EP0495078A1

EP0495078A1 - Corynebacteria integron, method of transformation of corynebacteria by said integron, and corynebacteria obtained

Info

Publication number: EP0495078A1
Application number: EP91915423A
Authority: EP
Inventors: Armel André Yves GUYONVARCH; Oscar Julio Reyes Alvarado; Jean Christian Jocelyn Labarre; Céline Anne-Marie BONAMY; Gérard Louis André LEBLON
Original assignee: Centre National de la Recherche Scientifique CNRS
Current assignee: Centre National de la Recherche Scientifique CNRS
Priority date: 1990-08-08
Filing date: 1991-08-08
Publication date: 1992-07-22
Also published as: CN1061624A; IL99126A0; ZA916216B; CA2067240A1; AU8442391A; FI921527A; FR2665711A1; HUT63656A; WO1992002627A1; MX9100560A; KR920702423A; FI921527A0; NZ239309A; IE912791A1; FR2665711B1; JPH05502797A; BR9105857A; PT98602B; AU646886B2; PT98602A

Abstract

La présente invention concerne un intégron de corynébactérie caractérisé en ce qu'il comporte: un gène assurant une sélection efficace dans ladite corynébactérie, une séquence homologue du génome de ladite corynébactérie, lesdites séquences ayant été adaptées à ladite bactérie. Application notamment à la production de protéines par culture de corynébactéries transformées par ledit intégron lorsqu'il comporte une séquence codant pour une protéine d'intérêt.The present invention relates to a corynebacterium integron characterized in that it comprises: a gene ensuring efficient selection in said corynebacterium, a sequence homologous to the genome of said corynebacterium, said sequences having been adapted to said bacterium. Application in particular to the production of proteins by culture of corynebacteria transformed by said integron when it comprises a sequence encoding a protein of interest.

Description

CORYNEBACTERIA INTEGRON, PROCESS FOR CONVERTING CORYNEBACTERIA BY SAID INTEGRON, AND CORYNEBACTERIA OBTAINED

The present invention relates to the integration of predetermined DNA sequences into the genome of corynebacteria.

This integration can have two main purposes. On the one hand, it may involve producing a particular protein from a corynebacterium strain, either that this protein is not normally expressed by the strain in question, or else making it overproduce a homologous protein. But, on the other hand, it can be a question of blocking the expression of a gene by carrying out an interruption of this one, which will lead to the cancellation of the corresponding enzymatic activity and to the accumulation of the substrate of the reaction in the cell.

The corynebacteria which group, in addition to Corynebacterium, Brevibacterium, are bacteria whose handling has so far proved to be rather delicate:

first of all because there was little or no metnode allowing their transformation, and

very significant restriction barriers exist which make the transformations using DNA from other bacterial sources most of the time ineffective.

Recently, we have been able to demonstrate the possibility of transforming corynebacteria by electroporation methods. However, if the techniques of transformation by self-replicating vector are interesting, it is preferable most of the time, and at the industrial level, to have bacteria which have been transformed by integration into the chromosome, i.e. strains which are stable over time, both in terms of the number of copies of the integrated element and in terms of its location.

It is the object of the present invention to provide systems for integration into the genome of corynebacteria.

The present invention relates to corynebacteria integrons characterized in that they comprise:

a gene ensuring selection, hereinafter called "selection gene", effective in said corynebacterium,

a homologous sequence of the genome of said corynebacterie.

said sequences having been adapted to said bacteria. By "integron" is intended to denote a non-replicating vector having the property of integrating into the genome of a corynebactery. this integron can be in linear or circular form.

However, the integron generally comes from a self-replicating piasmide which allows synthesis in a different host, Escherichia coli for example. However, before the integration step, it is preferable to delete all traces of DNA of non-corynebacterial origin, except the selection gene, and in particular all the sequences involved in replication.

The effective selection genes in said corynebacteria are:

- either resistance genes to a particular substance, antibiotic in particular,

- or genes conferring a clearly identifiable phenotype, coloring and / or complementation for example.

In the present case, the selection by resistance to an antibiotic is more particularly interesting. Thus, we can use:

- the AphlII genes conferring resistance to kanamyone. noted Km ^R , - the Cat genes conferring resistance to chloramphenicol, noted Cm ^R.

However, other genes can be used, notably resistance to erythromycin.

By "homologous sequence" is intended to denote sequences which correspond to those present in the transformed corynebacterium or which have a homology rate greater than 80%, they may be sequences of the same species or not, these sequences may moreover be synthetic.

The sequences will have to be adapted or not, that is to say to take into account the problems of restriction barriers existing in corynebacteria, by methods some of which are described below.

Preferably, this integron comes from a piasmid which comprises, in addition to the integron, a replicative part, the integron being flanked by restriction sites allowing its excision, and preferably from reverse repeat sequences corresponding to a restriction site not present in the integron. for example Notl, BstXI or SacI. Thus, by digestion using the enzyme which recognizes said restriction site, one can obtain directly the integron, which can, if necessary, be circularized, since the ends obtained are complementary, before being used for transformation of corynebacteria.

In the system according to the present invention, the integron therefore preferably forms part of a piasmidic vector which is capable of replicating with the aid of rephcation origins. or repiicon, placed in the repiicative part. According to the nature of the rephcon present in this replicative cassette. the composite plasmid can replicate either only in corynebacteria, if it consists of a single endogenous replicon, or both in corynebacteria and in a foreign host, Eschericnia coii for example, when associated with the plasmid two different rephcons, each allowing rephcation in its own host.

