Methods of increasing the production of cobalamins using cob gene expression

The present invention relates to new polypeptides involved in the biosynthesis of cobalamins and/or cobamides, and especially of coenzyme B₁₂. It also relates to the genetic material responsible for the expression of these polypeptides, as well as to a method by means of which they may be prepared. It relates, lastly, to a method for amplification of the production of cobalamins, and more especially of coenzyme B₁₂, by recombinant DNA techniques.

Vitamin B₁₂belongs to the B group of vitamins. It is a water-soluble vitamin which has been identified as the factor enabling patients suffering from pernicious anaemia to be treated. It is generally prescribed to stimulate haematopoiesis in fatigue subjects, but it is also used in many other cases comprising liver disorders and nervous deficiencies or as an appetite stimulant or an active principle with tonic activity, as well as in dermatology (Berck, 1982, Fraser et al., 1983). In the industrial rearing of non-ruminant animals, the feed being essentially based on proteins of vegetable origin, it is necessary to incorporate vitamin B₁₂in the feed rations in amounts of 10 to 15 mg per tonne of feed (Barrere et al., 1981).

Vitamin B₁₂belongs to a class of molecules known as cobalamins, the structure of which is presented in FIG. 1. Cobamides differ from cobalamins in the base of the lower nucleotide, which is no longer 5,6-dimethylbenzimidazole but another base, e.g. 5-hydroxybenzimidazole for vitamin B₁₂-factor III synthesised, inter alia, by Clostridium thermoaceticum and Methanosarcina barkeri (Iron et al., 1984). These structural similarities explain the fact that the metabolic pathways of biosynthesis of cobalamins and cobamides are, for the most part, shared.

Cobalamins are synthesised almost exclusively by bacteria, according to a complex and still poorly understood process which may be divided into four steps (FIG. 2):

i) synthesis of uroporphyrinogen III (or uro'gen III), then

ii) conversion of uro'gen III to cobyrinic acid, followed by

iii) conversion of the latter to cobinamide, and

iv) construction of the lower nucleotide loop with incorporation of the particular base (5,6-dimethylbenzimidazole in the case of cobalamins).

For coenzyme B₁₂, it is probable that the addition of the 5′-deoxyadenosyl group occurs shortly after the corrin ring-system is synthesised (Huennekens et al., 1982).

In the case of cobamides, only the step of synthesis and incorporation of the lower base is different.

The first part of the biosynthesis of cobalamins is very well known, since it is common to that of haemes as well as to that of chlorophylls (Battersby et al., 1980). It involves, successively, δ-aminolevulinate synthase (EC 2.3.137), δ-aminolevulinate dehydrase (EC 4.2.1.24), porphobilinogen deaminase (EC 4.3.1.8) and uro'gen III cosynthase (EC 4.2.1.75), which convert succinyl-CoA and glycine to uro'gen III. However, the first step takes place in some organisms [e.g. E. coli (Avissar et al., 1989) and in methanogenic bacteria (Kannangara et al., 1989), for example) by the conversion by means of a multi-enzyme complex of glutamic acid to δ-aminolevulinic acid.

Between uro'gen III and cobyrinic acid, only three intermediate derivatives have been purified to date; they are the factors FI, FII and FIII, which are oxidation products, respectively, of the three intermediates precorrin-1, precorrin-2 and precorrin-3, which correspond to the mono-, di- and trimethylated derivatives of uro'gen III (FIG. 3); these intermediates are obtained by successive transfers of methyl groups from SAM (S-adenosyl-L-methionine) to uro'gen III at positions C-2, C-7 and C-20, respectively. The other reactions which take place to give cobyrinic acid are, apart from five further transfers of methyl groups from SAM at C-17, C-12, C-1, C-15 and C-5, elimination of the carbon at C-20, decarboxylation at C-12 and insertion of a cobalt atom (FIG. 4). These biosynthetic steps have been deduced from experiments performed in vitro on acellular extracts of Propionibacterium shermanii or of Clostridium tetanomorphum. In these extracts, cobyrinic acid is obtained by conversion of uro'gen III after incubation under suitable anaerobic conditions (Batterby et al., 1982). No intermediate between precorrin-3 and cobyrinic acid capable of being converted to corrinoids by subsequent incubation with extracts of cobalamin-producing bacteria has been isolated to date. The difficulty of isolating and identifying these intermediates is linked to

i) their great instability,

ii) their sensitivity to oxygen, and

iii) their low level of accumulation in vivo.

In this part of the pathway, only one enzyme of Pseudomonas denitrificans has been purified and studied; it is SAM:uro'gen III methyltransferase (Blanche et al., 1989), referred to as SUMT.

Between cobyrinic acid and cobinamide, the following reactions are performed:

i) addition of the 5′-deoxyadenosyl group (if coenzyme B₁₂is the compound to be synthesised),

ii) amidation of six of the seven carboxyl functions by addition of amine groups, and

iii) amidation of the last carboxyl function (propionic acid chain of pyrrole ring D) by addition of (R)-1-amino-2-propanol (FIG. 2).

Whether there was really an order in the amidations was not elucidated (Herbert et al., 1970). Lastly, no assay of activity in this part of the pathway has been described, except as regards the addition of the 5′-deoxyadenosyl group (Huennekens et al., 1982).

The final step of the biosynthesis of a cobalamin, e.g. coenzyme B₁₂, comprises four successive phases described in FIG. 5 (Huennekens et al., 1982), namely:

i) phosphorylation of the hydroxyl group of the aminopropanol residue of cobinamide to cobinamide phosphate, then

ii) addition of a guanosine diphosphate by reaction with guanosine 5′-triphosphate; the compound obtained is GDP-cobinamide (Friedmann, 1975), which

iii) reacts with 5,6-dimethylbenzimidazole, itself synthesised from riboflavin, to give adenosylcobalamin 5′-phosphate (Friedmann et al., 1968), which

iv) on dephosphorylation leads to coenzyme B₁₂(Schneider and Friedmann, 1972).

Among bacteria capable of producing cobalamins, the following may be mentioned in particular:

Agrobacterium tumefaciens

Agrobacterium radiobacter

Bacillus megaterium

Clostridium sticklandii

Clostridium tetanomorphum

Clostridium thermoaceticum

Corynebacterium XG

Eubacterium limosum

Methanobacterium arbophilicum

Methanobacterium ivanovii

Methanobacterium ruminantium

Methanobacterium thermoautotrophicum

Methanosarcina barkeri

Pronionobacterium shermanii

Protaminobacter ruber

Pseudomonas denitrificans

Pseudomonas putida

Rhizobium meliloti

Rhodopseudomonas sphaeroides

Salmonella typhimurium

Spirulina platensis

Streptomvces antibioticus

Streptomyces aureofaciens

Streptomyces ariseus

Streptomvces olivaceus

At the industrial level, as a result of the great complexity of the biosynthetic mechanisms, the production of cobalamins, and especially of vitamin B₁₂, is exclusively microbiological. It is carried out by large-volume cultures of the bacteria Pseudomonas denitrificans, Propionibacterium shermanii and Propionibacterium freudenreichii (Florent, 1986). The strains used for the industrial production are derived from wild-type strains; they may have undergone a large number of cycles of random mutation and then of selection of improved clones for the production of cobalamins (Florent, 1986). The mutations are obtained by mutagenesis with mutagenic agents or by physical treatments such as treatments with ultraviolet rays (Barrere et al., 1981). By this empirical method, random mutations are obtained and improve the production of cobalamins. For example, it is described that, from the original strain of Pseudomonas denitrificans initially isolated by Miller and Rosenblum (1960, U.S. Pat. No. 2,938,822), the production of this microorganism was gradually increased in the space of ten years, by the techniques mentioned above, from 0.6 mg/l to 60 mg/l (Florent, 1986). For bacteria of the genus Propionibacterium [Pronionibacterium shermanii (ATCC 13673) and freudenreichii (ATCC 6207)), the same production values appear to be described in the literature; e.g. a production of 65 mg/l has been described (European Patent 87,920). However, no screen has yet been described enabling either mutants overproductive of cobalamins or mutants markedly improved in their production of cobalamins to be readily selected or identified.

At the genetic level, little work has been performed to date. The cloning of cob genes (coding for enzymes involved in the biosynthetic process) has been described in Bacillus megaterium (Brey et al., 1986). Eleven complementation groups have been identified by complementation of cob mutants of Bacillus megaterium with plasmids carrying different fragments of Bacillus megaterium DNA. These genes are grouped on the same locus, carried by a 12-kb fragment.

Studies have also been carried out on the cob genes of Salmonella typhimurium. Although the cloning of these has not been described, it has been shown that almost all the genes for cobalamin biosynthesis are grouped together between minutes 40 and 42 of the chromosome (Jeter and Roth, 1987). Only the cysG locus, which must permit the conversion of uro'gen III to precorrin-2, does not form part of this group of genes. However, the activity encoded by this locus and also its biochemical properties have not been described.

In addition, some phenotypes have been associated with cob mutations. In Salmonella typhimurium and in Bacillus megaterium, cob mutants no longer show growth on minimum medium with ethanolamine as a carbon source or as a nitrogen source (Roof and Roth, 1988). This is due to the fact that an enzyme of ethanolamine catabolism, ethanolamine ammonia-lyase (EC 4.3.1.7), has coenzyme B₁₂as a cofactor; the cob mutants no longer synthesise coenzyme B₁₂, and they can no longer grow with ethanolamine as a carbon source and/or as a nitrogen source. metE mutants of Salmonella typhimurium retain only a methylcobalamin-dependent homocysteine methyltransferase (EC 2.1.1.13). cob mutants of Salmonella typhimurium metE are auxotrophic for methionine (Jeter et al., 1984).

In Pseudomonas denitrificans and Agrobacterium tumefaciens, phenotypes associated with a total deficiency of cobalamin synthesis have not been described to date.

Finally, work on Pseudomosas denitrificans (Cameron et al., 1989) has led to the cloning of DNA fragments carrying cob genes of this bacterium. These are distributed in four complementation groups carried by at least 30 kb of DNA. At least fourteen complementation groups have been identified by heterologous complementation of cob mutants of Agrobacterium tumefaciens and of Pseudomonas outida with DNA fragments of Pseudomonas denitrificans carrying cob genes.

However, hitherto, none of these genes has been purified, and no nucleotide sequence has been described. Similarly, no protein identification nor any catalytic function attributed to the product of these genes has been described. Furthermore, no improvement in production of cobalamins by recombinant DNA techniques could be obtained. The amplification of cob genes of Bacillus megaterium does not bring about, in the strain from which they have been cloned, an improvement in production of cobalamins (Brey et al., 1986). In Salmonella tyrhimurium, physiological studies have been carried out in order to determine conditions under which a strong transcription of the cob genes studied was observed (Escalante and Roth, 1987). Under these conditions, there is no improvement in the production of cobalamins, although genes of the biosynthetic pathway are more expressed than under standard culture conditions.

The present invention results from the precise identification of DNA sequences coding for polypeptides involved in the biosynthesis of cobalamins and/or cobamides. A subject of the invention hence relates to the DNA sequences coding for the polypeptides involved in the biosynthesis of cobalamines and/or cobamides. More especially, the subject of the invention is the cobA, cobB, cobC, cobD, cobE, cobF, cobG, cobH, cobI, cobJ, cobK, cobL, cobM, cobN, cobO, cobP, cobQ, cobS, cobT, cobU, cobV, cobW, cobX and corA genes, any DNA sequence homologous with these genes resulting from the degeneracy of the genetic code, and also DNA sequences, of any origin (natural, synthetic, recombinant), which hybridise and/or which display significant homologies with these sequences or with fragments of the latter, and which code for polypeptides involved in the biosynthesis of cobalamins and/or cobamides. The subject of the invention is also the genes containing these DNA sequences.

The DNA sequences according to the present invention were isolated from an industrial strain, Pseudomonas denitrificans SC510, derived from strain MB580 (U.S. Pat. No. 3,018,225), by complementation of cob mutants of A. tumefaciens and P. putida; and of Methanobacterium ivanovii. The clones obtained could be analysed precisely, in particular by mapping using insertions of a derivative of transposon Tn5. These genetic studies have enabled the cob or cor genes to be localised on the restriction map and their sequencing to be carried out. An analysis of the open reading frames then enabled the coding regions of these DNA fragments to be demonstrated.

The subject of the present invention is also the use of these nucleotide sequences for cloning the cob genes of other bacteria. In effect, it is known that, for proteins catalysing the same activities, sequences are conserved, the divergence being the evolutionary divergence (Wein-Hsiung et al., 1985). It is shown in the present invention that there is a significant homology between the nucleotide sequences of different microorganisms coding for polypeptides involved in the biosynthesis of cobalamins and/or cobamides. The differences which are seen result from the evolutionary degeneracy, and from the degeneracy of the genetic code which is linked to the percentage of GC in the genome of the microorganism studied (Wein-Hsiung et al., 1985).

According to the present invention, a probe may be made with one or more DNA sequences of Pseudomonas denitrificans in particular, or with fragments of these, or with similar sequences displaying a specific degree of degeneracy in respect of the use of the codons and the percentage of GC in the DNA of the bacterium which it is desired to study. Under these conditions, it is possible to detect a specific hybridisation signal between the probe and fragments of genomic DNA of the bacterium studied; this specific hybridisation signal corresponds to the hybridisation of the probe with the isofunctional cob genes of the bacterium. The cob genes as well as their products may then be isolated, purified and characterised. The invention thus provides a means enabling access to be gained, by hybridisation, to the nucleotide sequences and the polypeptides involved in the biosynthesis of cobalamins and/or cobamides of any microorganism.

The subject of the present invention is also a recombinant DNA containing at least one DNA sequence coding for a polypeptide involved in the biosynthesis of cobalamins and/or cobamides, and in particular a recombinant DNA in which the said sequence or sequences are placed under the control of expression signals.

In this connection, promoter regions may, in particular, be positioned at the 5′ end of the DNA sequence. Such regions may be homologous or heterologous to the DNA sequence. In particular, strong bacterial promoters such as the promoter of the tryptophan operon Ptrp or of the lactose operon Plac of E. coli, the leftward or rightward promoter of bacteriophage lambda, the strong promoters of phages of bacteria such as Corynebacteria, the functional promoters in Gram-negative bacteria such as the Ptac promoter of E. coli, the PxylS promoter of the xylene catabolism genes of the TOL plasmid and the amylase promoter of Bacillus subtilis Pamy may be used. Promoters derived from glycolytic genes of yeasts may also be mentioned, such as the promoters of the genes coding for phosphoglycerate kinase, glyceraldehyde-3-phosphate dehydrogenase, lactase or enolase, which may be used when the recombinant DNA is to be introduced into a eukaryotic host. A ribosome binding site will also be positioned at the 5′ end of the DNA sequence, and it may be homologous or heterologous, such as the ribosome binding site of the cII gene of bacteriophage lambda.

Signals necessary to transcription termination may be placed at the 3′ end of the DNA sequence.

The recombinant DNA according to the present invention may then be introduced directly into a host cell compatible with the chosen expression signals, or be cloned into a plasmid vector to enable the DNA sequence in question to be introduced in a stable manner into the host cell.

Another subject of the invention relates to the plasmids thereby obtained, containing a DNA sequence coding for a polypeptide involved in the biosynthesis of cobalamins and/or cobamides. More specifically, these plasmids also contain a functional replication system and a selectable marker.

The subject of the invention is also the host cells into which one or more DNA sequences as defined above, or a plasmid as defined hereinbefore, has/have been introduced.

Another subject of the invention relates to a method for production of polypeptides involved in the biosynthesis of cobalamins and/or cobamides. According to this method, a host cell is transformed with a DNA sequence as described above, this transformed cell is cultured under conditions for expression of the said sequence and the polypeptides produced are then recovered.

The host cells which may be used for this purpose are either prokaryotes or eukaryotes, animal cells or plant cells. Preferably, they will be chosen from bacteria, and especially bacteria of the genus E. coli, P. denitrificans, A. tumefaciens or R. meliloti.

Another use of the DNA sequences according to the present invention lies in a method for amplification of the production of cobalamins and/or cobamides, by recombinant DNA techniques. In effect, if the limitation of the metabolic flux of the biosynthesis of cobalamins and/or cobamides is due to a limitation in the activity of an enzyme in the biosynthetic pathway, an increase in this activity by increasing the expression of this same enzyme using recombinant DNA techniques (gene amplification, substitution of the transcription/translation signals with more effective signals, etc.) will lead to an increase in the biosynthesis of cobalamins and/or cobamides. It is also possible that the limitation of the production of cobalamins and/or cobamides results from a biochemical regulation. In this case, the cob gene or genes corresponding to the regulated enzyme may be specifically mutagenised in vitro in order to obtain mutated genes whose products will have lost the regulation mechanisms impeding an improvement in the production.

The method according to the present invention consists in transforming a microorganism productive of cobalamins and/or cobamides, or only potentially productive of these compounds (i.e. deficient in one or more steps of the biosynthesis), with a DNA sequence as defined above, then in culturing this microorganism under conditions for expression of the said sequence and for synthesis of cobalamins and/or cobamides, and lastly in recovering the cobalamins and/or cobamides produced. Such a method is applicable, in particular, to all the productive microorganisms mentioned on pages 5 and 6, and more specifically to microorganisms of the genus P. denitrificans, Rhizobium meliloti, or Agrobacterium tumefaciens. In a preferred embodiment, the microorganism is P. denitrificans, and especially strain SC510. As regards potentially productive microorganisms, the DNA sequences used will be those corresponding to the steps of the biosynthesis which the microorganism cannot carry out.

Using the present invention, and by the various stragegies described above, an improvement in the production of cobalamins and/or cobamides may be obtained for any microorganism productive or potentially productive of cobalamins and/or cobamides. It will suffice to culture this recombinant microorganism under suitable conditions for the production of cobalamins and for the expression of the DNA sequences introduced. This culturing may be carried out batchwise or alternatively in continuous fashion, and the purification of the cobalamins may be carried out by the methods already used industrially (Florent, 1986). These methods comprise, inter alia:

i) solubilisation of the cobalamins and their conversion to their cyano form (e.g. by heat treatment of the fermentation must, with potassium cyanide in the presence of sodium nitrite), then

ii) purification of the cyanocobalamins in various steps which can be, e.g.

a) adsorption on different substrates such as Amberlite IRC-50, Dowex 1×2 or Amberlite XAD-2, followed by an elution with a water/alcohol or water/phenol mixture, then
b) extraction in an organic solvent, and lastly
c) precipitation or crystallisation from the organic phase, either by the addition of reagents or dilution in a suitable solvent, or by evaporation.

The present invention shows, furthermore, that it is possible by recombinant DNA techniques to improve the cobalamin production of a bacterium productive of cobalamins by cumulating improvements. This amounts to obtaining a first improvement as described above, and then in improving this improvement, still using recombinant DNA techniques, i.e., e.g. by amplifying genes for cobalamin biosynthesis.

Another subject of the present invention relates to the polypeptides involved in the biosynthesis of cobalamins and/or cobamides. In particular, the subject of the present invention is all polypeptides, or derivatives or fragments of these polypeptides, which are encoded by the DNA sequences described above, and which are involved in the pathway of biosynthesis of cobalamins and/or cobamides. The amino acid sequence of these polypeptides is described, as well as some of their physicochemical properties. An enzymatic activity or specific properties have also been associated with each of them.

In this connection, the subject of the invention is the polypeptides participating in the conversion of precorrin-3 to cobyrinic acid a,c-diamide, and more especially in the transfer of a methyl group from SAM to positions C-1, C-5, C-11, C-15 and C-17.

The subject of the invention is also the polypeptides:

participating in the conversion of cobyric acid to cobinamide, or

possessing an S-adenosyl-L-methionine:precorrin-2 methyltransferase (SP2MT) activity, or

possessing a cobyrinic and/or hydrogenobyrinic acid a,c-diamide synthase activity, or

possessing a precorrin-8x mutase activity, or

possessing a nicotinate-nucleotide: dimethylbenzimidazole phosphoribosyltransferase activity, or

possessing a cobalamin-5′-phosphate synthase activity, or

possessing a cobyric acid synthase activity, or

possessing a cob(I)alamin adenosyl-transferase activity, or

possessing a precorrin-6× reductase activity, or

participating in the conversion of hydrogenobyrinic acid a,c-diamide to cobyrinic acid a,c-diamide.

Advantageously, the subject of the invention is a polypeptide chosen from the COBA, COBB, COBC, COBD, COBE, COBF, COBG, COBH, COBI, COBJ, COBK, COBL, COBM, COBN, COBO, COBP, COBQ, COBS, COBT, COBU, COBV, COBW, COBX and CORA proteins presented in FIGS. 15, 16, 40, 41 and 47.

Furthermore, the use of the hybridisation probes described above makes it possible, from genes isolated in other microorganisms, to characterise and isolate the isofunctional polypeptides of other microorganisms. In this manner, the present invention shows that the sequence of a COB protein of Pseudomonas denitrificans is significantly homologous with the protein sequences of other microorganisms displaying the same type of activity. Between these COB proteins catalysing the same reaction in different microorganisms, only the evolutionary distances have introduced variations (Wein-Hsiung et al., 1985). The subject of the present invention is also these isofunctional polypeptides.

The assignment of a particular enzymatic activity is the result of an analysis which may be performed according to various strategies. In particular, in vitro affinity studies with respect to SAM (S-adenosyl-L-methionine) make it possible to assign a methyl transferase activity to a protein capable of binding SAM, and hence to assign its involvement in one of the steps of transfer of methyl groups which occur between uro'gen III and cobyrinic acid. Another means of assessing the activity of these polypeptides consists in assaying the intermediates in the pathway of biosynthesis of cobalamins which are accumulated in mutants incapable of expressing these polypeptides (identified by complementation experiments). These analyses enable it to be deduced that the polypeptide in question has the accumulated intermediate as its substrate, thereby enabling its activity in the biosynthetic pathway to be situated and defined. The present invention also describes a method for assaying the enzymatic activities of the biosynthetic pathway, applicable to any strain productive of cobalamins and/or cobamides. These assays enable the enzymatic activity assayed to be purified from any strain productive of these compounds. From this purified activity, the NH₂-terminal sequence of the COB protein in question, or alternatively that of the subunits of this protein, may be determined, thereby enabling the structural gene or genes which code for the activity in question to be identified. For Pseudomonas denitrificans, the structural genes which code for activities of the biosynthetic pathway are identified by finding, for each NH₂-terminal sequence, the COB protein having the same NH₂-terminal sequence.

The present invention also describes a method enabling intermediates in the pathway of biosynthesis of cobalamins or of other corrinoids to be identified and assayed in strains productive of cobalamins. These intermediates may be assayed both in culture musts and in the cells themselves. The intermediates which may be assayed are all the corrinoids which occur in the biosynthetic pathway after cobyrinic acid, namely, apart from cobyrinic acid, cobyrinic acid monoamide, cobyrinic acid diamide, cobyrinic acid triamide, cobyrinic acid tetraamide, cobyrinic acid pentaamide, cobyric acid, cobinamide, cobinamide phosphate, GDP-cobinamide, coenzyme B₁₂phosphate and coenzyme B₁₂. The non-adenosylated forms of these products may also be assayed by this technique.

Other subjects and advantages of the present invention will become apparent on reading the examples and the drawings which follow, which are to be considered as illustrative and not limiting.

Definition of the Terms Employed and Abbreviations.

ATP: adenosine 5′-triphosphate

bp: base pairs

BSA: bovine serum albumin

CADAS: cobyrinic acid a,c-diamide synthase

cluster: group of genes

Cob: corresponds to the phenotype with a reduced level (at least 10-fold lower than the control) of production of cobalamins

cob gene: gene involved in the biosynthesis of cobalamins and/or cobamides from uro'gen III

COB protein: protein participating either as a catalyst in the pathway of biosynthesis of cobalamins, or as a regulatory protein in the network of regulation of the cob genes, or both.

cor gene: gene involved in the biosynthesis of corrinoids from uro'gen III

COR protein: protein participating either as a catalyst in the pathway of biosynthesis of corrinoids, or as a regulatory protein in the network of regulation of the cor genes, or both

Corrinoids: cobyrinic acid derivatives possessing the corrin ring-system

dGTP: 2′-deoxyguanosine 5′-triphosphate

DMBI: dimethylbenzimidazole

dNTP: 2′-deoxyribonucleoside 5′-triphosphates

DTT: dithiothreitol

HPLC: high performance liquid chromatography

kb: kilobases

NN: DMBI PRT: nicotinate-nucleotide: dimethylbenzimidazole phosphoribosyltransferase

ORF: open reading frame

recombinant DNA: set of techniques making it possible either to combine within the same microorganism DNA sequences which are not naturally so combined, or to mutagenise specifically a DNA fragment

SAM: S-adenosyl-L-methionine

SDS: sodium dodecyl sulphate

SP₂MT: SAM-L-methionine:precorrin-2 methyltransferase

Stop codon: translation termination codon

SUMT: SAM:uro'gen III methyltransferase

Uro'gen III: uroporphyrinogen III

LEGENDS TO THE FIGURES

FIG. 1: Structure of coenzyme B₁₂; the 5′-deoxyadenosyl group is replaced by a CH₃group for methylcobalamin, by a cyano group for cyanocobalamin, by a hydroxyl group for hydroxocobalamin.

FIG. 2: Biosynthesis of cobalamins and various steps of this biosynthesis. X: axial ligands of the cobalt; the ligand at a may be different from the ligand at b. R: ligand at a of the cobalt which defines the cobalamin type (see FIG. 1).

FIG. 3: Structures of uro'gen III, precorrin-1, precorrin-2 and precorrin-3.

FIG. 4: Structural formulae of uro'gen III and cobyrinic acid. Between uro'gen III and cobyrinic acid, there occur 8 SAM-dependent methyl transfers successively at C-2, C-7, C-20, C-17, C-12, C-1, C-15 and C-5, a decarboxylation at C-12, elimination of the carbon at C-20 and insertion of the cobalt atom. X: axial ligands of the cobalt; the ligand at a may be different from the ligand at b.

FIG. 5: Final steps of the biosynthesis of cobalamins. In order to clarify the diagram, details of the corrin ring-system have been omitted. The five enzymatic steps are represented: 1, cobinamide kinase; 2, cobinamidephosphate guanylyltransferase; 3, cobalamin-5′-phosphate synthase; 4, cobalamin-5′-phosphate phosphohydrolase; 5, nicotinatenucleotide:DMBI phosphoribosyltransferase.

FIG. 6: Restriction maps of the 5.4-kb ClaI-HindIII-HindIII-HindIII, 8.7-kb EcoRI, 4748-bp SalI-SalI-SalI-SalI-SalI-BalI and 3855-bp SstI-SstI-BamHI fragments. Only the 20 restriction enzymes which cut the DNA least frequently are shown. The cleavage sites of each enzyme are indicated by a vertical line.

FIG. 7: Nucleotide sequence of both strands of the 5378-bp ClaI-HindIII-HindIII-HindIII fragment of Pseudomonas denitrificans. The strand situated at the top is to be read from 5′ to 3′ in the left-to-right direction which corresponds to the left-to-right orientation of the sequenced fragment presented in FIG. 6. The ClaI site occurs at position 23 (beginning of the cleavage site) since, in this sequence, there occur PstI, SalI and XbaI restriction sites which have appeared during clonings in multisites with a view to sequencing. The sequence of the ClaI-HindIII-HindIII-HindIII fragment hence begins at position 23.

FIG. 8: Nucleotide sequence of both strands of the 8753-bp EcoRI fragment of Pseudomonas denitrificans. The strand situated at the top is to be read from 5′ to 3′ in the left-to-right direction which corresponds to the left-to-right orientation of the fragment of the restriction map presented in FIG. 6.

FIG. 9: Analysis of the probabilities of the coding frames on the basis of codon preference using the programme of Staden and MacLachlan (1982) on the 6 reading frames of the 5378-bp ClaI-HindIII-HindIII-HindIII fragment. For the frames belonging to the same coding strand, the most probable frame corresponds to that in which a dotted line, not interrupted by stop codons, is placed under the probability line for this frame.

1. Sequence extending from nucleotide 1 to nucleotide 1200. By means of this analysis, open reading frame 1 is identified. It begins at the ATG at position 549 and ends at the TGA at position 1011.

2. Sequence extending from nucleotide 1000 to nucleotide 2200. By means of this analysis, open reading frame 2 is identified. It begins at the ATG at position 1141 and ends at the TGA at position 1981.

3. Sequence extending from nucleotide 1800 to nucleotide 3400. By means of this analysis, open reading frame 3 is identified. It begins at the ATG at position 1980 and ends at the TGA at position 3282.

4. Sequence extending from nucleotide 3000 to nucleotide 4500. By means of this analysis, open reading frame 4 is identified. It begins at the ATG at position 3281 and ends at the TGA at position 4280.

5. Sequence extending from nucleotide 3800 to nucleotide 5378. By means of this analysis, open reading frame 5 is identified. It begins at the GTG at position 4284 and ends at the TGA at position 5253.

