, the same results have been found except forthe cysG mutation, which indicates that most of the Cobgenes are located in the same region on the chromosome.For P. denitrificans, the cluster would cover at least 78 kbpif all of the Cob genes were linked on the chromosome.Furthermore, the inserts of the 14 subclones complementingCob mutants (Fig. 3) together represent 32 kbp, which ismuch more than the total length of 12 kbp for the clonedDNA fragments of B. megaterium containing 11 Cob genes(8); this fact may suggest that P. denitrificans Cob genes aremore spread out than are the corresponding genes from B.megaterium.

ACKNOWLEDGMENTS

We acknowledge A. Rambach, former president of Gdnetica S.A.,for his continued interest and support during this work. Weexpress our gratitude to M. Knapp for his help during the initial partof the work and for his advice. We thank P. E. Bost, G. Bourat, andJ. Lunel (Rh6ne-Poulenc Sante) for their constant interest in ourwork. We express our gratitude to G. Ditta, D. Helinski, F.Richaud, and K. Timmis for the gift of plasmids or strains; J. Davies(Institut Pasteur) and J.-F. Mayaux for critical reading of themanuscript and helpful discussions; and F. Blanche, D. Thibaut,and S. Schil. We also thank S. Bonnamy, L. Cauchois, A. Driver,N. Faulconnier, S. Rigault, and M.-C. Rouyez for their excellenttechnical assistance.

This work was supported by Rhone-Poulenc Sante.

LITERATURE CITED1. Babior, B. M. 1982. Ethanolamine ammonia-lyase, p. 107-144.

In D. Dolphin (ed.), B12, vol. 1. John Wiley & Sons, Inc., NewYork.

2. Bagdasarhn, M., R. Lurz, B. Riekert, F. C. Franklin, M. M.Bagdasarian, J. Frey, and K. TImmis. 1981. Specific-purposeplasmid cloning vectors. II. Broad host range, high copy num-ber, RSF1010-derived vectors, and a host vector system forgene cloning in Pseudomonas. Gene 16:237-247.

3. Battersby, A. R., and E. MacDonald. 1982. Biosynthesis of thecorrin macrocycle, p. 107-144. In D. Dolphin (ed.), B12, vol. 1.John Wiley & Sons, Inc., New York.

4. Beck, W. S. 1982. Biological and medical aspects of vitamin B12,p. 1-30. In D. Dolphin (ed.), B12, vol. 1. John Wiley & Sons,Inc., New York.

5. Beringer, J. E., J. L. Beynon, A. V. Buchanan-Wolaston, andA. W. B. Johnston. 1978. Transfer of the drug-resistance trans-poson TnS to Rhizobium meliloti. Nature (London) 276:633-634.

6. Birnboim, H. C., and J. Doly. 1979. A rapid alkaline extractionprocedure for screening recombinant plasmid DNA. NucleicAcids Res. 7:1513-1523.

7. Bradbeer, C. 1982. Cobalamin transport in microorganisms, p.145-167. In D. Dolphin (ed.), B12, vol. 2. John Wiley & Sons,Inc., New York.

8. Brey, R. N., C. D. B. Banner, and J. B. Wolf. 1986. Cloning ofmultiple genes involved with cobalamin (vitamin B12) biosyn-thesis in Bacillus megaterium. J. Bacteriol. 167:623-630.

9. Casadaban, M. J., A. Martinez-Arias, S. T. Shapira, and J.Chou. 1983. ,B-Galactosidase gene fusion for analysing geneexpression in Escherichia coli and yeast. Methods Enzymol.100:293-308.

10. Clarke, L., and J. Carbon. 1976. A colony bank containingsynthetic ColE1 hybrid plasmids representative of the entire E.coli genome. Cell 9:91-99.

11. Davis, B. D., and E. Mingioli. 1950. Mutants of Escherichia colirequiring methionine or vitamin B12. J. Bacteriol. 60:17-28.

12. De Graff, J., J. H. Crosa, F. Heffron, and S. Falkow. 1978.Replication of the nonconjugative plasmid RSF1010 in Esche-richia coli K-12. J. Bacteriol. 146:117-122.