In this case, the construction of the piasmid may be carried out in different systems these corynebacteria, which can sometimes be advantageous given the difficulties of this construction αans Corynebacterium and Brevibacterium.

Thanks to the presence of these homologous sequences present simultaneously in the integron and in the genome. the integron v has to merge by recombination in the chromosome. Amsi. in the primary transformant, the gene initially cloned at the multiple site of integration of the integron is found to be duplicated in the bacterial chromosome (see Figure 2a).

Such duplication can be of technical interest insofar as the doubling of the gene can lead to an increase in the activity created by this gene, in particular the corresponding enzymatic activity.

As the structure of the primary integrant corresponds to a direct tandem duplication of the homologous sequences flanking the selection gene, it is possible to provide for an amplification of this structure by selecting the growth of the primary integrant on a medium making it possible to detect the marigolds overexpressing the selection gene. Thus, when the selection gene is the antibiotic resistance gene, the most resistant strains can be selected, on media with an increasing antibiotic content, which strains must correspond to an overexpression of the genes corresponding to the homologous sequences, but also any gene or DNA sequence that has been inserted into the integron.

Of course, the integron will preferably comprise, in addition to the homologous sequence or sequences, a sequence coding for a sequence of interest, in particular for a peptide or protein of interest which may be homologous, that is to say come from a corynebacterium, or else heteroiogue, that is to say come from other bacterial species, but also be of eukaryotic or synthetic origin.

These sequences will preferably include the elements ensuring their expression in the corynebacteria or else will be inserted in phase so as to be able to be expressed by the expression elements of the host bacterium.

In the case of Corynebacterium, it may be advantageous for example to obtain an overexpression of certain enzymes, in particular gltA or gdhA.

As indicated previously, it is possible to use this system to ensure the interruption of a gene by using an integron according to the invention. By interruption or by substitution, as described in FIG. 2b, the corresponding gene is inactivated, this leads to an overproduction of the substrate of the enzyme encoded by the corresponding gene.

In general, the integron will be designed in the form of an integration cassette, that is to say that in addition to the selection gene and the homologous sequence, which in certain cases may be confused, provision will be made for sequence comprising several cloning sites which will allow the insertion of DNA sequences and / or genes at will.

In all cases, for the purpose of genetic confinement, an attempt will be made to provide an integron devoid of repiicative DNA from another species.

It is also possible to provide more complex integration systems, in particular integration systems which include in addition to the sequences of a transposable element. The sequences of transposable elements can be all of the sequences in. ensure the transposition, with the exception of the proteins coded by the corynebacterium, but they can also be transposables which are devoid of the sequences coding for the transposases.

Among the transposable elements. mention should be made of phage Mu, in particular in the form of a miniMu phage. In the case of transposable elements such as phage miniMu. as will be seen below, markers have been used which may be part of the toll or of different origin, for example colored markers.

The invention also relates to integrons using transposable elements originating from different corynebacteria. in particular this Brevibacterium. in particular ISaBl as described in figure 9.

Example 10 presents the characterization of the ISaBi element and Example 11 gives a general method for selecting and identifying this type of transposable element.

The present invention also relates to integrons comprising a sequence providing transposed elements, in particular elements coding for transposases and / or transposition repressors.

The integrons according to the inv ention can understand all or part of these sequences in question, in particular that corresponding to ISaBi.

This type of integration structure comprising a miniMu phage fragment. a homologous DNA sequence and a gene this selection can make it possible, as with the structures described above, to obtain either an overexpression of a particular gene, or the disruption of a gene when this proves necessary.

These latter structures, although different from the preceding structures will nevertheless be called for simplicity "integron".

The present invention also relates to strains of corynebacteria obtained by integrative transformation using the integrons described above, in particular when the integron has been introduced by electrotransformation.

Among the worries this corynebacteria utiiisacles. more particularly: B. lactofermentum,

B. flavum,

C. glutamicum,

C. melassecola,

for their industrial interest.

In the case where the cloning is carried out on a host different from the corynebacterium final host of the integron, the transfer of the integron can take advantage of the possible replicative properties of the construction in the corynebacterium in order to adapt the integron to the corynebacterium. Thus we will begin by introducing the piasmide comprising, in addition to the integron, a replicative part, in the same strain where the integration will take place, then this integron thus adapted will be recovered by extraction of the piasmide then digestion -by the enzyme (s) which release the integron, the purified fragment then being self-ligated or not and the ligation product used for the integrative transformation; in this case the restriction barriers no longer constitute a difficulty.

In some cases it may be useful to go through an intermediate corynebacterium, in particular in the case where the DNA is of E. coli origin, it may be advantageous to adapt the integron to Brevibacterium lactofermentum before adapting it to Corynebacterium melassecola.

Finally, the invention relates to the use of Corynebacteria according to the invention in the context of industrial processes, in particular for the preparation of proteins or meτabolites involving the integron according to the invention.

The examples below will better demonstrate the advantages of the present invention.

In the attached figures:

FIG. 1 shows diagrammatically the preparation of an integron starting from a replicative piasmid,

_ Figure 2 shows schematically the insertion of an integron.

