WO2006106193A1 - Method for obtaining soluble or better expressed variants of a protein of interest - Google Patents

Abstract

The invention concerns a method for selecting soluble or better expressed variants of a protein of interest of a peptide of interest, from a bank of mutants of the protein or of the peptide of interest fused to the gene of resistance to kanamycin.

Description

 METHOD OF OBTAINING SOLUBLE OR BETTER EXPRESS VARIANTS OF A PROTEIN OF INTEREST

Introduction The present invention relates to molecular biology and more particularly to the evolution of proteins in vitro. The subject of the invention is a method for selecting, from a mutant library of a protein generated in a first step, variants with improved solubility or higher level of expression.

The insolubility of recombinant proteins is a major problem in the field of biosciences: when expressed in bacteria, a large part of the proteins can not be expressed at a significant level without aggregating as inclusion bodies. They can not be purified and will be inactive in most cases.

When expressing a eukaryotic protein (especially a human protein) in a prokaryotic host, the various processing steps (glycosylation, peptide cleavage, etc.) do not take place. The protein therefore has a general structure quite far from its natural structure, and most often has a reduced solubility. In the case of bacterial recombinant proteins produced in bacteria, this phenomenon of insolubility is observed to a lesser extent. However, if it is desired to express a large quantity of the protein, a fraction of the proteins aggregates: there is a limiting concentration, a function of the solubility of the protein, which can not be exceeded without forming such bodies. 'inclusion. The nature of the aggregated proteins is sometimes reversible: if the inclusion bodies are treated in such a way as to dissolve them, either by using denaturants, or by modifying the pH or the temperature, then replacing the solution under conditions more physiological, we can find proteins in their native and soluble form, that is to say having recovered their functionalities. Nevertheless, most often only a very small fraction of the proteins can be recovered.

This limit to the expression of proteins represents an obstacle when it is desired to express a protein in large quantities: this is the case in particular with enzymes industrial, - used in the textile, food and chemical industries in particular - or that of fragments of recombinant antibodies to a single chain, the ScFv. This last class of molecules, which are very poorly expressed in bacteria for these reasons of solubility limit, can be used in human therapy and must then be produced in significant amounts.

This limitation on the expression of the insolubility of recombinant proteins is also a brake on the expression of proteins used in small quantities for research purposes. For example, the therapeutic targets must, in order to be characterized, be produced in sufficient quantity to allow structural studies to be carried out. These studies can not be carried out on insoluble proteins because the formation of the crystals intended for these analyzes is done in a liquid medium. The insolubility of proteins is therefore a phenomenon that limits progress in the characterization of many proteins.

Important efforts to improve the production of recombinant proteins have been conducted in recent years.

The general approach is to look for variants of the protein having acquired better solubility, while having retained its other functionalities (its activity for an enzyme, its ability to bind an antigen for a recombinant antibody). In a first step, mutants are generated from the gene encoding the starting protein. In a second step, these mutants are screened to identify those with the best level of solubility or expression. In a third step, the mutants having an improved level of solubility or expression are tested for their functionalities, in order to verify that the mutation (s) introduced have not affected them.

Different mutagenesis techniques exist; they make it possible to artificially modify the nucleotide sequence of a DNA fragment. They can be divided into four major groups: random mutagenesis, gene recombination, site-directed mutagenesis, and massive mutagenesis ™ (Massive Mutagenesis®).

In the random mutagenesis technique, the targeted DNA (containing the gene to be mutated) is amplified under particular conditions altering the ability of the polymerase to faithfully replicate the DNA. This introduces mutations over cycles, that is to say differences with respect to the initial sequence. At the end of the reaction, a large number of copies of the initial molecule are obtained, each of these molecules having different mutations. These molecules are present in the form of a bank, that is to say a mixture of molecules of different nature (differing in the nature and position of their mutations). The DNA is then inserted into an expression vector.

