JOURNAL OF BACTERIOLOGY, Apr. 1986, p. 51-580021-9193/86/040051-08$02.00/0Copyright © 1986, American Society for Microbiology

Vol. 166, No. 1

Isolation and Genetic Characterizations of Bacillus megateriumCobalamin Biosynthesis-Deficient Mutants

JULIE B. WOLFt AND ROBERT N. BREY*Genex Corporation, Gaithersburg, Maryland 20877

Received 1 November 1985/Accepted 23 January 1986

Ethanolamine is deaminated by the action of ethanolamine ammonia-lyase (EC 4.3.1.7), an adenosylcobal-amin-dependent enzyme. Consequently, to grow on ethanolamine as a sole nitrogen source, Bacillusmegaterium requires vitamin B12. Identification ofB. megaterium mutants deficient for growth on ethanolamineas the sole nitrogen source yielded a total of 34 vitamin Bi2 auxotrophs. The vitamin B12 auxotrophs weredivided into two major phenotypic groups: Cob mutants, which could use cobinamide or vitamin B12 to growon ethanolamine, and Cbl mutants, which could be supplemented only by vitamin B12. The Cob mutants wereresolved into six classes and the Cbl mutants were resolved into three, based on the spectrum of cobalt-labeledcorrinoid compounds which they accumulated. Although some radiolabeled cobalamin was detected in the wildtype, little or none was evident in the auxotrophs. The results indicate that Cob mutants contain lesions inbiosynthetic steps before the synthesis of combinamide, while Cbl mutants are defective in the conversion ofcobinamide to cobalamin. Analysis of phage-mediated transduction experiments revealed tight genetic linkagewithin the Cob class and within the Cbl class. Similar transduction analysis indicated the Cob and Cbl classesare weakly linked. In addition, cross-feeding experiments in which extracts prepared from mutants wereexamined for their effect on growth of various other mutants allowed a partial ordering of mutations within thecobalamin biosynthetic pathway.

The basic corrin ring structure (cobyrinic acid; Fig. 1)which. gives rise to vitamin B12 is biosynthetically derivedfrom utoporphyrinogen III (also a precursor to heme andsiroheme) by a series of reductive methylations, adecarboxylation at C-12, ring scission between C-19 andC-20, and insertion of cobalt (Fig. 1). Although it is. knownthat four melthylations of the uroporphyrinogen III ringprobably occur first in the biosynthesis of cobalamin, theorder of further steps in cobyrinic acid formation is notknown. Beyond cobyrinic acid there are six aniidations ofcarboxyl side groups and an addition of aminopropanol(biosynthetically derived from threonine) to yieldcobinamide. Cobinamide is phosphorylated by ATP to yieldcobinamide P04, which reacts with GTP to yield GDP-cobinamide. GMP is displaced by the 5' nucleotide ofdimethylbenzimidazole (DMBI; biosynthetically derivedfrom riboflavin) to produce cobalamin P04, which isdephosphorylated to yield cobalamin (vitamin B12).Adenylation of the corrin macrocycle, which is necessary forthe formation of the cofactor form of cobalamin (adenosyl-cobalamin; also known as coenzyme B12), may occur veryearly in the biosynthetic sequence but can occur after theformation of cobalamin. For more detailed description of thecobalamin biosynthetic pathway, the reader is referred torecent reviews (2, 11, 13). The correlation of a particularbiosynthetic step with a particular enzyme has not beenaccomplished. With the possible exception of thb partiaipurification of an enzyme(s) involved with the methylation ofuroporphyrinogen III, none of the biosynthetic enzymes hasbeen well characterized (17). Thus, knowledge of enzymol-ogy is insufficient even to conclude how many enzymes are

* Corresponding author.t Present address: Cell Biology and Metabolism Branch, National

Institute of Child Health and Human Developmerit, Bethesda, MD20892.

involved with cobalmin biosynthesis. Virtually nothing isknown about the genetic control of this pathway.To begin a genetic study of a complex biosynthetic path-

way, it is first desirable to establish conditions under whichan organism requires the pathway end product for growth. Alogical way to approach a study of cobalamin biosynthesis isto take advantage of the microbial enzymes that are knownto require either methylcobalamin or adenosylcobalamin toact, forcing an organism to utilize a substrate whose metab-olism is strictly dependent on vitamin B12. There are anumber of such enzymes known (1). Ethanolamine ammo-nia-lyase (EC 4.3.1.7) catalyzes the deamination of ethanol-amine to acetaldehyde and ammonia (1, 16). Bradbeer firstdiscovered this reaction in a choline-fermenting Clostridiumsp. and demonstrated that the enzyme required adenosylco-balamin (6, 7). Subsequently, this enzyme has been detectedin a number of bacterial species, including Klebsiella aero-genes, Escherichia coli, and Salmonella typhimurium, inwhich exogenous vitamin B12 is required for growth onethanolamine (3-5, 10, 19).Ethanolamine ammonia-lyase is also present in Bacillus

megaterium, an obligate aerobe which is capable of vitaminB12 synthesis under aerobic growth conditions. B.megaterium can use ethanolamine as a sole source of nitro-gen. Growth of B. megaterium on ethanolamine, but not onammonia, is inhibited by adeninylalkyl analogs (12, 18) ofadenosylcobalamin, suggesting that mutant bacteria unableto synthesize cobalamin would require, as do wild-type E.coli and S. typhimurium cells, exogenous cobalamin toutilize ethanolamine as a sole source of nitrogen (unpub-lished data).The establishment of growth conditions under which B.