. by simple recombination (a)

. by double recombination ((b),

FIG. 3 diagrams the structure of pCGL519,

FIG. 4 diagrams the percentage of resistance to kanamycin as a function of time for two transformed strains. - Figure 5 shows schematically the structure of the miniMu.

. MudII 1681,

. MudII 168 1 -Cat.

- Figure 6 shows schematically the plasmids

. pCGL107 and

. pCGL 107 :: Mud +,

- Figure 7 shows schematically the integration of the desired integrons whether or not containing the miniMu.

FIG. 8 represents the restriction map of the insertion element of Brevibacterium lactofermentum CGL2005 (B 115) cloned from an insertion in the 3 'terminal end of the lac operon,

FIG. 9 represents the ISaBi sequence,

FIG. 10 represents the restriction map of ISaBi,

- Figure 11 shows the restriction map of the pCSI330 piasmid, - Figure 12 shows the restriction map of the pCGL331 piasmide. EXAMPLE 1

Construction of a chromosomal DNA library of Corynebacterium melassecola ATCC 17965 and cloning of the gltA gene

The chromosomal DNA of the C. melassecola ATCC 17965 strain was prepared according to the method adapted from Ausubel et al. (1987). Digestion managed by the restriction enolonuclease Mbol (Boehringer) was carried out on 10 μg of this DNA following the protocol described in Maniatis et al. (1982). The DNA fragments were separated according to their size on a sucrose gradient as described by Ausubel et al. (1987). Fragments between 6 and 15 kb in size were used for the construction of the library.

The cloning plasmid pUN 121 (Nilsson et al., 1983) was prepared by the method of Birnboim and Doly, (1979) from the strain of E. coli GM2199 available freely. The piasmid was linearized by the Bell restriction endonuclease (Boehringer).

The bank was built by hgation with T4 DNA iigase

(Boehringer) under the conditions described by Ausubel et al. (1987), 1 μg of pUN 121 piasmid linearized by Bell and 2 μg of DNA fragments of 6 to 15 kb described above. The ligation mixture was introduced into the E. coli DH5alpha strain by electroporation following the protocol. described by Dower et al. (1988). The E. coli clones carrying the recombinant plasmids were selected directly by their capacity to grow on LB medium containing 10 μg / ml of tetracycline. The plasmids of all the tetracycline resistant clones were prepared by the method of Birnboim and Doly (1979). All of these plasmids correspond to the DNA library.

The E. coli W620 strain deficient for citrate synthase activity was transformed with the DNA library of C. melassecola ATCC17965. A transforming clone of E. coli W620 capable of growing on minimum selection medium containing tetracycline was selected. This clone carries a recombinant plasmid pCGL508. Different subcloning made it possible to shorten the DNA fragment of C. melassecola carrying the complete gltA gene to a DNA fragment of 3.5 kb delimited by two HindIII restriction sites.

EXAMPLE 2

Integron pCGL519

The integron preparation scheme is shown in Figure 1.

The aphlll gene was chosen as the selection gene; it confers resistance up to 600 μg / ml of kanamycin when it is integrated into a copy in the chromosome. We usually work at 25 μg / ml. The 3.5 kb HindIII fragment carrying the structural gene gltA coding for the citrate synthase of C. melassecola obtained in Example 1 was initially chosen as a DNA fragment homologous to the genome of the Corynebacteria and inserted into one unique sites of the multiple cloning site (the HindIII site). The restriction sites bordering the integron intended to be the most used in the integration strategy are the NotI sites which correspond to an 8 nucleotide sequence, and the BstXI sites. The enzyme BstXI recognizes the sequence (CCAN ₅ NTGG); therefore it cuts as frequently as a 6 nucleotide enzyme and we can expect to generate several fragments. However, the fragments released are reassociated according to the nature of the internal sequences of the different BstXI sites, which must ultimately lead to the only reconstitution of the starting fragment.

The pCGL519 piasmid (Figure 3) is an example of a versatile piasmid capable of generating an integron. We tested the integration in C. melassecola from its integrative cassette from a piasmid originating from E. coll. pCGL519 consists of two replicative and integrative fragments bordered by the NotI sites. The first fragment corresponding to the integron which comprises a multiple cloning site, the selection gene aph111 and a HindIII fragment homologous to the chromosome carrying the gltA gene which codes for citrate synthase. The second fragment includes the replicative part of the piasmid pBLI (SspI-HpaI fragment of 3 kb) which replicates in corynebacteria, the replicative part of the piasmid pACY184 which replicates in E. coli, ori p15A, and the origin of replication of the phage M13 complementable in trans. The plasmid pCGL519 was constructed in E. coli by insertion of a HindIII fragment of 3.5 kb containing the glt.A gene in a vector pCGL243.

As shown - in Figure 1, after cloning pCGL519 is used to transform B. lactofermentum. At the time of the transfer of pCGL519 into Brevibacterium lactofermentum CGL2002, the NotI fragment containing the origins of replication was substituted because the replicative part of pBLI had been inactivated in E. coli. This shows an additional advantage of the cassette structure.

EXAMPLE 3

Integration

The transfer of pCGL519 into Corynebacterium melassecola ATCC 17965 very restrictive vis-à-vis E. coli was carried out from the same piasmide extracted from B. lactofermentum. The proposed system makes it possible to isolate the piasmid pCGL519 from the strain of C. melassecola ATCC 17965 which is completely restrictive vis-à-vis E. coli but only partially vis-à-vis B. lactofermentum. Thus, an integrative cassette having the modification of the receptor strain was prepared.