DNA recombination mutagenesis is inspired by the recombination process at work in Darwinian evolution. Mutagenesis by DNA blending involves recombining partially homologous sequences, isolated from different organisms. For example, if we are working on an enzyme, the first step of a DNA mixing approach will be to isolate a large number of genes homologous to this enzyme, either from collections of strains or from genes. directly isolated from natural samples (by an approach now described under the term "metagenome"). Different approaches then exist for mixing the domains of these homologous genes and generating a library of "chimeric" DNA molecules, that is to say made up of several domains having different provenances (US Patent 6,132,970; Stemmer WP. et al., Nature, vol.370, p.389-391, 1994, Aguinaldo AM et al., Methods Mol Biol., vol.192, p.235-239, 2002, Zhao H. et al, Nat Biotechnol., vol.16, p.258-261, 1998, Shao Z. et al, Nucleic Acids Res., vol.26, p.681-683, 1998, Kawarasaki Y. et al, Nucleic Acids Res., vol.31, 2003; U.S. Patent 5,965,408; Patent WO 00/09679; Patent WO 02 48 351). It is expected that these molecules thus contain novel characteristics, and in particular cumulative properties of two or more parental genes. This recombinant DNA approach is based on the general idea that the novel combination of natural mutations, having thus been pre-screened by nature for the maintenance of enzyme activity, is more likely to 'to be a carrier of improvement than the introduction of mutations generated at random. At the same time that the generation of diversity is restricted to a 'reasonable' sequence space because 'preselected', it is however limited by the original sequences, which must be physically possessed, by the necessity of using genes having a sufficient level of homology, and the impossibility of generating sequences other than combinations of the initial sequences. These DNA mixing approaches have been found to be particularly effective in the field of enzyme activity enhancement or stability.

The purpose of site-directed mutagenesis is to introduce one or a few mutations (substitutions, but also deletions or insertions) of known nature and position into a DNA fragment. An oligonucleotide is used to introduce this mutation. This oligonucleotide is conventionally composed of about thirty bases, and is homologous in every respect to the sequence targeted on the DNA fragment, with the exception of one or a few localized positions in its middle part. This oligonucleotide is then used to initiate a replication reaction using the DNA fragment as a template. In many protocols, a second oligonucleotide, in the opposite sense, is used to improve the efficiency of replication. The DNA synthesized during this reaction is then selected, in particular according to the criterion of its content in methylated bases, so as to eliminate the non-mutant parental DNA. Different techniques of site-directed mutagenesis make it possible to obtain mutants quite rapidly, at a rate that remains limited, however: only a few mutants can be generated per day and per person. This low flow rate is only partially adapted to the problems of molecular evolution: it is indeed rare to be able to accurately predict the effect of introduced mutations and, when an improved mutant is sought on this or that criterion, it is generally better to generate a large number, most of which will not have the desired effect.

Massive Mutagenesis® allows to realize in a very short time a large number of mutations directed on the same target DNA fragment. The template containing the target DNA is prepared, and mixed with the mutant oligonucleotides initially manufactured. The matrix is denatured by heat so as to temporarily dispose of single-stranded DNA. When returning to a lower temperature, some or all of the oligonucleotides present in the mixture are fixed on the matrix at their site of homology. All the elements necessary to carry out a replication of the matrix from the mutant oligonucleotides are added, and in particular a polymerase, the buffer allowing its activity, nucleotide triphosphates in sufficient quantity, any necessary cofactors. The replication reaction then takes place under the temperature conditions corresponding to the maximum activity of the polymerase. The goal is to have mutants each containing a mutation or a combination of several different mutations. The average number of mutations per molecule can be modulated at will. The possible diversity increases very rapidly with the number of mutations per molecule: the use of Massive Mutagenesis® makes it possible to obtain very wide diversities. These various technologies make it possible to generate diversity. Depending on the specific context, one or the other may be preferred.

If one seeks mutants having an improvement in solubility or expression with respect to the original molecule, it is necessary to have a technology which makes it possible to rapidly measure this characteristic in each of the mutants generated. Of course, the measurements could be carried out one by one by observing in each of the mutants the level of soluble proteins. This approach would however be very laborious, and that is why techniques have been developed to allow this parameter to be tested in parallel, directly from the mutant libraries.