13. Ditta, G., S. Stanfield, D. Corbin, and D. R. Helnski. 1980.Broad host range DNA cloning system for Gram-negativebacteria: construction of a gene library of Rhizobium meliloti.Proc. Natl. Acad. Sci. USA 77:7347-7351.

14. Doudoroff, M., R. Contopoulou, R. Kunisaw, and N. J. PaLeroni.1974. Taxonomic validity of Pseudomonas denitrificans (Christ-ensen) Bergey et al. Int. J. Syst. Bacteriol. 24:294-300.

15. Doyen, N., A. J. Pesce, and C. Lapresle. 1981. Specificity of twoanti-human albumin monoclonal antibodies. Immunol. Lett. 3:365-370.

16. Engvall, E. 1980. Enzyme immunoassay ELISA and EMIT.Methods Enzymol. 70:419439.

17. Fa, Y. H., J. P. Kusel, and A. L. Demain. 1984. Dependence ofbetaine stimulation of vitamin B12 overproduction on proteinsynthesis. Appl. Environ. Microbiol. 47:1067-1069.

18. Figurskl, D. H., and D. R. HeUInskl. 1979. Replication of anorigin-containing derivative of plasmid RK2 dependent on aplasmid function provided in trans. Proc. Natl. Acad. Sci. USA76:1648-652.

19. Florent, J. 1986. Vitamins, p115-158. In H.-J. Rehm and G.Reed (ed.), Biotechnology, vol. 4. VCH VerlagsgesellschaftmbH, Weinheim, Federal Republic of Germany.

20. Foster, T. J. 1983. Plasmid-determined resistance to antimicro-bial drugs and toxic metal ions in bacteria. Microbiol. Rev. 47:361409.

21. Frey, J., M. Bagdaan, D. Feiss, F. C. H. Franklin, and J.Dehusses. 1983. Stable cosmid vectors that enable the introduc-tion of cloned fragments into a wide range of Gram-negativebacteria. Gene 24:299-308.

22. Friedmann, H. C. 1975. Biosynthesis of corrinoids, p. 75-109. InB. Babior (ed.), Cobalamin biochemistry and pathophysiology.John Wiley & Sons, Inc., New York.

23. Gilles, H., and R. K. Thauer. 1983. Uroporphyrinogen III, anintermediate in the biosynthesis of the nickel-containing F430 inMethanobacterium thermoautotrophicum. Eur. J. Biochem.135:109-112.

24. Gotlieb, C., K. S. Lau, and L. R. Wasserman. 1965. Rapid

J. BACTERIOL.

COENZYME B12 BIOSYNTHESIS IN P. DENITRIFICANS 557

charcoal assay for intrinsic factor (IF), gastric juice unsaturatedB12 binding capacity, antibody to IF, and serum unsaturated B12binding capacity. Blood 25:873-883.

25. Hoykaas, P. J. J., R. Peerbolt, A. J. G. Regensburg-Tuing, P. deVries, and R. A. Schilperort. 1982. A chromosomal linkage mapofAgrobacterium tumefaciens and a comparison with the mapsof Rhizobium sp. Mol. Gen. Genet. 188:12-17.

26. Huennekens, F. M., K. S. Vitols, K. Fujii, and D. W. Jacobsen.1982. Biosynthesis of cobalamin coenzymes, p. 145-167. In D.Dolphin (ed.), B12, vol. 1. John Wiley & Sons, Inc., New York.

27. Inukai, S., K. Sato, and S. Shimizu. 1977. The relationship ofcobalt requirement to vitamin B12-dependent methionine syn-thesis in Rhizobium meliloti. Agric. Biol. Chem. 41:2229-2234.

28. Inukai, S., K. Sato, and S. Shinizu. 1979. Purification andproperties of vitamin B12-dependent ribonucleotide reductasefrom Rhizobium meliloti. Agric. Biol. Chem. 43:637-646.