After digestion of the pCGL519 piasmid from C. melassecola ATCC 17965 with the restriction endonuclease NotI (Boehringer), the integron containing the gltA gene and the AphIII selection gene was isolated and purified from an agarose gel low melting point. The integron thus purified was then subjected to self-circulation by iigation. The ligation mixture was then introduced into the strain of C. melassecola ATCC 17965 by electroporation (Bonamy et al., 1990). The clones of C. melassecola resistant to 25 μg / ml of kanamycin were subjected to the analysis. 500 transformants were obtained. 31 of the 50 analyzed did not have the pCGL519 piasmide and 20 of them were analyzed by southern biot after XbaI digestion. All correspond to the same integration event taking place by homologous recombination in the gltA region. Depending on the nature of the ligation product (circular mortomeric molecule or linear or circular polymeric molecule), the primary integrants can be interpreted as coming from a single or double "crossing-over". A pulsed field analysis after NotI digestion followed by a hyridation with a probe corresponding to the 3.5 kb HindIII fragment containing gltA was carried out. The integration of a single copy of the integron is confirmed by this analysis.

The model involves the duplication of the wild copy of the gltA gene: the enzymatic activity has been measured, it is multiplied by a factor of 1.82 in agreement with the interpretation of the blots and shows that the integrated copy is not inactivated . The results are collated in Table 1 with respect to the wild strain and to the strain transformed with a replicating plasmid. The stability of the integrated structure was measured as for the percentage of resistant Kanamycin cells after about thirty generation (Figure 4) and as for the enzymatic activity of the citrate synthase which remain stable.

The amplification of the integrated structure was carried out by growth selection on a dish containing an excess of kanamycin, selection at 800 μg / ml then i000 μg / ml then 1000 μg / ml of kanamycin and neomycin. Direct tandem amplification was obtained. Despite. the stability of the amplified structure and of the resistance to kanamycin, the high level of enzymatic activity obtained at the start in the case of citrate synthase was not maintained subsequently. It is possible that a specific inactivation of the gltA gene has occurred.

EXAMPLE 4

MiniMu-based construction

The Mu derivatives chosen in this example are MudII 1681 and

MudII 1681-Cat (respectively Km ^R and Cm ^R ) which have a size of 14.8 kb and 16.6 kb respectively (Figure 5). The miniMu MudII 1681-Cat is a derivative of the MudII 1681 transposon (Castilho et al., 1984). They have the elements necessary for transposition stated above (except HU) as well as the gene for the heat-sensitive repressor C (regulator of the expression of transposases A and B), a gene for resistance to antibiotics (respectively aphll and cat) and the genes lacA lacY and lacZ '. The latter begins at the 8th codon and makes it possible to detect translational fusions of proteins when Mud fits into a reading frame.

We transferred these transposons to the Corynebacteria. After integration of the transposons into the chromosome, a method, described in Example 8, made it possible to amplify the integrated copy. Associated with this amplification, the target gene for the integration event (the structural gene for glutamate dehydrogenase) has in many cases also been amplified with an increase up to a factor of 25 in the corresponding enzymatic activity.

EXAMPLE 5

Construction of vectors for the transfer of miniMu

The integrative vector (pCGL 107 - Figure 6) contains the gdhA gene (Gdh ') interrupted by the kanamycin resistance marker

Km (aphlll), the replicon (ori) of pUN121 (Nilsson et al., 1983) and the genes conferring resistance to tetracycline (Tet ^R ) and to ampicillin (Amp ^R ).

This vector is not replicative in Brevibacterium lactofermentum and integrates by simple "crossing-over" at the site of homology of the gdhA gene.

MudII 16S l-Cat was introduced into the integrative vector pCGL107 by minimuduction from E. coli OR 1836 in the strain

MC4100. Among the various insertions obtained, one of them was selected giving a lac- to E. coli phenotype (pCGL 107 :: Mud +, Figure 6). From this same piasmide, a disarmed Mud pCGL 107 was prepared: Mud-,

Figure 6) in which the genes of transposases A and B were deleted by HindIII digestion.

Other transfer strategies have been tested; MudII 1681 was introduced in Corynebacteria by placing the transposon on two other types of vectors:

Suicide vector (non-replicative, non-integrative) pEV11 :: Mud, it is a derivative of pUC 18 lacking d3S lac genes into which MudII 168 1 has been introduced. Shuttle vector (pCGL229) having the replicon of pBL1 (HindIII-HpaI fragment), the replicon p15A and the cat gene of Tn9.

MudII 1681 was introduced into the shuttle vector by minimuduction in E. coli Rec +. Among the various insertions obtained, one of them was selected giving a Lac- phenotype (pCGL229 :: Mud +). From this vector, a disarmed pCGL229 :: Mud- Mud was prepared in which the genes of transposases A and B were deleted by Pst1 digestion.

EXAMPLE 6

Efficiency of electrotransformation of constructions and their transfer in Brevibacterium lactofermentum

With the vectors described above, we transform an. strain of E. coli (DH5alpha) and two strains of Brevibacterium lactofermentum CGL2002 and CGL2005 (B115). These two strains are partially permissive with respect to DNA from E. coli. These experiments brought the results presented in Table 2. The following observations can be made:

In E. coli, whether in the case of the shuttle vector pCGL229 or in the case of the vector pCGL 107 (which can replicate therein), the transformation efficiency is markedly reduced when Mud + is present on the vector which does not is not the case when the transposition genes have been deleted. This phenomenon typical of replicative plasmids carrying active Mu derivatives can be attributed to a very strong expression of the Mu transposase in its natural host. This shows that in the constructions presented above, the Mud + used (MudII 1681 and

MudII 1681-Cat) are very active.