These techniques are for the most part based on the fusion, in phase, of a mutant library of the protein of interest on the one hand, and on the other hand a reporter gene. These fusion proteins are then screened on the basis of the activity of the reporter gene.

Fusion proteins have been used for many years in contexts other than the search for soluble variants: mainly, these fusion proteins are used to localize proteins at the cellular or intracellular level. One of the first fusion proteins to be used is β-galactosidase, whose activity converts X-gal, an unnatural sugar, into a blue compound. This activity can therefore easily be traced. (Casadaban MJ et al., Methods Enzymol., VoL100, p.293-308, 1983)

CAT (chloramphenicol acetyl transferase) has also been used in this context for fusion proteins for many years. Chloramphenicol is an antibiotic that binds to the ribosomes of the bacteria, blocking the biosynthesis of proteins, resulting in rapid cell death. CAT is an enzyme that transfers an acetyl group from coenzyme A to chloramphenicol (which is then inactivated). From this difference of an acetyl group, colorimetric tests have been developed: the CAT activity is then traced by a yellow signal (Robben J. et al, Gene, vol. 126, p.109-113, 1993).

These approaches based on the use of enzymatic reporter proteins present a significant obstacle: the substrate must be able to access the compartment in which the reporter enzyme is located. In many cases, the labeling using these enzymes is therefore not possible. In addition, the colored product labeled with the enzyme is a small chemical molecule, which has a natural tendency to diffusion: the intracellular localization of the fusion proteins is therefore imprecise.

It is for these reasons that naturally fluorescent fusion proteins have been developed. The most commonly used reporter is today Green Fluorescent Protein (GFP) (Garamszegi N. et al, Biotechniques, vol.23, 1997). GFP, isolated in 1962 and cloned in 1992, is composed of 238 amino acids and is derived from the jellyfish Aequorea Victoria. It has the characteristic to have a chromophore allowing it to absorb the blue radiations (400nm) and to emit in the green (509nm). There are now many variants of GFP, which fluoresce at other wavelengths and thus emit different colors (yellow, blue, purple). Several fusion proteins can therefore be observed simultaneously. A second naturally fluorescent protein was isolated and cloned from the sea anemone: DsRed. This protein naturally emits a red signal. The combined use of this protein and GFP again makes it possible to trace several proteins at the same time.

Modified GFPs have been made to allow optimal expression in many organisms such as bacteria, yeasts, or mammalian cells, to enable studies in each of the commonly used systems in biology.

When making such fusion proteins, the binding between the gene of interest and that encoding the reporter protein must be done without interrupting the reading phase. It is also necessary to take care of the steric constraints of both proteins. Indeed, it is possible that the fusion of the protein of interest and the reporter protein does not allow both to adopt their natural conformation. Their structure would be modified, and their activity canceled. In order to avoid this possible pitfall, it may be necessary to introduce a spacer between the gene coding for the protein of interest and that coding for the reporter protein. This spacer is a DNA sequence coding for a peptide with the most neutral structure possible, and not exhibiting any particular activity in itself. It is usually composed of 8 or 9 neutral amino acids that should not affect the protein structure and should not be targeted by proteases. It is also preferable that it be flexible and hydrophilic (Sieber V., Methods Mol Biol., Vol.230, ρ.45-55, 2003, Waldo GS et al, U.S. Patent 6,448,087, 1999). This DNA fragment must of course be introduced in such a way that the phase is conserved on the set consisting of the protein of interest, the spacer, and the reporter protein.

The relative position of the protein of interest and the reporter protein is not neutral. In some cases, it is preferable that the fusion protein is positioned in

C-terminal. In some others, N-terminal positioning is preferable.

Frequently, the skilled person performs both constructions, and retains the one that has the best signal after a preliminary experimental step.

The main use of fusion proteins is therefore associated with protein localization work. However, these fusion proteins have also been used as a selection system for identifying more soluble variants of a protein: it has indeed been observed that when a protein of interest is fused to a reporter protein, the conformation of the one and the other are related. Incorrect conformation of the protein on the NH2 side is thus correlated with an equally incorrect conformation of the protein located at the COOH end. The opposite is generally not true: it is the conformation of the NH2 end of the protein which influences the whole. As the conformation of the protein is set up concomitantly with its synthesis, and this takes place from the NH 2 end to the COOH end, it can be understood that the overall conformation of the protein depends on the conformation acquired by the first segment synthesized. These conformational effects during synthesis are sometimes described as "nucleation".