FIG. 10: Analysis of the probabilities of the coding frames on the basis of codon preference using the programme of Staden and MacLachlan (1982) on the 6 reading frames of the 8753-bp EcoRI fragment. For the frames belonging to the same coding strand, the most probable frame corresponds to that in which a dotted line, not interrupted by stop codons, is placed under the probability line for this frame.

1. Sequence extending from nucleotide 650 to nucleotide 1650. By means of this analysis, open reading frame 6 is identified. It begins at the ATG at position 736 and ends at the TGA at position 1519.

2. Sequence extending from nucleotide 1400 to nucleotide 3100. By means of this analysis, open reading frame 7 is identified. It begins at the ATG at position 1620 and ends at the TAG at position 2997.

3. Sequence extending from nucleotide 2700 to nucleotide 3700. By means of this analysis, open reading frame 8 is identified. It begins at the ATG at position 3002 and ends at the TGA at position 3632.

4. Sequence extending from nucleotide 3500 to nucleotide 4100. By means of this analysis, open reading frame 9 is identified. It begins at the GTG at position 3631 and ends at the TGA at position 4366.

5. Sequence extending from nucleotide 4150 to nucleotide 5150. By means of this analysis, open reading frame 10 is identified. It begins at the ATG at position 4365 and ends at the TGA at position 5127.

6. Sequence extending from nucleotide 5000 to nucleotide 6000. By means of this analysis, open reading frame 11 is identified. It begins at the ATG at position 5893 and ends at the TAG at position 5110.

7. Sequence extending from nucleotide 5700 to nucleotide 7200. By means of this analysis, frame 12 is identified. It begins at the ATG at position 5862 and ends at the TAA at position 7101.

8. Sequence extending from nucleotide 7000 to nucleotide 8000. By means of this analysis, open reading frame 13 is identified. It begins at the ATG at position 7172 and ends at the TTG at position 7931.

FIG. 11: Construction of plasmids pXL556, pXL545 and pXL723.

A 2.4-kb ClaI-EcoRV fragment containing the cobA and cobE genes is excised from the 5.4-kb fragment and then purified. An EcoRI linker is added at the EcoRV site and the fragment is then inserted into pXL59 between the ClaI-EcoRI sites. The plasmid thereby constructed is designated pXL556.

The construction is comparable for pXL545: a 1.9-kb ClaI-HindIII-HindIII fragment is excised from the 5.4-kb fragment and then purified. This fragment contains only the CobE gene. An EcoRI linker is added at the HindIII site and the fragment is then inserted into pXL59 between the ClaI-EcoRI sites. pXL723 is constructed as follows: a 2.3-kb EcoRI-HindIII fragment is excised from the 5.4-kb fragment and purified, and the ends are then filled in with the large fragment of E. coli DNA polymerase I. This fragment is cloned into pRK290 (Ditta et al., 1981) digested with EcoRI and then treated with the large fragment of E. coli DNA polymerase I in order to fill in the ends.

The restriction sites which are shown in brackets correspond to sites which have disappeared after treatment with the large fragment of E. coli DNA polymerase I.

1, PstI-SstI fragment of RSF1010 (De Graff et al., 1978); 2, PstI-BamHI fragment of pACYC177 (Bagdasarian et al., 1981); 3, BamHI-SstI fragment containing the lactose operon of E. coli without its promoter, the operator, the translation initiation site and the first 8 non-essential codons of lacZ (Casadaban et al., 1983); 4, Sau3AI fragment of Pseudomonas putida KT2440 (Bagdasarian et al., 1981); ori, origin of replication; nic, relaxation site; mob, locus essential for mobilisation; Km′, kanamycin resistance gene (Bagdasarian et al., 1981); B. BamHI; C, ClaI; E, EcoRI; H, HindIII; P, PstI; S. SstI; Sa, SalI; X, XhoI; Xb, XbaI.

FIG. 12: Studies of the insertions of transposons Tn5Sp′ and Tn5 into the 5378-bp fragment. The insertions of transposon Tn5 into plasmid pXL723 are shown as in FIG. 14; those of transposon Tn5Sp′, into the chromosome of strain G2 Rif′, are boxed; the insertions into the chromosome of SC510 Rif′ of cassettes carrying the kanamycin resistance gene (1630 and 1631) are shown with an arrow, according to the orientation of transcription of the kanamycin resistance gene, under the insertion number. The open reading frames deduced from the sequence are given in this figure (from cobA to cobE); + or − signs are shown under each insertion of transposon or of resistance cassette to indicate that the insertion is inactivating (−) or otherwise (+), i.e. for the complementation of different mutants (the case with the insertions of transposons Tn5), or that the insertion abolishes the cobalamin production of the strain in which it takes place. There is an absence of complementation when the recombinant mutant synthesises less than threefold less cobalamins than the level of synthesis of the strain from which the mutant is derived. The inserts of plasmids pXL545, pXL1500, pXL1397 and pXL302 are shown with the restriction sites occurring at their ends. These inserts are cloned into broad host range plasmids, pXL435 and pXL59 (Cameron et al., 1989):

plasmid pXL545 corresponds to plasmid pXL545 described in FIG. 11 with, in addition, the 2-kb BamHI fragment of pHP45 (Prentki and Krisch) containing a spectinomycin resistance gene cloned at the BamHI site of pXL545;

plasmid pXL1500 corresponds to the 4.2-kb BqlII-SstI fragment presented in this figure, cloned at the BamHI and SstI sites of pKT230 (Bagdasarian et al., 1981); presented in FIG. 30;

plasmid pXL1397 corresponds to the 2.4-kb HindIII-SstI fragment indicated in the figure, inserted between the HindIII and SstI sites of the multisite of pXL435 (Cameron et al., 1989) described in FIG. 30; plasmid pXL302 corresponds to the 2.3-kb EcoRI-HindIII fragment as described in the figure, inserted between the EcoRI and HindIII sites of pXL59 (Cameron et al., 1989) described in FIG. 30, the HindIII site used being the site occurring in the cloning multisite of pXL59;

pXL723 is described in FIG. 11, like pXL545.

+or − signs are shown above each of these inserts to indicate whether there is complementation by the plasmid in question of the chromosomal insertions shown underneath. C, ClaI; E, EcoRI; H, HindIII; RV, EcoRV; Sau, Sau3AI; S, SstI.

FIG. 13: Construction of plasmids pXL253 and pXL367. The 8.7-kb EcoRI fragment is excised and then purified from plasmid pXL151. It is cloned at the EcoRI site of pKT230 to give pXL253. This same fragment is inserted at the EcoRI site of pRK290 (Ditta et al., 1981) to give pXL3E7. 1, PstI-SstI fragment of RSF1010 (De Graff et al., 1978); 2, PstI-BamHI fragment of pACYC177 (Bagdasarian et al., 1981); ori, origin of replication; nic, relaxation site; mob, locus essential for mobilisation (Bagdasarian et al., 1981); B, BamHI; C, ClaI; E, EcoRI; H. HindIII; P, PstI; S, SstI; Sa, SalI; X, XhoI; Xb, XbaI; tet′, tetracycline resistance gene; Km′, kanamycin resistance gene.

FIG. 14: Studies of the insertions of transposons Tn3lacZ and Tn5 into the 8.7-kb EcoRI fragment cloned into pRK290 (Ditta et al., 1980). The insertions of transposons Tn3lacZ are underlined, in contrast to those of transposons Tn5. The open reading frames deduced from the sequence (cobF to cobM) are given in this figure, and the eight groups of inactivating insertions (numbered from 1 to 8) are presented; +or − signs are shown under each transposon insertion to indicate that the insertion is inactivating (−) or otherwise (+) for the complementation of different mutants. There is an absence of complementation when the recombinant mutant synthesises less than threefold less cobalamins than the level of synthesis of the strain from which the mutant is derived. These groups of inactivating insertions correspond to the following mutants: 1, G615; 2, G614 and G616; 3, G613 and G614; 4, G620; 5, G638; 6, G610 and G609; 7, G612; 8, G611. These mutants are Cob mutants of Agrobacterium tumefaciens already described (Cameron et al., 1989). A restriction map of the 8.7-kb fragment is given at the bottom of the figure.

FIG. 15: The coding sequences of each of the genes of the 5.4-kb fragment, cobA to cobE, respectively, are indicated. The sequences of the proteins COBA to COBE encoded by these sequences appear under their respective coding sequence, cobA to cobE. The amino acid composition of each protein, in number and in percentage, respectively, of COBA to COBE, is presented, as well as the molecular weight, the index of polarity, the isoelectric point and the optical density at 260 nm and 280 nm of a solution containing 1 mg/ml of purified protein. The hydrophilicity profile of each COBA to COBE protein, respectively, is shown; it was calculated on the basis of the programme of Hopp and Woods (1981). Positive values correspond to regions of the protein which are hydrophilic. The position of the amino acids is indicated as abscissa, while the value of the index of hydrophilicity is shown as ordinate; when this value is positive, this indicates that the region of the protein is hydrophilic.

FIG. 16: The coding sequences of each of the genes of the 8.7-kb fragment, cobF to cobM respectively, are indicated. The sequences of the COBF to COBM proteins encoded by these sequences appear under their sequence. The legend is identical to that for FIG. 15. NB. We have shown the COBF protein as beginning at the ATG located at position. 736; it is possible that the ATG located at position 751 is the true initiation codon of this protein.

FIG. 17: Reaction catalysed by cobyrinic acid a,c-diamide synthase. CADAS catalyses the amidation of the carboxylic acid functions of the peripheral acetate chains a and c of cobyrinic acid to give cobyrinic acid diamide; the donor of the amine group used in the enzymatic test is L-glutamine; it gives L-glutamic acid on deamination. X corresponds to the axial ligands of the cobalt, which may be different from one another.

FIG. 18: Reaction catalysed by SP₂MT. SP₂MT catalyses the transfer of a methyl from SAM to dihydrosirohydrochlorin or precorrin-2 to give precorrin-3. The methyl group is transferred to position C-20 of the porphyrin ring-system.

FIG. 19: Structure of hydrogenobyrinic acid and of hydrogenobyrinic acid a,c-diamide.

FIG. 20: Affinities of the COBA and COBF proteins for SAM. The curves give in arbitrary units the radioactivity at emergence from the TSK-125 column for each protein applied to this column. The retention times are indicated in minutes and the radioactivity peak corresponding to free SAM is observed at the time of 10 min 30 sec.

FIG. 21: Comparison of the sequences of COBA and COBI. Only the regions 1, 2 and 3, of strong homology, are presented. =signs are placed between identical residues and—signs between homologous residues (H K R, LIVM, A G S T, Y F W, D E Q N B Z, P, C).

FIG. 22: Comparison of the primary sequences of the proteins COBA of Pseudomonas denitrificans and CYSG of E. coli. The alignment has been done according to the programme of Kanehisa, 1984. =signs are placed between identical residues and − signs between homologous residues (HKR, L I V M, A G S T, Y F W, D E Q N B Z, P, C). The regions 1, 2 and 3 correspond to zones of strong homology between the proteins.

FIG. 23: Comparison of the sequences of CYSG of E. coli with COB proteins of Pseudomonas denitrificans (COBA, COBF, COBI, COBJ, COBL and COBM). The comparisons concern the regions 1, 2 and 3, of strong homologies, which exist between CYSG, COBA and COBI. The positions in the protein sequences of the regions displaying homologies are presented in the figure. We have considered the following groups of homologous residues: H K R, L I V M, A G S T, Y F W, D E Q N B Z, P, C. If there are at least 3 homologous residues at the same position, we have boxed these amino acids.

FIG. 24: Construction of plasmids pXL1148 and pXL1149. pXL1148 is constructed as follows: the 1.9-kb BamHI-BamHI-SstI-SstI fragment of the 8.7-kb fragment containing the cobH and cobI genes is purified, and XbaI and EcoRI linkers are placed at the BamHI and SstI ends respectively. This fragment is then inserted between the XbaI and EcoRI sites of the broad host range plasmid pXL59 (Cameron et al., 1989) to give plasmid pXL1148.

pXL1149 is constructed like pXL1148, apart from the fact that the fragment initially purified is the 1.5-kb BamHI-BamHI-SstI fragment instead of the fragment additionally containing the small 400-bp SstI fragment used for pXL1148. The fragment then undergoes the same enzymatic treatments and the same cloning into pXL59.

1, PstI-SstI fragment of RSF1010 (De Graff et al., 1978); 2, PstI-BamHI fragment of pACYC177 (Bagdasarian et al., 1981); 3, BamHI-SstI fragment containing the lactose operon of E. coli without promoter, operator, translation initiation site and the first 8 non-essential codons of lacZ (Casadaban et al., 1983); 4, Sau3AI fragment of Pseudomonas putida KT2440 (Bagdasarian et al., 1981); ori, origin of replication; nic, relaxation site; Km′, kanamycin resistance gene; mob, locus essential for mobilisation (Bagdasarian et al., 1981); B, BamHI; C, ClaI; E, EcoRI; H, HindIII; P, PstI; S, SstI; Sa, SalI; X, XhoI; Xb, XbaI.

FIG. 25: Total proteins of strains SC510 Rif′, SC510 Rif′ pKT230, SC510 Rif′ pXL1148, SC510 Rif′ pXL1149 analysed in 10% SDS-PAGE as described. The bacteria were cultured for 4 days in PS4 medium, and lysates of the total proteins were then made. Lane 1, SC510 Rif′; lane 2, SC510 Rif′ pXL1149; lane 3, SC510 Rif′ pXL1148; lane 4, SC510 Rif′ pKT230. The molecular masses of the molecular mass markers are indicated. The positions to which the COBI and COBH proteins migrate are indicated.

FIG. 26: Construction of plasmids pXL1496 and pXL1546. Plasmid pXL1496 enables the COBF protein to be overexpressed in E. coli, and plasmid pXL1546 enables COBF to be overexpressed in Pseudomonas denitrificans.

The 2.2-kb EcoRI-XhoI fragment is excised and purified from the 8.7-kb fragment. It is cloned at the EcoRI site of phage M13mp19 to give plasmid pXL1405. An NdeI site is then introduced by directed mutagenesis, as described above, at position 733 of this fragment; in this manner, an NdeI site occurs exactly on the presumed initiation codon of the cobF gene. The new plasmid thereby obtained is designated pXL1406. A 1.5-kb NdeI-SphI-sphI fragment, containing the cobF gene starting from its presumed initiation codon, is purified after partial digestion with the appropriate enzymes and ligated with the appropriate fragments of plasmid pXL694 (120-bp EcoRI-NdeI fragment containing expression signals of E. coli—see text—and 3.1-kb EcoRI-SphI fragment containing the ampicillin resistance gene, the replication functions of the plasmid and also the terminators of the rrnB operon of E. coli, as described in the text). The plasmid thereby constructed is designated pXL1496. pXL1546 is constructed as follows: the 2-kb EcoRI-BamHI-BamHI fragment of pXL1496 is purified by partial digestion with the appropriate enzymes; this fragment contains the expression signals of E. coli, followed by the cobF gene and then the 5′ portion of the cobG gene, this portion itself being followed by terminators of the rrnB operon of E. coli, as described in the text. This fragment is cloned into the multihost plasmid pKT230 (Bagdasarian et al., 1981) described in FIG. 30. B, BamHI; C, ClaI; E, EcoRI; H, HindIII; P. PstI; S, SstI, Sa, SalI; X, XhoI; Xb, XbaI; Km′, kanamycin resistance gene; Amp, ampicillin resistance gene.

FIG. 27: Total proteins of strains SC510 Rif′, SC510 Rif′ pKT230, SC510 Rif′ pXL1546 analysed in 10% SDS-PAGE as described. The bacteria were cultured for 4 days in PS4 medium, and lysates of the total proteins were then made. Lane 1, SC510 Rif′; lane 2, SC510 Rif′ pKT230; lane 3, SC510 Rif′ pXL1546. The molecular masses of the molecular mass markers are indicated. The position to which the COBF protein migrates is indicated.

FIG. 28: Total proteins of the strains E. coli B and E. coli B pXL1496 analysed in 10% SDS-PAGE as described. Lane 1, E. coli pXL1496 cultured in the absence of tryptophan; lane 2, E. coli pXL1496 cultured under the same conditions in the presence of tryptophan; lane 3, E. coli cultured in the absence of tryptophan; lane 4, E. coli cultured under the same conditions in the presence of tryptophan. The molecular masses of the markers are indicated. The position of migration of the COBF protein is indicated.

FIG. 29: Construction of plasmids pXL525 and pXL368. Plasmid pXL368 is constructed as follows: the 2.4-kb EcoRV-ClaI fragment (containing the cobA and cobE genes) is purified from plasmid pXL556, thereby enabling this fragment to be obtained with a BamHI site and an XbaI site at the ends; this fragment is cloned into pXL203 at the BamHI and XbaI sites.

For the construction of pXL525, an XbaI linker is added at the EcoRI site situated at the right-hand end of the 8.7-kb EcoRI fragment; this 8.7-kb EcoRI-XbaI fragment is then cocloned with the 2.4-kb EcoRI-XbaI fragment originating from pXL556 and containing cobA and cobE.

The restriction sites which are shown in brackets correspond to sites which have disappeared after treatment with the large fragment of E. coli DNA polymerase I. 1, PstI-SstI fragment of RSF1010 (De Graff et al., 1978); 2, PstI-BamHI fragment of pACYC177 (Bagdasarian et al., 1981); ori, origin of replication; nic, relaxation site; mob, locus essential for mobilisation; Km′, kanamycin resistance gene (Bagdasarian et al., 1981); B, BamHI; C, ClaI; E, EcoRI; H, HindIII; P, PstI; S. SstI; Sa, SalI; X, XhoI; Xb, XbaI; tet, tetracycline resistance gene; Amp′ and Amp, ampicillin resistance gene.

FIG. 30: Plasmids of the incompatibility group Q having a broad host range in Gram-negative bacteria. These plasmids are described in a previous publication (Cameron et al., 1989) and are used in the present invention.

1, PstI-SstI fragment of RSF1010 (De Graff et al., 1978); 2, PstI-BamHI fragment of pACYC177 (Bagdasarian et al., 1981); 3, BamHI-SstI fragment containing the lactose operon of E. coli without promoter, operator, translation initiation site and the first 8 non-essential codons of lacZ (Casadaban et al., 1983); 4, Sau3AI fragment of Pseudomonas putida KT2440 (Bagdasarian et al., 1981); ori, origin of replication; nic, relaxation site; Km′, kanamycin resistance gene; Sm′, streptomycin resistance gene; mob, locus essential for mobilisation (Bagdasarian et al., 1981); B, BamHI; C, ClaI; E, EcoRI; H, HindIII; P, PstI; S, SstI; Sa, SalI; X, XhoI; Xb, XbaI.

FIG. 31: Retention time of different corrinoid standards (1 mg/standard) on the separation system described in Example 7. The column used is a Nucleosil C-18 column (Macherey-Nagel). Against each absorbance peak, a number is shown corresponding to the corrinoid described below. The retention time is shown as abscissa and the absorbance at 371 nm appears as ordinate.

1, cobyrinic acid; 2, cobyrinic acid a-amide; 3, cobyrinic acid g-amide; 4, cobyrinic acid a,g-diamide; 5, cobyrinic acid c-amide; 6, cobyrinic acid c,g-diamide; 7, cobyrinic acid a,c-diamide; 8, cobyrinic acid triamide; 9, cobyrinic acid tetraamide; 10, cobyrinic acid pentaamide; 11, cobyric acid; 12, GDP-cobinamide; 13, cobinamide phosphate; 14, cobinamide; 15, cyanocobalamin 5′-phosphate; 16, cyanocobalamin.

FIG. 32: Nucleotide sequence of both strands of the 4748-bp SalI-SalI-SalI-SalI-SalI-BqlI fragment of Pseudomonas denitrificans. The strand situated at the top is to be read from 5′ to 3′ in the left-to-right direction which corresponds to the left-to-right orientation of the fragment of the restriction map presented in FIG. 6.

FIG. 33: Nucleotide sequence of both strands of the 3855-bp SstI-SstI-BamHI fragment of Pseudomonas denitrificans. The strand situated at the top is to be read from 5′ to 3′ in the left-to-right direction which corresponds to the left-to-right orientation of the fragment of the restriction map presented in FIG. 6.

FIG. 34: Analysis of the probabilities of the coding frames on the basis of codon preference using the programme of Staden and MacLachlan (1982) on the six reading frames of the 4748-bp SalI-SalI-SalI-SalI-SalI-BglI fragment. For the frames belonging to the same coding strand, the most probable frame corresponds to that in which a dotted line, not interrupted by stop codons, is placed under the probability line for this frame. 4a. Analysis of the sequence corresponding to nucleotides 200 to 800. This analysis enables open reading frame 14 to be identified. It begins at the ATG at position 660 and ends at the TGA at position 379. 4b. Analysis of the sequence corresponding to nucleotides 800 to 1500. This analysis enables open reading frame 15 to be identified. It begins at the GTG at position 925 and ends at the TAA at position 1440. 4c. Analysis of the sequence corresponding to nucleotides 1450 to 2600. This analysis enables open reading frame 16 to be identified. It begins at the ATG at position 1512 and ends at the TGA at position 2510. 4d. Analysis of the sequence corresponding to nucleotides 2500 to 4650. This analysis enables open reading frame 17 to be identified. It begins at the GTG at position 2616 and ends at the TGA at position 4511.

FIG. 35: Analysis of the probabilities of the coding frames on the basis of codon preference using the programme of Staden and MacLachlan (1982) on the six reading frames of the 3855-bp SstI-SstI-BamHI fragment. For the frames belonging to the same coding strand, the most probable frame corresponds to that in which a dotted line, not interrupted by stop codons, is placed under the probability line for this frame. 5a. Analysis of the sequence corresponding to nucleotides 1 to 905. This analysis enables open reading frame 18 to be identified. It begins at the ATG at position 809 and ends at the TGA at position 108. 5b. Analysis of the sequence corresponding to nucleotides 955 to 2105. This analysis enables open reading frame 19 to be identified. It begins at the ATG at position 1971 and ends at the TGA at position 1063. 5c. Analysis of the sequence corresponding to nucleotides 2000 to 3300. This analysis enables open reading frame 20 to be identified. It begins at the ATG at position 2099 and ends at the TAG at position 3115. 5d. Analysis of the sequence corresponding to nucleotides 3250 to 3855. This analysis enables open reading frame 21 to be identified. It begins at the ATG at position 3344 and ends at the TGA at position 3757.

FIG. 36: Construction of plasmids pXL233, pXL843 and pXL1558 from pXL154.

The plasmids are constructed in the following manner. The 3.5-kb EcoRI fragment containing the truncated cobs gene and the sequence upstream is excised from pXL154, then purified and cloned to the EcoRI site of pKT230. The plasmid thereby constructed is designated pXL233. The 3.5-kb EcoRI-XhoI-XhoI fragment containing the cobT gene and the sequence downstream is excised and purified from pXL154 by partial digestions. The 4.3-kb EcoRI-EcoRI-EcoRI fragment containing the cobs gene and the sequence upstream is excised and purified from pXL154 and then ligated to the above 3.5-kb fragment. The approximately 8-kb EcoRI-XhoI fragment thereby attained is cloned into the EcoRI and SalI sites of pXL59 to generate plasmid pXL843. Plasmid pXL1558 is constructed in the following manner: the 12-kb HindIII-HindIII fragment is excised from pXL154 and purified, and the ends are then filled in with the large fragment of E. coli DNA polymerase I. This insert is cloned in pRK29O (Ditta et al., 1981) digested with EcoRI and then treated with the large fragment of E. coli DNA polymerase I in order to make the ends blunt. Restriction sites which are shown in brackets correspond to sites which have disappeared during cloning. 1, PstI-SstI fragment of RSF1010 (Degraff et al., 1978); 2, PstI-BamHI fragment of pACYC177 (Bagdasarian et al., 1981); B, BamHI; C, ClaI; E, EcoRI; H, HindIII; P, PstI; S, SstI; Sa, SalI; X, XhoI; Xb, XbaI;

Tet tetracycline resistance gene; Km′, kanamycin resistance gene; Sm′, streptomycin resistance gene.

FIG. 37: Study of the insertions of the transposon Tn5Sp into the 12-kb HindIII-HindIII insert of pXL154.

The insertions of the transposon are mapped on the 12-kb HindIII-HindIII insert cloned into pXL1558. The chromosomal insertions into strain SC510 Rif′ are boxed, that which is not is introduced into strain SBL27 Rif′. A plus or minus sign is shown under each insertion to indicate the Cob phenotype of the strain having this insertion. Absence of complementation (or complementation) of strain G2035 by plasmids pXL1558::Tn5Sp is indicated by minus (or plus) signs below each insertion. The inserts of the plasmids described in FIG. 36 are shown. The plus (or minus) signs over these plasmids, and aligned with the transposon insertions, show diagrammatically the complementation (or absence) of the transposon-mutated strain by the plasmid. The open reading frames deduced from the sequence are also given in this figure (ORF14 to 17, as well as the corresponding cob genes (cobS and cobT)). E: EcoRI; H: HindIII; X: XhoI.

FIG. 38: Construction of plasmids pXL1286, pXL1303, pXL1324, pXL1490B and pXL1557 from pXL519. The position of the sequenced fragment appears in the upper part of the figure above the restriction map of the cluster; it is a 3.9-kb SstI-SstI-SstI-BamHI fragment. The plasmids are constructed in the following manner. The 2-kb BqlII-EcoRI fragment containing the cobU gene and the sequence downstream is excised from pXL519, then purified and cloned at the BamHI and EcoRI sites of pKT230 to generate plasmid pXL1286. The 2.7-kb SstI-EcoRI fragment containing the truncated cobV gene, cobU gene and the sequence downstream is excised on pXL519, then purified and cloned at the SstI and EcoRI sites of pKT230 to generate plasmid pxL1324. The 1.6-kb SstI-SstI fragment containing the truncated cobV gene and the sequence upstream is excised from pXL519, then purified and cloned at the SstI site of pKT230 to generate plasmid pXL1303. The 3.85-kb SstI-SstI-BamHI fragment is purified after total digestion of pXL519 with BamHI and partial digestion with SstI. This fragment is then cloned at the BamHI and SstI sites of pKT230 to generate pXL1490B. Plasmid pXL1557 is constructed in the following manner: the 9-kb HindIII-BamHI fragment is excised from pXL519 and purified, and the ends are then filled in with the large fragment of E. coli DNA polymerase I. This insert is cloned into pRK290 (Ditta et al., 1981) digested with EcoRI and then treated with the large fragment of E. coli DNA polymerase I to make the ends blunt. The restriction sites which are shown in brackets correspond to sites which have disappeared during cloning. 1, PstI-SstI fragment of RSF1010 (Degraff et al., 1978); 2, PstI-BamHI fragment of pACYC177 (Bagdasarian et al., 1981); B, BamHI;Bg, BqlII; C, ClaI; E, EcoRI; H, HindIII; P, PstI; S, SstI; Sa, SalI; X, XhoI; Xb, XbaI; Tet′, tetracycline resistance gene; Km′ kanamycin resistance gene; Sm′, streptomycin resistance gene.

FIG. 39: Study of the insertions of the transposon Tn5Sp into the 9-kb HindIII-BamHI insert of pXL519. The insertions of the transposon are mapped on the 9-kb HindIII-BamHI insert cloned into pXL1557. The chromosomal insertions into strain SC510 Rif′ are boxed, those which are not-are introduced into strain SBL27 Rif′. A plus or minus sign is shown under each insertion to indicate the Cob phenotype of the strain having this insertion. Absence of complementation (or complementation) of strain G2040 by plasmids pXL1557::Tn5Sp is indicated by minus (or plus) signs below each insertion. The inserts of the plasmids described in FIG. 6 are shown. The plus (or minus) signs over these plasmids and aligned with the transposon insertions, show diagrammatically the complementation (or absence) of the transposon-mutated strain by the plasmid. The open reading frames deduced from the sequence are also given in this figure (ORF18 to 21), as well as the corresponding cob genes (cobU and cobV).

FIG. 40: Coding sequences of each of the genes of the 4.8-kb fragment, cobx, cobS and cobT, respectively, are indicated. The sequence of the COBX, COBS and COBT proteins encoded by these sequences appears under the respective coding sequences cobX, cobS and cobT. The legend is identical to that for FIG. 15.

FIG. 41: Coding sequences of each of the genes of the 3.9-kb fragment, cobU and cobV, respectively, are indicated. The sequence of the COBU and COBV proteins encoded by these sequences appears under the respective coding sequences cobU and cobV. The legend is identical to that of FIG. 15.

FIG. 42: A. Total proteins of the strains E. coli BL21 pLysS pET3b and E. coli BL21 pLysS pXL1937 analysed in 10% SDS-PAGE. Lane 1, BL21 pLyspET3b; lane 2, E. coli BL21 pLysS pXL1937. B. Total proteins of the strains E. coli BL21, E. coli BL21 pXL1874 and E. coli BL21 pXL1875 analysed in 10% SDS-PAGE. Lane 1, E. coli BL21; lane 2, E. coli BL21 pXL1874; lane 3, E. coli BL21 pXL1875.

The molecular masses of the markers are indicated. The band corresponding to the overexpressed protein is indicated by an arrow.