In none of the Corynebacterium strains tested it is not possible to obtain transformants with the shuttle vectors pCGL229 :: Mud + or - (nor with the suicide vector pEV11 :: Mud). In the case of the pCGL229 derivatives, it is possible that during minimuduction in E. coli Rec + the pBLI replicon was inactivated, explaining the inability of the vector to multiply in B. lactofermentum.

Transformants were obtained in B. lactofermentum CGL2005 (B115) with the integrative vector pCGL107 and its derivatives. The transformation efficiency obtained with pCGL107:; Mud- and pCGL107 :: Mud- is similar, indicating that the MiniMu MudII 1 681 -Cat A + B + does not transpose at high efficiency when introduced into

B. lactofermentum using an integrative vector.

The decrease observed (factor 20) in the transformation efficiency of the two derivatives of pCGL 107 compared to pCGL 107 itself is probably due to the increase in size of the transforming piasmide (10 kb in the case of pCGL 107, 26.7 kb in the case of pCGL 107 :: Mud + and 22.1 kb in the case of pCGL 1 07 :: Mud-).

EXAMPLE 7

Study of pCGL107 integration events:; Mud + in

B. lactofermentum CGL2005 (B115)

The transformation of the CGL20C 5 strain (B1 1 5) by pCGL 1 07 :: Mud- (hereinafter called pCGL320) made it possible to obtain 1,47 clones resistant to kanamycin (25 μg / ml) but no transformant by selecting for resistance to chloramphenicol (5 μg / ml). However, after replication on chloramphenicol, 103 of these clones are resistant to chloramphenicol. It therefore appears that the cat αe Tn9 gene does not express itself sufficiently to a single copy to allow primary selection in colonies; on the other hand, this expression is sufficient to determine a resistant phenotype by subsequent streak tests. All the clones obtained are of the lac- phenotype, which corresponds to the phenotype observed in E. coll.

Those types of integrants obtained (transformants of type

I. Km ^R Cm ^S and type 2. Km ^R Cm ^R ) may correspond respectively to the substitution of the gdh A gene by double crossing-over event and to the integration of the complete piasmid at the gdhA site by single event

"crossing-over" (Figure 7). The data in favor of this interpretation are as follows:

The dosage of glutamate dehydrogenase (according to the technique of Meers et al., 1970) in transformants of type 1 and 2 (Table 3): 5 out of 7 transformants of type 1 (for example K2 and K3, Table 3) have a zero gdh activity, which is in agreement with the substitution of the gdh.A gene by the interrupted gene, 4 of the 5 type 2 transformants have gdh activity similar to the control (KC2 and KC4- for example, Table 3). Molecular analyzes in Southern blot corresponding to the BamHI digestion of transformants type 1 (K2) and type 2 (KC2 and KC4) and revelation of the characteristic bands (visualized in Figure 7) by the plasmid probe pCGL107 (results not shown) . These indicate that the molecular structure of the gdh- transformants

(K2) conforms to a gene substitution. It also confirms that the gdh + transformants (type 2 events (KC2 and KC4) and some type 1 (K l) are obtained by integration of the piasmide at the gdhA site.

EXAMPLE 8

Selection of possible transpositions

The type 2 transformants (Km ^R and Cm ^R ) do not sufficiently express the chloramphenicol resistance gene to obtain the growth of isolated colonies in the presence of chloramphenicol 5 μg / ml. We therefore sought in these transformants Cm ^R subclones in the hope of selecting Mud transpositions (which carries the cat gene). These subclones were obtained at a frequency close to 1 per 10 cells. After replying to Xgal, these clones show a whole gradation of colors from white to blue (30% are clearly blue). This result is confirmed by the measurement of beta-galactosidase activities (Table 4). This could indicate a Mud transposition, enhancing resistance to chloramphenicol and showing beta-galactosidase activities by protein fusion at the insertion site. In fact, the same experiments carried out with pCGL107 :: Mud- have led to the same result indicating that these events are not due to transposition events. EXAMPLE 9

Demonstration of tandem amplification on the chromosome of the piasmid pCGL107;: Mud +

It turns out that almost all of these Lac + clones isolated previously (with the exception of KC3T4) also have amplified gdh activity (Table 4). In addition, with one exception (KC3T4), the beta-galactosidase (dosed according to Miller, 1972) and gdh activities are almost proportional. The same observation can be made regarding the determination of chloramphenicol acetyl transferase (according to the method of Shaw, 1975, Table 5). This result is contradictory with the transposition of Mud which never takes adjacent sequences when transposing. Rather, it indicates a tandem amplification on the chromosome of the pUN-Mud-gdh repeat unit which is bordered by sequence homologies. This point was confirmed by digestion of the genomic DNA of the KC3, KC3T1 and KC3T3 strains with BamHI, Notl and Xbal. Not only the BamHI band internal to Mud (and equal to 7 kb) is amplified but also the bands containing the ends of Mud (11 kb and 2.4 kb) which is consistent with a tandem amplification and what opposes to a transposition event. In addition, the digestions Notl (which cuts only once in MudII 168 1-Cat) and Xbal (which cuts only once in gdh ') clearly amplify the 24 kb band of the repeating unit.