megaterium requires vitamin 1312 is analogous to the behav-ior of metE mutants of S. typhimurium and E. coli. Retainingthe methylcobalamin-dependent tetrahydropteroylglutamatemethyltransferase (EC 2.1.1.13), metE mutants can utilizeexogenous vitamin B12 to bypass their requirement for

51

on January 2, 2019 by guesthttp://jb.asm

.org/D

ownloaded from

52 WOLF AND BREY

Succinyl CoA + Glycine

d - Aminolevulinic Acid

Porphobilinogen

Uroporphyrinogen III

-~ --_ 2CH3

HemeSirohydrochlorin2

5CH3

Siroheme CO2 Co + +

aHCOH r

(Sulfite Reductase) CobyriniC Acid

Riboflavin

_.-6NHd4

,--aminopropanol

Cobinamide

ATP

Cobinamide P04

GTP

\ GDP-Cobinamide

DMBI

Vitamin B12

FIG. 1. Cobalamin biosynthetic pathway according to cuunderstanding (2, 10). The region bracketed by cob indicateportion of the pathway affected in Cob mutants; region bracketicbl indicates the portion of the pathway affected in Cbl mutar

exogenous methionine. Because E. coli and S. typhimumetE mutants require exogenous vitamin B12 or methioiit had been thought that these organisms were incapabsynthesizing vitamin B12. Hence, genetic studiercobalamin biosynthesis in general have lagged until recgwhen Roth and co-workers, showing de novo synthescobalamin under anaerobic growth conditions, characteia series of cobalamin auxotrophs of S. typhimuriumThe basis for their work stemmed from the observationmetE mutants of S. typhimurium were able to grow anae

cally in the absence of vitamin B12 or methionine.We initiated a genetic study of the cobalamin biosynt

pathway in B. megaterium to attempt to correlate bioclical steps with specific genetic lesions. B. megateriuespecially amenable to studies on cobalamin biosyntibecause it synthesizes the vitamin aerobically and can Iwith ethanolamine as the sole nitrogen source. In addiB. megaterium can be used for genetic studies since thea known transducing bacteriophage for it and its protopcan be transformed with several plasmids (9, 21). Instudy we isolated 34 mutants deficient in cobalamin bicthesis by their failure to grow on ethanolamine as a sournitrogen in the absence of vitamin B12, deriving auxotrcrecipients for cloning genes which may be rate limititcobalamin biosynthesis, a tactic for the constructionstrain for the industrial production of vitamin B12.describe the experiments to differentiate mutant phenot3

to analyze linkage of mutations with respect to each other,and to define a biosynthetic ordering of mutant phenotypes.

MATERIALS AND METHODS

Bacterial strains and bacteriophage. Bacterial strains usedin this study are shown in Table 1. All B. megaterium strainswere derived from the prototroph B. megaterium ATCC10778. S. typhimurium LT2 strains used as vitamin B12indicator strains were TT8723 (metE205 metP760 ara-9cob-il::TnlO), which utilizes vitamin B12 or cobinamide tosatisfy requirements for methionine, or TT7573 (metE205metP760 ara-9 cob4::TnlO), which can only utilize vitaminB12 for growth in the absence of methionine (15).Bacteriophage MP13, a B. megaterium generalized transduc-ing phage, was obtained from P. Vary (21).Media and culture conditions. Minimal media consisted of

cob 14.0 g of K2HPO4 per liter, 6 g of KH2PO4 per liter, 1 g ofsodium citrate, 1 mM MgCl2, and 0.5% D-glucose with either10 mM NH4SO4 or 0.2% (vol/vol) ethanolamine as thenitrogen source. When ethanolamine was used as the nitro-gen source, the medium was also supplemented with 10 mMNa2SO4. MGY medium contained 4.18 g of MOPS [3-(N-morphilino)ethanesulfonic acid] per liter, 0.375 g of N-tris(hydroxymethyl)methyl glycine (Tricine), 10 mM NH4Cl,0.27 mM K2HPO4, 0.1 mM MgCl2, 5.0 g of D-glucose per

cbl liter, and 10 g of yeast extract (Difco) per liter, adjusted to

pH 7.3 with KOH. All minimal media and MGY mediumwere supplemented with the following micronutrients (perliter): 27 mg of FeCl3 * 6H20, 7.2 mg of ZnSO4- 7H20, 5 mgof MnCl2 * 4H2O, 1.25 mg of CuSO4 * 5H20, and 0.3 mg ofH3BO3. When cells were grown in minimal medium, 1.25 mgof CoCl2 per liter was also added. Concentrations of nutri-

t tional supplements, as required, were 15 nM cyanocobala--s the min (vitamin B12), 20 ,ug of dicyanocobinamide per liter, 5Sed by mg of DMBI per liter, and 20 mg of cysteine per liter. Allnts. incubations were performed at 37°C. Routine culturing of B.

megaterium was done in L broth.Mutagenesis and enrichment for vitamin B12 auxotrophs.

rrium Early-exponential-phase cultures of B. megaterium ATCCnine, 10778 (2 x 107 cells per ml) in L broth were treated withle of N-methyl-N'-nitro-nitrosoguanidine at a final concentrations on of 100 ,ug/ml for 20 min at 37°C, washed, and grown in LB forently 1 h to allow expression of mutations. Cultures were washed,is of and suspended in minimal ethanolamine medium supple-rized mented with 15 nM vitamin B12 and 20 mg of cysteine per(15). liter to allow growth of mutants also blocked in the bipsyn-that thesis of siroheme. After overnight growth, the cells wererobi-

hetic TABLE 1. Bacterial strains used in this studyhem-tm ishesisgrowition,-re is)laststhis

)syn-*ce of)phicng inof aWe

ypes,

Strain aGntp hntpdesignationa GenOtYPe Phenotype

ATCC 10778 Wild type CobI Cbl1GX5101 cob-i Cob IGX5139 cob-39 Cob IGX5141 cob4i Cob IIGX5151 cob-51 Cob IIIGX5137 cob-37 cys-i Cob IV Cys-GX5134 cob-34 Cob VGX5127 cob-27 Cob VIGX5157 cbl-57 Cbl XGX5160 cbl-60 Cbl XIGX5143 cbl43 Cbl XII

a Strain ATCC 10778 was obtained from the American Type CultureCollection, Rockville, Md. All other strains originated from this study.