FIG. 43: Nucleotide sequence of both strands of the 13144-bp SstI-SstI-SstI-SstI-BqlII-BqlII fragment of Pseudomonas denitrificans. The strand situated at the top is to be read from 5′ to 3′ in left-to-right direction which corresponds to the left-to-right orientation of the fragment of the restriction map presented in FIG. 46.

FIG. 44: Restriction map of the 13144-bp SstI-SstI-SstI-SstI-BqlII-SstI-BqlII fragment of Pseudomonas denitrificans. The position or positions of restriction sites occurring are indicated in increasing order of the cut number on the fragment sequenced; the positions correspond to the sequence presented in FIG. 43.

FIG. 45: Analysis of the probabilities of the coding frames on the basis of codon preference using the programme of Staden and MacLachlan (1982) on the six reading frames of the 13144-bp SstI-SstI-SstI-SstI-BqlII-SstI-BqlII fragment of Pseudomonas denitrificans. For the frames belonging to the same coding strand, the most probable frame corresponds to that in which a dotted line, not interrupted by stop codons, is placed under the probability line for this frame.

1. Sequence corresponding to nucleotides 1 to 2266. This analysis enables open reading frame 22 to be identified. It begins at the ATG at position 429 and ends at the TAG at position 1884.

2. Sequence corresponding to nucleotides 2266 to 4000. This analysis enables open reading frame 23 to be identified. It begins at the ATG at position 3364 and ends at the TGA at position 3886.

3. Sequence corresponding to nucleotides 3800 to 5000. This analysis enables open reading frame 24 to be identified. It begins at the ATG at position 3892 and ends at the TAG at position 4954.

4. Sequence corresponding to nucleotides 5000 to 9000. This analysis enables open reading frame 25 to be identified. It begins at the ATG at position 5060 and ends at the TAG at position 8885.

5. Sequence corresponding to nucleotides 9000 to 9700. This analysis enables open reading frame 26 to be identified. It begins at the ATG at position 9034 and ends at the TGA position 9676.

6. Sequence corresponding to nucleotides 9600 to 13144. This analysis enables open reading frames 27, 28, 29 and 30 to be identified. They begin, respectively, at the ATGs at positions 9678, 10895, 11656 and 13059, and end at the stop codons at positions 10101, 10304, 12181 and 12366. Open reading frames 28 and 30 occur on the strand complementary to the coding strand corresponding to all the other open reading frames.

FIG. 46: 13.4-kb EcoRI-BqlII-EcoRI-BglII fragment, positions of the insertions of transposons Tn5Sp into the 9.1-kb EcoRI fragment, positions of the insertions of transposons Tn5 into the insert of plasmid pXL189 as well as the inserts of the various plasmids used during the experiments on complementation of strains SC510 Rif′::Tn5Sp. The complementations of the mutants SC510 Rif′::Tn5Sp by the plasmids are indicated (+)—between 5% and 100% of the level of the parent strain SC510 Rif′−, (.)—partial complementation, between 0.5 and 5% of the level of SC510 Rif′−, or (−)—absence of complementation, i.e. less than one thousand times less than SC510 Rif′−, positioned immediately above the lines showing diagrammatically the insert of the plasmids and aligned with the insertion sites of the corresponding mutants. Below the mapping of the insertions of transposons Tn5 into the insert of plasmid pXL189, the complementation (+) or absence of complementation (−) of these mutant plasmids for the Agrobacterium tumefaciens mutants G632 and G633 is shown. On the right-hand part of the figure, there is a table showing the complementation of the mutants G622, G623 and G630 (Cameron et al., 1989) by different plasmids; (+)—total complementation, 100% of the level of the parent strain C58C9 Rif′″, (.)—partial complementation, between 10 and 50% of the level of C58C9 Rif′−, or (−)—absence of complementation.

The different plasmids whose insert is shown are constructed as follows (the fragments are excised either from pXL156 or from pXL157):

pXL618 corresponds to the 2.5-kb EcoRI-BamHI fragment cloned at the same sites of pKT230 (Bagdasarian et al., 1981);

pXL593 corresponds to the 3.1-kb BamHI fragment cloned at the BamHI site of pKT230 (Bagdasarian et al., 1981);

pXL623 corresponds to the 1.9-kb BamHI-XhoI fragment cloned at the BamHI-SalI sites of pXL59 (Cameron et al., 1989);

pXL1909 corresponds to the 8.4-kb BamHI-BamHI-BamHI fragment cloned at the BamHI of pKT230 (Bagdasarian et al., 1981);

pXL221 corresponds to the 1.6-kb EcoRI-ClaI fragment cloned at the same sites of pXL59 (the ClaI site into which this fragment is cloned is the ClaI site of the multisite of pXL59) (Cameron et al., 1989);

pXL1908 and 1938 correspond to the same insert, 6.5-kb XhoI-BamHI-BamHI fragment, to which XbaI linkers have been added; this insert is cloned in both orientations at the XbaI site of pXL435 (Cameron et al., 1989); an arrow positioned on the figure indicates the position of the kanamycin resistance gene with respect to the ends of the insert of the two plasmids;

pXL208 corresponds to the 5.2-kb BamHI fragment cloned at the BamHI site of pKT230 (Bagdasarian et al., 1981);

pXL297 corresponds to the 9.1-kb EcoRI fragment cloned at the EcoRI site of pKT230 (Bagdasarian et al., 1981).

The open reading frames (ORF) defined by the sequencing of the fragment (ORF 22 to 30) are shown, as well as the corresponding cob genes; an arrow indicates the polarity of the transcription.

E, EcoRI; B, BamHI; Bg, BqlII; Cl, ClaI; Sau, Sau3AI; X, XhoI;

FIG. 47: Coding sequences of each of the genes of the 13.4-kb fragment, cobQ, cobP and cobW, cobN and cobO, respectively, are indicated. The sequences of the COBQ, COBP, COBW, COBN and COBO proteins encoded by these sequences appear under their respective coding sequence cobQ, cobP, cobW, cobN and cobO. The legend is identical to that for FIG. 15.

FIG. 48: A-NH₂-terminal sequence of SUMT of M. ivanovii and sequence of the oligonucleotides 923, 946, 947; −, means that, at this position, the residue could not be determined; for the antisense oligonucleotide, the amino acids indicated below the sequence correspond to the anticodons shown. B—Presentation of the enzymatic amplification of a fragment internal to the structural gene of SUMT of M. ivanovii with the oligonnucleotides 946 and 947.

FIG. 49: Construction of the recombinant replicative form pG10. The 615-bp fragment obtained by amplification is digested with HindIII and EcoRI and then purified as described. This fragment is then ligated with the replicative form of phage M13mp19 digested with the same enzymes. The recombinant clone is found as described in the text.

FIG. 50: Autoradiograph of a genomic DNA blot of M. ivanovii digested with various enzymes, separated by agarose gel electrophoresis and then transferred onto a nylon membrane as described previously. The membrane is hybridised with the pG10 probe as described previously. 1, HindIII-BqlII; 2, KpnI-BclII; 3, EcoRI-BqlII; 4, BqlII-PstI. The sizes of the different fragments which hybridise with the probe are shown in kb.

FIG. 51: Nucleotide sequence of both strands of the 955-pb fragment of M. ivanovii. The strand situated at the top is to be read from 5′ to 3′ in the left-to-right direction.

FIG. 52: Coding sequence of the corA gene of M. ivanovii obtained from the 955-bp sequence. The primary sequence of the CORA protein is also shown. The amino acids are shown above their codon and the stop codon is designated by a star. The main physical properties of the CORA protein of M. ivanovii, namely the amino acid composition, in number and in percentage, the molecular weight, the index of polarity, the isoelectric point and the optical density at 280 nm of a solution containing 1 mg/l of purified protein. The hydrophobicity profile of the CORA protein of M. ivanovii; this profile was obtained on the basis of the programme of Hopp and Woods (1981). Positive values correspond to regions of the protein which are hydrophilic. The position of the amino acids is indicated as abscissa, and the value of the index of hydrophilicity as ordinate; when this value is positive, this indicates that the protein is hydrophilic in this region.

FIG. 53: Comparison of the primary sequences of the proteins COBA of P. denitrificans and CORA of M. ivanovii. The proteins have been aligned by means of the programme of Kanehisa (1984). =, identical amino acids; −, homologous amino acids on the basis of the criteria defined above (see FIGS. 22 and 23).

FIG. 54: Construction of plasmids pXL1832 and pXL1841. The legends described, placed on the figure, enable the constructions to be followed.

GENERAL TECHNIQUES OF CLONING, MOLECULAR BIOLOGY AND BIOCHEMISTRY

The classical methods of molecular biology, such as centrifugation of plasmid DNA in a caesium chloride/ethidium bromide gradient, digestions with restriction enzymes, gel electrophoresis, electroelution of DNA fragments from agarose gels, transformation in E. coli, and the like, are described in the literature (Maniatis et al., 1982, Ausubel et al., 1987).

Restriction enzymes were supplied by New-England Biolabs (Biolabs), Bethesda Research Laboratories (BRL) or Amersham Ltd (Amersham). Linker oligonucleotides were supplied by Biolabs.

For the ligations, the DNA fragments are separated according to their size on 0.7% agarose or 8% acrylamide gels, purified by electroelution, extracted with phenol, precipitated with ethanol and then incubated in 50 mM Tris-HCl buffer pH 7.4, 10 mM MgCl, 10 mM DTT, 2 mM ATP, in the presence of phage T4 DNA ligase (Biolabs).

If necessary, DNA fragments having protuberant 5′ ends are dephosphorylated by a treatment with calf intestinal alkaline phosphatase (CIP, Pharmacia) at 37° C. for 30 min in the following buffer: 100 mM glycine, 1 mM MgCl₂, 1 mM ZnCl₂, pH 10.5. The same technique is used for dephosphorylation of protuberant or blunt 3′ ends, but the treatment is for 15 min at 37° C. and then 15 min at 56° C. The enzyme is inactivated by heating the reaction mixture to 68° C. for 15 min in the presence of 1% SDS and 100 mM NaCl, followed by a phenol/chloroform extraction and an ethanol precipitation.

Filling-in of protuberant 5′ ends is performed with the Klenow fragment of E. coli DNA polymerase I (Biolabs). The reaction is performed at room temperature for 30 min in 50 mM Tris-HCl buffer pH 7.2, 0.4 mM dNTPs, 10 mM MgSO₄, 0.1 mM DTT, 50 mg/ml BSA. Filling-in of protuberant 3′ ends is performed in the presence of phage T4 DNA polymerase (Biolabs) according to the manufacturer's recommendations. Digestion of the protuberant ends is performed by limited treatment with S1 nuclease (BRL) according to the manufacturer's recommendations. Linker oligonucleotides are added onto ends of DNA fragments as already described (Maniatis, 1982).

In vitro mutageneis with oligodeoxynucleotides is performed according to the method developed by Taylor et al., 1985, using the kit distributed by Amersham.

The ligated DNAs are used for transforming the strain rendered competent: E. coli MC 1060 [D(lacIOPZYA)X74, galU, galK, strA′, hsdR] for plasmids or E. coli TG1[D(lac proA,B), supE, thi, hsdD5/ F′ traD36, proA⁺, B⁺, lacI^q, lacZDM15) for replicative forms of phages derived from bacteriophage M13.

Plasmid DNAs are purified according to the technique of Birnboim and Doly, 1979. Minipreparations of plasmid DNA are made according to the protocol of Klein et al., 1980. Preparations of chromosomal DNA of Gram-negative bacteria are produced as already described (Cameron et al., 1989).

Radioactive probes are prepared by nick translation according to the method already detailed (Rigby et al., 1977). Hybridisations between DNA sequences as well as the immobilisation of nucleic acids on nitrocellulose membranes are performed as already described (Cameron et al., 1989). In clonings for which there is a small probability of finding the desired recombinant clone, the latter are found after hybridisation on filters as already described (Maniatis et al., 1982).

The nucleotide sequence of DNA fragments is determined by the chain-termination method (Sanger et al., 1977). In the reaction mixture, dGTP is replaced by 7-deaza-dGTP, in order to avoid compression of bands during acrylamide gel electrophoresis caused by the high percentage of GC in the DNA.

The culture media used for the bacteriological part have already appeared (Maniatis et al., 1982). Culturing in PS4 medium is carried out as already described (Cameron et al., 1989); Pseudomonas denitrificans strains SC510 Rif′ and G2 Rif′ are cultured in PS4 medium as follows: 250-ml Erlenmeyers containing PS4 medium (25 ml), with, if necessary, the selective antibiotic for the plasmid carried by each strain, are inoculated-with a 1/100 dilution of saturated preculture in L medium (Miller 1972), with, if necessary, the selective antibiotic for the plasmid carried by each strain; these cultures are incubated for 6 days at 30° C. and the musts are then analysed for their cobalamin content or alternatively the enzymatic activity of some enzymes of the pathway. Strains of Agrobacterium tumefaciens, Pseudomonas putida and Rhizobium meliloti are cultured at 30° C.; except where otherwise stated, they are cultured in L medium.

Bacterial conjugations are carried out as already described (Cameron et al., 1989).

Extracts of total proteins are produced as already described (Ausubel et al., 1987).

Analytical electrophoresis (SDS-PAGE) of proteins in acrylamide gel under denaturing conditions is performed as already described (Ausubel et al., 1987). The PhastSystem apparatus (Pharmacia) using Laemli's discontinuous-buffer system (Laemli, 1970) is also used; different gels are used in accordance with the molecular weights of the proteins to be analysed as well as their purity:

PhastGel gradient 8-25

PhastGel Homogeneous 12.5

Staining is performed either with Coomassie blue with the aid of PhastGel Blue R (Pharmacia), or with silver nitrate using the PhastGel silver Kit (Pharmacia) in accordance with the manufacturer's instructions.

NH₂-terminal sequences of the proteins are determined by the Edman degradation technique, using an automated sequencer (Applied Biosystems model 407A) coupled to an HPLC apparatus for identification of the phenylthiohydantoin derivatives.

EXAMPLE 1
Isolation of DNA fragments of P. denitrificans Containing Cob Genes

This example describes the isolation of DNA fragments of Pseudomonas denitrificans carrying Cob genes. These fragments were demonstrated by complementation experiments on Cob mutants of A. tumefaciens and P. putida (Cameron et al., 1989).

These Cob mutants were obtained by mutagenesis with N-methyl-N′-nitro-N-nitrosoguanidine according to the technique of Miller (Miller et al., 1972), or by insertions of transposon Tn5. In this manner, strains incapable of synthesising cobalamins were demonstrated, and especially the Cob mutant G572 of P. putida and the Cob mutants G159, G161, G164, G169, G171, G258, G609, G610, G611, G612, G613, G614, G615, G616, G620, G622, G623, G630, G632, G633, G634, G638, G642, G643, G2034, G2035, G2037, G2038, G2039, G2040, G2041, G2042 and G2043 of A. tumefaciens.

At the same time, a library of genomic DNA of P. denitrificans is produced in a mobilisable broad host range vector pXL59, by digestion of 5 μg of DNA in the presence of restriction enzymes (Cameron et al., 1989).

By complementation, several plasmids could be isolated, enabling the Cob mutants of P. putida and of A. tumefaciens to be complemented. Among these, plasmids pXL151, pXL154, pXL156, pXL157 and pXL519 will be noted more especially.

These plasmids were isolated and DNA fragments could be excised, purified and analysed by restriction. These fragments are presented in FIGS. 6 and 44: a 5.4-kb ClaI-HindIII-HindIII-HindIII fragment, an 8.7-kb EcoRI-EcoRI fragment, a 4.8-kb SalI-SalI-SalI-SalI-SalI-BglI fragment, a 3.9-kb SstI-SstI-BamHI fragment and a 13.4-kb EcoRI-BglII-EcoRI-BglII fragment.

EXAMPLE 2
Sequencing of the DNA Fragments Isolated

This example illustrates the sequencing of DNA fragments carrying cob genes of Pseudomonas denitrificans SC510.

2.1. Sequencing of a 5.4-kb ClaI-HindIII-HindIII-HindIII Fragment.

This fragment is contained in plasmid pXL157 described in Example 1. After excision, the subfragments of the 5.4-kb fragment were cloned into phages M13mp18 or M13mp19 (Norrander et al., 1983) or M13tg130 or M13tg131 (Kieny et al., 1983) in both orientations. Deletions were then produced in vitro by the method of Henikoff (1987). These deletions were then sequenced with the “universal primer” as a synthetic primer of chain-termination reactions. The overlap between these different deletions enabled the total sequence, over both strands, of the 5.4-kb fragment to be established (FIG. 7). This fragment comprises 5378 bp. In the sequence described in FIG. 7, there are seen, before the ClaI site, three restriction sites (PstI, SalI and XbaI) which have appeared during the cloning of the fragment in question with a view to sequencing in cloning multisites. When subsequent reference is made, in the present invention, to the sequence of this ClaI-HindIII-HindIII-HindIII fragment, this will be to the sequence presented in FIG. 7 in which the first 22 bases do not correspond to the DNA of Pseudomonas denitrificans (thus, all the positions of restriction site or of beginning of open reading frame refer to the sequence presented in FIG. 7).

2.2. Nucleotide Sequence of an 8.7-kb EcoRI-EcoRI Fragment.

This fragment is carried by pXL151 described in Example 1. The EcoRI site as well as the adjacent 70 bp located to the right of this fragment originate from pXL59, which is the vector used for constructing pXL151 by cloning an Sau3AI fragment of Pseudomonas denitrificans SC510. After excision, subfragments of the 8.7-kb fragment were cloned into phages M13mp18 or M13mp19 (Norrander et al., 1983) or M13tg130 or M13tg131 in both orientations (Kieny et al., 1983). Deletions were then produced in vitro by the method of Henikoff (1987). These deletions were then sequenced with the “universal primer” as a synthetic primer of chain-termination reactions. The overlap between these different deletions enabled the total sequence, over both strands, of the 8.7-kb fragment to be established (FIG. 8). This fragment comprises 8753 bp.

2.3. Sequencing of a 4.8-kb SalI-SalI-SalI-SalI-SalI-BglI Fragment.

This fragment is contained in plasmid pXL154 described in Example 1. The protocol is identical to that used in Example 2.2. The total sequence on both strands of the 4.8-kb fragment is presented in FIG. 32. This fragment contains 4749 bp.

2.4. Nucleotide Sequence of a 3.9-kb SstI-SstI-BamHI Fragment.

This fragment is included in plasmid pXL519 described in Example 1. The protocol is identical to that used in Example 2.2. The total sequence on both strands of the 3.9-kb fragment is presented in FIG. 33. This fragment contains 3855 bp.

2.5. Nucleotide Sequence of a 13.4-kb EcoRI-BglII-EcoRI-BglII Fragment.

This fragment is contained in plasmids pXL156 and pXL157 described in Example 1. The protocol used is identical to that of Example 2.2. The sequence on both strands of the 13.15-kb fragment is presented in FIG. 43. It corresponds to the total sequence of the 13.4-kb fragment except for 250 bp, corresponding to an EcoRI-SstI fragment, occurring at the left-hand end of the fragment.

From these nucleotide sequences, restriction maps were obtained for the enzymes which cut least frequently (FIGS. 6 and 44). The percentage of GC bases in Pseudomonas denitrificans SC150 DNA is relatively high (65.5%) and manifests itself in compression on the sequencing gels. To avoid these problems, two approaches are adopted:

i) the use of 7-deaza-dGTP instead of dGTP in the sequencing reactions to decrease the secondary structures which form during electrophoresis in the sequencing gel, and

ii) the sequencing of both strands.

EXAMPLE 3
Analysis of these Nucleotide Sequences Determination of the Open Reading Frames

The nucleotide sequences of the 5.4-kb ClaI-HindIII-HindIII-HindIII (FIG. 7), 8.7-kb EcoRI-EcoRI (FIG. 8), 4.8-kb SalI-SalI-SalI-SalI-SalI-BglI (FIG. 32), 3.9-kb SstI-SstI-BamHI (FIG. 33) and 13.4-kb EcoRI-BglII-EcoRI-BglII (FIG. 43) fragments enable open reading frames to be defined. Since the DNA in question contains a high percentage of GC, the open reading frames are numerous in view of the low frequency of translation stop codons. A study of the probability of the coding frames on the basis of codon preference using the method of Staden and MacLachlan (1982) is carried out. It characterises the open reading frames which have the greatest probability of being coding relative to the other frames of the same DNA strand, this probability being dependent on the codon preference of genes already sequenced originating from bacteria of the genus Pseudomonas. In this manner:

3.1. Five open reading frames are characterised for the 5.4-kb ClaI-HindIII-HindIII-HindIII fragment. They are designated frames 1 to 5, and their positions in the sequence of the 5.4-kb fragment are as follows (in the 5′-3′ sequence from the ClaI site to the HindIII sites): Table: Probable open reading frames of the 5.4-kb ClaI-HindIII-HindIII-HindIII fragment. The positions in the sequence correspond to the positions in the sequence described in FIG. 7; the coding strand is the 5′-3′ strand corresponding to the upper strand in this figure.

TranslationMolecular weightFrameinitiationStopin kD of thenumbercodoncodonencoded protein1549101115.521141198029.23198032825.743281428035.054284525334.1

The representations of the probabilities that these open reading frames are coding frames, with those observed on the other frames (5 in total) in parallel, are given in FIG. 9. These five frames are encoded by the same strand. Four of them (open reading frames 1 to 4) display the characteristics of coding frames in translational coupling (Normak et al., 1983), namely, the translation initiation codon of frame x+1 overlaps the translation termination codon of frame x, or else these codons are very close.

3.2. Eight frames are characterised for the 8.7-kb EcoRI-EcoRI fragment. They are designated frames 6 to 13 and their positions in the sequence of the 8.7-kb fragment are given in the table below. Table: Probable open reading frames of the 8.7-kb EcoRI fragment. The positions in the sequence correspond to the positions in the sequence described in FIG. 8; in this figure, the coding strand is the upper strand.

Molecular weightTranslationInitiationStopin kD of theframe numbercodoncodonencoded protein6736151928.971620299746.783002363222.093631436625.8104365512727.1115126586726.8125862710142.9137172793126.8

The representations of the probabilities of these open reading frames, with those observed on the other frames (6 frames in total) in parallel, are given in FIG. 10. With the exception of frame 11, these eight frames are encoded by the same strand. Four of them (from 7 to 10) display the characteristics of coding frames in translational coupling (Normark et al., 1983), namely, the translation initiation codon of frame x+1 overlaps the translation termination codon of frame x, or else these codons are very close.

3.3. Four open reading frames are characterised for the 4.8-kb SalI-SalI-SalI-SalI-SalI-BalI fragment. They are designated phases 14 to 17 and their positions in the sequence of the 4.8-kb fragment are as follows (in the 5′-3′ sequence from the SalI sites to the BglI site): Table: Probable open reading frames of the 4.8-kb SalI-SalI-SalI-SalI-SalI-BglI fragment. The positions in the sequence corresponds to the positions described in FIG. 32, where the upper strand is given in its 5′- 3′ orientation. Frames 15, 16 and 17 are encoded by the upper strand, in contrast to frame 14.

TranslationMolecular weightFrameinitiationStopin kD of thenumbercodoncodonencoded protein146603791028615925144018941161512251036983172616451170335

The representations of the probabilities that these open reading frames are coding, with those observed on the other frames (4 in total) in parallel, are given in FIG. 34. Frames 15, 16 and 17 are encoded by the same strand, frame 14 by the complementary strand.

3.4. Four frames are characterised for the 3.9-kb SstI-SstI-BamHI fragment. They are designated 18 to 21 and their positions in the sequence of the 3.9-kb fragment are given in the table below. Table: Probable open reading frames of the 3.9-kb SstI-SstI-BamHI fragment. The positions in the sequence correspond to the positions described in FIG. 33, where the polarity of the upper strand is 5′- 3′. Frames 18 and 19 are encoded by the lower strand, in contrast to frames 20 and 21.

TranslationMolecular weightFrameinitiationStopin kD of thenumbercodoncodonencoded protein1880910825148191971106330662202099311534682213344375714802

The representations of the probabilities that these open reading frames are coding, with those observed on the other frames (4 in total) in parallel, are given in FIG. 35. Frames 19 and 20 are transcribed in a differing manner.

3.5. Nine open reading frames are characterised for the 13.1-kb EcoRI-BglII-EcoRI-BglII fragment. They are designated frames 22 to 30 and their positions in the sequence of the 13.1-kb fragment are as follows (in the 5′-3′ sequence from the EcoRI site to the BglII site): Table: Probable open reading frames of the 13.1-kb EcoRI-BglII-EcoRI-BgIII fragment. The positions in the sequence correspond to the positions described in FIG. 43, where the upper strand is given in its 5′- 3′ orientation. The frames 22, 23, 24, 25, 26, 27 and 29 are encoded by the upper strand, in contrast to the frames 28 and 30.

Molecular weightTranslationInitiationStopin kD of theframe numbercodoncodonencoded protein2242918845198223336438861944224389249543812125506088851380552690349676240272796781010114990281083510306210572911656121811918330130591236824321

The representations of the probabilities that open reading frames 22, 23, 24, 25 and 26 are coding, with those observed on the other frames (5 in total) in parallel, are given in FIG. 45. These 5 frames are encoded by the same strand.

EXAMPLE 4
Genetic Studies on the DNA Fragments Carrying cob Genes

This example shows the relationship which exists between the different open reading frames identified above and the genes involved in the biosynthesis of cobalamins and/or cobamides carried by these same fragments. These genes are identified by a genetic study as described below.

4.1—Genetic Study of the 5.4-kb Fragment

Plasmid pXL723 is plasmid pRK290 (Ditta et al., 1980) containing the 2264-bp EcoRI-HindIII fragment corresponding to the right-hand portion of the fragment studied, cloned at the EcoRI site of pRK290 (FIG. 11). The construction of the other plasmids used in this study (pXL302, pXL1397, pXL545, pXL545Ω, pXL556 and pXL1500) is described in the legend to FIGS. 11 and 12.

Insertions were obtained in plasmid pXL723 using the technique of de Bruijn and Lupski, 1984. Insertions of transposon Tn5 into plasmid pXL723 were selected and then mapped in the 5.4-kb fragment (FIG. 12). pXL723 complements the Cob mutant G572 of Pseudomonas putida and the Cob mutant G634 of Agrobacterium tumefaciens. These insertions are classified in two groups of inactivating insertions: either those which no longer permit complementation of the Cob mutant G572, or those which abolish the complementation of the Cob mutant G634 (FIG. 12). Insertions which inactivate the complementation of mutant G572 are mapped in open reading frame 4 (these are insertions 15, 27, 68, 81 and 97); open reading frame 4 hence corresponds to a cob gene. The latter is designated cobC. Insertions which inactivate the complementation of mutant G634 are mapped in frame 5 (these are insertions 66 and 107, FIG. 12); open reading frame 5 hence corresponds to a cob gene. The latter is designated cobD. Moreover, insertions with a transposon Tn5Sp′ were produced. Transposon Tn5Sp′ was constructed in the laboratory by cloning a BamHI cassette, containing the spectinomycin resistance gene originating from plasmid pHP45Ω (Prentki and Krisch, 1984), at the BamHI site of transposon Tn5 (Jorgensen et al., 1979). These insertions were made in the chromosome of Pseudonomas denitrificans strain SBL27 Rif′. Strain SBL27 is a strain of Pseudomonas denitrificans from which SC510 is derived by several mutageneses. SBL27 produces 10-fold less cobalamins than SC510 on PS4 medium. Of 10,000 clones of strain SBL27 Rif′ each carrying an insertion of transposon, more than 30 had lost the capacity to synthesise cobalamins. Some of these clones possessed an insertion in the fragment studied in this example. These insertions were mapped by restriction analysis according to Southern's method (Southern, 1975). The sites of insertions of the transposon in these different mutants are given in FIG. 12. One of these insertions, number 2639, occurs in the cobC gene; this insertion is complemented by plasmid pXL302, which carries a fragment containing the cobC gene (FIG. 12). Two insertions, designated 2636 and 2638, are in open reading frame 3. These mutants are blocked in the biosynthesis of cobalamins, and they are complemented by plasmid pXL1397 which contains only open reading frame 3, but non-complemented by plasmid pXL302 which contains the cobC and cobD genes (FIG. 12). Both of these insertions are hence in another gene. With open reading frame 3, we associate the cobB gene. An insertion 2933 is placed in open reading frame 2; it is complemented by plasmid pXL1500 which contains open reading frame 2; this insertion is non-complemented by plasmid pXL1397, which contains the cobB gene and which complements the two insertions in cobB. In this instance, the insertion is hence in another gene; with open reading frame 2, we associate a gene designated cobA.