This correlation between the conformation of the protein of interest and that of the protein being fused to it could thus be used to set up systems for selecting more soluble variants. Of course, the most traditional fusion proteins have been used: CAT, β-galactosidase, and GFP.

The team of Maxwell et al. (Protein science, vol. 8, p.1908-1911, 1999), was able to express a fusion protein comprising an insoluble protein with CAT: the catalytic domain of an insoluble HFV integrase was fused to the N-terminal domain of the enzyme. On the other hand, the same construct containing this time a mutant of the more soluble protease was made. A significant increase in resistance to the antibiotic chloramphenicol is observed in the mutated form. This technique would therefore make it possible to select a soluble protein in the middle of a set of insoluble proteins. Few publications describing work based on this approach have been made to date.

In EP 1479694, there is described a method for isolating a soluble scFv fragment. The first step is to create a fragment mutant library. Then, there is construction of two types of fusion proteins. The genes of these variants are, in the first construct, fused with a part of the gene necessary for the expression of β-galactosidase. The second construct consists in fusing the gene coding for the corresponding antigen with the last part of the β-galactosidase gene. Thus, under the selection conditions allowing the expression of the mutant antigen and ScFv, only the most soluble forms of ScFv will interact and allow the functioning of β-galactosidase. Differentiation between two different solubility clones here rests on the fact that the specificity of the scFv fragment is proportional to its solubility. The more soluble the fragment, the higher the recognition of the antigen and therefore the more β-Galactosidase will work. This technique, based on the complementation of two domains of β-galactosidase, is limited to use in the context of an antibody-antigen binding. The need to co-transform two plasmids presents an additional complication compared to systems based on a single plasmid. In addition, this system selects both the more soluble ScFvs and those with a better affinity for the antigen, which can also introduce an undesirable complexity.

The most used system today for selecting more soluble proteins is based on the use of GFP.

In US6448087 (Waldo G.S. et al), the authors describe a technique in which GFP is fused with a mutant library of a protein of interest. Bacteria expressing the genes corresponding to more soluble proteins emit the strongest fluorescence, and can easily be selected using a cell sorter.

This approach has allowed several teams to obtain positive results, that is, to identify mutants with improved solubility. However, this technique is limited by the fact that this is not a system based on the survival of the positive clones, but on a variation of a signal which will have to be measured using a complex apparatus.

In patent application 20040170976 (Scott LA et al), there is described a system in which the expression of the protein of interest in insoluble form causes a change in expression of the reporter gene, in the absence of fusion protein .

To do this, a promoter described as specifically activatable is used if soluble proteins are present in the bacterium. The reporter gene placed under the control of this particular promoter may for example be chosen to allow observation of a fluorescence signal. If the reporter gene encodes an enzyme, it will be necessary to quantify the product of enzymatic modifications of a substrate compound. The level of protease proteins produced will be directly proportional to the level of soluble proteins of interest in the cytoplasm.

US Pat. No. 6,727,070 (Thomas PJ et al) describes a complementation system between two fragments of the same reporter protein: a fusion protein combining the protein of interest with the first fragment of a reporter protein which, by itself, same, is not active, is used. The second part of the marker is expressed within the host cell from the chromosome. There is manifestation of the expression of the reporter protein in the case where the two parts can complement each other. This reporter protein can be an enzyme, a protein inhibitor, a fluorophore or a chromophore. Better solubility or increased level of expression can be detected using the necessary apparatus.

Thus, the observation of fluorescence, the expression of an enzyme catalyzing a reaction, or simple cell survival, can make it possible to segregate between the types of soluble and insoluble variants of a library of mutants.

The survival-based systems have the following superiority over those based on fluorescence or colorimetry: they do not require the measurement of a signal. They thus allow the direct selection of the desired mutants.