J. BACTERIOL.

on January 2, 2019 by guesthttp://jb.asm

.org/D

ownloaded from

VITAMIN B12 IN B. MEGATERIUM 53

washed and suspended in minimal ethanolamine withoutvitamin B12 supplementation. After the cell population haddoubled, 1 g of carbenicillin per liter and 2 g of cycloserineper liter were added, and cultures were incubated until celllysis had occurred. The survivors of the antibiotic treatmentwere plated on minimal ethanolamine medium supplementedwith cysteine and vitamin B12. Bacterial colonies werereplica plated on ethanolamine minimal medium withoutsupplementation, and auxotrophs were identified as thoseisolates which required vitamin B12 for growth.

57CoC12 labeling of vitamin B12 and corrinoids. The wild-type strain and each of the vitamin B12 auxotrophs weregrown for 16 h at 37°C in 5 ml of MGY medium containing0.2 ,uCi of 57CoC12 per ml until cultures had attained approx-imately 108 cells per ml (late-logarithmic-growth phase).Corrin compounds labeled with 57Co were extracted byboiling these cultures for 15 minutes in the presence of 0.2 Macetic acid. Corrin compounds were concentrated furtherfrom boiled cultures by extraction into m-cresol-CCL4 (4:1)as described by Bray and Shemin (8). The corrinoid com-pounds were back extracted into 0.2 N NH40H in thepresence of enough 2-butanol to allow phase separation. Thetotal water-soluble corrinoid fraction was subjected tolyophilization and was finally redissolved in 50 ,ul of 0.01 NNH40H containing traces of KCN to convert corrins intotheir dicyano forms. Samples (5 RI) of the extracts wereanalyzed on high-performance silica gel thin-layer chroma-tography plates by developing them in a solvent consisting of2-butanol-NH40H (2:1). The 57Co-labeled species were de-tected by autoradiography.

Bacteriophage transductions. Lysates of several vitaminB12 auxotrophs were prepared b using B. megateriumbacteriophage MP13. Reciprocal tv )-factor transductionalcrosses were carried out as described by Vary et al. (21).Recipient bacteria were grown in 10 ml of SNB broth (21) at37°C to mid-log phase (approximately 5 x 107 cells per ml),washed twice, and suspended in 5 ml of minimal ethanol-amine medium. Minimal ethanolamine plates were spreadwith 2 x 107 to 2 x 108 phage particles inactivated byirradiation for 30 s. After irradiation, plates were spread with3 x 107 recipient bacteria and incubated for 48 h at 37°C. Theresults of two-factor transductional crosses were calculatedas the number of prototrophic transductants per 2 x 107PFU. Cotransduction studies were carried out by transduc-ing Cbl mutants to Cbl+ with MP13 phage grown on mutantsrequiring B12 or cobinamide (Cob-) and selecting for growthon ethanolamine minimal medium supplemented withcobinamide. Among the Cbl+ transductants, those that wereCob- were identified as those colonies unable to grow onethanolamine minimal medium without cobinamide.

Bioassay for vitamin B12. To determine the amount ofvitamin B12 produced by a particular strain, the appropriateB. megaterium strains were cultured in 50 ml of minimalammonia medium until they had reached a density of ap-proximately 7 x 107 cells per ml. The cells were harvested,washed twice in 100 mM potassium phosphate (pH 7.0), andresuspended in a final volume of 2 ml in the same buffer andboiled for 20 min to release vitamin B12 (and other corrinoidcompounds). Portions (20 ,u) of the boiled extracts wereplaced on sterile antibiotic sensitivity disk blanks on minimalammonia plates superimposed with soft agar lawns of ap-proximately 5 x 106 cells of the S. typhimurium vitaminB12-requiring strains TT7573 or TT8723. After 16 h ofincubation at 37°C, the growth zone surrounding the indica-tor strain was measured and compared to a vitamin B12standard curve. The bioassay was sensitive in the range of 1

ng to 10 ,ug of vitamin B12. A standard curve of cobinamidewas identical with the curve for vitamin B12 (data notshown).

In cross-feeding experiments with vitamin B12 auxotrophsof B. megaterium, the same procedure was used to extractcorrinoid compounds from the mutants. The biologic activitywas determined on soft agar lawns of B. megaterium mu-tants containing approximately 5 x 105 cells on minimalethanolamine medium.

Materials, chemicals, and radiochemicals. High-perfor-mance silica gel thin-layer chromatography plates (5 by 10cm) were obtained from EM Science, Gibbstown, N.J.Sterile antibiotic sensitivity disk blanks were obtained fromBBL Microbiology Systems, Cockeysville, Md.

Carrier-free 57CoC12 (4 mCi/mg) was purchased from NewEngland Nuclear Corp., Boston, Mass. Cyanocobalamin,dicyanocobinamide, DMBI, ethanolamine, and other nutri-tional biochemicals were from Sigma Chemical Co., St.Louis, Mo.