A kanamycin resistance cassette originating from plasmid pUC4K (Barany et al., 1985) was introduced at the NotI site of the ClaI (position 0 in the sequence)-RsaI (position 1686 in the sequence) fragment cloned into a plasmid pUC8 (Viera and Messing, 1982); the NotI site in question is located at position 771 in frame 1 (see the sequence in FIG. 7); two insertions were adopted, each corresponding to a different orientation of the resistance cassette. These fragments, each carrying an insertion of the resistance cassette, were cloned into plasmid pRK404 (Ditta and al.) to give plasmids pXL1630 and 1631. These plasmids were introduced by conjugative transfer into Pseudomonas denitrificans strain SC510 Rif′, and then, by a series of cultures/dilutions in the absence of the selective antibiotic for the plasmid (tetracycline), double recombinants which had exchanged the plasmid fragment with the chromosomal fragment and had lost the plasmid were found. Two strains were thereby characterised:

i) one is designated SC510:1631 Rif′; in this strain, the kanamycin resistance cassette is inserted in the chromosome at the NotI site (occurring in frame 1); the polarities of the transcriptions of the kanamycin resistance gene and that of open frame 1 are opposite,

ii) the other insertion is designated SC510:1630 Rif′; the resistance cassette is inserted at the same site, but the transcription of the resistance gene has the same polarity as that of the complete open reading frame 1.

These two strains both have a rate of synthesis of cobalamins at least 100-fold lower than that of SC510.

Plasmid pXL545Ω corresponds to plasmid pXL545 into which the spectinomycin resistance cassette of plasmid pHP45Ω has been inserted at the BamHI site. This plasmid (FIG. 12), which contains the 814-bp ClaI-HindIII fragment (in which only open reading frame 1 is complete) complements only mutant SC510:1630 Rif′. This suffices to define a new gene, since this mutant is complemented by a plasmid which only contains the complete open reading frame 1. Open reading frame 1 corresponds to a gene of the pathway of biosynthesis of cobalamins and/or cobamides. This gene is designated cobE. The absence of complementation of mutant SC510:1631 Rif′ by plasmid pXL545Ω is possibly due to the fact that the cobA, cobB, cobC, cobD and cobE genes, or a part of them, belong to the same operon, and that the insertion in cobE which preserves a transcription in the direction of transcription of the operon may be complemented only by trans expression of the cobE gene. In contrast, mutant SC510:1631 Rif′, for its part, can be complemented only by a plasmid which permits trans expression of the cobA to cobE genes.

The 5.4-kb ClaI-HindIII-HindIII-HindIII fragment hence contains five cob genes designated cobA, cobB, cobC, cobD and cobE.

4.2—Genetic Studies of the 8.7-kb Fragment

Plasmid pXL367 is pRK290 (Ditta et al., 1980) containing the 8.7-kb EcoRI fragment cloned at the EcoRI site (FIG. 13).

Insertions of transposon Tn5 into plasmid pXL367 were selected using the technique already described (de Bruijn and Lupski, 1984). The insertions in the 8.7-kb fragment were mapped. In the same manner, insertions of transposon Tn3lacZ were obtained according to the method already described (Stachel et al., 1985) and mapped. 29 insertions of transposon Tn5 and 13 insertions of transposon Tn3lacZ were thus mapped. The precise position of these insertions in the 8.7-kb fragment is given in FIG. 14. Plasmids each carrying a single insertion in the 8.7-kb fragment were introduced by conjugative transfers into the Cob mutants of Agrobacterium tumefaciens G164, G609, G610, G611, G612, G613, G614, G615, G616, G620 and G638. These mutants are all complemented by pXL367. Insertions which no longer permit the complementation of the different mutants were sought. They correspond to an insertion in the gene responsible for complementation of the corresponding mutant. The results of the complementations of the different mutants for their character of production of cobalamins (Cob phenotype) are given in FIG. 14. If the recombinant mutant produces less than threefold less cobalamins than are produced by the same mutant with plasmid pXL367, it is considered to be non-complemented. Of the mutants studied, G164, G609, G610, G611, G612, G613, G614, G615, G616, G620 and G638, eight different classes of inactivating insertions of transposons leading to a mutated phenotype are observed. These classes characterise insertions by the absence of complementation of one or more mutants by plasmids pXL367 carrying these same insertions. Each class hence corresponds to a mutated gene. It is observed that the insertions belonging to the same class are positioned beside one another. Eight classes of insertions are thus observed, which enable eight genes to be defined. Each class of insertions defines a minimum fragment which must be contained in the corresponding gene. FIG. 14 demonstrates a perfect correlation between the regions bounded by each class, in respect of the restriction map, and the open reading frames described above (Example 3). It is found, in effect, that, for each class of insertions, the transposons are always inserted in a portion of the 8.7-kb fragment which is contained in a single open reading frame. Each class of insertions is hence associated with one, and only one, open reading frame. The open reading frames indicated above hence each code for a protein involved in the pathway of biosynthesis of cobalamins and/or cobamides. The open reading frames each correspond to genes involved in the biosynthesis of cobalamins and/or cobamides. These open reading frames are referred to as cobF, cobG, cobH, cobI, cobJ, cobK, cobL and cobM for frames 6 to 13, respectively. The position of these genes relative to the restriction map is shown in FIG. 14.

4.3—Genetic Study of the 4.8-kb Fragment

Plasmid pXL1558 is plasmid pRK290 (Ditta et al., 1980) containing the 12-kb HindIII-HindIII fragment of pXL154 (Cameron et al., 1989) cloned at the EcoRI site of pRK290 (FIG. 36). The construction of the other plasmids used in this study (pXL233 and pXL843) is described in the legend to FIG. 36.

Tn5Sp insertions were obtained in plasmid pXL1558. First, a strain containing a transposon Tn5Sp was constructed; this was done by transforming strain C2110 (Stachel et al., 1985) using plasmid pRK2013Tn5Sp (Blanche et al., 1989); since it has a ColEl origin of replication, plasmid pRK2013Tn5Sp does not replicate in strain C2110, which is polA-. The colonies obtained after transformation, which are resistant to spectinomycin, hence have transposon Tn5Sp in their chromosome; a colony is then reisolated, after which the insertion of the transposon is then transduced using phage P1 in strain MC1060 as described previously (Miller, 1972). Strain MC1060 Tn5Sp is transformed with plasmid pXL1558; plasmid pXL1558 is then mobilised by conjugation using pRK2013 in C600 Rif′. Conjugants resistant to tetracycline (for plasmid pXL1558) and to spectinomycin (for the transposon) are then selected. Such conjugants must contain plasmid pXL1558 in which transposon Tn5Sp has been inserted. Insertions carried in plasmid pXL1558, and more precisely in the 12-kb fragment, are then mapped by restriction digestion; 23 insertions are thereby obtained and mapped on the 12-kb fragment; the position of these different insertions in the fragment is presented in FIG. 37. These 23 insertions were introduced into the chromosome of strain SC510 Rif′ after conjugative transfer of p-XL1558::Tn5Sp, followed by introduction of plasmid pR751. Plasmid pR751 is a trimethoprim-resistant plasmid of the same incompatibility group as pXL1558 (incP, Thomas and Smith, 1987). By culturing non-selectively for pXL1558 (absence of tetracycline) but selectively for pR751 and the transposon (presence of trimethoprim and of spectinomycin), the exchange of the mutation carried by pXL1558::Tn5Sp with the chromosome and also the segregation of pXL1558 are obtained by the technique of marker exchange by double homologous recombination, as already described (Schell et al., 1988). The strains thereby selected carry the transposon in their chromosome. The double homologous recombination is verified by Southern's method (Southern, 1975). In this way, 23 SC510 Rif′::Tn5Sp strains in the 12-kb fragment were identified.

Furthermore, another Tn5Sp insertion obtained by random mutagenesis of transposon Tn5Sp in strain SBL27 Rif′ (Blanche et al., 1989) was mapped on the 12-kb fragment by restriction analysis according to Southern's method (Southern, 1975), see FIG. 37; this strain is designated SBL27 Rif′::Tn5Sp 1480.

The level of cobalamin synthesis is determined for these 24 strains cultured in PS4 medium according to the protocol already described (Cameron et al., 1989), and the Cob-phenotype is assigned to strains producing at least 1000 (or 100) times less vitamin B₁₂than the parent strain SC510 Rif′ (or SBL27 Rif′), FIG. 37. It is thus observed that 6 of these chromosomal insertions lead to a Cob-phenotype in P. denitrificans; they are the insertions 31.1, 41.3, 45, 55, 22.1 and 1480.

Three plasmids pxL233, pXL837 (Cameron et al.) and pXL843 are introduced by conjugative transfers into three strains possessing the Cob-phenotype, namely SC510 Rif′::Tn5Sp 31.1, SC510 Rif′::Tn5Sp 45 and SBL27 Rif′::Tn5Sp 1480. These three mutants each have a different complementation profile for cobalamine synthesis. In effect, SBL27 Rif′::Tn5Sp 1480 is complemented by pXL837 and pXL843 but not by pXL233; the mutant SC510 Rif′::Tn5Sp 45 is complemented only by pXL843; the mutant SC510 Rif′::Tn5Sp 31.1 is complemented by plasmid pXL843 and also by plasmid pXL233 (see FIG. 37). The data presented hence enable it to be concluded, on the basis of the results of the complementations of the three mutants, that the three mutants are different and that, for each of them, transposon Tn5Sp has been inserted into a different cob gene.

Furthermore, plasmids pXL1558::Tn5Sp 41.3, pXL1558::Tn5Sp 45 and pXL1558::Tn5Sp 22.1 are introduced by conjugative transfers into strain G2035 (Cameron et al., 1989), and do not complement it. Plasmid pXL1558 complements this mutant, in contrast to plasmid pXL1558::Tn5Sp 31.1.

The phenotype and complementation data enable us to define 3 classes of insertions; each of these classes is represented by the following insertions : 31.1, class 1; 45, 41.3, 55 and 22.1, class 2; 1480, class 3.

For each class of insertions, the transposons are always inserted in a portion of the 4.8-kb fragment which is contained in a single open reading frame (ORF14, ORF16 and ORF17, as defined in Example 3). Each class of insertions is associated with a single open reading frame. The open reading frames indicated above hence code for a protein involved in the pathway of biosynthesis of cobalamins and/or cobinamides. These open reading frames are referred to as cobX, cobS and cobT for frames 14, 16 and 17. The position of these genes relative to the restriction map is shown in FIG. 37. Open reading frame 15 is not a gene involved in the biosynthesis of coenzyme B₁₂.

4.4—Genetic Studies of the 3.9-kb Fragment

Plasmid pXL1557 is plasmid pRK290 (Ditta et al., 1980) containing the 9-kb HindIII-BamHI fragment of pXL519 cloned at the EcoRI site of pRK290 (FIG. 38). The construction of the other plasmids used in this study (pXL1286, pXL1303, pXL1324) is described in the legend to FIG. 38. Moreover, the 2-kb BglII-XhoI fragment (positions in the sequence presented in FIG. 33: 251 and 2234) of plasmid pXL519 is cloned at the BamHI-SalI sites of plasmid pXL435 (Cameron et al) to generate plasmid pXL699.

Tn5Sp insertions were obtained in plasmid pXL1557 according to the technique described in Example 4.3. Insertions of transposon Tn5Sp into plasmid pXL1557, then designated pXL1557::Tn5Sp, were selected. Those which are mapped in the 9-kb fragment (FIG. 39) were introduced into the chromosome of strain SC510 Rif′ after conjugative transfer of pXL1557::Tn5Sp and marker exchange by double homologous recombination as described in 4.3.

The double homologous recombination is verified by Southern's method (Southern, 1975). In this way, 20 SC510 Rif′::Tn5Sp strains were identified.

Furthermore, two other Tn5Sp insertions obtained by random mutagenesis of transposon Tn5Sp in strain SBL27 Rif′ (Blanche et al., 1989) were mapped on the 9-kb fragment by restriction analysis according to Southern's method (Southern, 1975), see the insertions 1003 and 1147 in FIG. 39.

The level of cobalamin synthesis is determined for these 22 strains cultured in PS4 medium according to the protocol already described (Cameron et al., 1989), and the Cob-phenotype is assigned to strains producing 1000 (or 100) times less vitamin B₁₂than the parent strain SC510 Rif′ (or SBL27 Rif′), FIG. 39. Only the 4 insertions 1, 1003, 23 and 1147 result in a Cob-phenotype in P. denitrificans.

Four plasmids pXL699, pXL1286, pXL1303 and pXL1324 are introduced by conjugative transfers into the four strains possessing the cob-phenotype, namely SC510 Rif′::Tn5Sp 1, SBL27 Rif′::Tn5Sp 1003, SC510 Rif′::Tn5Sp 23 and SBL27 Rif′::Tn5Sp 1147. Plasmid pXL699 complements the first two mutants (SC510 Rif′::Tn5Sp 1, SBL27 Rif′::Tn5Sp 1003), but plasmid pXL1303 does not complement them, plasmid pXL1324 complements the other two mutants (SC510 Rif′::Tn5Sp 23 and SBL27 Rif′::Tn5Sp 1147) but plasmid pXL1286 does not complement them.

Furthermore, plasmid pXL1557::Tn5Sp 1, is introduced by conjugative transfer into strain G2040, and does not complement it, whereas plasmids pXL1557, pXL1557::Tn5Sp 6A, pXL1557::Tn5Sp 54, pXL1557::Tn5Sp 48, pXL1557::Tn5Sp 21, pXL1557::Tn5Sp 8, pXL1557::Tn5Sp 23, also introduced by conjugative transfers, complement it (see FIG. 39).

The phenotype and complementation data enable 2 classes of insertions to be defined. For each class of insertions, the transposons are always inserted in a portion of the 3.9-kb fragment which is contained in a single open reading frame (ORF19 and ORF20 as defined in Example 3).

Each class of insertions is associated with a single open reading frame. The open reading frames indicated above code for a protein involved in the pathway of biosynthesis of cobalamins and/or cobinamides. These open reading frames are referred to as cobV and cobU for frames 19 and 20. Frames 18 and 21 are not genes involved in the pathway of biosynthesis of coenzyme B₁₂. The position of these genes relative to the restriction map is shown in FIG. 39. The insertions 48, 21 and 8 are mapped between the cobU and cobV genes.

4.5—Genetic Studies of the 13.4-kb Fragment

4.5.1. Studies on the 4327-bp EcoRI-BqII Fragment.

Plasmid pXL189 (Cameron et al., 1989), which contains at least one cob gene, carries a 3.1-kb insert which, except for 300 bp, corresponds to a 4.26-kb EcoRI-ClaI fragment (see FIG. 45). pXL189 was subjected to a mutagenesis with transposon Tn5, as described previously (De Bruijn and Lupski (1984)). 13 insertions were thereby mapped in the insert of pXL189, as presented in FIG. 46. These 13 mutant plasmids, as well as pXL189, were conjugated in two A. tumefaciens mutants, G632 and G633, which are mutants complemented by pXL189 (Cameron et al., 1989). Only the insertion 58 proved to be an inactivating insertion. This result shows that the two mutants G632 and G633 correspond to a mutation in the same gene, and that, moreover, the only gene of P. denitrificans which could be responsible for their complementation corresponds to open reading frame 26 (see FIG. 46), since insertion 58 is mapped in this open reading frame; in addition, it is the only insertion of the 13 which is mapped in this open reading frame. A cob gene, designated cobO, is hence associated with open reading frame 26.

To know whether the four open reading frames (open reading frames 27 to 30) identified in this fragment correspond to cob genes, a spectinomycin resistance cassette from plasmid pHP45Ω (Prentki and Krisch, 1984) was specifically inserted into each of these genes, and then introduced into the chromosome of P. denitrificans SC510 Rif′ by homologous recombination so as to obtain mutants of insertions in each of these open reading frames. For this purpose, the EcoRI-ClaI fragment (respective positions 8818 and 13082 in the sequence presented in FIG. 43) was used. This fragment, which carries the open reading frames 27 to 30, was purified from pXL157 (Cameron et al., 1989); an EcoRI linker was added to the ClaI end after the latter had been filled in with the Klenow fragment of E. coli DNA polymerase. This fragment was then cloned into plasmid pUC13 (Viera et al., 1982) at the EcoRI site. The plasmid thus constructed was referred to as pXL332. Insertions of the spectinomycin resistance cassette from plasmid pHP45Ω (Prentki and Krisch, 1984) were carried out on pXL332. These insertions were done separately at the SmaI (position 9868, open reading frame 27), BamHI (position 10664, open reading frame 28), ClaI (position 11687, open reading frame 29) and NcoI (position 12474, open reading frame 30) sites by total or partial digestions of pXL332 with the corresponding enzymes, and then, if necessary, filling-in of these ends with the Klenow fragment of E. coli DNA polymerase, followed by ligation with the 2-kb SmaI fragment of pHP45≠ (Prentki and Krisch, 1984) containing a spectinomycin resistance gene; these insertions are designated Ω2, Ω1, Ω3 and Ω4, respectively, as presented in FIG. 46. The EcoRI fragments carrying these different insertions were then cloned into pRK404 (Ditta et al., 1985) at one of the two EcoRI sites. The 4 plasmids carrying these different insertions were then introduced by conjugation in SC510 Rif′, as described above. Plasmid pR751 (Thomas and Smith, 1987) was then introduced into the transconjugants. The exchange of mutations carried by the 4 different derivatives of pRK404 and the chromosome of SC510 Rif′ could be selected as described (see Example 4.3). 4 strains were thereby obtained. These strains each carry an insertion of the resistance cassette in one of the four open reading frames 27 to 30. These insertions were verified by analysis of the genomic DNA by Southern blotting (Southern, 1975). The cobalamin production of these different strains was studied. They all showed a Cob+ phenotype on culturing in PS4 medium. This result indicates that these four open reading frames do not participate in the biosynthesis of coenzyme B₁₂. However, it is possible that one or more of these frames code for proteins which participate, e.g., in the conversion of coenzyme B₁₂to methylcobalamin for example, i.e. the synthesis of another cobalamin or even of another corrinoid.

4.5.2. Study of the 9.1-kb EcoRI-EcoRI Fragment.

Various plasmids are used in this study; plasmid pXL1560 is plasmid pRK290 (Ditta et al., 1980) containing the 9.1-kb EcoRI-EcoRI fragment of pXL156 (Example 1) cloned at the EcoRI site of pRK290 (see FIG. 46). The construction of the other plasmids used in this study (pXL618, pXL593, pXL623, pXL1909, pXL1938, pXL1908, pXL221, pXL208, pXL297) is described in the legend to FIG. 45.

Tn5Sp insertions were obtained in plasmid pXL1560. Strain MC1060 Tn5Sp transformed with plasmid pXL1560 was used to obtain insertions of transposon Tn5Sp into the pXL1560 fragment; 27 insertions were thereby obtained and mapped on the 9.1-kb fragment; the position of these different insertions in the fragment is presented in FIG. 4. These 27 insertions were introduced into the chromosome of strain SC510 Rif′ after conjugated transfer of pXL1560::Tn5Sp, followed by introduction of plasmid pR751. Plasmid pR751 is a trimethoprim-resistant plasmid of the same incompatibility group as pXL1560 (incP, Thomas and Smith, 1987). By culturing non-selectively for pXL1560 (absence of tetracycline) but selectively for pR751 and the transposon (presence of trimethoprim and of spectinomycin), the exchange of the mutation carried by pXL1560::Tn5Sp with the chromosome and also the segregation of pXL1560 are obtained; this technique of marker exchange by double homologous recombination is equivalent to that already described by Schell et al., 1988. The strains thus selected carry the transposon in their chromosome.

The double homologous recombination is verified by Southern's method (Southern, 1975). In this way, 27 SC510 Rif′::Tn5Sp strains each possessing a different insertion of transposon Tn5Sp in the 9.1-kb fragment were identified.

The level of cobalamin synthesis is determined for these 27 strains cultured in PS4 medium, and the Cob-phenotype is assigned to strains producing at least 1000 times less vitamin B₁₂than the parent strain SC510 Rif′, FIG. 46. It is thus observed that 18 out of the 27 of these chromosomal insertions lead to a Cob-phenotype in P. denitrificans, as shown in FIG. 46.

The insertions 19, 32, 24, 27, 37, 39, 26, 11 and 14 are mapped in open reading frame 22 (see FIG. 46). All these insertions are complemented by plasmid pXL618, which contains only open reading frame 22. We deduce from this that open reading frame 22 corresponds to a cob gene, which we referred to as cobO. No insertion was obtained in open reading frame 23; however, plasmid pXL623, which contains only this open reading frame (see FIG. 46), complements two cob mutants of Agrobacterium tumefaciens, G642 and G2043 (Cameron et al., 1989). Open reading frame 23 hence corresponds to a cob gene designated cobP. The insertions 23, 13, 12, 30, 22, 40, 35, 10 and 17 which are mapped in open reading frames 24 and 25 lead to a Cob-phenotype in SC510 Rif′. There hence appear to be two open reading frames whose product is involved in the biosynthesis of cobalamins. However, it cannot be ruled out that these insertions have polar effects on the genes positioned on the 3′ side, such as cobO. It is hence appropriate to study the complementation of these mutants in order to determine whether the Cob-phenotype does not result from a polar effect.

The Cob mutants of Agrobacterium tumefaciens, G622, G623 and G630, complemented by pXL156, were studied. These mutants are not complemented by plasmid pXL189 (Cameron et al., 1989), which contains cobO as the only cob gene. In contrast, they are complemented by plasmid pXL1908, which contains cobO and open reading frame 25 in addition to the open reading frames 27 to 30 (see FIG. 45). The latter frames cannot be responsible for the complementation of these mutants, since the proteins for which they code do not participate in the coenzyme B₁₂pathway. Hence, the observed complementations can only result from open reading frame 25. In addition, the SC510 Rif′ Tn5Sp mutants mapped in this same open reading frame (these are the mutants 22, 40, 35, 10 and 17) are complemented by plasmid pXL1908, see FIG. 46, (carrying cobO and frame 25), whereas at least two of them are not complemented by pXL189, which contains only cobO as a cob gene. These results show clearly that open reading frame 25 is a cob gene; this cob gene is designated cobN.

The SC510 Rif′ Tn5Sp mutants 23, 13 and 12, which have the Cob-phenotype, are mapped in open reading frame 24. These mutants are not complemented by plasmid pXL623, which contains only the cobP gene. In contrast, these mutants are complemented by plasmid pXL593 which contains cobP and open reading frame 24, thereby indicating that open reading frame 24 is responsible for their complementation. Open reading frame 24 is hence a cob gene, which is designated cobW.

EXAMPLE 5
Genes and Proteins

5.1-5.4-kb Fragment

Five genes (cobA, cobB, cobC, cobD and cobE) are hence defined on the 5.4-kb ClaI-HindIII-HindIII-HindIII fragment. They code, respectively, for the following COB proteins: COBA, COBB, COBC, COBD and COBE. The coding portions of the genes (cobA to cobE) are described in FIG. 15, as well as the sequences of the COBA to COBE proteins. Properties of each of these proteins are also presented (amino acid composition, isoelectric point, index of polarity and hydrophilicity profile).

5.2-8.7-kb Fragment

Eight genes are hence defined on the 8.7-kb fragment. These cobF to cobM genes code, respectively, for the following COB proteins: COBF, COBG, COBH, COBI, COBJ, COBK, COBL, and COBM. The coding portions of the genes (cobF to cobM) are described in FIG. 16, as well as the sequences of the COBF to COBM proteins. Properties of each of these proteins are also presented (amino acid composition, molecular weight, isoelectric point, index of polarity and hydrophilicity profile).

5.3-4.8-kb Fragment

Three genes (cobX, cobS, cobT) are defined on the 4.8-kb SalI-SalI-SalI-SalI-SalI-BglI fragment. They code, respectively, for the following proteins: COBX, COBS and COBT. The coding portions of these genes are described in FIG. 40, as well as the sequences of the COBX, COBS and COBT proteins. Arbitrarily, the ATG at position 1512 of cobs has been chosen as the initiation codon, rather than that located at position 1485 (see FIG. 32). Properties of each of these proteins are also shown (amino acid composition, isoelectric point, index of polarity and hydrophobicity profile). COBT possesses a hydrophilic pocket corresponding to amino acids 214 to 305.

5.4-3.9-kb Fragment

Two genes (cobU and cobV) are defined on the 3.9-kb SstI-SstI-BamHI fragment. They code, respectively, for the following proteins: COBU and COBV. The coding portions of these genes are described in FIG. 41, as well as the sequences of the COBU to COBV proteins. Properties of each of these proteins is also shown (amino acid composition, isoelectric point, index of polarity and hydrophobicity profile).

5.5-13.4-kb Fragment

Five cob genes are defined on the 13.4-kb fragment (cobQ, cobP, cobW, cobN and cobO and cobV). They code, respectively, for the following proteins: COBQ, COBP, COBW, COBN and COBO. The coding portions of these genes (cobQ, cobP, cobW, cobN and cobO) are described in FIG. 46, as well as the sequences of COBQ, COBP, COBW, COBN and COBO proteins. Properties of each of these proteins are also shown (amino acid composition, isoelectric point, index of polarity and hydrophobicity profile).

From the hydrophilicity profiles, which were produced according to the programmes of Hopp and Woods (1981), all the COB proteins with the exception of COBV are probably soluble proteins, as opposed to membrane proteins, since the absence of large hydrophobic domains is noted. COBV is either a membrane protein, since 4 long hydrophobic domains are noted (see FIG. 41), or a cytoplasmic protein having large hydrophobic domains.

For all the amino acid sequences of the COB proteins, a methionine is indicated as the first amino acid at the NH₂-terminal position. It is understood that this methionine may be excised in vivo (Ben Bassat and Bauer, 1984). Rules relating to the in vivo excision of NH₂-terminal methionine by methionine aminopeptidase are known to have been proposed. (Hirel et al., 1989).

Moreover, these protein sequences were compared with the Genpro proteins, Genpro being a Genbank protein extraction (version 59) augmented by putative coding portions larger than 200 amino acids, according to the programme of Kanehisa (1984). No significant homology could be demonstrated with the parameters used on Genbank version 59, except for COBT. In effect, the COBT protein possesses a “core of acidic amino acids” between (amino acid) positions 224 and 293 (see FIG. 40); in this portion of the protein, more than one amino acid out of 2 is a glutamic or aspartic acid residue; this core of acidic amino acids renders the protein homologous over this region, according to the programme of Kanehisa (1984), with other proteins also having such an acidic core. The most homologous proteins are: GARP protein of Plasmodium falciparum (Triglia et al., 1988), rat cardiac troponin T (Jin and Lin, 1989), human and rat prothymosin (Eschenfeld and Berger, 1986), an androgen-dependant rat protein that binds to spermine (Chang et al., 1987), and the human, rat and chicken “mid-size neurofilament subunit”, proteins (Myers et al., 1987, Levy et al., 1987, Zopf et al., 1987). The function of these cores rich in acidic residues is unknown; however, this acidic core should either permit the binding of metal cations such as Co⁺⁺, which would give the COBT protein the role of a cobalt metallothionein, or else permit interactions with other proteins.

EXAMPLE 6
Enzymatic Studies

6.1—Identification of COB Proteins and Their Genes from Purified Enzymatic Activities

This example describes how, from a purified protein, after its NH₂-terminal sequence has been established, it is possible to find the corresponding structural gene among sequenced cob genes.

6.1.1. Identification of the COBA Protein Encoded by the cobA Gene

The purification of Pseudomonas denitrificans SUMT has been described (F. Blanche et al., 1989). The NH₂-terminal sequence of the protein thus purified could be determined according to the technique described above. The first ten amino acids were identified:

12345678910MetIleAspAspLeuPheAlaGlyLeuPro

The NH₂terminal sequence of the COBA protein (FIG. 15) corresponds exactly to this sequence. The molecular weight of the purified SUMT, estimated by 12.5% SDS-PAGE electrophoresis, is 30,000. The COBA protein has a molecular weight deduced from its sequence of 29,234 (FIG. 15). The correspondences between the NH₂-terminal sequences and the molecular weights indicate clearly that the COBA protein corresponds to SUMT. The cobA gene is the SUMT structural gene.

6.1.2. Identification of the COBB Protein Encoded by the cobB Gene

a) Assay of Cobyrinic Acid a,c-Diamide Synthase Activity

This example illustrates the assay of an activity of the pathway of biosynthesis of corrinoids which has never yet been described. The enzyme in question is cobyrinic acid a,c-diamide synthase (CADAS), which catalyses the amidation of two carboxylic acid functions of the corrin or decobalt-ocorrin ring-system at positions a and c (FIG. 17). The donor of the NH₂group is L-glutamine, and the reaction consumes 1 molecule of ATP per amidation of each carboxylic acid function. The assay which is described below applies to the diamidation reaction of cobyrinic acid; with a few modifications (detection in HPLC at 330 nm in particular), it applies to the diamidation reaction of hydrogenobyrinic acid.