The only simple technique for selecting solubility or expression mutants to be based on survival is based in the prior art on the use of the chloramphenicol resistance. This approach has only appeared in very few documents, suggesting low efficiency.

It is therefore useful to develop a simple technique for the sensitive and reliable selection of the solubility or expression mutants present in a bank.

Description of the invention

The present invention is a technique for obtaining mutants of a protein or peptide of interest having improved solubility or associated with a higher level of expression. The method according to the invention is characterized by generating a mutant library of the gene encoding the protein or peptide of interest fused in phase with the kanamycin resistance gene, and by using it as a selective screen for a solid or liquid medium containing a stringy kanamycin concentration. Kanamycin is an inexpensive antibiotic and commonly used in research laboratories. It is a potent bactericide, acting independently of bacterial density, and having a very intense and very rapid effect in vitro.

Many protein families can confer resistance to the antibiotic kanamycin by various mechanisms. Among them, three families of proteins act by directly modifying the antibiotic, thus destroying its anti-bacterial activity:

Kanamycin nucleotidyltransferases by O-nucleotidylation,

Kanamycin phosphotransferases by O-phosphorylation, and kanamycin acetyltransferases by N-acetylation.

The family most commonly encountered and used is the first family, that of kanamycin nucleotidyltransferases (Kat). These enzymes have very different structures and all confer resistance to kanamycin. The inventors have established that they are not all usable in the method of the present invention. Indeed, they do not all work when they are fused with a 5 'protein, even soluble. In particular, kanamycin nucleotidyltransferase (Kat) functions very poorly in the form of fusion whereas kanamycin phosphotransferase works very well. Therefore, the kanamycin resistance gene used in the present invention encodes kanamycin phosphotransferase. Preferably, kanamycin phosphotransferase is kanamycin kinase II, a variant or functional fragment thereof. Preferably, the variant is an improved kanamycin kinase, having for example greater resistance to the antibiotic.

By "kanamycin kinase II" is meant the Klebsiella pneumoniae protein whose enzyme code is EC 2.7.1.95, the Genbank access number is U32991, and the SwissProt access number is P00552. Kanamycin kinase II is also called aminoglycoside-3 '-phospho-transferase-IIa, neomycin-kanamycin phosphotransferase II, APH (3') II, nptII or APH (3 ') - II. Preferably, the protein has the sequence described in SEQ ID No. 14 and an exemplary nucleotide sequence coding for it is described in SEQ ID No. 13.

By "functional fragment" is meant a fragment capable of ensuring resistance to kanamycin. In particular, since the resistance to kanamycin is due to the ability of the enzyme to transfer a phosphate group to kanamycin, a functional fragment is capable of phosphorylating kanamycin. Preferably, the fragment comprises the catalytic site of the enzyme at position 190. Preferably, the fragment comprises at least 100 consecutive amino acids. In particular, a preferred fragment may comprise residues 23-257 (APH-phosphotransferase domain). In a preferred embodiment, kanamycin kinase II may comprise a deletion of methionine at position 1. This kanamycin kinase II is described in SEQ ID Nos. 1 and 2.

By "variant" of kanamycin kinase II is meant a polypeptide capable of ensuring resistance to kanamycin, in particular capable of transferring a phosphate group to kanamycin, and which has at least 90% identity with kanamycin kinase II. Klebsiella pneumoniae or a sequence of 100, 150 or 200 consecutive amino acids thereof. Preferably, the variant has at least 95%, 96%, 97%, 98% or 99% thereof or a sequence of 100, 150 or 200 consecutive amino acids thereof. The percentages of identity referred to here are established, as indicated above, using the BlastX software, using the default settings.

In a particular embodiment of the present invention, the kanamycin kinase II used is an improved variant thereof to obtain resistance to kanamycin concentrations higher than the wild-type protein. More particularly, the kanamycin kinase II used is an improved kanamycin kinase II, variant or functional fragment thereof comprising at least one substitution selected from the group consisting of Q155W and Sl14H.