This amplification seems relatively stable insofar as the beta-galactosidase activity remains at a level of 70% of the initial activity after 15 generations without selection pressure. The Cat and Gdh activities also show a 30% loss after 25 generations. This tandem amplification recalls the results of Albertini et al. (1985) and de 3annière et al. (1985) in Bacillus subtilis. The beta-galactosidase activity of the amplified clones is attributed to an amplified parasitic translation (outside the reading frame of the ampicillin resistance gene where the lac operon is fused) present in B. lactofermentum and undetectable in E. coll.

EXAMPLE 10

Study and characterization of the ISaBl insertion element

An insert was isolated by recovering plasmids in E. coli DH5aipha from the amplified DNA of KC3T4. The genomic DNA of the B. lactofermentum CGL2005 (B1I 5) strain, of a few derivatives and of other corynebacteria strains, was probed by a probe containing the insertion element previously isolated. First, the 3.5 kb PvuII fragment internal to Mu, derived from DNA amplified in KC3T4 and containing the insertion element, was used to probe the initial blot which contains the digests of DNA by BamHI various integrants and amplified strains (including KC3T4). As the insertion does not contain a BamHI site, this experiment makes it possible to reveal the BamHI genomic fragments containing at least one insertion or an insertion fragment. In the strain K l - (which corresponds to a substituted type 1 integrant obtained from pCGL 107: Mud-), five bands appear revealing the presence of several copies (whole or not) of the insertion. BamHI digestion of genomic DNA from

B. lactofermentum CGL2005 (BI15) showed four common bands with K l- (of respective size 18 kb, 5.9 kb, 5 kb and 4.5 kb); the fifth band present in the other strains K1-, KC 1- and KC3 (size 6.5 kb) can translate a transposition into a segregant of the strain CGL2005 (B115) at the origin of these strains.

Between two lines of different brevibacteria (i)

Brevibacterium lactofermentum CGL2005 (B1 15), and (ii) Brevibacterium lactofermentum CGL2002, two common bands appear in size 18 kb and 4.5 kb; on the other hand, the CGL2002 strain has no other bands. This reflects mobility of the element. The strains of Corynebacterium melassecola give weak hybridization signals with the insertion reflecting the existence of other different but related sequences in this species.

A first specific mobile insertion element of

B. lactofermentum has been identified and cloned; called ISaB1, it can be present in several copies (2 to 5 copies) in the genome. It is capable of transposing several times to different sites in an amplified region. Its fine restriction map is known. Different but related sequences exist in the genome of other corynebacteria.

ISaBi consists of 1288 base pairs; the ends can be identified because ISaBi is inserted into a fragment which corresponds to the terminal region (3 ') of the lac operon which has been sequenced by Hediger et al. (Biochemistry Proc. Natl. Acad. Sci. USA 82, 1985). ISaBi was inserted between nucleotide 5575 and 5576 by duplicating a target sequence of 5 bp (CCGAT) (Figure 8). The entire sequence of ISaBi is given in Figure 9 and the restriction map in Figure 10. Two open reading phases have been identified which could correspond, given the sequence analyzes, to the transposase structural gene and to the gene transposition repressor.

EXAMPLE 11

Vector trap for insertion elements and transposons

The ISaBi insertion was obtained during gene amplification studies in the chromosome of B. lactofermentum (B1 15). In order to isolate all the elements of insertion likely to be transposed in the corynebacteria, we built specific integrative vectors: pCGL330 and pCGL331 (Figures 11 and 12). These two vectors consist of a first fragment derived from the vector pUN 121. The piasmide pUN 121 is replicative in E. coli and carries resistance to ampicillin; it carries a sequence coding for the repressor cI of phage lambda which inhibits the expression of an operon fusion between the promoter P _L of phage lambda and a resistance gene of tetracycline. Insertion into the cl gene will thereby inactivate the repressor which will then allow expression of the tetracycline resistance gene which is expressed in corynebacteria under the control of P _L. The insertion events can thus be selected by resistance to tetracycline. We linearized pUN 121 at the Sspl site and fused to a second fragment obtained by digestion of the vector pCGL 107 with EcoRI filled with Klenovv to give, after transformation of E. coli DH5alpha, the vectors pCGL330 and pCGL331 which differ between them in the direction of cloning. The EcoRI fragment derived from pCGL 107 contains a selection gene for the direct transformation of coryneoacteria (the aphlll gene which confers resistance to kanamycin) and a fragment containing a homologous part of the gdhA gene (structural gene for ia glutamate dehydrogenase) which will serve as a point of integration into the chromosome of corynebacteria.

The B. lactofermentum CGL2005 (BL 15) and CGL2002 strains were electrotransformed for resistance to kanamycin by the plasmids pCGL330 and pCGL331. The transformation frequency was approximately 103 per μg in each of the cases, which is compatible with integration of the plasmids. The transformants (like

CGL2005 :: pCGL330 and CGL2005 :: pCGL331) are sensitive to tetracycline, which confirms that the regulation of P _L by cl works well in transformants. The frequency of tetracycline resistant segregants was measured after approximately 10 generations.

Mutations inactivating the C1 gene appear in strains with the integrated plasmids pCGL330 and pCGL331 at a frequency of 2 × 10 ⁶ per generation.