The incubation mixture (0.1 M Tris-HCl pH 8 (250 μl)) containing ATP (1 mM), MgCl₂(2.5 mM), glut-amine (100 μm), cobyrinic acid (50 μM) or hydrogeno-byrinic acid (50 μM) and cobyrinic a,c-diamide synthase (approximately 1 unit of activity) is incubated for 1 hour at 30° C. At the end of the incubation, an aqueous solution (125 μl) of KCN (2.6 g/l) and 0.2 M HCl (125 μl) are added to the mixture, which is then heated to 80° C. for 10 minutes and thereafter centri-fuged for 5 minutes at 5,000 g. An aliquot (50 μl) of the centrifugation supernatant is analysed in HPLC. It is injected onto a 25-cm Nucleosil 5-C₁₈column and eluted with a gradient from 0 to 100% of buffer B in A in the course of 30 minutes; buffer A: 0.1 M potassium phosphate pH 6.5, 10 mM KCN; buffer B: 0.1 M potassium phosphate pH 8, 10 mM KCN/acetonitrile (1:1). The corr-inoids are detected by means of their UV absorption at 371 nm. The unit of enzymatic activity is defined as the quantity of enzyme necessary for synthesising 1 nmol of amide groups per hour under the conditions described.

b) Purification of Pseudomonas Denitrificans Cobyrinic Acid a,c-Diamide Synthase Activity

This experiment illustrates how a Pseudomonas denitrificans protein participating in the pathway of biosynthesis of cobalamins may be purified.

Using the assay described in Example 6.1.2 a), the purification of Pseudomonas denitrificans cobyrinic acid a,c-diamide synthase is carried out as described below.

In a typical purification experiment, wet cells (7 g) of strain SC 510 Rif′, into which plasmid pXL1500 has been introduced (see Example 4.1. for the description of pXL1500, as well as FIG. 12), are suspended in 0.1 M Tris-HCl pH 7.7 (30 ml) and sonicated for 15 minutes at 4° C. The crude extract is then recovered by centrifugation for 1 hour at 50,000 g, and a portion (10 ml) of this extract is injected onto a Mono Q HR 10/10 column equilibrated with the same buffer. The proteins are eluted with a linear KCl gradient (0 to 0.5 M). The fractions containing the enzymatic activity (demonstrated by means of the test described in Example 6.2 b)) are combined and concentrated to 2.5 ml. After dilution with 25 mM Tris-HCl pH 7.7 (1 ml), the proteins are fractionated on a Mono Q HR 5/5 using the above KCl gradient (0 to 0.5 M). The active fractions are combined, and 0.1 M Tris-HCl pH 7.7 (1 ml) containing 1.7 M ammonium sulphate is added to the sample, which is then chromatographed on a Phenyl-Superose (Pharmacia) column with a decreasing ammonium sulphate gradient (1.0 M to 0 M). The fractions containing the desired activity are combined and chromatographed on a Bio-Gel HPHT (Bio-Rad) column with a potassium phosphate gradient (0 to 0.35 M).

After this step, the enzyme is more than 95% pure. It shows no contaminant protein in SDS-PAGE. The purity of the protein is confirmed by the uniqueness of the NH₂-terminal sequence. Its molecular weight in this technique is 45,000. The different steps of purifi-cation of CADAS, with their purification factor and their yield, are given in the table below.

TABLEPurification of CADASSp. activityPurificationVolProteins(u/mg ofPurificationstep(ml)(mg)proteins)Yieldfactor¹Crude extract102008.5——MonoQ 10/101215.11089612.7MonoQ 5/533.752726032Phenyl-10.86585043100SuperoseBio-Gel HPHT20.451132035155
¹This factor is calculated from the increase in the specific activity of the fractions during the purification.

c) NH₂-terminal Sequence of Pseudomonas Denitrificans Cobyrinic Acid a,c-Diamide Synthase and Identification of the Pseudomonas Denitrificans Structural Gene Coding for this Activity

This example illustrates how the NH₂-terminal sequence of a protein which participates in the pathway of biosynthesis of cobalamins enables the structural gene which codes for this protein to be identified.

The NH₂-terminal sequence of Pseudomonas denitrificans cobyrinic acid a,c-diamide synthase, purified as described in Example 6.1.2 b), was determined as described above. 15 residues were identified:

123456789101112131415SerGlyLeuLeuIleAlaAlaProAlaSerGlySerGlyLysThr

The NH₂-terminal sequence of the COBB protein (FIG. 15) corresponds exactly to this sequence except that, in the sequence presented in FIG. 15, a methionine precedes the peptide sequence determined by direct sequencing. It follows from this that the amino-terminal methionine is definitely excised in vivo by methionine aminopeptidase (Ben Bassat and Bauer, 1987). The molecular weight of the purified CADAS, estimated by 12.5% SDS-PAGE electrophoresis, is 45,000. The COBB protein has a molecular weight deduced from its sequence of 45,676 (FIG. 15). The correspondences between the NH₂-terminal sequences and the molecular weights indicate clearly that the COBB protein corresponds to CADAS. The cobB gene is the CADAS structural gene.

6.1.3. Identification of the COBI Protein Encoded by the cobI Gene

a) Assay of an S-Adenosyl-L-Methionine:Precorrin-2 Methyltransferase Activity

This example illustrates the assay of an enzymatic activity of the pathway of bipsynthesis of corrinoids which has never yet been described. The enzyme in question is S-adenosyl-L-methionine: precorrin-2 methyltransferase (SP₂MT), which catalyses the transfer of a methyl group from S-adenosyl-L-methionine (SAM) to precorrin-2 to give precorrin-3 (FIG. 18). Factors II and III, oxidation products of precorrin-2 and precorrin-3, respectively, have already been purified from cell extracts of Propionibacterium shermanii (Battersby and MacDonald, 1982, Scott et al., 1984); precorrin-2 and precorrin-3 are recognised as presumed intermediates of coenzyme B₁₂biosynthesis, but they have never been purified as such. For this reason, the corresponding activity has never been either assayed or purified beforehand. The substrate of the enzymatic reaction, precorrin-2, is a very labile molecule which it is not possible to store, since it oxidises spontaneously in the presence of even infinitesimal traces of oxygen (Battersby and MacDonald, 1982). The principle of this enzymatic test hence lies in the possibility of generating precorrin-2 from SAM and δ-aminolevulinic acid at the required moment using an enzymatic extract of strain SC510 Rif′ into which plasmid pXL1500 has been introduced. The incubation must be performed under strictly anaerobic conditions.

The fractions containing SP₂MT are incubated in 0.1 M Tris-HCl pH 7.7 (1 ml) in the presence of 5 mM DTT, 1 mM EDTA, 100 μM [methyl-³H]SAM (1 μCi), 0.8 mM δ-aminolevulinic acid and crude enzyme extract (6 mg) of Pseudomonas denitrificans strain SC510 Rif′ pXL1500 for 3 hours at 30° C. Strain SC510 Rif′ pXL1500 contains a strong SUMT activity (F. Blanche et al., 1989). The tetrapyrrole compounds produced during the incubation are bound to a DEAE-Sephadex anion exchange column and esterified in methanol containing 5% of sulphuric acid in the absence of oxygen. The dimethylated and trimethylated derivatives of uro'gen III are then separated by thin-layer chromatography on silica using dichloromethane/methanol (98.3:1.7) as an eluent system (F. Blanche et al., 1989). The SP₂MT activity is expressed as the ratio of the quantity of trimethylated derivatives obtained to the total of (di- and tri-) methylated derivatives produced, referred to the quantity of protein. The SC510 Rif′ pXL1500 extract introduced in the test does not display detectable SP₂MT activity under the assay conditions (the ratio of precorrin-3 produced to precorrin-2 produced during the test is less than 0.05).

b) Purification of Pseudomonas Denitrificans S-Adenosyl-L-Methionine:Precorrin-2 Methyltransferase

This experiment illustrates how a Pseudomonas denitrificans protein participating in the pathway of biosynthesis of cobalamins may be purified when an assay for the activity in question exists.

The protein is purified from SC510 Rif′ cells containing plasmid pXL253. This is plasmid pKT230 into which the 8.7-kb EcoRI fragment has been inserted (FIG. 13). In a typical purification experiment, wet cells (50 g) of strain SC150 Rif′ into which plasmid pXL253 has been introduced are suspended in 0.1 M potassium phosphate pH 7.7, 5 mM DTT (250 ml) and sonicated for 15 minutes at 4° C. After centrifugation at 50,000 g for 1 hour, the supernatant is passed through a DEAE-Sephadex column (10 ml of gel) to remove the tetrapyrrole compounds. The pH of the crude extract thereby obtained is adjusted to pH 7.7 with 0.1 M KOH. The proteins precipitating at between 33% and 45% ammonium sulphate saturation are collected and dissolved in 0.1 M Tris-HCl pH 7.7, 5 mM DTT (40 ml). This solution is passed through a Sephadex G-25 column eluted with 10 mM Tris-HCl pH 7.7, 5 mM DTT, and the proteins collected are injected onto a DEAE-Trisacryl-M column. The proteins are eluted with a linear gradient of 0 to 0.25 M KCl, and the fractions containing the SP₂MT activity are combined and passed a second time through a Sephadex G-25 column as above. The protein fraction is injected onto an Ultrogel HA (IBF) column equilibrated in 10 mM Tris-HCl pH 7.7, 5 mM DTT. The proteins are eluted with a linear gradient of 0 to 50 mM potassium phosphate pH 7.8 containing 5 mM DTT. The fractions containing the desired activity are combined and injected onto a MonoQ HR 5/5 (Pharmacia) column equilibrated with 50 mM Tris-HCl pH 7.7, 5 mM DTT. The SP₂MT is eluted with a linear gradient (0 to 0.25 M) of KCl. At emergence from the MonoQ step, 12.5% SDS-PAGE electrophoresis with staining with silver salts reveals the enzyme is more than 99% pure. This is confirmed by the uniqueness of the NH₂-terminal sequence of the protein. The molecular weight calculated from the electrophoresis under denaturing conditions (12.5% SDS-PAGE) is 26,500. The steps of purification of SP₂MT with their yields are described in the table below.

TABLEPurification of SP₂MTPurificationVolProteinsPurificationstep(ml)(mg)factor¹Crude extract3006000—Precipitation4015303.9(33-45%)DEAE-Tris-5735516.9acryl-MUltrogel HA307185MonoQ HR 5/51233.5179
¹This factor is calculated from the yield of protein.

c) NH₂-Terminal Sequence of SP₂MT and Identification of the Structural Gene Coding for this Activity

This example illustrates how the NH₂-terminal sequence of a protein participating in the biosynthetic pathway enables the structural gene which codes for this protein to be identified. In the present example, the structural gene in question is that for SP₂MT.

The NH₂-terminal sequence of the purified protein was determined as described above. The first 15 amino acids were identified:

123456789101112131415SerGlyValGlyValGlyArgLeuIleGlyValGlyThrGlyPro

The NH₂-terminal sequence of the COBI protein (FIG. 16) corresponds exactly to this sequence except that, in the sequence presented in FIG. 16, a methionine precedes the peptide sequence deduced from the nucleotide sequence. It follows from this that the amino-terminal methionine is definitely excised in vivo by methionine aminopeptidase (Ben Bassat and Bauer, 1987). The molecular weight of the purified SP₂MT, estimated by 12.5% SDS-PAGE electrophoresis, is 26,500. The COBI protein has a molecular weight deduced from its amino acid sequence of 25,878 (FIG. 16). The correspondences between the NH₂-terminal sequences and the molecular weights indicate clearly that the COBI protein corresponds to SP₂MT. The cobI gene is the SP₂MT structural gene.

6.1.4. Identification of the COBH Protein Encoded by the cobH Gene

a) Assay of Precorrin-8x Mutase Activity

This example illustrates the assay of an enzymatic activity of the pathway of biosynthesis of cobalamine which has never been described hitherto. The enzyme in question is precorrin-8x mutase. This enzyme catalyses the transfer of the methyl group from position C-11 to position C-12 during the conversion of precorrin-8x to hydrogenobyrinic acid (see the nomenclature of the carbon atoms in FIG. 19; PL 68). More generally, it is the enzyme which catalyses the transfer of the methyl group C-11 to C-12, thereby leading to the corrin ring-system. The enzyme is referred to here as a mutase, although it has not been formally demonstrated that the transfer of the methyl group is intramolecular, even though this is very probable.

The enzymatic activity is demonstrated by the conversion of precorrin-8x (5 μM) to hydrogenobyrinic acid during incubations in the presence of enzyme fractions in 0.1 M Tris-HCl pH 7.7, 1 mM EDTA, at 30° C. for 1 h. At the end of the incubation, the reaction is stopped by heating to 80° C. for 10 min and, after centrifugation at 3000×g for 10 min, the hydrogenobyrinic acid formed, present in the supernatant, is analysed by HPLC (see Example 6.1.2.a).

b) Purification of Precorrin-8x Mutase.

The purification of Pseudomonas denitrificans precorrin-8x mutase is carried out as described below.

During this purification, all the buffer solutions are adjusted to pH 7.7.

In a typical purification experiment, cells (50 g) of strain SC510 Rif′, carrying plasmid pXL253 (plasmid pKT230 into which the 8.7-kb fragment has been cloned at the EcoRI site, FIG. 13) and obtained after culture in PS4 medium, are resuspended in 0.1 M potassium phosphate buffer (200 ml) and sonicated for 12 minutes. After centrifugation at 50,000 g for 1 hour, the supernatant is passed through a DEAE-Sephadex column (10 ml of gel) to remove the tetrapyrrole compounds. The pH of the solution is immediately adjusted to 7.7 with 1 M KOH solution. The protein fraction precipitating at between 40 and 60% ammonium sulphate saturation is collected by centrifugation and dissolved in 0.1 M Tris-HCl (50 ml). This sample is then injected onto an Ultrogel AcA 54 (IBF, France) column (gel volume 1,000 ml) and the proteins are eluted at a flow rate of 60 ml/h with 50 mM Tris-HCl. The fractions containing the activity are pooled and injected onto a DEAE-Trisacryl M (IBF, France) column equilibrated with 50 mM Tris-HCl, and the proteins are eluted with a gradient of 0 to 0.2 M KCl. The fractions containing the protein to be purified are pooled and passed through a Sephadex G-25 column equilibrated in 10 mM Tris-HCl. The protein fraction is injected onto an Ultrogel HA (IBF, France) column equilibrated with 10 mM Tris-HCl, the proteins are eluted with a gradient of 0 to 0.1 M potassium phosphate, and the active fraction is then chromatographed on a Phenyl-Sepharose CL (Pharmacia) 4B column in 10 mM potassium phosphate, the column being eluted with a gradient of 0.65 to 0 M ammonium sulphate. The active fractions are pooled. The protein thereby obtained is more than 95% pure (according to the results of 12.5% SDS-PAGE electrophoresis and staining with silver salts). The purity of the protein is confirmed by the uniqueness of the N-terminal sequence. Its molecular weight calculated using this technique is 22,000. The steps of purification of precorrin-8x mutase with their purification yields are described in the table below.

TABLEPurification of precorrin-8x mutasePurificationVolProteinsPurificationstep(ml)(mg)factor¹Crude extract2506000—Precipitation5023502.6(40-60%)Ultrogel ACA 54706559.2DEAE-Tris-3027122acryl-MUltrogel HA229365Phenyl-Sepharose1231194
¹This factor is calculated from the yield of protein.

c) NH₂-Terminal Sequence of Precorrin-8x Mutase and Identification of its Structural Gene

This example illustrates how the NH₂-terminal sequence of a protein participating in the biosynthetic pathway enables the structural gene which codes for this protein to be identified.

The NH₂-terminal sequence of this protein was determined as described above. 15 residues were identified:

123456789101112131415ProGluTyrAspTyrIleArgAspGlyAsnAlaIleTyrGluArg

The NH₂-terminal sequence of the COBH protein (FIG. 16) corresponds exactly to this sequence except that, in the sequence presented in FIG. 16, a methionine precedes the peptide sequence determined by the sequencing described above. It follows from this that the amino-terminal methionine is definitely excised in vivo by methionine aminopeptidase (Ben Bassat and Bauer, 1987). Since the second residue is a proline, this excision is in keeping with the rules already stated (Hirel et al., 1989). The molecular weight of the purified precorrin-8x mutase, estimated by 12.5% SDS-PAGE electrophoresis, is 22,000. The COBH protein has a molecular weight deduced from its sequence of 22,050 (FIG. 16). The correspondences between the NH₂-terminal sequences and the molecular weights of these proteins indicate clearly that the COBH protein corresponds to precorrin-8x mutase. cobH is the precorrin-8x mutase structural gene.

d) Preparation, Isolation and Identification of Precorrin-8x.

In a typical experiment for preparation of precorrin-8x, a crude enzyme extract of strain SC510 Rif′ pXL253 (1000 mg of proteins) is incubated anaerobically for 20 h at 30° C. in 0.1 M Tris-HCl buffer pH 7.7 (100 ml) with trimethylisobacteriochlorin (1000 nmol) prepared as described previously (Battersby et al., 1982), EDTA (1 mM), ATP (100 μmol), MgCl₂(250 μmol), NADH (50 μmol), NADPH (50 μmol), SAM (50 μmol) and hydrogenobyrinic acid (20 μmol). At the end of the incubation, precorrin-8x is the preponderant tetrapyrrole product formed. It is isolated and purified by HPLC on a μBondapak C18 (Waters) column using a linear elution gradient of 0 to 50% of acetonitrile in a potassium phosphate buffer pH 5.8. The mass of precorrin-8x (m/z=880) and the mass of its methyl ester derivative (m/z=978) indicate that it is a compound having the same empirical formula as hydrogenobyrinic acid. The UV/visible and fluorescence characteristics are very different from those of hydrogenobyrinic acid, and indicate that the molecule possesses two separate chromophors. Since the only enzymatic isomerisation reaction between precorrin-6x (Thibaut et al., 1990) and hydrogenobyrinic acid is the migration of the methyl from C-11 to C-12, precorrin-8x is the last intermediate before hydrogenobyrinic acid, and the corresponding reaction is the migration of the methyl from C-11 to C-12, catalysed by precorrin-8x mutase.

6.1.5. Identification of the COBU Protein Encoded by the cobU Gene

a) Assay of nicotinate-nucleotide:dimethylbenzimidazole phosphoribosyltransferase activity (FIG. 5, reaction 5). This example illustrates the assay of an enzymatic activity directly linked to the pathway of biosynthesis of cobalamins. The enzyme in question is nicotinate-nucleotide:dimethylbenzimidazole phosphoribosyl-transferase (NN:DMBI PRT) (EC 2.4.2.21). The fractions containing NN:DMBI PRT activity (approximately 5 units) are incubated at 30° C. for 8 min in 0.1 M glycine-NaOH buffer pH 9.7 (500 μl) in the presence of 1 mM NaMN (nicotinic acid mononucleotide) and 10 μM DMBI. The reaction is then stopped by heating to 80° C. for 10 min, the reaction mixture is diluted with water (4 volumes) and this solution (100 μl) is injected onto a 15-cm Nucleosil 5-C8 HPLC column eluted with a 0.1 M potassium phosphate pH 2.9/acetonitrile (93:7) mixture at a flow rate of 1 ml/min. The α-ribazole 5′-phosphate is detected and quantified by fluorimetry (excitation: 260 nm; emission>370 nm). The unit of enzymatic activity is defined as the quantity of enzyme necessary for generating 1 nmol of α-ribazole 5′-phosphate per hour under these conditions.

b) Purification of Pseudomonas denitrificans NN:DMBI PRT activity. This experiment illustrates how a P. denitrificans protein participating in the pathway of biosynthesis of cobalamins may be purified. Using the assay described in Example 6.1.5.a), the purification of Pseudomonas denitrificans NN:DMBI PRT is carried out as described below. In a typical purification experiment, wet cells (10 g) of strain SC510 Rif′, into which plasmid pXL1490B has been introduced as described above, are used. Plasmid pXL1490B is described in FIG. 38; this plasmid was obtained by cloning the 3.85-kb BamHI-SstI-SstI fragment of pXL519 (see FIG. 38). This plasmid hence carries the cobU and cobV genes of P. denitrificans. The cells, cultured in PS4 medium supplemented with lividomycin, as described previously, are harvested after 96 hours of culture in PS4 medium. They are resuspended in 0.1M Tris-HCl buffer pH 7.2 (25 ml) and sonicated for 15 min at 4° C. The crude extract is then recovered by centrifugation for 1 h at 50,000 g, and thereafter passed through a DEAE-Trisacryl M (IBF, France) column equilibrated with the same buffer. 10% of the eluate (120 mg of proteins) is fractionated on a mono Q HR 10/10 column using a KCl gradient (0 to 0.6 M). The active fractions are pooled and concentrated to 2 ml by ultrafiltration, and then, after mixing with 30 mM Tris-HCl buffer pH 7.2 (one volume), the sample is fractionated a second time on a Mono Q HR 5/5 column as before. The active fractions are pooled, and the sample is then brought to a molarity of 1 M using ammonium sulphate and chromatographed on a Phenyl-Superose HR 5/5 column eluted with a decreasing ammonium sulphate gradient (1 M to 0 M). The fractions containing the desired activity are pooled, concentrated by ultrafiltration and chromatographed on a Bio-Sil 250 gel permeation column eluted with 20 mM sodium phosphate/50 mM sodium sulphate pH 6.8.

After this step, the enzyme is more than 95% pure. It shows no contaminant protein in SDS-PAGE. This purity is confirmed by the uniqueness of the NH₂-terminal sequence. Its molecular weight in this technique is 35,000. The different steps of purification of the NN:DMBI PRT are given in the table below.

TABLEPurification of P. denitrificans NN:DMBI PRTSp. activityPurificationVolProteins(u/mg ofPurificationStep(ml)(mg)proteinsYieldfactor¹Crude extract6.01202650——MonoQ 10/106.012.71351551.3 5.1MonoQ 5/53.06.192014039.2 7.6Phenyl-Superose1.52.603551029.013.4Bio-Sil 2501.21.923975024.015.0

c) NH₂-terminal sequence of P. denitrificans NN:DMBI PRT and identification of the Pseudomonas denitrificans structural gene coding for this activity. The NH₂-terminal sequence of Pseudomonas denitrificans NN:DMBI PRT, purified as described in Example 6.1.5b), was carried out according to the technique described above. The first 15 residues-were identified:

123456789101112131415SerAlaSerGlyLeuProPheAspAspPheArgGluLeuLeuArg

The NH₂-terminal sequence of the COBU protein (FIG. 41) corresponds to this sequence except that, in the sequence presented in FIG. 41, a methionine precedes the first amino acid of the peptide sequence determined by direct sequencing. It follows from this that the amino-terminal methionine is definitely excised in vivo by methionine aminopeptidase (Ben Bassat and Bauer, 1987). The molecular weight of the purified N-transglycosidase, estimated by 12.5% SDS-PAGE electrophoresis, is 35,000. The COBU protein has a molecular weight deduced from its sequence of 34,642 (FIG. 41). The correspondences between the NH₂-terminal sequences and the molecular weights clearly indicate that the COBU protein corresponds to NN:DMBI PRT. The cobU gene is the NN:DMBI PRT structural gene.

d) Specificity of NN:DBI PRT for DBI. This example illustrates how a study of the specificity of P. denitrificans NN:DMBI PRT enables P. denitrificans to be made to biosynthesise various cobamides, using the catalytic properties of P. denitrificans NN:DMBI PRT to perform the synthesis of the nucleotide base in question.

The enzyme substrate for synthesising cobalamine is 5,6-dimethylbenzimidazole. Benzimidazole and 5-methylbenzimidazole, respectively, are substrates for the reaction with reaction rates of 157% and 92%, respectively, compared to the natural substrate (5,6-dimethylbenzimidazole), the NaMN concentration being fixed at 2 mM. The specificity of P. denitrificans NN:DMBI PRT is hence low for substrates containing a benzimidazole ring-system. It is hence possible to use P. denitrificans strain SC510 Rif′ (Cameron et al., 1989), and to culture it in PS4 medium in which 5,6-dimethylbenzimidazole is replaced by benzimidazole or 5-dimethylbenzimidazole, respectively, in order to make the bacterium synthesise Coα-(benzimidazolyl)-Coβ-cyanocobamide or Coα-(5-methylbenzimidazolyl)-Coβ-cyanocobamide, respectively. There is no doubt that other cobamides could be synthesised in this way.

6.1.6. Identification of the COBV Protein Encoded by the cobV Gene

This example illustrates how the assay of an activity of the pathway of biosynthesis of coenzyme B₁₂in P. denitrificans, and then the partial purification of this activity, can enable the structural gene for this enzyme to be identified in P. denitrificans.

a) Assay of GDP-cobinamide:α-ribazole-5′-phosphate cobinamidephosphotransferase (or cobalamin-5′-phosphate synthase) activity. This example illustrates the assay of an activity directly linked to the pathway of biosynthesis of cobalamines. The enzyme in question is cobalamin-5′-phosphate synthase. The fractions containing the activity (approximately 5 to 10 units) are incubated in darkness at 30° C. in 0.3 Tris-HCl buffer pH 9.0 (500 μl) in the presence of 1 mM EDTA, 12.5 MM MgCl₂, 50 μM α-ribazole 5′-phosphate and 20 μM GDP-cobinamide [in 5′-deoxy-5′-adenosyl (Ado) or coenzyme form]. After 15 min of incubation, 20 mM potassium cyanide (500 μl) is added and the solution is heated to 80° C. for 10 min. After centrifugation to remove the precipitated matter, the vitamin B₁₂5′-phosphate present in the supernatant is assayed as described in Example 9. One unit of cobalamin-5′-phosphate synthase is defined as the quantity of enzyme necessary for generating 1 nmol of cobalamine 5′-phosphate per h under the conditions described above.

Ado-GDP-cobinamide is obtained by incubation of Ado-cobinamide phosphate (Blanche et al., 1989) with a SC510 Rif′ pXL623 extract under the conditions of assay of cobinamidephosphate guanylyltransferase (see 6.1.11.b). The α-ribazole and α-ribazole 5′-phosphate are isolated from SC510 Rif′ cultures and purified by HPLC under the assay conditions described in Example 6.1.5a).

b) Partial Purification of Cobalamin-5′-Phosphate Synthase

This experiment illustrates how a P. denitrificans enzymatic activity participating in the pathway of biosynthesis of cobalamines of P. denitrificans can be partially purified. Using the assay described above, the purification of cobalamin 5′-phosphate synthase is carried out. For this purpose, in a typical purification experiment, wet cells (10 g) of strain SC510 Rif′, into which plasmid pXL1490B has been introduced as described previously, are used. Plasmid pXL1490B is described in FIG. 38: this plasmid corresponds to the 3.85-kb SstI-SstI-BamHI fragment cloned into PKT230. This plasmid carries the P. denitrificans cobU and cobV genes. The presence of this plasmid in P. denitrificans SC510 Rif′ leads to an amplification of the cobalamin-5′-phosphate synthase activity by a factor of approximately 100; it is hence probable that the insert carried by plasmid pXL1490B contains the structural gene for this enzyme; hence this gene can be only cobU or cobV. The SC510 Rif′ pXL1490B cells are obtained by culture in PS4 medium supplemented with lividomycin, as described above. The cells are centrifuged and then resuspended in 0.1 M Tris-HCl (pH 8.3)/1 mM EDTA buffer (buffer A) (25 ml) and sonicated for 15 min at 4° C. The crude extract is then recovered by centrifugation for 1 h at 50,000 g and passed through a Sephadex G-25 column equilibrated with buffer A. The protein fraction is recovered and injected in 300-μl fractions (7.5 mg of proteins) onto a Superose 12 HR 10/30 column eluted in buffer A. The excluded fraction is recovered, mixed with an equal volume of buffer A/1.0 M ammonium sulphate and chromatographed on a Phenyl-Superose HR 5/5 column. The proteins are eluted with a decreasing ammonium sulphate gradient (0.5 M to 0 M) in buffer A, followed by a plateau at 0 M ammonium sulphate with the object of eluting the cobalamin-5′-phosphate synthase activity. The partial purification of this enzyme is described in the table below, on the basis of 75 mg of proteins introduced at the start in the purification process.

TABLEPartial purification of P. denitrificans cobalamin-5′-phosphate synthaseSp. activityPurificationVolProteins(u/mg ofPurificationstep(ml)(mg)proteins)Yieldfactor¹Crude extract3.075325——Superose 12 HR50.02.96,8108121Phenyl-Superose4.50.3517,8502655

c) Specificity of cobalamin-5′-phosphate synthase. The Km for (Ado)GDP-cobinamide is 0.9 μm. However, the enzyme possesses the same affinity and a virtually identical reaction rate for the (CN, aq) form of the substrate. The Km of the enzyme for α-ribazole 5′-phosphate is approximately 2.7 μM. In addition, the purest preparations of cobalamine-5′-phosphate synthase catalyse the reaction of Ado-GDP-cobinamide with α-ribazole to give coenzyme B₁₂and, under these conditions, no accumulation of cobalamin 5′-phosphate is observed. The Km of the enzyme for α-ribazole is 7.8 μM. Intracellular α-ribazole 5′-phosphate and α-ribazole concentrations of 30 and 700 μM, respectively, were measured by HPLC during the production of cobalamins by SC510 Rif′ in PS4 medium under the culture conditions described in Example 6.1.5a). This shows that coenzyme B₁₂may be generated directly from Ado-GDP-cobinamide by cobalamin-5′-phosphate synthase without the participation of a cobalamin 5′-phosphatase.

The absence of accumulation or the presence of traces of cobalamin 5′-phosphate in the P. dinitrificans SC510 Rif′ cultures confirms that coenzyme B₁₂is produced by the direct reaction of Ado-GDP-cobinamide with α-ribazole in vivo.