By "Q155W" it is understood that glutamine at position 155 in the sequence described SEQ ID No. 14, or in corresponding position in another sequence, is substituted with tryptophan. By "S114H" is meant that the serine at position 114 in the sequence described SEQ ID No. 14, or in a corresponding position in another sequence, is substituted with a histidine.

In a first embodiment of the present invention, improved kanamycin kinase II, a variant or a functional fragment thereof comprises a Q 155 substitution. An example of kanamycin kinase II comprising a Q155W substitution is described in SEQ ID No 16. The sequence may further comprise another substitution, preferably S114H.

In a second embodiment of the present invention, enhanced kanamycin kinase II, a variant, or a functional fragment thereof comprises an S114H substitution. An example of kanamycin kinase II comprising an S114H substitution is described in SEQ ID No. 18. The sequence may further comprise another substitution, preferably Ql 55 W.

Variants expressed in a soluble form allow the bacterium to survive in the presence of kanamycin. Surprisingly, unlike the CAT fusion system, this direct survival selection is remarkable in that a very clear discrimination is obtained between the different variants of the protein, which allows the direct selection of the clones expressing the soluble variants. or better expressed of the protein or peptide of interest.

Thus, the present invention relates to a process for obtaining soluble or better expressed variants of a protein or peptide of interest, characterized by the generation of a mutant library of the gene coding for said protein or said peptide. interest fused to the kanamycin resistance gene in an expression vector, and by selecting mutant genes corresponding to more soluble or better expressed proteins or peptides by culturing the bacteria transformed by said expression vector library into a culture medium containing kanamycin. Preferably, the kanamycin resistance gene is a polynucleotide encoding kanamycin phosphotransferase, preferably kanamycin kinase II, a variant or a functional fragment thereof. In a preferred embodiment, the kanamycin resistance gene is a polynucleotide encoding an improved kanamycin kinase II comprising at least one selected from the group consisting of Q 155 W and S114H. Preferably, the gene codes for a protein of interest. The polypeptide of interest may be for example a cytokine, an enzyme, or any other polypeptide / protein of industrial or therapeutic interest. In particular, the expression vector is a prokaryotic expression vector. Optionally, this may include elements that also allow expression in a eukaryotic host. In a particular embodiment, the method comprises: i) generating mutants of a gene encoding the protein or peptide of interest; ii) Preparation of a mutant library by cloning the mutants obtained in i) in phase with the kanamycin resistance gene in an expression vector; iii) transformation of competent bacteria by the mutant bank obtained in ii); and iv) selecting, using a kanamycin-containing culture medium, bacteria transformed by a vector comprising a mutant corresponding to a protein or peptide of mutant interest that is more soluble or better expressed than the starting protein of interest.

Preferably, the mutants are generated in step i) by random mutagenesis. In a also preferred alternative, the mutants are generated in step i) by recombination of genes.

In another embodiment, the method comprises: i) producing an expression vector containing the gene encoding the fusion protein or peptide of interest in phase with the kanamycin resistance gene; ii) generating a mutant library of the gene coding for the protein or peptide of interest on the expression vector obtained in i); iii) transformation of competent bacteria by the mutant bank obtained in ii); iv) Selection, using a culture medium containing kanamycin, of bacteria transformed by a vector comprising a mutant corresponding to a mutant protein or peptide of interest that is more soluble or better expressed than the starting protein of interest.

Preferably, the mutants are generated in step ii) by massive mutagenesis. Optionally, the gene coding for the protein or peptide of interest and the kanamycin resistance gene are separated by a spacer that does not change the reading phase.

Preferably, the bacteria are transformed by electroporation or by thermal shock. The culture medium containing kanamycin may be solid or liquid. Preferably, the culture medium contains a quantity of stringy kanamycin. Preferably, the expression vector is a plasmid.

The present invention further relates to the use of an expression vector encoding a fusion protein between a mutant of a protein or peptide of interest and the kanamycin resistance gene for selecting a variant of this protein or peptide having improved solubility or increased expression. Preferably, the kanamycin resistance gene is a polynucleotide encoding a kanamycin phosphotransferase, preferably kanamycin kinase II, a variant or a functional fragment thereof. In a preferred embodiment, the kanamycin resistance gene is a polynucleotide encoding an improved kanamycin kinase II comprising at least one substitution selected from the group consisting of Q155W and Sl14H.