RESULTSIsolation and characterization of vitamin B12 auxotrophs.

Because ethanolamine ammonia lyase requires adenosylco-balamin (coenzyme B12) as a cofactor, mutant B. megater-ium cells that are incapable of utilizing ethanolamine as asole source of nitrogen should be defective in production ofethanolamine ammonia lyase, the biosynthesis of vitaminB12, or the adenylation of vitamin B12.Mutagenized cultures of B. megaterium 10778 were en-

riched for vitamin B12 auxotrophs by counterselecting in anethanolamine minimal medium in the presence of carbenicil-lin and cycloserine. Among the survivors, cobalaminauxotrophs were identified as those strains unable to useethanolamine as a nitrogen source except when suppliedwith exogenous vitamin B12. Thirty-four such mutants wereidentified. All of the mutants isolated required vitamin B12for growth on ethanolamine as the sole nitrogen or carbonsource. All strains grew normally on minimal ammoniamedium, except GX5137, which also required cysteine.To distinguish different classes of mutants, the auxotrophs

were tested for their ability to use cobinamide, a vitamin B12biosynthetic precursor, instead of vitamin B12 for growth onethanolamine. Six of the mutants could not use cobinamideand were designated Cbl mutants. Twenty-eight of theauxotrophs were able to use cobinamide or vitamin B12 forgrowth on ethanolamine and were designated Cob mutants.None of the auxotrophs was capable of utilizing DMBIinstead of vitamin B12, indicating that none had a biosyn-thetic block in DMBI synthesis. In addition, no mutantsblocked in sirohydrochlorin synthesis were obtained. Thesemutants would evince a simultaneous requirement for re-duced sulfur (cysteine) and vitamin B12 to grow on ethanol-amine minimal medium, or a requirement for reduced sulfuralone to grow on minimal ammonia medium. All revertantsof GX5137 to vitamin B12 independence still required cyste-ine, indicating an unrelated mutation in cysteine biosynthe-sis in that strain (data not shown).

Cobalt incorporation into corrinoids in Cob and Cbl mu-tants. Phenotypic classification of the auxotrophs based onfeeding of vitamin B12 precursors was limited to the use ofcobinamide, the only commercially available cobalaminbiosynthetic precursor. Hence, further subdivision of theCob and Cbl mutant classes depended on developing criteriaother than growth properties.When B. megaterium was cultured in the presence of

57CoC12, the cobalt was incorporated exclusively into vita-

VOL. 166, 1986

on January 2, 2019 by guesthttp://jb.asm

.org/D

ownloaded from

54 WOLF AND BREY

1 2 3 4 5

9

FIG. 2. Separation of 57Co-labeletained from representative Cob mutar2, GX5101; lane 3, GX5141; lane 4, G:GX5134; lane 7, GX5127. The approp]each mutant is noted below each lan(

min B12 and related corrinoids (dcorrinoid compounds were extrailabeled cultures of the wild typemigrated on thin-layer chromatogB12. To subdivide the Cob and (similar cobalt labeling experimeeach of the auxotrophs (Fig. 2 an

6 7 2, lane 6). Indeed, when a concentrated cell extract of amutant of the Cob V class was made, it was found to contain

-cobinamide enough active cobalamin to allow growth of other B.megaterium vitamin B12 auxotrophs as well as S.typhimurium vitamin B12 auxotrophs (Tables 2 and 3). Al-though the amount of the vitamin that the Cob V strain

v -cobalamin synthesized was approximately 20% of the wild-type level, itrequired exogenous vitamin for growth on ethanolamine.The six mutants with the Cbl phenotype exhibited three

characteristic 57Co-labeling patterns (Fig. 3). Three of thesemutants were designated Cbl X, two were designated CblXI, and one was designated Cbl XII. Two of the Cbl mutantsaccumulated a 57Co-labeled compound which migrated on athin-layer chromatogram in a manner identical to that ofcobinamide. Cbl mutants also accumulated labeled com-pounds which had mobilities identical with compounds thataccumulated in some of the Cob mutants.

3,wPresenceof vitamin B12 in concentrated extracts of Cob andCbl mutants. To confirm that the auxotrophic phenotype was

;;> t_ due to an inability to synthesize active cobalamin, theo a> amount of intracellular vitamin B12 was determined for each8 0 mutant. When concentrated extracts prepared from each of

the Cob and Cbl mutants were used to feed lawns of the S.ed corrinoid compounds ob- typhimurium vitamin B12 auxotrophs TT8723 and TT7573,nts. Lane 1, ATCC 10778; lane no growth was detected, indicating that no active cobalaminX5151; lane 5, GX5137; lane 6, was present in any of the extracts with the exception of theriate phenotypic designation of extract from GX5134 (Cob V) which synthesized reduced

e* amounts of the vitamin. Extracts of the six Cbl mutants wereable to support the growth of S. typhimurium TT8723, a

lata not shown). When the mutant whose auxotrophy can be satisfied by exogenouslycted from cells of 57CoC12- supplied cobinamide. This observation indicates that B., the major corrin species megaterium Cbl auxotrophs are defective in steps whichIrams identically to vitamin occur late in the vitamin B12 biosynthetic pathway, after the2bl mutant classes further, synthesis of cobinamide (Table 2).-nts were performed with Biosynthetic order of Cob and Cbl mutations. It wasid 3). As expected, most of possible to deduce an order for the biosynthetic steps

the strains classified as vitamin B12 auxotrophs did notaccumulate significant amounts of 57Co-labeled vitamin B12.The mutants (e.g., GX5134; see below) that did appear toincorporate 57Co into vitamin B12 produced reduced levels ascompared with the wild type. In addition, many of the Coband Cbl mutants exhibited characteristic, albeit different,labeling patterns of cobalt-containing compounds. Analysisof these labeling patterns resulted in a subclassification ofmutants. Mutants of the Cob phenotype showed at least sixdifferent patterns and were classified as Cob I, II, III, IV, V,and VI (Fig. 2); three different patterns were observed in theCbl mutants, and these were designated Cbl X, XI, and XII(Fig. 3).