This direct reaction has already been observed and described in vitro in Propionibacterium shermanii (Ronzio et al., 1967; Renz, 1968). As the cobalamin-5′-phosphate synthase structural gene can be only cobU or cobV, since the amplication in P. denitrificans of a fragment carrying these two P. denitrificans cob genes leads to an increase in cobalamin-5′-phosphate synthase activity by a factor of 100, and since the cobU gene is the NN:DMBI PRT structural gene, cobV is hence the cobalamin-5′-phosphate synthase structural gene.

6.1.7. Identification of the COBK Protein Encoded by the cobK Gene

a) Assay of Precorrin-6x Reductase Activity.

This example illustrates the assay of a novel enzymatic activity directly linked to the pathway biosynthesis of cobalamins. The enzyme in question is precorrin-6x reductase.

The fractions containing precorrin-6x reductase activity (approximately 0.05 unit, U) are incubated at 30° C. for 60 min in 0.1 M Tris-HCl buffer pH 7.7 (250 μl) in the presence of 1 mM EDTA, 500 μM NADPH, 25 μM [methyl-³H]SAM (80 μCi/μmol), 4 μM precorrin-6x (Thibaut et al., 1990) and partially purified dihydroprecorrin-6x methylase (0.5 U) (see preparation below). The reaction is then stopped by heating to 80° C. for 5 min and, after centrifugation at 5000×g for 5 min, the supernatant is injected onto a DEAE-Sephadex column (containing 200 μl of gel). The column is then washed extensively with the Tris-HCl buffer, and the compounds bound are eluted with 1 M HCl (4 ml). The radio activity in this eluent is counted by liquid scintillation counting. The unit of enzymatic activity is defined as the quantity of enzyme necessary for reducing 1 nmol of precorrin-6x per h under these conditions.

Dihydroprecorrin-6x methylase is partially purified from a crude extract of SC510 Rif′ pXL253 on a Mono Q HR 5/5 (Pharmacia) anion exchange column. The column is eluted with a linear gradient of 0 to 0.4 M KCl in 0.1 M Tris-HCl buffer pH 7.7. The enzymatic activity is eluted at 0.35 M KCl. This activity is detected and quantified by means of the precorrin-6x reductase activity test defined above (in the presence of precorrin-6x reductase (0.5 U) in the incubation medium). After the Mono Q step, the fractions containing dihydroprecorrin-6x methylase activity are completely devoid of precorrin-6x reductase activity. The unit of methylase activity is defined as the quantity of enzyme necessary for transferring 1 nmol of methyl groups to dihydroprecorrin-6x per h under the conditions described above.

b) Purification of Precorrin-6x Reductase Activity

Using the assay described above, the purification of Pseudomonas denitrificans precorrin-6x reductase is carried out as described below.

In a typical purification experiment, wet cells (100 g) of strain SC510 Rif′, into which plasmid pXL253 (plasmid pKT230 into which the 8.7-kb fragment has been cloned at the EcoRI site, FIG. 13) has been introduced, are suspended in 0.1 M Tris-HCl pH 7.7/1 mM EDTA buffer (buffer A) (200 ml) and sonicated for 15 min at 4° C. The crude extract is then recovered by centrifugation at 1 h at 50,000×g and passed in three portions through a Sephadex G-25 column equilibrated with buffer A. The three fractions excluded from the gel are pooled and adjusted to 1 l with buffer A. The proteins precipitating at between 25 and 40% ammonium sulphate saturation are collected by centrifugation and resuspended in buffer A (50 ml), and this solution is desalted through a Sephadex G-25 column equilibrated with buffer B (25 mM Tris-HCl/500 μM DTT/15% glycerol). The protein solution is then injected at 2.5 ml/min onto a Q Sepharose Fast Flow (Pharmacia) column equilibrated with buffer B, and the proteins are eluted with a buffer B/0.2 M KCl mixture. This fraction is desalted on a Sephadex G-25 column equilibrated with buffer C (50 mM Tris-HCl/500 μM DTT/15% glycerol). The protein solution is then fractionated (100 mg of proteins at each chromatographic run) on a Mono Q HR 10/10 (Pharmacia) column using a gradient of 0 to 0.4 M KCl in buffer C, and the fraction containing the activity is thereafter chromatographed on a Phenyl-Superose HR 10/10 (Pharmacia) column in a linear decreasing ammonium sulphate gradient (1 to 0 M). The active fraction is desalted and the precorrin-6x reductase is repurified on a Mono Q HR 5/5 column. It is eluted in 50 mM Tris-HCl pH 8.1/500 μM DTT/15% glycerol buffer with a gradient of 0 to 0.2 M KCl. To complete the purification, the protein is finally chromatographed on a Bio-Sil 250 (Bio-Rad) column eluted with 20 mM potassium phosphate/50 mM sodium sulphate pH 6.8/500 μM DTT/15% glycerol. After this step, the enzyme is more than 95% pure. It shows no contaminant protein in SDS-PAGE, the proteins being visualised with silver nitrate. This degree of purity is confirmed by the uniqueness of the NH₂-terminal sequence. Its molecular weight in this technique is 31,000. The different steps of purification of precorrin-6x reductase with their purification factor and their yield, are given in the table below.

TABLEPurification of precorrin-6x reductaseSp. activityPurificationVolProteins(u/mg ofPurificationstep(ml)(mg)proteins)Yieldfactor¹Crude extract27096000.535——A.S. 25 40%10041601.14922.1Q Sepharose15010443.64746.8Mono Q 10/10556724.53246Phenyl-Superose102.232514607Mono Q 5/52.50.08257509.210750Bio-sil 2501.00.05576508.214300

c) NH₂-Terminal Sequence and Partial Internal Sequences of Pseudomonas Denitrificans Precorrin-6x Reductase and Identification of the Pseudomonas Denitrificans Structural Gene Coding for this Activity

The NH₂-terminal sequence of Pseudomonas denitrificans precorrin-6x reductase, purified as described above, was determined as described before. Six residues were identified: Ala-Gly-Ser-Leu-Phe-Asp

Similarly, after tryptic digestion and separation of the fragments by HPLC on a C-18 reversed-phase column, three internal sequences were obtained:

Ile-Gly-Gly-Phe-Gly-G ly-Ala-Asp-Gly-LeuArg-Pro-Glu-Trp-Val-Pro-Leu-Pro-Gly-Asp-ArgVal-Phe-Leu-Ala-Ile-Gly

The NH₂-terminal sequence of the COBK protein (FIG. 16) corresponds exactly to the NH₂-terminal sequence of precorrin-6x reductase except that, in the sequence presented in FIG. 16, a methionine precedes the peptide sequence determined by direct sequencing. It follows from this that the amino-terminal methionine is definitely excised in vivo by methionine aminopeptidase (Ben Bassat and Bauer, 1987). Similarly, the three internal sequences correspond to the three sequences 60 to 69, 112 to 122 and 143 to 148 of the COBK protein. The molecular weight of the purified precorrin-6x reductase is estimated by SDS-PAGE electrophoresis at 31,000. The COBK protein has a molecular weight deduced from its sequence of 28,000 (FIG. 16). The correspondences between the internal NH₂-terminal sequences and the molecular weights indicate clearly that the COBK protein corresponds to precorrin-6x reductase. The cobK gene is the precorrin-6x reductase structural gene.

d) Reaction Catalysed by Precorrin-6x Reductase

The enzymatic reaction of reduction of precorrin-6x is strictly NADPH-dependant in P. denitrificans. NADPH cannot be replaced by NADH. When the purified enzyme (or an active fraction during purification, or even a crude enzyme extract) is incubated under the conditions of the assay of activity, but in the absence of SAM and of dihydroprecorrin-6x methylase, the product of the reaction can then be purified by HPLC in the system described for the purification of precorrin-6x (see Example 6.1.4.d). After desalting and esterification (4% methanolic sulphuric acid, 20° C., 24 h, argon atmosphere), the corresponding ester has a mass m/z=1008. The product of the reaction catalysed by precorrin-6x reductase is hence dihydroprecorrin-6x, also known as precorrin-6y.

6.1.8. Identification of the COBQ Protein Encoded by the cobQ Gene

a) Assay of Cobyric Acid Synthase Activity

This example illustrates the assay of an enzymatic activity of the pathway of biosynthesis of cobalamins which has never been described hitherto. The enzyme in question is cobyric acid synthase. This enzyme catalyses the amidation of the peripheral carboxylic acid functions at positions b, d, e and g on the corrin ring-system (see FIG. 19; PL. 68). The NH₂-group donor is L-glutamine, and each amidation reaction is accompanied by the consumption of one ATP molecule.

The fraction to be assayed is incubated in darkness at 30° C. for 60 min in 0.1 M Tris hydrochloride buffer pH 7.5 (250 μl) containing 1 mM DTT, 1 mM EDTA, 1 mM ATP, 2.5 mM MgCl₂, 1 mM glutamine and 10 μM Ado-cobyrinic acid di- or pentaamide. The reaction is stopped by adding 0.1 M aqueous potassium cyanide solution (25 μl). After heating to 80° for 10 min and centrifugation at 3000×g for 10 min, the compounds formed, present in the supernatant, are analysed by HPLC. The unit of activity is defined as the quantity of enzyme necessary for generating 1 nmol of amide functions per h under these conditions.

5′-Deoxy-5′-adenosyl(Ado)-cobyrinic acid diamide and pentaamide are isolated from cultures of strain SC510 in PS4 medium, using the method the principle of which is described in Example 9.

b) Purification of Cobyric Acid Synthase

Using the assay described in Example 6.1.8 a), purification of Pseudomonas denitrificans cobyric acid synthase is carried out as described below.

In a typical purification experiment, wet SC510 Rif′ cells (6 g), into which strain plasmid pXL618 (see Example 4.5.2) has been introduced, are sonicated in 0.1 M Tris-HCl pH 7.7, 1 mM DTT, 1 mM EDTA buffer (15 ml). After centrifugation (50,000×g for 1 h), the extract is brought to 20% of glycerol (vol/vol). 10 mM Tris-HCl, 1 mM DTT, 20% glycerol buffer (24 ml) are added to the crude extract (8.5 ml; 203.5 mg of proteins). The solution is injected onto Mono Q HR 10/10 (Pharmacia) at 2 ml/min, equilibrated with 50 mM Tris-HCl pH 7.7, 1 mM DTT, 20% glycerol buffer. The proteins are eluted with a linear gradient of 0.5 M NaCl and the active fractions are pooled and brought to 1 mM EDTA. The solution is brought to 0.85 M with respect to ammonium sulphate and injected onto a Phenyl-Superose HR 5/5 (Pharmacia) column equilibrated in Tris-HCl pH 7.7, 1 mM DTT, 0.85 M ammonium sulphate buffer, and the proteins are eluted with a linear decreasing gradient of 0.85 M to 0 M ammonium sulphate. The fractions are immediately brought to 20% of glycerol. The active fraction is concentrated to 2.5 ml by ultrafiltration and chromatographed on a PD 10 (Pharmacia) column equilibrated and eluted with 50 mM Tris-HCl pH 8.3, 1 mM DTT, 20% glycerol (vol/vol) buffer. The protein fraction is collected and injected onto a mono Q HR 5/5 column equilibrated with the same buffer, and the proteins are eluted with a linear gradient of 0.5 M NaCl. Gel permeation chromatography on Bio-Sil 250 (Bio-Rad) gel in 50 mM Tris-HCl pH 7.5, 1 mM DTT, 20% glycerol, 0.1 M NaCl buffer medium finally enables a protein which is more than 97% pure to be obtained. It shows no contaminant protein in SDS-PAGE. This purity is confirmed by the uniqueness of the NH₂-terminal sequence. Its molecular weight in this technique is 57,000. The different steps of purification of cobyric acid synthase with their purification factor and their yield are given in the table below.

TABLEPurification of cobyric acid synthaseSp. activityU/mgPurificationVolProteinsabPurificationstep(ml)(mg)ABYield*factor¹Crude extract8.5203 114/ 118——Mono Q 10/108.035.5 388/ 425603.4Phenyl-8.0*3.231988/20212817SuperoseMono Q 5/51.01.204549/40852440Bio-Sil 2500.750.884992/N.D.1944
a / with Ado-cobyrinic acid a,c-diamide as substrate

b / with Ado-cobyrinic acid pentaamide as substrate

ND = Not Determined

The very high degree of purity of the purified protein, together with the constancy of the ratio of the activities of amidation of cobyrinic acid diamide and pentaamide throughout the process of purification of the protein (see table above), indicate unambiguously that one and the same protein is responsible for the four activities of amidation of the corrin ring-system at positions b, d, e and g.

c) NH₂-Terminal Sequence of Pseudomonas Denitrificans Cobyric Acid Synthase and Identification of the Pseudomonas Denitrificans Structural Gene Coding for this Activity

The NH₂-terminal sequence of Pseudomonas denitrificans cobyric acid synthase was determined as described above. Sixteen residues were identified:

Thr-Arg-Arg-Ile-Met-Leu-Gln-Gly-Thr-Gly-Ser-Asp-Val-Gly-Lys-Ser

The NH₂-terminal sequence of the COBQ protein (FIG. 47) corresponds exactly to this sequence except that, in the sequence presented in FIG. 47, a methionine precedes the peptide sequence determined by direct sequencing. It follows from this that the amino-terminal methionine is definitely excised in vivo by methionine aminopeptidase (Ben Bassat and Bauer, 1987). The molecular weight of the purified cobyric acid synthase is estimated by SDS-PAGE electrophoresis at 57,000. The COBQ protein has a molecular weight deduced from its sequence of 52,000 (FIG. 47). The correspondences between the NH₂-terminal sequences and the molecular weights indicate clearly that the COBQ protein corresponds to cobyric acid synthase. The cobQ gene is the cobyric acid synthase structural gene.

6.1.9. Identification of the COBO Protein Encoded by the cobO Gene

a) Assay of cob(I)alamin Adenosyltransferase (EC 2.5.1.17) Activity

This example illustrates the assay of an enzymatic activity directly linked to the pathway of biosynthesis of cobalamins. The enzyme in question is cob(I)alamin adenosyltransferase (EC 2.5.1.17). This enzyme was demonstrated in bacterial cells (Ohta et al., 1976, Brady et al. , 1962) and animal cells (Fenton et al., 1978). It was purified from Clostridium tetanomorphum (Vitols et al., 1966).

The fractions containing cob(I)alamin adenosyltransferase activity (approximately 20 units) are incubated anaerobically at 30° C. for 15 min protected from light in 0.2 M Tris-HCl buffer pH 8.0 (1 ml) in the presence of 5 mM DTT, 400 μM [8-¹⁴C]-ATP (2.5 μCi/μmol), 800 MM MnCl₂, 50 μM hydroxocobalamin or diaquacobinamide and KBH₄(3 mg). The reaction is then stopped by heating to 80° C. for 10 min and, after centrifugation at 15000×g for 5 min, the supernatant (200 μl) is analysed by HPLC (Gimsing et al., 1986, Jacobsen et al., 1986).

The unit of enzymatic activity is defined as the quantity of enzyme necessary for generating 1 nmol of adenosylcorrinoid per min under these conditions.

b) Purification of cob(I)alamin Adenosyltransferase Activity

Using the assay described in Example 6.1.9 a), the purification of Pseudomonas denitrificans cob(I)alamin adenosyltransferase is carried out as described below.

In a typical purification experiment, wet cells (10 g) of strain SC510 Rif′ in which the cobO gene has been amplified are suspended in 0.2 M Tris-HCl buffer pH 8.0 (20 ml) and sonicated for 40 min at 4° C. The crude extract is then recovered by centrifugation for 1 h at 50,000×g and desalted on PD10 (Pharmacia) columns equilibrated with 50 mM Tris-HCl pH 8.0, 5 mM DTT buffer (buffer A). The protein solution is then fractionated (280 mg of proteins at each chromatographic run) on a Mono Q HR 10/10 (Pharmacia) column using a gradient of 0 to 0.5 M KCl in buffer A, and the fractions containing the activity are then pooled, concentrated by ultrafiltration and chromatographed on a Phenyl-Superose HR 10/10 (Pharmacia) column in a linear decreasing ammonium sulphate gradient (1.7 to 0 M), the column being equilibrated in 0.1 M Tris-HCl pH 8.0, 5 mM DTT buffer. To complete the purification, the protein is finally chromatographed, after concentration by ultrafiltration, on a Bio-Sil 250 (Bio-Rad) column eluted with 50 mM Tris-HCl pH 7.5, 0.1 M NaCl, 5 mM DTT buffer.

After this step, the enzyme is more than 95% pure. It does not show any contaminant protein in SDS-PAGE. Its molecular weight in this technique is 28,000. This degree of purity is confirmed by the uniqueness of the NH₂-terminal sequence. The different steps of purification of cob(I)alamin adenosyltransferase, with their purification factor and their yield, are given in the table below for the following two substrates: diaquacobinamide (a) and hydroxocobalamin (b). These results demonstrate the absence of specificity of this enzyme for the nature of the corrinoid substrate. Moreover, all corrinoids of the biosynthetic pathway between cobyrinic acid diamide and B₁₂have been isolated (Blanche et al., unpublished results) in their native form, and have proved to be in coenzyme form. This demonstrates that the natural substrate of cob(I)alamin adenosyltransferase is cobyrinic acid a,c-diamide.

TABLEPurification of cob(I)alamin adenosyltransferaseSp. activityU/mgPurificationVolProteinsabPurificationstep(ml)(mg)ABYield*factor¹Crude extract^c10014005.4/3.4——Mono Q 10/109014034.9/14.1656.5Phenyl-3015.984.5/49.51816SuperoseBio-Sil 2506.52.9182.4/88.77.034
^cafter desalting on PD10

c) NH₂-Terminal Sequence of Pseudomonas Denitrificans cob(I)alamin Adenosyltransferase and Identification of the Pseudomonas Denitrificans Structural Gene Coding for this Activity.

The NH₂-terminal sequence of Pseudomonas denitrificans cob(I)alamin adenosyltransferase, purified as described in Example 6.1.9 b), was determined as described above. 13 residues were identified: Ser-Asp-Glu-Thr-?-Val-Gly-Gly-Glu-Ala-Pro-Ala-Lys-Lys

The NH₂-terminal sequence of the COBO protein (FIG. 47) corresponds exactly to the NH₂-terminal sequence of cob(I)alamin adenosyltransferase except that, in the sequence presented in FIG. 47, a methionine precedes the peptide sequence determined by direct sequencing. It follows from this that the amino-terminal methionine is definitely excised in vivo by methionine aminopeptidase (Ben Bassat and Bauer, 1987). The molecular weight of the purified cob(I)alamin adenosyltransferase is estimated by SDS-PAGE electrophoresis at 28,000. The COBO protein has a molecular weight deduced from its sequence of 24,000 (FIG. 47). The correspondences between NH₂-terminal sequences and the molecular weights indicate clearly that the COBO protein corresponds to cob(I)alamin adenosyltransferase. The cobO gene is the cob(I)alamin adenosyltransferase structural gene.

6.1.10. Identification of the COBN Protein Encoded by the cobN Gene

a) Demonstration of the Activity of Conversion of Hydrogenobyrinic Acid a,c-Diamide to Cobyrinic Acid a,c-Diamide

This example illustrates the demonstration of an enzymatic activity directly linked to the pathway of biosynthesis of cobalamins which has never been described hitherto. The activity in question is that of conversion of hydrogenobyrinic acid a,c-diamide to cobyrinic acid a,c-diamide.

This activity is demonstrated, inter alia, by the following typical experiment. A crude extract of strain SC510 Rif′ is obtained by sonication of wet cells (10 g) in 0.2 M Tris-HCl buffer pH 8.0 (20 ml), followed by removal of the cell debris by centrifugation for 1 h at 50,000×g. Proteins (1000 mg) of this extract are incubated for 1 h at 30° C. with carbon-14-labelled hydrogenobyrinic acid diamide (32 nmol; 50 μCi/μmol) in 0.2 M Tris-HCl buffer pH 8.0 (40 ml) containing 7 mM ATP and 200 μM CoCl₂. The reaction is stopped by adding 1 M KH₂PO₄(7.5 ml) and 0.3 M KCN (6 ml), followed by heating for 10 min at 80° C. After centrifugation at 15000×g for 50 min, HPLC analysis of the supernatant shows: (1) the formation during the incubation of cobyrinic acid a,c-diamide (19.2 nmol) having the same specific radioactivity as the starting hydrogenobyrinic acid a,c-diamide, and (2) the disappearance of a corresponding quantity of the latter. To confirm that the product is indeed cobyrinic acid a,c-diamide, the product is purified by HPLC and then esterified in methanol containing 5% of sulphuric acid (18 h, 20° C.). The authenticity of the cobyrinic acid a,c-diamide pentamethyl ester produced is demonstrated by TLC (relative to a reference sample) and mass spectrometry. It should be noted that, under similar incubation conditions in which the radioactive labelling is introduced, not into the hydrogenobyrinic acid a,c-diamide, but into the cobalt (using cobalt-57), cobalt-57-labelled cobyrinic acid a,c-diamide is biosynthesised and the same conclusions could be drawn.

Carbon-14-labelled hydrogenobyrinic acid a,c-diamide is obtained in the following manner: hydrogenobyrinic acid is biosynthesised in vitro using [methyl-¹⁴C] SAM, then converted to hydrogenobyrinic acid a,c-diamide and purified by HPLC as described in Example 6.1.2.

This study demonstrates that the insertion of cobalt takes place at hydrogenobyrinic acid a,c-diamide level in P. denitrificans. Under the conditions described, hydrogenobyrinic acid is not a substrate for enzymatic chelation with cobalt.

b) Assay and Purification of a Protein of Strain SC510 Rif′ Involved in the Conversion of Hydrogenobyrinic Acid a,c-Diamide to Cobyrinic Acid a,c-Diamide

The fraction to be assayed (0.5 to 2 units) is incubated for 60 min at 30° C. with crude extract (50 μl) of strain SC510 Rif′ obtained as described above, 7 mM ATP, 200 μM CoCl₂, and 7 μM carbon-14-labelled hydrogenobyrinic acid a,c-diamide (50 μCi/μmol) in 0.1 M Tris-HCl buffer pH 8.0 (400 μl). The reaction is stopped by adding 1 M KH₂PO₄(75 μl) and 0.3 M KCN (60 μl), followed by heating for 10 min at 80° C. After centrifugation at 15000×g for 15 min, the supernatant is analysed by HPLC in order to quantify the cobyrinic acid a,c-diamide formed (see Example 9). The unit of enzymatic activity is defined as the quantity of enzyme necessary for generating 1 nmol of cobyrinic acid a,c-diamide per h under these conditions. Under these conditions, it is apparent that extracts of strain SC510 Rif′ into which plasmid pXL1909 has been introduced (see Example 4.5.2) possess an activity between 20 and 50 times as high as extracts of strain SC510 Rif′. It is on this basis that a protein which is alone responsible for this amplication of activity is purified.

In a typical purification experiment, wet cells (10 g) of strain SC510 Rif′, into which plasmid pXL1909 has been introduced, are suspended in 0.2 M Tris-HCl buffer pH 8.0 (20 ml) and sonicated for 30 min at 4° C. The crude extract is then recovered by centrifugation for 1 h at 50,000×g and desalted on PD10 (Pharmacia) columns equilibrated with 0.1 M Tris-HCl buffer pH 8.0 (buffer A). The protein solution is then fractionated (213 mg of proteins at each chromatographic run) on a Mono Q HR 10/10 (Pharmacia) column using a gradient of 0 to 0.5 M KCl in buffer A, and the fractions containing the activity are then pooled, concentrated by ultrafiltration, desalted on PD10 (Pharmacia) columns equilibrated with 0.1 M Tris-HCl buffer pH 7.2 (buffer B) and chromatographed on a Mono Q HR 10/10 (Pharmacia) column using a gradient of 0 to 0.5 M KCl in buffer B. The fractions containing the activity are pooled, concentrated by ultrafiltration, desalted on PD10 (Pharmacia) columns equilibrated with buffer B and chromatographed on a Mono Q HR 5/5 (Pharmacia) column using a gradient of 0 to 0.5 M KCl in buffer B. To complete the purification, the protein is finally chromatographed on a Bio-Sil 250 (Bio-Rad) column eluted with 20 mM potassium phosphate/50 mM sodium sulphate pH 6.8.

After this step, the enzyme is more than 95% pure. It does not show any contaminant protein in SDS-PAGE. Its molecular weight in this technique is 135,000. This degree of purity is confirmed by the uniqueness of the NH₂-terminal sequence. The different steps of purification of the protein of strain SC510 Rif′ involved in the conversion of hydrogenobyrinic acid a,c-diamide to cobyrinic acid a,c-diamide, with their purification factor and their yield, are given in the table below.

TABLEPurification of a protein of strain SC510 Rif^rinvolved in the conversionof hydrogenobyrinic acid a,c-diamide cobyrinic acid a,c-diamideSp. activityPurificationVolProteins(u/mg ofPurificationstep(ml)(mg)proteins)Yieldfactor¹Crude extract31.512780.23——Mono Q 10/104479.22.46410Mono Q 10/102133.66.87830Mono Q 5/536.616.03670Bio-Sil 2501.85.916.33371

c) NH₂-terminal sequence of the Pseudomonas denitrificans protein involved in the conversion of hydrogenobyrinic acid a,c-diamide to cobyrinic acid a,c-diamide, and identification of the Pseudomonas denitrificans structural gene coding for this activity

The NH₂-terminal sequence of this protein, purified as described in Example 6.1.10b), was determined as described above. Six residues were identified:

His-Leu-Leu-Leu-Ala-Gln

The NH₂-terminal sequence of the COBN protein (FIG. 47) corresponds exactly to the NH₂-terminal sequence of the purified protein except that, in the sequence presented in FIG. 47, a methionine precedes the peptide sequence determined by direct sequencing. It follows from this that the amino-terminal methionine is definitely excised in vivo by methionine aminopeptidase (Ben Bassat and Bauer, 1987). The molecular weight of the purified protein is estimated by SDS-PAGE electrophoresis at 135,000. The COBN protein has a molecular weight deduced from its sequence of 138,000 (FIG. 47). The correspondences between the NH₂-terminal sequences and the molecular weights indicated clearly that the COBN protein corresponds to the protein involved in the conversion of hydrogenobyrinic acid a,c-diamide to cobyrinic acid a,c-diamide. The cobN gene is hence the structural gene for this protein.

6.1.11. Identification of the COBP Protein Encoded by the cobP Gene

a) Assay of Cobinamide Kinase Activity

This example illustrates the assay of an enzymatic activity of the pathway of biosynthesis of cobalamins which has never been studied hitherto. The activity in question is that of cobinamide kinase. It catalyses the ATP-dependent phosphorylation of the hydroxyl group of the (R)-1-amino-2-propanol residue of Ado-cobinamide to generate cobinamide phosphate.

The fraction to be assayed is incubated in darkness at 30° C. for 60 min in 0.1 M Tris-HCl buffer pH 8.8 (500 μl) containing 1 mM EDTA, 1 mM ATP, 2.5 mM MgCl₂16 μM Ado-cobinamide (Blanche et al., 1989). The reaction is stopped by adding 20 mM aqueous potassium cyanide solution (500 μl). After heating to 80° C. for 10 min and centrifugation at 5,000×g for 10 min, the cobinamide phosphate formed, present in the supernatant, is assayed by HPLC (see Example 9) using the following simplified linear gradient: 25% to 30% of B in A in the course of 15 min, then 30% to 100% of B in the course of 12 min, and 3 min at 100% of B.

The unit of activity is defined as the quantity of enzyme necessary for generating 1 nmol of cobinamide phosphate from cobinamide per h under these conditions.

b) Assay of Cobinamidephosphate Guanylyltransferase Activity

This example illustrates the assay of an enzymatic activity of the pathway of biosynthesis of cobalamins which has never been studied hitherto. The activity in question is that of cobalamidephosphate guanylyltransferase. It catalyses the addition of the GMP portion of a GTP molecule to Ado-cobinamide phosphate, thereby generating one molecule of GDP-cobinamide and liberating one molecule of pyrophosphate.

This activity is assayed under the same conditions as cobinamide kinase, except that Ado-cobinamide phosphate (16 μM) (Blanche et al., 1989) and GTP (2 mM) replace Ado-cobinamide and ATP, respectively, during the incubation.

The unit of activity is defined as the quantity of enzyme necessary for generating 1 nmol of GDP-cobinamide from cobinamide phosphate per h under these conditions.

c) Purification of Cobinamide Kinase

Using the assay described in Example 6.1.11a), the purification of Pseudomonas denitrificans kinase is carried as described below.