29. Jeter, R. M., B. M. Olivera, and J. R. Roth. 1984. Salmonellatyphimurium synthesizes cobalamin (vitamin B12) de novo underanaerobic growth conditions. J. Bacteriol. 159:206-216.

30. Jeter, R. M., and J. R. Roth. 1987. Cobalamin (vitamin B12)biosynthetic genes of Salmonella typhimurium. J. Bacteriol.169:3189-3198.

31. Jorgensen, R. A., S. J. Rothstein, and W. R. Reznikoff. 1979. Arestriction enzyme cleavage map ofTnS and location of a regionencoding neomycin resistance. Mol. Gen. Genet. 177:62-65.

32. Kleckner, N., J. Roth, and D. Botstein. 1977. Genetic engineer-ing in vivo using translocatable drug resistance element. J. Mol.Biol. 116:125-159.

33. Kussel, J. P., Y. H. Fa, and A. L. Demain. 1984. Betainestimulation of vitamin B12 biosynthesis in Pseudomonas deni-trificans may be mediated by an increase in activity of 8-aminolevulinic acid synthetase. J. Gen. Microbiol. 130:835-841.

34. Lago, B. D., and A. L. Demain. 1969. Alternate requirement forvitamin B12 or methionine of Pseudomonas denitrificans, avitamin B12-producing bacterium. J. Bacteriol. 99:347-349.

35. Leong, L. S., G. S. Ditta, and D. R. Helinski. 1982. Identificationof a cloned gene coding for 8-aminolevulinic acid synthetasefrom Rhizobium meliloti. J. Biol. Chem. 257:8724-8730.

36. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecular

cloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

37. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

38. Muller, G. R., R. Deeg, K. D. Gneuss, G. Gungerand, and H. P.Kriemler. 1979. In B. Zagalak and W. Friedrich (ed.), VitaminB12, p. 279-291. Walter de Gruyter, Berlin.

39. Nexo, E., and H. Olensen. 1982. Quantification of cobalamins inhuman serum, p. 87-104. In D. Dolphin (ed.), B12. John Wiley &Sons, Inc., New York.

40. Rigby, P. W., J. M. Dieckmann, C. Rhodes, and P. Berg. 1977.Labelling deoxyribonucleic acid to high specific activity in vitroby nick translation with DNA polymerase I. J. Mol. Biol. 113:395-403.

41. Scott, A. I. 1976. The biosynthesis of vitamin B12. Phil. Trans.R. Soc. London 273:303-316.

42. Scott, A. I., N. E. Mackenzie, P. J. Santander, P. E. Fagerness,G. Muller, E. Schneider, R. Seldmeier, and G. Worner. 1984.Biosynthesis of vitamin B12-timing of the methylation stepsbetween uro'gen III and cobyrinic acid. Bioorg. Chem. 12:356-352.

43. Shapira, S. K., J. Chou, F. V. Richaud, and M. C. Casadaban.1983. New versatile plasmid vector for expression of hybridproteins coded by a cloned gene fused to lacZ gene sequencesencoding enzymatically active carboxy-terminal portion of ,B-galactosidase. Gene 25:71-83.

44. Shepard, H. M., and B. Polisky. 1979. Measurement of plasmidcopy number. Methods Enzymol. 68:503-506.

45. Srivastava, D. K., and S. A. Bernhard. 1986. Metabolite transfervia enzyme-enzyme complexes. Science 234:1081-1086.

46. Uzar, H. C., A. R. Battersby, T. A. Carpenter, and F. J. Leeper.1987. Biosynthesis of porphyrins and related macrocycles. J.Chem. Soc. Perkin Trans. 1:1689-1696.

47. White, R. F., and A. L. Demain. 1971. Catabolism of betaine andits relationship to cobalamin overproduction. Biochim.Biophys. Acta 237:112-119.

48. Wolf, J. B., and R. N. Brey. 1986. Isolation and geneticcharacterization of Bacillus megaterium cobalamin biosynthe-sis-deficient mutants. J. Bacteriol. 166:51-58.

VOL. 171, 1989