Specifically, Cob mutants classified as Cob I did notaccumulate any cobalt-containing corrinoid. Ten mutantstrains were classified as Cob I based on this phenotype. CobII, III, IV, V, and VI strains did not synthesize any vitaminB12 or cobinamide but did accumulate other cobalt-containing corrinoids that were not found in detectablequantities in the wild-type strain. Based on particular label-ing patterns, five separate mutants were classified as Cob II,whereas two were classified as Cob III (e.g., GX5151), withone each classified as Cob IV (GX5137), Cob V (GX5134),and Cob VI (GX5127). The remaining eight mutants havingthe Cob phenotype could not be classified definitively intoany of the above categories and were not further character-ized in this study. A mutant of the Cob V class (GX5134), inaddition to accumulating cobalt-containing corrinoids, alsoappeared to synthesize significant amounts of a compoundthat migrated in a manner similar to that of vitamin B12 (Fig.

1 2 3

-cobinamide

-cobalamin

n..._S _.

_L _. ~L

FIG. 3.from CblGX5143.

Separation of 57Co-labeled corrinoid compounds isolatedmutants. Lane 1, GX5157; lane 2, GX5160; lane 3,

J. BACTERIOL.

on January 2, 2019 by guesthttp://jb.asm

.org/D

ownloaded from

VITAMIN B12 IN B. MEGATERIUM 55

TABLE 2. Presence of vitamin B12 or bioactive corrinoids inconcentrated extracts of Cob and Cbl mutants

Vitamin B12 activityPhenotypic Extract source assayed ona:

classTT8723 TT7573

Cob I GX5101 0 0Cob I GX5139 0 0Cob II GX5141 0 0Cob III GX5151 0 0Cob IV GX5137 0 0Cob V GX5134 0.4 0.3Cob VI GX5127 0 0Cbl X GX5157 3.6 0Cbl XI GX5160 2.0 0Cbl XII GX5143 1.6 0Wild type ATCC 10778 1.6 1.6

a Vitamin B12 activity is expressed in nanograms of vitamin B12 or equiva-lent (in the case of cobinamide) per 10' cells. Cultures were grown to a densityof 7 x 10i cells per ml; 1.6 ng of vitamin B12 is thus equivalent to 12 .g/liter.

affected by some of the Cob and Cbl mutants by cross-feeding experiments in which extracts prepared from themutants strains were tested in all possible combinations forthe ability to support growth (Table 3). The results were thatextracts prepared from each of the Cbl mutants were capableof forming compounds which supplemented each of themutants of the Cob phenotype. None of the extracts pre-pared from the Cob mutants was capable of supportinggrowth of any of the Cbl mutants. Because the extracts ofGX5157 (Cbl X) was capable of feeding both GX5160 (CblXI) and GX5134 (Cbl XII), the lesion in GX5157 may likelyoccur in a biosynthetic step(s) after steps determined bylesions in GX5160 and GX5143. Further support for thisorder is evidenced by the fact that extracts prepared fromneither GX5143 nor GX5160 could support the growth ofGX5157.

Within the Cob phenotypic grouping, each of the membersof the Cob II class, typified by GX5141, was capable ofsupporting growth of all other mutants of the Cob pheno-type, indicating a biosynthetic lesion(s) possibly just beforethe synthesis of cobinamide. Extracts prepared from mu-tants of the Cob III, IV, V, and VI classes were incapable ofcross-feeding each of the others of those classes, but ex-tracts of each of those supported the growth of some of theCob I mutants, exemplified by GX5139. Of 10 mutants of theCob I phenotype, 2 (e.g., GX5101) failed to be supplemented

by extracts of Cob III, IV, V, or VI mutants, while theremaining 8 behaved identically with GX5139 (data notshown). The failure of extracts of Cob III, IV, V, and VImutants to feed GX5101 indicated that the effect of the cob-imutation might be pleiotropic or that GX5101 harbors mul-tiple mutations. As expected, none of the extracts from the10 Cob I strains could support the growth of Cob II, III, IV,V, or VI mutants. Further, none of the Cob I mutants wascapable of cross-feeding any of the other Cob I mutants. Theresults of the cross-feeding experiments permitted the order-ing of mutations or mutant classes (Fig. 4).