In a typical purification experiment, wet SC510 Rif′ cells (5 g), into which strain plasmid pXL623 has been introduced (see Example 4.5.2) are sonicated in 0.1 M Tris buffer pH 7.6 (buffer A) (20 ml). After centrifugation (50,000×g for 1 h) and dialysis for 4 h against buffer A, the retentate (4.5 ml) is injected onto Mono Q HR 10/10 (Pharmacia) equilibrated with buffer A. The proteins are eluted with a linear gradient of 0.4 M NaCl, and the pooled active fractions are passed through a PD-10 (Pharmacia) column equilibrated in 30 mM Tris-HCl/5 mM potassium phosphate/5 μM calcium chloride pH 7.6 (buffer B). The protein solution is fractionated on a Bio-Gel HPHT (Bio-Rad) column equilibrated in buffer B and eluted with a gradient of 5 to 350 mM potassium phosphate. The active fractions are pooled and brought to 500 mM with respect to ammonium sulphate, and then fractionated on a Phenyl-Superose HR 5/5 (Pharmacia) column eluted with a decreasing ammonium sulphate gradient. The fraction containing the activity is finally repurified on a Mono Q HR 5/5 column in Tris-HCl at pH 7.3. After this step, the protein is more than 97% pure. It shows no contaminant protein in SDS-PAGE. This purity is confirmed by the uniqueness of the NH₂-terminal sequence. Its molecular weight in this technique is 20,000. The different steps of purification of cobinamide kinase, with their purification factor and their yield, are given in Table A.

The fractions containing cobinamide kinase activity also possess cobinamidephosphate guanylyltransferase activity. Moreover, as shown by the results presented in the table above, the ratio of these two activities remains constant in the fractions throughout the purification. Lastly, the purified protein possesses a very high degree of purity, exceeding 97%. These results collectively hence indicate unambiguously that one and the same protein is responsible for both successive activities, namely cobinamide kinase and cobinamidephosphate guanylyltransferase of the pathway of biosynthesis of cobalamins in Pseudomonas denitrificans.

d) NH₂-Terminal sequence of Pseudomonas denitrificans cobinamide kinase/cobinamidephosphate gunaylyltransferase, and identification of the Pseudomonas denitrificans structural gene coding for this activity

The NH₂-terminal sequence of Pseudomonas denitrificans cobinamide kinase/cobinamidephosphate guanylyltransferase was determined as described above. Ten residues were identified:

Ser-Ser-Leu-Ser-Ala-Gly-Pro-Val-Leu-Val

The NH₂-terminal sequence of the COBP protein (FIG. 47) corresponds exactly to this sequence except that, in the sequence presented in FIG. 47, a methionine precedes the peptide sequence determined by direct sequencing. It follows from this that the amino terminal methionine is definitely excised in vivo by methionine aminopeptidase (Ben Bassat and Bauer, 1987). The molecular weight of the purified cobinamide kinase/cobinamidephosphate guanylyltransferase is estimated by SDS-PAGE electrophoresis at 20,000. The COBP protein has a molecular weight deduced from its sequence of 19,500 (FIG. 47). The correspondences between the NH₂-terminal sequences and the molecular weights indicate clearly that the COBP protein corresponds to cobinamide kinase/cobinamidephosphate guanylyltransferase. The cobP gene is the cobinamide kinase/cobinamidephosphate guanylyltransferase structural gene.

6.2—Determination of the Properties of COB Proteins by Measurement of Accumulated Biosynthesis Intermediates

This example illustrates how it is possible to assign an enzymatic activity to a COB protein of Pseudomonas denitrificans. This activity is assigned on the basis of data obtained relating to accumulated biosynthesis intermediates in the Cob mutant or mutants blocked in the step in question. In effect, if a mutant accumulates a biosynthesis intermediate, it is very probable that this mutant is blocked in the step which has the intermediate in question as its substrate.

6.2.1. Properties of the COBC and COBD Proteins

The Cob mutants G643 (Agrobacterium tumefaciens) and G572 (Pseudomonas putida) already described in Examples 1 and 4 are blocked in the step corresponding to the COBC protein. In effect, these two mutants are not complemented by the inactivating insertions of transposons Tn5 which occur in the cobC gene. The two strains G643 and G572, as well as the unmutated parent strains [C58-C9 Rif′ and KT 2440 Rif′ (Cameron et al., 1989)), were cultured in PS4′ medium for A. tumefaciens and PS4″ medium for P. putida (PS4′ and PS4″ correspond to PS4 medium containing 100-fold and 1000-fold, respectively, less cobalt than PS4 described above) for 3 days as described above. ⁵⁷CoCl₂was added to the cultures (2.5 μCi/0.1 μm for a 25-ml culture). The intracellular corrinoids were isolated in their native form and identified by their HPLC behaviour. The parent strains do not accumulate corrinoids other than coenzyme B₁₂. The two mutants G643 and G572 accumulate adenosylated cobyric acid in respective proportions of 11% and 6%. These % proportions are calculated relative to the level of coenzyme B₁₂synthesised by the parent strain. Apart from cobyric acid, mutant G643 accumulates cobyrinic acid pentaamide in a proportion of 2%; cobyrinic acid pentaamide is the intermediate which precedes cobyric acid. A study of these mutants brings out the fact that they are blocked after cobyric acid. All these Cob mutants are blocked either between uro'gen III and cobinamide, or between cobinamide and the cobalamins. The mutants G643 and G572 are blocked between uro'gen III and cobinamide. Now, if these mutants are blocked before cobinamide, and both accumulate cobyric acid, the proteins for which they code can participate only in the enzymatic step (referred to as cobinamide synthase) which catalyses the amidation of cobyric acid with an aminopropanol residue to give cobinamide; they can also possibly participate in the synthesis of the substrate of the reaction which provides aminopropanol, if not aminopropanol itself. The cobC gene codes for a protein which is either cobinamide synthase or one of its subunits.

The Cob mutant G634 of Agrobacterium tumefaciens which is blocked in the step corresponding to the cobD gene was analysed in the same manner. This mutant is not complemented by the inactivating insertions in the cobD gene (Example 4.1). The only intracellular corrinoid found in this mutant is adenosylated cobyric acid. Like the above mutants, this mutant codes for a protein participating in the conversion of cobyric acid to cobinamide, or else possibly in the synthesis of the other substrate of the reaction.

These two different genes (cobC and cobD) code for two proteins which participate in the same step.

6.2.2. Properties of the COBF to COBM Proteins

The Agrobacterium tumefaciens mutants already described were studied, the study described in Example 4.2 having shown in which genes each of these mutants is blocked. They are the following mutants: G612 (cobF), G615 (cobG), G616 (cobH), G613 (cobI), G611 (cobJ), G620 (cobK), G638 (cobL) and G609 (cobM); we have shown in brackets the Pseudomonas denitrificans gene responsible for the complementation of these mutants (Example 5), which hence corresponds to the gene mutated in this mutant. These mutants were cultured in PS4 medium as described above with labelled cobalt. After four days' incubation, the mutants were analysed for their intracellular content of corrinoids and decobaltocorrinoids (see Examples 6.1.2 and 9). Table: Intermediates accumulated by Agrobacterium tumefaciens mutants blocked in the genes of the 8.7-kb fragment of Pseudomonas denitrificans

IntracellularIntracellularMu-decobaltocorrinoids in %¹corrinoids astatedStrainHBAHBAMHBAD% of cobalaminsgeneC58-C9*100100100coenzyme B₁₂100—G612<5<564cobinamide 2.2cobFcoenzyme B₁₂34G615<5<584coenzyme B₁₂17cobGG61635<10<10coenzyme B₁₂13cobHG613<5<557coenzyme B₁₂<1cobIG611<5<565coenzyme B₁₂<1cobJG62012<5<10coenzyme B₁₂<1cobKG638<5<547coenzyme B₁₂<1cobLG609<5<533coenzyme B₁₂<1cobM
HBA: hydrogenobyrinic acid

HBAM: hydrogenobyrinic acid monoamide

HBAD: hydrogenobyrinic acid diamide

*in fact, this is strain C58-C9 Rif^rNal^ralready described (Cameron et al., 1989)

¹the values are expressed as % of the same intermediates accumulated in the unmutated parent strain C58-C9 Rif^rNal^r.

These results show that none of the mutants accumulate any corrinoid (with the exception of the mutant inactivated in the cobF gene, G612, which, for its part, accumulates cobinamide but at a low level equivalent to 2.2% of the cobalamins synthesised by the unmutated strain). However, some mutants (G612, G615 and G616) have levels of cobalamins which represent more than 10% of the cobalamin level of the parent strain. It is probable that all these mutants are blocked at least before cobyrinic acid diamide. All the mutants accumulate hydrogenobyrinic acid and hydrogenobyrinic acid diamide in smaller quantities than the unmutated strain; they are hence very probably blocked before hydrogenobyrinic acid. It may be concluded that all the cobF to cobG genes code for proteins which participate before hydrogenobyrinic acid. Mutant G613 is known to be mutated in the cobI gene which codes for SP₂MT, participating well before hydrogenobyrinic acid. For this mutant, the results of the present example relating to the accumulation of intermediates are in complete agreement with the step inactivated in this mutant, namely, this mutant accumulates no intermediate after hydrogenobyrinic acid at a level higher than that observed with the unmutated strain. This result is, for the cobF, cobJ, cobL, and cobM genes, consistent with those of Example 6.4, where it proposed that these genes code for proteins which catalyse SAM-dependent transfers of methyl and hence which participate before hydrogenobyrinic acid. With the exception of cobI, which is the SP₂MT structural gene, these genes participate after precorrin-3. In effect, since they are neither the structural genes for SUMT nor for SP₂MT, they inevitably participate later, that is to say after precorrin-3 (all the cob genes described in the present invention participate between uro'gen III and the cobalamins). These cobF to cobH and cobJ to cobM genes code for enzymes which participate between precorrin-3 and hydrogenobyrinic acid.

6.2.3. Properties of the COBS and COBT Proteins

The mutant G2035 described in Examples 1 and 4.3 is blocked in the step corresponding to the COBS protein. The mutant G2037 described in Example 1 is blocked in this step corresponding to the COBT protein. These strains, as well as the parent strain (Agrobacterium tumefaciens C58C9Rif′), are cultured in PS4′ medium (this is PS4 medium in which the cobalt chloride concentration is 100-fold lower than in PS4 medium) in the presence of radioactive cobalt 57CoCl₂for 3 days, and their intracellular content of decobaltocorrinoids is analysed, as is the corrinoid content, as already described above (see Example 6.2.2). The strains G2035 and G2037 do not accumulate corrinoids, and large concentrations (greater than those observed with the parent strain) of hydrogenobyrinic acid and hydrogenobyrinic acid mono- and diamide are present only with strain G2035. This mutant is probably blocked in a step located after hydrogenobyrinic acid diamide and before cobyrinic acid diamide. Consequently, the cobs gene is considered to code for one of the enzymes involved in the conversion of hydrogenobyrinic acid diamide to cobyrinic acid diamide; this protein may hence participate either in the insertion of cobalt, or in the reduction of the cobalt of unadenosylated cobyrinic acid a,c-diamide. In contrast, the mutant G2037 is considered to be blocked in a step located upstream of hydrogenobyrinic acid. The cobT gene is considered to code for a protein involved in an enzymatic step upstream of hydrogenobyrinic acid and downstream of precorrin-3 (other structural genes coding for enzymes involved downstream of precorrin-3 have already been identified). Another possibility for the COBT protein is that it participates, as proposed in Example 5, as a cobalt-binding protein and/or as a protein which interacts with other protein(s) via its acidic portion.

6.2.4. Properties of the COBV Protein

The mutants G2039 and G2040 described in Examples 1 and 4.4 are blocked in the step corresponding to the COBV protein. These strains, as well as the parent strain, are cultured in PS4′ medium for 3 days in the presence of radioactive cobalt ⁵⁷CoCl₂, and their intracellular content of decobaltocorrinoids is then analysed and the corrinoid content is determined as described in Example 9. Strains G2039 and G2040 accumulate cobyric acid, cobinamide, cobinamide phosphate and GDP-cobinamide. These mutants are probably blocked in an enzymatic step downstream of GDP-cobinamide. The cobV gene is considered to code for an enzyme involved in the conversion of GDP-cobinamide to cobalamin, see FIG. 5. This result is in complete agreement with the cobalamin-5′-phosphate synthase activity of the COBV protein which possesses Ado-GDP-cobinamide as a substrate.

6.3—Determination of the Activity of COB Proteins by Studies of Affinity for SAM

This example illustrates how it is possible, using COB proteins purified from Pseudomonas denitrificans, to demonstrate in vitro a SAM-binding activity. If a COB protein possesses such an activity, it means that this COB protein is a methyltransferase of the pathway, and that it participates in one of the transfers of the eight methyl groups which occur between the uro'gen III and cobyrinic acid.

6.3.1. Test of Affinity for SAM on a Purified Protein

The test is based on the principle according to which methyl transferases of the pathway of biosynthesis of cobalamins definitely have an SAM-binding site. This site must be demonstrated by a higher affinity of SAM than for any protein which does not specifically bind SAM. After incubation of the protein under study in the presence of an excess of radioactive SAM, the latter is separated from the free SAM by gel permeation chromatography. The radioactivity appearing in the fraction having the molecular weight of the protein corresponds to the SAM bound during the incubation. The chromatography is performed at 2° C. in order to limit to the maximum the release of bound SAM during the separation.

The protein (approximately 10 μg) is incubated for 10 minutes at 30° C. in 0.1 M Tris-HCl pH 7.7 (200 μl) with [methyl-³H]SAM (5 nmol; 1 μCi). After incubation, a portion (100 μl) of the mixture is immediately injected onto a TSK-125 (Bio-Rad) column eluted at 1 ml/minute with the 50 mM sodium sulphate/20 mM sodium dihydrogen phosphate mixture, pH 6.8, recommended by the distributor of this column. 0.5-ml fractions are collected and subjected to liquid scintillation counting. The retention times of the protein and the SAM are obtained directly from the recording of the absorbance of the eluate at 280 nm.

6.3.2. In vitro Study of the Binding of SAM to the COBA and COBF Proteins of Pseudomonas denitrificans

a) Purification of the COBF and COBA Proteins

The COBF protein of Pseudomonas denitrificans is purified as described below. In a typical purification experiment, wet cells (5 g) of strain SC150 Rif′ into which plasmid pXL1546 has been introduced (see Ex. 7.3), obtained after culturing in PS4 medium, are resuspended in 0.1 M Tris-HCl pH 7.7 (30 ml) and sonicated for 15 minutes at 4° C. The crude extract is then recovered by centrifugation for 1 hour at 50,000 g, and the supernatant is passed through a DEAE-Sephadex column (1 ml of gel) to remove the tetrapyrrole compounds present. Proteins (10 mg; 0.7 ml) of this extract are then injected onto a MonoQ HR 5/5 column equilibrated with the same buffer. The proteins are eluted with a linear KCl gradient (0 to 0.25 M). The COBF protein is eluted with 0.20 M KCl. It is diluted twofold with 0.1 M Tris-HCl pH 7.7 and purified a second time on a MonoQ HR 5/5. SDS-PAGE electrophoresis with visualisation with Coomassie blue is used to reveal the protein. This technique shows, moreover, that COBF is approximately 95% pure after this purification step. The NH₂-terminal sequence of the purified protein was determined as described above. Two NH₂-terminal sequences appear at the same time in each degradation cycle; they are the following sequences, in the proportions indicated:

Sequence 1 (abundance 34%)1234567891011AlaGluAlaGlyMetArgLysIleLeuIleIleSequence 2 (abundance 66%)1234567891011MetArgLysIleLeuIleIleGlyIleGlySer

Sequence 1 corresponds to the NH₂-terminal sequence of the COBF protein which is given in FIG. 16, except that the amino-terminal methionine is excised according to rules already stated (Hirel et al., 1989) by methionine aminopeptidase (Ben Bassat and Bauer, 1989). Sequence 2, present in the larger amount, corresponds to the same protein but having its translation initiation apparently done not at the translation initiation ATG codon we had assumed, but at that located 5 codons downstream on the coding frame (FIG. 16). In effect, the amino acids of this sequence are exactly those which are found in the sequence of the COBF protein starting from the second methionine (amino acid No. 6) of this sequence (FIG. 16). In this case, the amino-terminal methionine is not excised, which confirms the rules already stated (Hirel et al., 1989). In strain SC510 Rif′ carrying plasmid pXL1546, there are two translation initiations, on the one hand that corresponding to the methionine codon positioned at the correct distance, in our construction, from the Shine-Dalgarno sequence, and on the other hand that which is carried out at the second methionine codon occurring in the sequence of the cobF gene presented in FIG. 16. It emerges from this that the COBF protein proably begins not at the methionine indicated in FIG. 16, but at that occurring 5 amino acids further on.

At all events, this result shows that the COBF protein is, indeed, the one expressed, and that the latter is expressed in a form elongated by 4 amino acids. During purification, both protein forms are purified. In this example, the mixture of these two purified proteins is referred to by us as purified COBF protein.

The COBA protein of Pseudomonas denitrificans is purified as described above (Blanche et al., 1989).

b) Binding of SAM

The binding of SAM to these two proteins is studied as described above in Example 6.3 a). Bovine serum albumin and the purified COBH protein are used as negative controls. For the COBA and COBF proteins, a peak of radioactivity is observed at emergence from the TSK-125 column at the emergence time of these proteins (FIG. 20). In this test, the COBI protein displays the same property of binding of SAM. In contrast, there are no such peaks of radioactivity with BSA and the COBH protein. This test demonstrates the in vitro binding of SAM to the COBA, COBI and COBF proteins. These results show that COBA, COBI and COBF are SAM methyltransferases. This result is in complete agreement with the COBA and COBI activities, since they are the SUMT and the SP₂MT, respectively, of Pseudomonas denitrificans. The COBF protein is hence probably an SAM methyltransferase of the pathway of biosynthesis of cobalamins. This test confirms that COBF is a methyltransferase.

6.4—Determination of the Activity of COB Proteins by Sequence Homology Studies

This example illustrates how it is possible to find the COB proteins which are SAM methyl-transferases of the pathway of biosynthesis of cobalamins by comparisons between the sequences of various COB proteins of Pseudomonas denitrificans.

The COBI and COBA proteins are both SAM methyltransferases of the biosynthetic pathway. These two proteins were compared according to the programme of Kanehisa, 1984. This comparison brings out three regions of strong homology (FIG. 21). In each of these regions, there is more than 45% strict homology between the two proteins. Three regions of strong homology between COBA and CYSG are also presented (FIG. 22); they are the same regions of COBA which display a strong homology with COBI. These regions of strong homologies between COBA, CYSG and COBI display homology with other COB proteins. The proteins in question are COBF, COBJ, COBL and COBM (FIG. 23). As regards the region 1, the COBF, COBL and COBM proteins display significant homologies with respect to all the Genpro proteins, Genpro being a Genbank (version 59) protein extraction augmented by putative coding portions larger than 200 amino acids, according to the programme of Kanehisa (1984). As regards the region 2, the COBJ, COBL and COBM proteins display significant homologies with respect to all the Genpro (version 59) proteins. As regards the third region of homology, COBJ, COBL and COBM display significant homologies with respect to all the Genpro (version 59) proteins. The sequence comparisons hence enable it to be demonstrated that four proteins, COBF, COBJ, COBL and COBM, display significant homologies with the conserved regions of the sequences of three types of methyltransferases, COBA, COBI and COBF. The COBG, COBH and COBK proteins do not display significant homologies with the conserved regions of the methylases. The COBF protein displays a significant homology with the other proteins only in the region 1. These homologies must probably correspond to the fact that all these proteins are methyltransferases. This result ties up with the biological data described for COBF, relating to the capacity possessed by this protein for binding SAM in vitro (Example 6.3). These homologies on the one hand enable it to be confirmed that COF is an SAM methyltransferase of the pathway of biosynthesis of cobalamins, and on the other hand demonstrate that COBJ, COBL and COBM could be SAM methyltransferases of the pathway of biosynthesis of cobalamins. These results also show the homology existing between the COB proteins of P. denitrificans and the isofunctional proteins of other microorganisms.

EXAMPLE 6(B)
Purification and Cloning of the Methanobacterium ivanovii SUMT Structural Gene

This example illustrates how it is possible to obtain, in other microorganisms, COB enzymes and cob genes corresponding to those identified in P. denitrificans.

6(B).1. Purification of Methanobacterium ivanovii SUMT

This example describes the purification of Methanobacterium ivanovii SUMT and a study of its catalytic properties.

Methanobacterium ivanovii strain DSM2611 is cultured as described (Souillard et al., 1988). Wet cells (12 g) are obtained. The latter are resuspended in 0.1 M Tris-HCl buffer pH 7.6 (80 ml) containing 5 mM DTT and 1 mM EDTA, and sonicated for 1 h 30 min at 4° C. and then centrifuged for 1 h at 50,000 g. Free tetrapyrrole compounds are then cleared from the extract by passage through a small DEAE-Sephadex A25 column set up in the same buffer. The proteins precipitating at between 55 and 75% ammonium sulphate saturation are solubilised in a 0.1 M Tris-HCl pH 7.5, 0.5 mM DTT, 1.7 M ammonium sulphate buffer and injected onto a Phenyl-Superose HR 10/10 (Pharmacia France/SA) column eluted with a decreasing gradient (1.7 M to 0 M with respect to ammonium sulphate). The active fractions are passed through a Sephadex G-25 column equilibrated with 0.1 M Tris-HCl pH 7.5, 0.5 mM DTT, 25% glycerol buffer (buffer A), then injected onto a Mono Q HR 5/5 (Pharmacia France SA) column equilibrated with buffer A and eluted with a KCl gradient of 0 to 0.3 M; this step is repeated a second time under the same conditions. Gel permeation chromatography of the active fraction of the preceding step on Bio-Sil TSK-250 (BioRad France SA) enables a protein which is homogeneous in SDS-PAGE and in RP-HPLC (C-18 μBondapak) to be obtained. The different steps of purification, with their yield, as well as their purification factor, are described in the table below.

As shown in this table, the total purification factor is more than 4,500. Some properties of the pure enzyme have been studied according to methods already described (Blanche et al., 1989). This enzyme does indeed have SUMT activity, i.e. it does indeed catalyse the SAM-dependant transfer of two methyl groups at C-2 and at C-7 of uro'gen III. The molecular weight of the enzyme estimated by gel permeation is 60,000±1,500, while by SDS-PAGE it is 29,000, which shows clearly that it is a homodimeric enzyme. Under conditions already described (Blanche et al., 1989), the enzyme has a Km for uro'gen III of 52±8 nM. In addition, this enzyme does not show inhibition by substrate at concentrations below 20 μM, whereas Pseudomonas denitrificans SUMT shows an inhibition by uro'gen III at a concentration above 2 μM (Blanche et al., 1989).

TABLEPurification of M. ivanovii SUMTSp. activityPurificationVolProteins(u/mg ofPurificationstep(ml)(mg)proteins)Yieldfactor¹Crude extract927310.337——55-75% AS7.11531.215763.6Phenyl-Superose9.58.3415.355246Mono Q 5/51.00.252422431252Bio-Sil TSK1.00.0611537384561
¹calculated from the yield of proteins.

The Vmax of M. ivanovii SUMT was determined. It is 1537 U/mg of proteins. This value is greater than that found for P. denitrificans SUMT, already determined under optimal conditions for the reaction (taking account of its inhibition by uro'gen III), 489 U/mg of proteins (Blanche et al., 1989).

6(B).2. Cloning of the M. ivanovii SUMT Structural Gene in E. coli

6(B).2.1. Cloning of a fragment internal to the M. ivanovii SUMT structural gene. For this purpose, the procedure is as follows: 200 picomols of M. ivanovii SUMT are used for the NH₂-terminal sequencing of the protein as described above. In addition, a peptide fragment obtained by tryptic digestion of the protein is likewise subjected to a sequencing of its NH₂-terminal portion. The sequences obtained are presented in FIG. 48. The sense and antisense oligonucleotides 946, 923 and 947, respectively (see FIG. 48) are synthesised as described above; these oligonucleotides contain a restriction site at their 5′ end, which is either EcoRI for the sense oligonucleotides or HindIII for the antisense oligonucleotide. These oligonucleotides are used for an enzymatic DNA amplification experiment (Saiki et al., 1988) as shown diagrammatically in FIG. 48.B.

M. ivanovii genomic DNA is prepared in the following manner: M. ivanovii (DSM 2611) cells (0.4 g) are washed with 0.15 M NaCl solution. The cells are then incubated in a 25% sucrose, 50 mM Tris-HCl pH 8, lysozyme (40 mg) solution (4 ml), and thereafter for 2 to 3 h at 50° C. after the addition of proteinase K (40 mg) and a 0.2% SDS, 0.1 M EDTA pH 8 solution (5 ml). The DNA is then extracted with phenol/chloroform (50%/50%) twice and then twice with chloroform, and thereafter precipitated with isopropanol and taken up in TNE (10 mM Tris-HCl pH 8, 1 mM EDTA, 100 mM NaCl) (3 ml).

Enzymatic amplification of M. ivanovii DNA is performed according to the protocol of Saiki et al., 1988, in a volume of 0.1 ml with M. ivanovii genomic DNA (600 ng), using the primers 946 and 947 (reaction 1) or 923 and 947 (reaction 2). The buffer used for this reaction is 1 mM MgCl₂, 50 mM KCl, 0.001% gelatin and each dNTP at a concentration of 0.2 mM; for each amplification reaction, 10 mg of each oligonucleotide are used, as well as Taq DNA polymerase (2.5 units) (Cetus Corporation). Amplification is carried out over 30 cycles in the Perkin-Elmer Cetus DNA Amplication system; during each cycle, the DNA is denatured for 1 min at 95° C., the oligonucleotide primers are hybridysed with single-stranded DNA for 2 min at 38° C. and the newly formed strands are polymerised for 3 min at 72° C. The amplification products are then extracted with chloroform and thereafter undergo ethanol precipitation; they can then be visualised after migration on acrylamide gel, and thereafter be digested with restriction enzymes such as EcoRI and HindIII.

In the case of reaction 1, two fragments are observed: at 615 bp as well as at 240 bp. As regards reaction 2, two fragments are also observed: at 630 and 170 bp. The whole of the product of an enzymatic amplification reaction between the oligonucleotides 946-947 is separated by migration on acrylamide gel; the 615-bp fragment is purified as described above. This fragment is then digested with EcoRi and HindIII in order to make the ends of the fragment cohesive. This fragment is then ligated with the DNA of the replicative form of phage M13mp19. The ligation is transformed into E. coli TG1. Six recombinant clones containing a 615-bp insert are analysed by sequencing with the universal primer-20 (Pharmacia SA, France). As shown in FIG. 49, when the single-standed DNA of the recombinant phages which contain 615-bp insert is sequenced, there must be observed, downstream of the EcoRI site, a non-degenerate sequence corresponding to that of the oligonucleotide 946 followed, in the same frame, by a sequence coding for the amino acids LITLKAVNVLK?ADVVL (? means that, at this position, the residue could not be determined); this sequence corresponds to that which, in the NH₂-terminal sequence of SUMT, follows the amino acids corresponding to the oligonucleotide 946 (see FIG. 48). For two clones, there was actually observed, after the EcoRI site, a sequence able to code for the NH₂-terminal region of Methanobacterium ivanovii SUMT, this sequence beginning with the arrangement Pro-Gly-Asp-Pro-Glu-Leu which are the amino acids encoded by a sequence containing the oligonucleotide 946. This observation shows that these two recombinant replicative forms contain an insert which corresponds to a fragment internal to the Methanobacterium ivanovii SUMT structural gene. The replicative form carrying this fragment internal to the M. ivanovii structural gene is referred to as pG10. 6(B).2.2. Cloning of the Methanobacterium ivanovii SUMT structural gene

Methanobacterium ivanovii genomic DNA is digested with several restriction enzymes (single or double digestions). After digestion, the fragments are separated by agarose gel electrophoresis and are then transferred onto a nylon membrane as described above. After denaturation of the fragments thus transferred and prehybridisation, a hybridisation is performed with the replicative form pG10 as a ³²P-labelled probe, as described above. It is thus found that a 3.2-kb fragment emanating from an EcoRI-BqlII digestion of Methanobacterium ivanovii hybridises with the probe (see FIG. 50). Genomic DNA (40 μg) of M. ivanovii are then digested with EcoRI and BqlII and thereafter separated by migration on agarose gel. The fragments having a size of between 3 and 3.5 kb are electroeluted as described above. The fragments thus purified are ligated with the vector pBKS+ (Stratagene Cloning Systems, La Jolla) digested with BamHI-EcoRI. The ligation is transformed into E. coli DH5α (Gibco BRL). The transformants are selected on LB medium supplemented with ampicillin and X-gal. 800 white colonies are subcultured on filters; after growth and then lysis of the bacteria, a colony hybridisation is performed according to the technique of Grünstein and Hogness (1975). The probe used is the replicative form pG10 labelled with ³²p A single positive clone after this hybridisation test with the probe is found. The plasmid DNA of this clone is referred to as pXL1809 (see FIG. 56). A digestion of this DNA with EcoRI-XbaI enables a 3.2-kb insert to be visualised, as expected. Plasmid pXL1809 is sequenced on both strands by the technique of Chen and Seeburg (1985). A sequence of 955 bases is obtained (FIG. 51). An analysis of the open reading frames leads us to identify an open reading frame from base 34 (ATG) to base 729 (TGA). This open reading frame codes for a protein whose sequence is presented in FIG. 52. This protein has a molecular weight of 24,900 (see FIG. 53), which is close to the molecular weight of the protein purified from M. ivanovii. The NH₂-terminal sequence of this protein is exactly that determined for purified M. ivanovii SUMT (see FIG. 48 and FIG. 52). These observations establish unambiguously that the cloned and sequenced gene is indeed the M. ivanovii SUMT structural gene. Since this activity is assumed to participate in the biosynthesis of corrinoids in all bacteria, this gene is designated corA gene, and the protein encoded by this same gene CORA protein. The hydrophobicity profile of the CORA protein of M. ivanovii, produced from the programme of Hopp and Woods (1981), shows that it is, as expected, a hydrophilic protein, as presented in FIG. 54. The CORA protein of M. ivanovii shows a degree of strict homology of more than 40% with respect to COBA of P. denitrificans (FIG. 53). This homology extends over practically the whole of both proteins, since it relates to residues 3 to 227 of CORA of M. ivanovii and residues 17 to 251 of COBA of P. denitrificans. This homology reflects the structural homologies existing between two proteins that catalyse the same reaction. The regions which are most highly conserved between CORA and COBA of P. denitrificans are the same ones as are conserved between COBA of P. denitrificans and CYSG of E. coli (FIG. 22).