A piasmid recovery was carried out from genomic DNA extracted from resistant tetracycline segregants isolated from the CGL2005 :: pCGL331 and CGL2005 :: pCGL330 strains and digested with the enzyme Pstl. These digested DNAs were ligated and the ligation product was used to transform the DH5alpha strain of E. coli for resistance to tetracycline. The recovered plasmids were analyzed and in 7 cases out of 9 tetracycline-resistant clones, an insert was identified. In one case (from CGL2005 :: pCGL331), an insertion element identical to ISaBl was identified (1.2 kb in size, having the unique AccI sites,

EcoRV and XhoI) with internal localization to cI. In another case (from CGL2005 :: pCGL330), an insertion element different from ISaBl (size of 1.0 kb, having an AccI site but no EcoRV or Xhol site) was identified.

The transposon trap vector works; in the majority of cases, the mutation obtained is an insertion; the frequencies of resistance to tetracycline therefore fairly correctly measure the transposition frequencies; the most mobile elements have therefore been identified; of these, ISaBl has been re-isolated and another element of insertion different from ISaBl has been identified.

The strains cited have the following origins:

Escherichia coli

. DH5alpha: Gibco BRL

. MC4100: Casadaban (1976)

. OR 1836: Reyes

Brevibacterium lactofermentum

. CGL2002: Bonamy et al. ( 1990)

. CGL2005 (BU 5): Bonnassie et al. (1990)

Corynebacterium melassecola

. ATCC17965: ORSAN

. ATCC 17965 :: gltA: (this request)

Four of these strains were deposited in the National Collection of Cultures of Microorganisms (CNCM) of the Institut Pasteur (Paris) on July 23, 1991:

- Corynebacterium melassecola ATCC 17965 :: gltA: n ° 1- 1124

- Escherichia coli OR 1836: n ° 1-1125

- Brevibacterium lactofermentum CGL2005 (B115): n ° I- 1126

- Brevibacterium lactofermentum CGL2002: n ° I-1127.

The DH5alpha strainer is available in the catalog of

Clontech Laboratories, n ° C 1021-1, (Palo Alto, CA. USA) and the strain MC4100 at ATCC under n ° 35695.

Table 2: Transformation efficiency of vectors carrying MiniMu

Bacterial strains

E.coli DH5α B .lactofermentum B .lactofermentu

CGL2002 CGL2005 (B115) pCGL229 5 × 10 ⁷ 10 ⁵ 10 ⁶

pCGL229 :: Mud + 2 × 10 ⁴ 0 nd

pCGL229 :: Mud- 4 × 10 ⁶ 0 nd

pCGL107 10 ² 3 × 10 ³ 10 ⁴ pCGL107 :: Mud + 5 × 10 ² 0 4 × 10 ² pCGL107 :: Mud- 10 ⁵ nd 5 × 10 ²

TABLE 3: DETERMINATION OF GLUTAMATE DEHYDROGENASE

IN PRIMARY TRANSFORMANTS Specific activity Gdh in μmole of

A bacterial strain

NADPH ₂ consumed / min / mg protein

starting strain Km ^S Cm ^S

CGL2005 (B115) 2.18 transforming Km ^S Cm ^S

K1 2.0

K2 10.06

K3 10.06

K4 10.06

K5 10.06

K6 2.18

K7 10.06 transforming Km ^S Cm ^S

KC2 2.24

KC3 2.12

KC4 2.18

KC5 3.45

KC7 2.06 TABLE 4: DETERMINATION OF BGALACTOSIDASE ACTIVITIES AND

GLUTAMATE DEHYDROGENASE IN AMPLIFIED INTEGRANTS

Bacterial strain Activity Specific activity ßgalactosidase glutamate dehydroge

starting strain Km ^s Cm ^s

CGL2005 (B115) 3.7 2.48 primary integrants Km ^R Cm ^R

KC2 6.2 2.24

KC3 4.5 2.06

amplified integrants

KC3T1 27.9 18.2

KC3T2 20.7 16.0

KC3T3 48.2 49.6

KC3T4 32.2 2.24

KC3T5 19.7 8.55

KC3T6 19.0 11.9

KC3T7 34.5 28.0

KC2T1 7.4 2.73 TABLE 5: DETERMINATION OF CAT, BGA1 and GDH ACTIVITIES

IN THE AMPLIFIED STRAINS

Bacterial strain ßgal activity Activity Activity

specific Cat specific Gdh

starting strain Km ^s cm ^s

CGL2005 (BH5) 3.7 ≤0.07 2.18 integrating primary Km ^R Cm ^s

K2 10.07 10.006

K3 10.07 10.006 integrating primary Km ^R Cm ^R

KC3 4.5 2.72 2.06

KC7 5.4 1.82 amplified integrants

KC3T1 27.9 19.7 18.2

KC3T3 48.2 55.5 49.6

KC3T4 32.2 18.4 2.24

KC3T5 19.7 13.6 8.55

REFERENCES

Albertini A. M. and Galizzi A. (1985). Amplification of a chromosomal region in Bacillus subtilis; 3. Bacteriol., 162: 1203-1211)

Ausubel MA, Brent R., Kingston RE, Moore DD, Seidman JG, Smith JA and Struhl K. (1987) in "Current protocols in Molecular Biology", Greene Publishing Associates and Wiley - Interscience Birnboim HC and Doly 3., (1979 ) A rapid alkaline extraction procedure for recombinant screening piasmid DNA. Nucleic Acid Res. 7: 1513-1523