Transductional linkage of Cob and Cbl mutations. Todetermine whether the mutations that define the Cob and Cblmutants are genetically linked, reciprocal two-factortransductional crosses were performed on representativestrains of several of the Cob and Cbl subclasses. Thetransductions were performed with the B. megaterium gen-eralized transducing bacteriophage MP13 (21). Phage lysatesprepared from five of the Cob strains and from three of theCbl strains were tested in all possible combinations for theability to transduce recipient strains to vitamin B12 indepen-dence on ethanolamine. The Cob I mutants having thephenotype of GX5139 were not included in this analysis. Theresults of the reciprocal crosses are shown in Table 4. Ingeneral, a low number of prototrophic recombinants indi-cated the probability of few crossovers between defectiveallele(s) and, hence, tight linkage between markers. Becausereversion of each of the recipient strains was less than 0.33x 107, values which are low but not background mightindicate some crossing over between markers or multiplemutations in some of the strains. These data suggest that themutations fall into two linkage groups in that those mutationsthat characterize the Cob class form one linkage group andthe mutations that characterize the Cbl class form another.In independent experiments with GX5141 (Cob II) in twofactor crosses, weak linkage between cob-41 and the othercob and cbl loci was inferred, suggesting that the Cob IIphenotype belongs in yet another linkage group (data notshown).To determine whether the Cob group is linked to the Cbl

group, Cbl- strains were transduced to growth on ethanol-amine in the presence of cobinamide with phage lysatesprepared from Cob- strains. Cbl+ transductants were thenscored for inheritance of the defective Cob allele by growthon ethanolamine in the absence of cobinamide or vitaminB12. Only several Cob- colonies were detected when each ofthe Cob mutants was transduced into each of the Cbl

TABLE 3. Cross-feeding of Cob and Cbl mutantsGrowth stimulation by extract from the following source (phenotype)a:

Indicator strain(Phenotype) GX5101 GX5139 GX5141 GX5151 GX5137 GX5134 GX5127 GX5157 GX5160 GX5143

(Cob I) (Cob I) (Cob II) (Cob III) (Cob IV) (Cob V) (Cob VI) (Cbl X) (Cbl XI) (Cbl XII)

GX5101 (Cob I) - - + - - + - + + +GX5139 (Cob I) - - + + + + + + + +GX5141 (Cob II) - - - - - + - + + +GX5151 (Cob III) - - + - - + - + + +GX5137 (Cob IV) - - + - - + - + + +GX5134 (Cob V) - - + - - + - + + +GX5127 (Cob VI) - - + - - + - + + +GX5157 (Cbl X) - - - - - + - - - -GX5160 (Cbl XI) - - - - - + - + -GX5143 (Cbl XII) - - - - - + - + -

a Extracts of the strains listed horizontally were fed to lawns of the strains listed vertically. Symbols: +, significant growth stimulation of the lawns; -, nogrowth stimulation observed; ±, poor but significant growth observed.

VOL. 166, 1986

on January 2, 2019 by guesthttp://jb.asm

.org/D

ownloaded from

56 WOLF AND BREY

Cobl II, CoblV, CobV, CobV](cob-i9)j

Cobil j---i-----C-CbNH2--5bm{CbIXI,CbIXII C-X---{CbIX{ ---B12

FIG. 4. Positioning of biosynthetic defects in the cobalamin biosynthetic pathway. Abbreviations: SHC, sironydrochlorin; CbNH2,cobinamide. Phenotypic classes or particular mutations within a class are surrounded by brackets indicating a single grouping within thepathway.

mutants (Table 5). This result indicates a weak linkagebetween the two mutant classes.

DISCUSSIONThe work that we have reported in this paper should

provide a rational framework for further studies incobalamin biosynthesis and regulation. The biosyntheticpathway for cobalamin surely involves the construction ofthe most complex biological molecules other than proteinsand involves an unknown number of gene products, themethod of regulation of which is also unknown. The isolationof mutants provides a first discreet step in understandinggenetic regulation of cobalamin biosynthesis in an aerobicorganism in which regulation of trace quantity synthesis ofvitamin B12 might have evolved differently from the pathwayin facultative organisms. The isolation of mutants will even-tually permit the correlation of a particular enzyme with adefined biosynthetic conversion and will facilitate the clon-ing and further definition of the genes.The mutations that result in the Cob or Cbl phenotypes

cannot be assigned to defects in any known biosyntheticenzymes. However, some genetic defects could be assignedto narrow regions of the cobalamin biosynthetic pathwaybased on the combined results of cross-feeding and cobalt-labeling experiments. For example, mutants of the Cob Iclass did not accumulate any cobalt-containing corrinoids;hence, some of those mutants were probably defective in theconversion of sirohydrochlorin to cobyrinic acid, at pointsbefore the insertion of cobalt into the ring structure. Because

none of the Cob I mutants required reduced sulfur for growththey must be able to synthesize sirohydrochlorin and hencesiroheme, the prosthetic group for sulfite and nitritereductases (14, 20; Fig. 1). The exact point at which cobalt isinserted into the ring structure is unknown. Thus, membersof the Cob I class could be deficient in any of a number ofbiosynthetic steps in the conversion of sirohydrochlorin tocobyrinic acid. Cobalt insertion could occur at any point inthat series of steps, possibly by enzymatic means; it couldalso occur nonspecifically or nonenzymatically at multiplepoints. Failure to accumulate cobalt ion could also result inthe inability to synthesize cobalt-containing compounds.Among the 20 Cob mutants, six different cobalt-labelingpatterns were obtained; three different patterns were ob-tained in the Cbl mutants. An interpretation for these resultsis that lesions in particular biosynthetic enzymes, presum-ably caused by mutations in separate genes, caused accumu-lation of pathway intermediates before the lesions. Each ofthe 57Co-labeled compounds could be intermediates in thecobalamin pathway, but some of the compounds could beside products.