EXAMPLE 7
Expression of COB Proteins

7.1—Expression in Pseudomonas denitrificans

This example illustrates that the amplification of a structural gene for a COB protein of Pseudomonas denitrificans in Pseudomonas denitrificans leads to amplification of the activity of the COB protein.

7.1.1—Expression of the COBA Protein

Plasmid pXL557 corresponds to plasmid pXL59 into which the 2.4-kb BqlII-EcoRV fragment (at positions 80 and 2394, respectively, in the sequence of FIG. 7) of the 5.4-kb fragment has been cloned. This fragment contains the cobA and cobE genes.

Plasmid pXL545 contains only the cobE gene. Its construction has been described in Example 4.1.

These two plasmids were introduced by conjugative transfer into SC510 Rif′. Strains SC510 Rif′, SC510 Rif′ pXL59, SC510 Rif′ pXL557 and SC510 Rif′ pXL545 were cultured in PS4 medium. At 4 days, culturing was stopped and the SUMT activities were assayed according to a standard protocol already described (F. Blanche et al., 1989). The activities are given below.

TABLESUMT activity of SC510 Rif^rand of some of itsderivativesSUMT assayednmol/h/mg ofStrainproteinSC510 Rif^r0.05SC510 Rif^rpXL590.04SC510 Rif^rpXL5572.10SC510 Rif^rpXL5450.05

It emerges clearly from these results that only plasmid pXL557 brings about a marked increase in SUMT activity (a factor of 50) in SC510 Rif′. This increase results from the amplification of cobA and not of cobE, since plasmid pXL545, which permits the amplification of only cobE, does not produce an increase in SUMT activity. This result confirms that cobA is the structural gene for SUMT of Pseudomonas denitrificans. This result shows that it is possible to obtain an amplification of the SUMT activity in Pseudomonas denitrificans by amplification of the structural gene for SUMT of Pseudomonas denitrificans.

7.1.2—Expression of the COBI Protein

A fragment originating from the 8.7-kb DNA fragment containing the structural gene for SP₂MT (cobI) is cloned into a plasmid having a broad host range in Gram-negative bacteria, and this plasmid is then introduced by conjugation into Pseudomonas denitrificans SC510 Rif′. The S-adenosyl-L-methionine:precorrin-2 methyltransferase activity of the strain is then measured relative to that of the strain carrying the vector.

The 1.9-kb BamHI-BamHI-SstI-SstI fragment containing the cobH and cobI genes is purified from the 8.7-kb fragment. XbaI and EcoRI linkers are placed at the BamHI and SstI ends, respectively, after the latter have been filled in with bacteriophage T4 DNA polymerase. The fragment is then inserted between the XbaI and EcoRI sites of the broad host range plasmid pXL59. It carries kanamycin resistance. The plasmid thereby obtained is designated pXL1148 (FIG. 24).

Separately, a related plasmid was constructed: the 1.5-kb BamHI-BamHI-SstI fragment containing only the whole cobH gene and the 5′ portion of the cobI gene was purified from the 8.7-kb fragment. XbaI and EcoRI linkers were added at the BamHI and SstI sites, respectively, after the latter had been filled in or digested with phage T4 DNA polymerase. This fragment was then inserted between the EcoRI and XbaI sites of pXL59 to give plasmid pXL1149. Plasmids pXL1148 and pXL1149 differ only in the presence in pXL1148 of the 0.3-kb SstI-SstI fragment which contains the 3′ end of the cobI gene. pXL1148 possesses the whole structural gene for cobI, in contrast to pXL1149. Both plasmids contain the cobH gene.

These two plasmids were introduced by conjugation into SC510 Rif′. Strains SC510 Rif′, SC510 Rif′ pXL59, SC510 Rif′ pXL1148 and SC510 Rif′ pXL1149 are cultured in PS4 medium. After 4 days of culture, the cells are harvested and the SP₂MT activities are assayed as described in Example 6.1.3 a).

The result of these assays is given below, with the SP₂MT activities defined as in Example 6.1.3 a).

TABLESP₂MT activities of various strains derived fromPseudomonas denitrificansSP₂MT activity¹Strainin %SC510 Rif^r<5SC510 Rif^rpXL59<5SC510 Rif^rpXL114875SC510 Rif^rpXL1149<5
¹per 500 μg of crude extract introduced in the test.

The activity is expressed in % as defined in Example 6.1.3 a).

Only plasmid pXL1148 brings about a substantial increase in SP₂MT activity. In contrast, plasmid pXL1149 does not give results different from those observed with the controls SC510 Rif′ and SC510 Rif′ pXL59. pXL1148 is the only plasmid to contain the cobI gene, and it is the only one to amplify SP₂MT activity; this result confirms that the structural gene for SP₂MT of Pseudomonas denitrificans is the cobI gene. Furthermore, if the total proteins of these different strains are separated by electrophoresis under denaturing conditions (SDS-PAGE with 10% of acrylamide), the presence of a band which corresponds to a protein having a molecular weight of 25,000 is observed specifically in the case of pXL1148 (FIG. 25). The molecular weight of this protein corresponds to that of the COBI protein. Plasmid pXL1148 enables overproduction of the COBI protein to be obtained in

Pseudomonas Denitrificans.

7.1.3—Expression of COBF

The expression is obtained by positioning the Ptrp promoter of E. coli and the ribosome-binding site of the cII gene of bacteriophage lambda upstream of the cobF gene. The expression thereby obtained is much higher than that observed by simple gene amplification using the same multicopy plasmid.

The 2-kb EcoRI-BamHI-BamHI fragment of pXL1496 (Example 7.2.1 below) is purified (FIG. 26). This fragment contains the Ptrp promoter of E. coli and the ribosome-binding site of the cII gene of bacteriophage lambda upstream of the cobF gene. Dowstream of the cobF gene, there is the terminator of the rrnB operon of E. coli. This fragment is cloned at the EcoRI-BamHI sites of plasmid pKT230 to give pXL1546 (FIG. 26). pKT230 is a plasmid of the incompatibility group Q which replicates in almost all Gram-negative bacteria (Bagdasarian et al., 1981); this plasmid carries kanamycin resistance. Plasmid pXL1546 and pKT230 are introduced by conjugation into SC510 Rif′. Strains Sc510 Rif′, SC510 Rif′ pKT230 and SC510 Rif′ pXL1546 are cultured in PS4 medium as described above. After four days of culture, the total proteins of the different strains are analysed in 10% SDS-PAGE. As shown in FIG. 27, a protein of molecular weight 32,000 which is overexpressed is observed in the extract of SC510 Rif′ pXL1546; this protein comigrates with the protein which is overexpressed by E. coli B pXL1496 (Example 7.2.1 below). Furthermore, this protein is specifically expressed in strain SC510 Rif′ containing pXL1546, where it represents at least 20% of the total proteins. In contrast, this protein is not observed in the total proteins of strains SC510 Rif′ and SC510Rif′ pKT230. This overexpressed protein is hence the COBF protein.

7.1.4—Expression of COBH

This example describes the amplification of a DNA fragment of Pseudomonas denitrificans containing the cobH gene. The protein which is encoded by this gene is purified; it is the COBH protein. Plasmid pXL1149, described in Example 7.1.2, contains in the DNA insert originating from the 8.7-kb fragment only the whole cobH gene. In SC510 Rif′, this plasmid, in contrast to the vector, brings about the overexpression of a protein of molecular weight 22,000 (FIG. 25).

7.1.5—Expression of COBV

This example describes the amplification of cobalamin-5′-phosphate synthase activity by a plasmid carrying only cobV (pXL699, see FIG. 38). The cobalamin-5′-phosphate synthase activity is amplified in SC877 Rif′ by plasmid pXL699 by a factor of 50 relative to the same strain with the vector pXL435, pXL1303, pXL1324 or pKT230. This plasmid contains in its insert only the whole of cobV plus the 5′-terminal portions of ORF18 and of cobU. In such a strain (SC877Rif′ pXL699), the COBV protein is definitely overexpressed; this overexpression is by a factor of 50 relative to the expression of strain SC877Rif′.

7.1.6—Expression of the CORA Protein

The 1.5-kb EcoRI-BamHI-BamHI fragment of pXL1832 (see Example 7.2.4), containing the Ptrp promotor and then the RBS cII of bacteriophage λ, the M. ivanovii SUMT structural gene and the terminator region of the rrnB operon of E. coli, is cloned at the EcoRI-BamHI sites of pKT230 (Bagdasarian et al., 1981). In this manner, plasmid pXL1841 is obtained (see FIG. 56). This plasmid is mobilised in P. denitrificans SC510 Rif′ as described above. A transconjugant is studied in greater detail. This strain is cultured in PS4 medium, and the SUMT activity of the bacterial extracts is assayed at the same time as that of the control strain SC510 Rif′ pXL435 (Cameron et al., 1989). The activities of these strains are presented below.

Strain SUMT specific activity in pmol/h/mg of proteins

SC510 Rif′pXL435 50-100

SC510 Rif′pXL1841 1700

This result shows clearly that there is expression of the SUMT activity of M. ivanovii in P. denitrificans as a result of plasmid pXL1841, since the SUMT activity of strain SC510 Rif′ pXL1841 is markedly greater than that of SC510 Rif′ pXL435.

7.2—Expression in E. coli

This example illustrates how a COB protein of Pseudomonas denitrificans can be overproduced in E. coli.

7.2.1—Expression of COBF

The 2250-bp EcoRI-XhoI fragment of the 8.7-kb EcoRI fragment (at the respective positions 0 and 2250 in the sequence presented in FIG. 8) was cloned into phage M13mp19 (Norrander et al., 1983) between the EcoRI and SalI sites. The plasmid thereby constructed is designated pXL1405. An NdeI site was introduced by directed mutagenesis so that the last three bases (ATG) of this restriction site constitute the translation initiation site of the cobF gene. This amounts to modifying the three bases which precede the ATG of the cobF gene, GAA (the G is at position 733 in the sequence presented in FIG. 8), to CAT. The NdeI-SphI-SphI fragment (FIG. 26) containing the cobF gene is then purified; this 1.5-kb fragment is then cloned between the NdeI-SphI sites of plasmid pXL694 (Denèfle et al., 1987). The plasmid thereby constructed is designated pXL1496 (FIG. 26). Signals for regulation of genetic expression in E. coli are present in the 120-bp EcoRI-NdeI fragment (which originates from pXL694) which precedes the cobF gene. These signals consist of the [−40+1] region of the Ptrp promoter of E. coli, and then of 73 bp which contain the ribosome-binding site of the cII gene of bacteriophage λ (Denèfle et al., 1987). Downstream of the cobF gene, there are the terminators of the rrnB operon of E. coli (in the HindIII-BamHI fragment). Plasmid pXL1496 was introduced by transformation into the E. coli strain (Monod and Wollman, 1947). Expression of the cobF gene was studied as already described (Denèfle et al., 1987) under conditions where the Ptrp promoter is either repressed (in the presence of tryptophan) or not repressed (absence of tryptophan). The medium in which the expression was carried out is M9 minimum medium (Miller, 1972) supplemented with 0.4% of glucose, 0.4% of casamino acids, 10 mM thiamine and 40 μg/ml of tryptophan in the case where it is desired to repress the Ptrp promoter. E. coli strain B pXL1496 was cultured at 37° C. in the medium described above with ampicillin (100 μg). As shown in FIG. 28, the absence of tryptophan brings about the expression of a protein of molecular weight 32,000. In effect, in the extract of total proteins of E. coli B pXL1496 analysed in SDS-PAGE (FIG. 28), a protein of molecular weight 32,000 D which represents between 1 and 4% of the total proteins is clearly observed. This protein is present in markedly smaller quantities in the extract of the total proteins of E. coli B pXL1496 cultured under the same conditions but in the presence of tryptophan. The molecular weight of the protein which is expressed under these conditions is close to the molecular weight of the COBF protein deduced from the amino acid sequence of the protein, which is 28,927 (FIG. 16). The protein which is thus expressed in E. coli is the COBF protein.

7.2.2—Expression of COBT

Overproduction is obtained by fusing the lac promotor and the first three codons of lacZ of E. coli to the 5′ end of the cob gene.

The EcoRI site located at position 2624 in the sequence presented in FIG. 32 of the 4.8-kb fragment contains the fourth codon of the cobT gene. The 3.5-kb EcoRI-XbaI fragment of pXL837 (see FIG. 36) is cloned at the EcoRI and XbaI sites of pTZ18R or pTZ19R (Pharmacia) to generate pXL1874 or pXL1875, respectively; these two plasmids differ in the orientation of the truncated cobT gene with respect to the promoter of the lactose operon of E. coli (Plac). Plac is upstream of cobT in pXL1874 while the opposite is true in pXL1875. Cloning of the EcoRI-XbaI fragment of pXL837 at the EcoRI-XbaI sites of pTZ18R enables a protein fusion to be carried out between the first 4 amino acids of E. coli β-galactosidase and the cobT gene from its 4^thcodon. Expression of this lacZ′ ′cobT gene is under the control of the expression signals of lacZ. Plasmids pXL1874, pXL1875 and pTZ18R are introduced by transformation into E. coli strain BL21. Expression of the cobT gene is studied as already described (Maniatis et al., 1989).

As shown in FIG. 42B, a protein of molecular weight 72,000 is expressed only with pXL1874 and represents, in the extract of total proteins of BL21, pXL1874 analysed in SDS-PAGE, 1 to 4% of the total proteins. The molecular weight of the protein which is expressed under these conditions is close to the molecular weight of the COBT protein deduced from the amino acid sequence, which is 70,335, in FIG. 40. This experiment shows clearly that, from the EcoRI site located in the fourth codon of the cobT gene, an open reading frame compatible with that found for the cobT gene can be expressed.

7.2.3—Expression of a Truncated COBS Protein

A BamHI site is located at the 45th codon of the COBS gene. The 1.2-kb BamHI-BamHI fragment containing the 3′ portion of the cobs gene and sequences downstream of this gene is excised from pXL843 and cloned at the BamHI site of plasmid pET-3b (Rosenberg et al., 1987) to generate pXL1937. The BamHI fragment is oriented in such a way that the truncated portion of the cobS gene is fused, in frame, with the first 12 codons of the major capsid protein of bacteriophage T7 or gene 10 (Rosenberg et al., 1987). This hydbrid gene is under the control of the Φ10 promotor of bacteriophage T7. Plasmid pXL1937 and also pET-3b are introduced by transformation into E. coli BL21 pLysS (W. Studier, personal communication). After reisolation on selective medium, both strains are cultured in L liquid medium to an OD at 610 nm of 1; at this stage, the medium is adjusted to an IPTG (isopropyl β-thiogalactoside) concentration of 1 mM in order to induce expression of the polymerase of bacteriophage T7 (Rosenberg et al., 1987). The culture is then incubated for 3 h at 37° C. and bacterial lysates are thereafter prepared. The total proteins of the bacteria thus cultured are separated by PAGE under denaturing conditions. As seen in FIG. 42A, there is specifically overexpression of a 33,000 protein with the culture BL21 pLysS pXL1937. This molecular weight is entirely compatible with the expected molecular weight for the fusion protein (33 kD). This experiment shows clearly that, from the BamHI site located at the 45th codon of the cobs gene, an open reading frame compatible with that found for the cobs gene can be overexpressed.

7.2.4. Expression of the CORA Protein

The following oligonucleotides were synthesised as described above:

oligonucleotide 12775′ GGC CGA ATT CAT ATG GTA GTT TAT TTA 3′ ------- 1 2 3 4 5 (1 to 5 first 5 EcoRI _{———————————} codons of M. ivanovii NdeI SUMT)oligonucleotide 12785′ GGC CGA GCT CTA TTA CAT AAT T _{————————}=============== SstI

(=sequence appearing in FIG. 51, positions 926 to 915, in the strand complementary to the coding strand)

Oligonucleotide 1277 possesses the recognition sequences for the restriction enzymes EcoRI and NdeI. The last three bases of the NdeI site (ATG), which corresponds to a translation initiation codon, are directly followed by codons 2 to 5 of the M. ivanovii SUMT structural gene as appear in the sequence presented in FIG. 52. The oligonucleotide 1278 contains the recognition sequence for SstI, followed directly by the sequence TATTACATAATT which corresponds to a sequence present in the 955-bp fragment containing the corA gene presented in FIG. 51; this sequence occurs at position 926 to 915 (see FIG. 51) in the strand complementary to the strand coding of the CORA protein. The two oligonucleotides 1277 and 1278 hence contain sequences in their 3′ portion corresponding, respectively, to the coding strand of the corA gene and to the complementary strand downstream of this gene. These two oligonucleotides may be used to carry out an enzymatic amplification experiment with plasmid pXL1809 as template. This experiment makes it possible to obtain a 910-bp fragment containing the corA gene of M. ivanovii possessing an NdeI site at the ATG of the corA gene, and an SstI site at the other end of the fragment after the end of the corA gene. Enzymatic amplification is carried out as described above for the enzymatic amplification performed on the genomic DNA of M. ivanovii, except that the template consists of DNA (10 ng) of plasmid pXL1809; the temperatures used are the same, but only 20 amplification cycles are carried out. As described above, the amplification products are digested with NdeI and SstI before being separated by migration on agarose gel. As expected, a fragment 910 bp in size is indeed visualised. This fragment is purified as already described. This fragment is cloned at the NdeI and SstI sites of pXL694 (Denèfle et al., 1987). The resulting plasmid, designated pXL1832, is described in FIG. 56. In this plasmid, in the same way as described in Example 7.2, the M. ivanovii SUMT structural gene is preceded by the ribosome binding site of the cII gene of bacteriophage λ. Upstream of this RBS there is the Ptrp promotor. Plasmid pXL832 is introduced into E. coli B5548, which is an E. coli strain carrying the mutation cysG44 (Cossart and Sanzey, 1982) by transformation. The SUMT activities of the strains E. coli B5548 pUC13 and E. coli B5548 pXL1832 are assayed on extracts obtained from cells cultured in LB medium supplemented with ampicillin. The assay of SUMT activity is carried out as already described (Blanche et al., 1989). The results of this assay are given below.

SUMT specific activityStrainin pmol/h/mg of proteinsE. coli B5548 pUC135.9E. coli B5548 pXL1832310

The results presented in the table above show clearly that there is expression of a SUMT activity in E. coli strain B5548 when the latter contains a plasmid pXL1832 which expresses M. ivanovii SUMT. The M. ivanovii SUMT can hence be expressed in E. coli.

EXAMPLE 8
Amplification of the Production of Cobalamins by Recombinant DNA Techniques

8.1—Amplification in P. denitrificans

This example illustrates how an improvement in the production of cobalamins is obtained in Pseudomonas denitrificans SC510 Rif′ by amplification of cob genes of Pseudomonas denitrificans SC510.

8.1.1 Improvement in the production of Cobalamins in Pseudomonas denitrificans by Removal of a Limiting Step in the Biosynthesis of Cobalamins

This example illustrates how the productivity of cobalamins in Pseudomonas denitrificans strains may be improved by amplification of cob genes of Pseudomonas denitrificans. This improvement results from the removal of a limiting step of the biosynthetic pathway.

Plasmid pXL367 is described in Example 4.2 (FIG. 13). This plasmid corresponds to pRK290 (Ditta et al., 1981) into which the 8.7-kb EcoRI fragment has been inserted. This plasmid pXL367 effects an improvement in the biosynthesis of cobalamins in strain SC510 Rif′. Strains SC510 Rif′, SC510 Rif′ pRK290 and SC510 Rif′ pXL367 are cultured in an Erlenmeyer in PS4 medium according to the conditions described in the experimemntal protocols. An improvement in the production titre due to the presence of plasmid pXL367 is observed. In effect, strain SC510 Rif′ pXL367 produces 30% more cobalamins than strains SC510 Rif′ and SC510 Rif′ pRK290. This improvement is not due to the amplification of unspecified genes of Pseudomonas denitrificans, but to the specific amplification of the genes carried by the 8.7-kb EcoRI fragment. In effect, plasmid pXL723 described in FIG. 11 gives no improvement, and the same production titre is observed with this plasmid as with strains SC510 Rif′ and SC510 Rif′ pRK290.

8.1.2 Improvement in the production of coenzyme B1₂in Pseudomonas denitrificans by removal of two limiting steps in the biosynthesis of cobalamins

This example illustrates how the productivity of cobalamins in strains of Pseudomonas denitrificans may be improved by amplification of cob genes of Pseudomonas denitrificans. This improvement results from the removal of two limiting steps of the biosynthetic pathway.

The 2.4-kb ClaI-Eco RV fragment derived from the 5.4-kb fragment (containing the cobA and cobE genes) is cocloned with the 8.7-kb EcoRI fragment into the broad host range plasmid pXL203. The plasmid thereby constructed is referred to as pXL525 (FIG. 29). This plasmid is introduced into SC510 Rif′ by conjugation. Strain SC510 Rif′ pXL525 produces 20% more cobalamins than SC510 Rif′ pXL367. Amplification of the cobA and cobE genes enables a further limiting step in SC510 Rif′ in the biosynthesis of cobalamins to be removed. Pseudomonas denitrificans strain SC510 Rif′ is improved in the present example by the successive removal of two limiting steps. This example shows that the removal of two limiting steps in the biosynthesis of cobalamins can lead to further improvements in production.

8.2—Improvement in the Productivity of Cobalamins in Agrobacterium tumefaciens

This example illustrates the improvement in the production of cobalamins in a strain productive of cobalamins by amplification of the cob genes of Pseudomonas denitrificans SC510.

The strain used is a strain of a Gram-negative bacterium; it is a strain of Agrobacterium tumefaciens.

The plasmids described in Examples 4.2 and 8.1, pXL367 and pXL525, as well as the vector pRK290 (Ditta et al., 1981) and plasmid pXL368 (FIG. 29), are introduced by conjugative transfer into Agrobacterium tumefaciens strain C58-C9 Rif′ (Cameron et al., 1989). Strains C58-C9 Rif′, C58-C9 Rif′ pRK290, C58-C9 Rif′ pXL367, C58-C9 Rif′ pXL368 and C58-C9 Rif′ pXL525 are cultured in PS4 medium at 30° C. as described above. The cobalamins produced are assayed as described above. The production titres are given in the table below.

TABLETitres of vitamin B₁₂produced by differentrecombinant strains of Agrobacterium tumefaciensVitamin B₁₂Strainin mg/lC58-C9 Rif^r0.4C58-C9 Rif^rpRK2900.4C58-C9 Rif^rpXL3670.8C58-C9 Rif^rpXL3680.8C58-C9 Rif^rpXL5251.2

As is clearly apparent in the above table, the production of cobalamins is improved in the Agrobacterium tumefaciens strain used. Two different plasmids improve the production of cobalamins in the Agrobacterium tumefaciens strain used: pXL367 and pXL368. These plasmids contain the 8.7-kb EcoRI fragment (cobF to cobM genes) and the 2.4-kb ClaI-EcoRV fragment (cobE and cobA gene), respectively. Separately, they improve the production of cobalamins by Agrobacterium tumefaciens C58-C9 Rif′ by a factor of 2; this result shows that it is possible to improve the production of cobalamins by a strain of Agrobacterium tumefaciens by amplifying fragments carrying cob genes of Pseudomonas denitrificans. In the present case, it is possible to speak of heterologous improvment, that is to say improvement of the production of cobalamins by one strain by means of the amplification of cob genes of another strain.

The improvements in production of cobalamins provided by the different Pseudomonas denitrificans fragments containing cob genes are capable of cumulation, i.e., by putting into the same plasmid the two fragments which are separately cloned into pXL367 and pXL368, additive improvements are observed, in the sense that plasmid pXL525 provides in Agrobacterium tumefaciens C58-C9 Rif′ an improvement in the production greater than that provided by each of the fragments cloned separately into the same vector.

8.3—Improvement in the Productivity of Cobalamins in Rhizobium meliloti

This example describes the improvement in the production of cobalamins by another strain productive of cobalamins.

The plasmid described in Example 8.2, pXL368, as well as the vector pRK290 (Ditta et al., 1981), are introduced by conjugative transfer into Rhizobium meliloti strain 102F34 Rif′ (Leong et al., 1982). The transconjugants, namely 102F34 Rif′, 102F34 Rif′ pRK290 and 102F34 Rif′ pXL368, are cultured in PS4 medium at 30° C. as described above. The cobalamins produced are assayed as described above. The production titres are given in the table below.

TABLETitres of cobalamins produced by differentrecombinant strains of Rhizobium melilotiVitamin B₁₂Strainin mg/l102F34 Rif^r0.4102F34 Rif^rpRK2900.4102F34 Rif^rpXL3680.8

As is clearly apparent in the above table, the production of cobalamins is improved in the Rhizobium meliloti strain used. Plasmid pXL368 improves the production of cobalamins by the Rhizobium meliloti strain used. This plasmid contains the 2.4-kb ClaI-EcoRV fragment (cobA and cobE genes); it improves the production of cobalamins by Rhizobium meliloti 102F34 Rif′ by a factor of 2. This result shows that it is possible to improve the production of cobalamins by a strain of Rhizobium meliloti by amplifying fragments carrying cob genes of Pseudomonas denitrificans. In the present case, it is possible to speak of heterologous improvement, that is to say improvement of the production of cobalamins by one strain by means of the amplification of cob genes of another strain.

EXAMPLE 9
Assay of Corrinoids and Decobaltocorrinoids in Musts and Cells of Strains Productive of Corrinoids

This example illustrates how it is possible to identify and assay the different corrinoids and decobaltocorrinoids produced by different strains productive of cobalamins. This assay makes it possible, inter alia, to assay coenzyme B₁₂.

The musts (or the cells alone) are cyanide-treated as already described (Renz, 1971). After centrifugation, an aliquot of the supernatant is passed through a DEAE-Sephadex column which is then washed with 0.1 M phosphate buffer. The collected fractions are combined and desalted on a Sep-Pak C-18 (Waters) cartridge. After evaporation and resuspension in water (100 μl to 1 ml depending on the quantity of corrinoids present), the corrinoids are identified and assayed by HPLC on a Nucleosil C-18 column (Macherey-Nagel). The column is eluted at 1 ml/min with an acetonitrile gradient (from 0% to 100%) in 0.1 M potassium phosphate buffer containing 10 mM KCN.

The corrinoids are visualised by UV detection at 371 nm and/or by specific detection of ⁵⁷Co (if culturing has been performed in the presence of ⁵⁷CoCl₂) using a Berthold LB 505 detector. They are hence identified by comparison of their retention times with standards. Similarly, the “metal-free corrinoids” (hydrogenobyrinic acid, hydrogenobyrinic acid monoamide and hydrogenobyrinic acid diamide) are visualised by UV detection at 330 nm. By this technique, the following intermediates are separated: cobyrinic acid, cobyrinic acid monoamide, cobyrinic acid diamide, cobyrinic acid triamide, cobyrinic acid tetraamide, cobyrinic acid pentaamide, cobyric acid, cobinamide, cobinamide phosphate, GDP-cobinamide, B₁₂phosphate and vitamin B₁₂. The adenosylated forms of these products are also separated and assayed by this technique. For this purpose, the initial step of the cyanide treatment is cut out and the HPLC column is eluted with buffer devoid of KCN. FIG. 31 gives the retention times of different standards separated by this system and identified at emergence from the column by UV absorbance.

A sample of strain SC510 Rif′ was deposited on 30th Jan. 1990 at the Centraal Bureau voor Schimmelcultures at Baarn (Netherlands), where it was registered under reference CBS 103.90.

Number	Date	Country	Kind
90 01137	Jan 1990	FR	national
PCT/FR91/00054	Jan 1991	WO	international

	Number	Date	Country
Parent	08426630	Apr 1995	US
Child	10724598	Dec 2003	US
Parent	07916151	Sep 1992	US
Child	08426630	Apr 1995	US

Methods of increasing the production of cobalamins using cob gene expression

Information

Publication Number

Date Filed

Date Published

Inventors

Original Assignees

CPC

US Classifications

International Classifications

Abstract

Description

Claims

Priority Claims (2)

Continuations (2)