Bonamy. C, Guyonvarch A., Reyes O., David F. and Leblon G. (1990) Interspecies eiectro-transformation in Corynebacteria. FEMS Microbiology Letters 66: 263-270

Bonnassie S. Oreglia 3. Trautwetter A. and Sicard A. M. (1990) Isolation and characterization of a restriction and modification deficient mutant of Brevibacterium lactofermentum, FEMS Microbiol. Letters, vol. 72: 143-146

Casadaban M. J. Transposition and fusion of the lac genes to selected promoters in E. coli using bacteriophages lambda and Mu. 3. Mol. Bioi: 104 (1976) 541-555 Castilho B.A., Olfson P. and Casabadan M.3. (1984) Plasmid insertion mutagenesis and lac gene fusion with miniMu bacteriophage transposons, 3. Bacteriol. 158: 488-495

Dower W. 3., Miller 3.F. and Ragsdale C.W., High Efficiency transformation of E. coli by high voltage electroporation. Nucleic Acid Res. 16 (1988) 6127-6145

Jannière L., Niaudet B., Pierre E., Ehrlich SD (1985) Stable gene amplification in the chromosome of Bacillus subtilis, Gene 40: 47-55 Maniatis T., Fritsch Ed. And Sambrook 3. (1982). Molecular cloning: A laboratory manual. Cold Spring Harbor Laborator \ Press, Cold Spring Harbor, NY Meers 3.L. Tempest DW and Brown CM. Glutamine (amide): 2-Oxoglutaraτe Amino Transferase Oxido-reductase (NADP), an Enzyme Involved in the Synthesis of Glutamate by Some Bacteria 3. General Microbiology 64 (1970) 187-194

Miller 3. (1972) Experiments in Molecular Genetics (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY) p352-355

Nilsson B., Uhlén M., 3osephson S .. Gatenbeck S. and Pnilipson L., 1983. An improved positive selection plasmid vector constructed by oligonucleotide mediated mutagenesis. Nuc. Acid. Res. 11, 8019-8030

Shaw WV (1975) Chloramphenicol acety l transferase from chloramphenicol resistant bacteria. Metho. in Enz. 43: 737-755

Claims

1) Corynebacterium integron characterized in that it comprises:

- a gene ensuring efficient selection in said corynebacterium, - a sequence homologous to the genome of said corynebacterium, said sequences having been adapted to said bacterium.

2) Integron according to claim 1, characterized in that it comes from a piasmid which comprises, in addition to the integron, a replicative part, the integron being flanked by reverse repeat sequences corresponding to a restriction site not present in the 'integron.

3) Integron according to claim 2, characterized in that the replicative part comprises an origin of effective replication in a non-Coryneform bacterium.

4) Integron according to claim 3, characterized in that the origin of replication is effective in E. coli.

5) Integron according to one of claims 2 to k, characterized in that the replicative part comprises an origin of effective replication in Coryneform bacteria.

6) Integron according to claim 1, characterized in that it further comprises the sequences of a transposable element.

7) Integron according to one of claims 1 to 6, characterized in that the transposable element comprises sequences originating from a corynebacterium.

8) Integron according to claim 7, characterized in that the sequence comes from Brevibacterium.

9) Integron according to claim 8, characterized in that the sequence comes from ISaB1 as described in Figure 9.

10) Integron according to one of claims 6 to 9, characterized in that it further comprises the sequences of a transposable element ensuring transposition with the exception of the proteins coded by the coryneform bacterium.

11) Integron according to one of claims 6 to 10, characterized in that the sequences of the transposable element are devoid of the sequences coding for the transposases.

12) Integron according to one of claims 1 to 11, characterized in that it comprises a sequence of interest. 13) Integron according to claim 12, characterized in that the sequence of interest codes for a peptide or a protein of interest.

14) Integron according to claim 13, characterized in that the protein of interest is a homologous protein.

15) Integron according to claim 14, characterized in that the protein of interest is a heterologous protein.

16) Integron according to one of claims 13 to 15, characterized in that the protein of interest is an enzyme.

17) Integron according to claim 16, characterized in that the sequence coding for a protein of interest is chosen from the genes: - gltA

- gdhA.

18) Integron according to one of claims 2 to 17, characterized in that the restriction site not present in the integron is chosen from:

- NotI

- BstXI.

19) Integron according to one of claims 1 to 15, characterized in that said sequences have been adapted for a strain of Corynebacterium.

20) Integron according to rev endication 19, characterized in that the said sequences were transferred to a strain of Brevibacterium before being adapted in Corynebacτerium.

21) Integron according to claim 20, characterized in that said sequences have been amplified.

22) Amplified integron according to claim 21, characterized in that the amplification gave stable amplified sequences.

23) Integron according to claim i. characterized in that it contains a homologous sequence of Corynebacterium melassecola which integrates into the chromosome of Brevibacterium lactofermentum.

24) Process for transforming a strain of coryneform bacteria with an integron according to one of claims 1 to 23, characterized in that the integron is introduced into said strain by electroporation. 25) Corynebacterium capable of being obtained by the process according to claim 24.

26) Corynebacterium according to claim 25, characterized in that it is a strain of:

- B. lactofermentum

- B. flavum

- C. glutamicum

- C. melassecola.