Biosynthetic defects present in Cob II, III, IV, V, and VImutants must occur at points after the insertion of cobaltand, thus, after those of the Cob I mutants. Defects in thefive Cob II mutants could be positioned after thebiosynthetic defects of the Cob III, IV, V, and VI mutantsbecause extracts prepared from Cob II mutants were able tofeed all other Cob mutants. The Cob III, IV, V, and VImutants formed a group in that none of those four mutants

TABLE 4. Transductional linkage of B. megaterium cob and cbl mutations

No. of transductants obtained with the following donor (phenotype)a:Recipient(genotype) GX5157 GX5160 GX5143 GX5137 GX5134 GX5151 GX5127 GX5101 10778

(Cbl X) (Cbl XI) (Cbl XII) (Cob IV) (Cob V) (Cob III) (Cob VI) (Cob I) (WT)

GX5157 (cbl-57) 0 5 1 68 108 45 15 39 16GX5160 (cbl-60) 0 0 0 28 840 52 ob 783 136GX5143 (cbl43) 1 2 0 16 51 106 7 81 30

GX5137 (cob-37) 10 18 36 0 6 1 2 9 11GX5134 (cob-34) 3 12 140 0 0 3 5 12 38GX5151 (cob-SI) 9 39 48 0 9 0 0 8 39GX5127 (cob-27) 6 20 30 1 3 1 0 3 9GX5101 (cob-i) 30 153 126 6 41 12 7 0 50

a Values represent the number of prototrophic transductants obtained per 2 x 107 PFU. In each of these crosses, a single lysate of each of the donor strainswas used to transduce each of the recipients; values in a column are therefore comparable. The values for the wild-type (10778) control were compiled from severalexperiments with different phage lysates. Reversion of each of the recipients was <0.33 x 107. WT, Wild type.

b Although no prototrophic recombinants were observed when a lysate of GX5127 was transduced into GX5160, the reciprocal cross indicated that GX5127belonged in this linkage group.

J. BACTERIOL.

on January 2, 2019 by guesthttp://jb.asm

.org/D

ownloaded from

VITAMIN B12 IN B. MEGATERIUM 57

TABLE 5. Cotransduction of cob and cbl mutations

No. of Cob- colonies/no. of Cbl+ transductants, obtained with the following donor (phenotype)a:Recipient

(phenotype) GX5141 GX5134 GX5127 GX5137 GX5101 GX5151(Cob II) (Cob VI) (Cob V) (Cob IV) (Cob I) (Cob III)

GX5157 (Cbl X) 0/53 (53) 0/117 (736) 0/93 (93) 0/111 (628) 2/107 (1,176) 0/125 (520)GX5160 (Cbl XI) 1/134 (134) 0/216 (216) 0/114 (114) 0/100 (520) 0/64 (239) 0/100 (640)GX5143 (Cbl XII) 0/207 (207) 1/143 (143) 0/45 (45) 0/102 (180) 0/40 (56) 0/110 (256)

a Values represent the number ofCob- colonies among Cbl+ transductants. Values in parentheses represent the total number of Cbl+ colonies on the selectionplate. Uninfected cell controls for each of the recipients strains yielded no Cbl+ revertants; phage controls also yielded no Cbl+ colonies. The input phage titerin each of these crosses was approximately 2 x 108 PFU per 3 x 10' recipient cells (per plate).

could cross-feed each other. This grouping of mutants waspositioned as it was in Fig. 4 because extracts prepared fromeach of those strains could feed 8 of 10 Cob I mutants. Thisresult indicated that the Cob I class consisted of twosubclasses; it also indicated that at least one compoundpresent in each of the extracts of Cob III, IV, V, and VImutants could be taken up by GX5139 and other Cob Imutants of its ilk, suggesting that a failure of those mutantsto cross-feed each other was not due to impermeability ofthose corrinoids.At least one compound present in extracts of Cob III, IV,

V, and VI mutants can be converted into cobalamin in vivoby Cob I mutants similar to GX5139, bypassing the Cob Ibiosynthetic defects, whereas Cob I mutants similar toGX5101 cannot convert those compounds into cobalamin.This suggests that the cob-i mutation(s) may affect a numberof biosynthetic steps in that GX5101 does not synthesizecobalt-containing corrinoids and also behaves as though itwere a member of the Cob III, IV, V, and VI group.Combined results of cobalt-labeling and' cross-feeding

experiments revealed that most of the mutants which wereincapable of growth on ethanolamire failed to synthesizecobalamin. However, one particular mutant, GX5134 (CobV), synthesized a compound which migrated identically withcobalamin and formed bioactive vitamin at 20% the level ofthe wild type, yet this strain required exogenous vitamin B12for growth on ethanolamine. The material from GX5134which migrated as cobalamin may not be authenticcobalamin and may differ also in its ability to serve as acofactor for ethanolamine ammonia-lyase. If the compoundis authentic vitamin B12, the rate of its biosynthesis in themutant may be much lower than in the wild type, resulting inlevels of cobalamin in the mutant that are too low to act ascofactor during logarithmic growth on ethanolamine.The cob-37, cob-34, cob-SI, cob-27, and cob-i mutations

appeared to be genetically linked, based on transductiondata. However, in the absence of suitable outside geneticmarkers, quantitative transduction could only point to trendsin linkage; actual linkage between markers or their ordercould not be determined. The trend in the linkage dataindicated that the cbl-43, cbl-57, and cbl-60 alleles werelinked to each other and not tightly to other mapped coballeles.The major corrinoid compound that accumulated in wild-

type cells was cobalamin; none of the cobalt-labeled inter-mediates that accumulated in the Cob or Cbl class mutantswas detectable in the wild type. Loss of vitamin B12 biosyn-thesis in some of the Cbl mutants sometimes resulted in atwo- to threefold increase over the wild-type levels ofcorrinoids which could support growth of an S. typhimuriumindicator strain, suggesting that cobalamin or a closelyrelated compound is involved in regulation of the pathway.Points at which some compounds accumulated in the mu-

tants may suggest which steps are rate limiting in biosynthe-sis.An obvious strategy for overcoming the regulation of

rate-limiting steps is to increase the gene dosage of therate-limiting enzymes by cloning. Recent experiments sug-gest that some of the auxotrophs can be used to isolate DNAfragments that complement the mutations (unpublisheddata). Given the apparently tight genetic linkage betweenmutations that characterize the Cob mutants and the Cblmutants, a relative small number of cloned fragments couldencompass many of the biosynthetic genes. Hence, thisapproach could lead to a better understanding of the regula-tion of a biosynthetic pathway which is composed ofperhaps20 or more enzymes and genes that encode them.

ACKNOWLEDGMENTSWe thank all ofthe members of the former Microbial Biochemistry

Department at fGenex Corp. for their advice; in particular, we thankJim Jacobson, Ed Stellwag, and Mike Brenner. We thank HarryHogenkamp for the gift of analogs and John Roth for strains andinformation before publication.

This work was conducted at Genex Corp. as part of a proprietaryproject to develop a Vitamin B12 production strain.

LITERATURE CITED1. Barker, H. A. 1972. Corrinoid-dependent enzyme reactions.

Annu. Rev. Biochem. 41:55-90.2. Battersby, A. R., and E. McDonald. 1982. Biosynthesis of the

corrin macrocycle, p. 107-144. In D. Dolphin (ed.), B12. JohnWiley & Sons, Inc., New York.

3. Blackwell, C. M., F. A. Scarlett, and J. M. Turner. 1977.Microbial metabolism of amino alcohols. Control of formationand stability of partially purified ethanolamine ammonia-lyase inEscherichia coli. J. Gen. Microbiol. 98:133-139.

4. Blackwell, C. M., andJ. M. Turner. 1978. Microbial metabolismof amino alcohols. Purification and properties of coenzymeB12-dependent ethanolamine ammonia-lyase of Escherichia coli.Biochem. J. 175:555-567.

5. Blackwell, C. M., and J. M. Turner. 1978. Microbial metabolismof anilrio alcohols. Formation of coenzyme B12-dependent eth-anolamine ammonia-lyase and its concerted induction in Esch-erichia coli. Biochem. J. 176:751-757.

6. Bradbeer, C. 1965. The clostridial fermentation of choline andethanolamine. I. Preparation and properties of cell free extracts.J. Biol. Chem. 240:4669-4674.

7. Bradbeer, C. 1965. The clostridial fermentation of choline andethanolamine. II. Requirement for a cobamide coenzyme by anethanolamine deaminase. J. Biol. Chem. 240:4675-4681.

8. Bray, R. C., and D. Shemin. 1963. On the biosynthesis ofvitamin B12. The derivation of the corrin structure from 8-aminolevulinic acid and the methyl group of methionine. J. Biol.Chem. 238:1501-1508.

9. Brown, B. J., and B. C. Carlton. 1980. Plasmnid-mediatedtransformation in Bacillus megaterium. J. Bacteriol. 142:508-512.

10. Chang, G. W., and J. T. Chang. 1975. Evidence for the

VOL. 166, 1986

on January 2, 2019 by guesthttp://jb.asm

.org/D

ownloaded from

58 WOLF AND BREY

1l12-dependent enzyme ethanolamine deaminase in Salmonella.Nature (London) 254:150-151.

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

12. Hogenkainp, H. R. C. 1974. The chemical synthesis of analoguesof adenosylcobalamin. Biochemistry 13:2736-2740.

13. 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. John Wiley & Sons, Inc., New York.

14. Jackson, R. H., A. Cornish-Bowden, and J. A. Cole. 1981.Prosthetic groups of the NADH-dependent nitrite reductasefrom Escherichia coli Biochem. J. 193:861-867.

15. Jeter, R. M., B. M. Oliveria, and J. R. Roth. 1984. Salmonellatyphimurium synthesizes cobalamin (Vitamin B12) de novounder anaerobic growth tonditions. J. Bacteriol. 159:206-213.

16. Kaplan, B. H., and E. R. Stadtman. 1968. Ethanolaminedeaminase, a cobamide coenzyme dependent enzyme. I. Puri-fication, assay, and properties of the enzyme. J. Biol. Chem.

243:1787-1793.17. Muller, G., R. Deeg, K. D. Gneuss, G. Gunzer, and H.-P.

Kriemler. 1979. On the methylation process in cobyrinic acidbiosynthesis, p. 279-291. In B. Zagalak and W. Friederich,(ed.), Vitamin B12. Walter de Gruyter and Co., New York.

18. Sando, G. N., M. E. Grant, and H. P. C. Hogenkamp. 1976. Theinteractions of adeninylalkylcobalamins with ribonucleotidereductase. Biochim. Biophys. Acta 428:228-232.

19. Scarlett, F. A., and J. M. Turner. 1976. Microbial metabolism ofamino alcohols. Ethanolamine catabolism mediated by coen-zyme B12-dependent ethanolamine ammonia-lyase in Esche-richia coli and Klebsiella aerogenes. J. Gen. Microbiol.95:175-176.

20. Siegel, L. M., P. S. Davis, and M. J. Murphy. 1977. Incorpora-tion of methionine-derived methyl groups into sirohaem byEscherichia coli. Biochem. J. 167:669-674.

21. Vary, P. S., J. C. Garbe, M. Franzen, and E. W. Frampton.1982. MP13, a generalized transducing bacteriophage for Bacil-lus megaterium. J. Bacteriol. 149:1112-1119.

J. BACTERIOL.

on January 2, 2019 by guesthttp://jb.asm

.org/D

ownloaded from