The method of the invention is a process for obtaining or selecting mutants of a protein or peptide of interest that are more soluble or better expressed in bacteria, characterized in that it contains the following steps in In this order or in a different order: i) A plasmid containing the gene of interest fused to the kanamycin resistance gene is constructed. The whole is placed under the control of a bacterial promoter. ii) Mutations are introduced at the level of the gene of interest. iii) competent bacteria are transformed with these constructs, iv) The mutants having acquired an increased solubility or level of expression are selected using a stringant kanamycin concentration. In particular, steps i) and ii) can be reversed.

In a particular embodiment, it is possible to carry out an additional step iii) bis before step iv) in order to preselect the bacteria having integrated a plasmid. Resistance to an antibiotic other than kanamycin, such as ampicillin, is then used. In this embodiment, the vector used must therefore contain, in addition to the kanamycin resistance gene fused to mutant genes of interest, an expression cassette containing the antibiotic resistance gene retained (ampicillin by example), and the promoter by allowing the expression.

In step i), the kanamycin resistance gene can be fused to the N-terminal or C-terminal protein (FIG. 1) of the protein of interest. In a preferred embodiment, this resistance gene is fused to the C-terminal of the protein of interest, the general conformation of the protein being mainly governed by its NH2 end. The coding sequence for the kanamycin resistance gene is well known to those skilled in the art. Examples of sequences useful in the present invention are given in SEQ ID NO: 1, 13, 15 and 17. It is understood that a variant or fragment retaining the biological activity of the kanamycin resistance gene may also be used in the process according to the present invention.

In a particular embodiment, a peptide spacer is inserted between the protein of interest and the expression product of the kanamycin resistance gene so that the two proteins are distant from one another and can fold back without interfere.

The introduction of the mutations in step ii) can be carried out using one of the existing mutagenesis methods.

In a first embodiment, random mutagenesis is used. In the case where no element makes it possible to define a priori the nature and the position of the changes likely to bring about an improvement of the physical property studied, it is necessary to produce a large number of molecules having mutations of different position and nature. , in order to maximize the chances that there is among them a molecule corresponding to an improved solubility. Mutants are generated under form of a PCR amplification, intended then to be cloned in phase with the kanamycin resistance gene (FIG. 1). As in any double-strand ligation step, it can not be avoided that some of the clones obtained consist of a vector religated on itself. These clones are very troublesome here, because they will be associated with the expression of the kanamycin resistance gene, and will appear in the subsequent selection step as false positives. It is nevertheless possible to limit this rate of false positives by performing a cloning strategy such that the clones religated on themselves can not be expressed, for lack of initiator codon or promoter, for example, these being then present in the fragment to be cloned. These methods are well known to those skilled in the art. However, these maneuvers complicate the cloning strategy.

In a second embodiment, DNA recombination is used to generate the diversity. This approach is similar to random mutagenesis because we do not precisely control the diversity we introduce, but differs from it, however, because this diversity is selected to contain a high rate of enhancement mutations. Indeed, only mutations preselected by nature to retain at least part of the activity are retained. As in the case of random mutagenesis, a double-stranded ligation step is usually necessary, with the consequences in terms of the false positive rate described above. In a third embodiment, site-directed mutagenesis can be used to generate diversity. Site-directed mutagenesis makes it possible to introduce mutations in a perfectly precise manner, but at a low rate. An additional advantage of context-directed mutagenesis is the absence of a double-stranded ligation step, and the rate of false positives is therefore limited. However, the generation of the mutants one by one is not really adapted to a subsequent screening system by direct selection on the criterion of survival. In addition, knowledge of mutation sites for increasing the solubility of a protein is an uncommon situation. Site-directed mutagenesis is therefore not the ideal technique for improving the soluble character of a polypeptide. In a fourth embodiment, the Massive can be used

Mutagenesis® (FR2813314). This method combines the advantages of site-directed mutagenesis (control of the nature of the modifications obtained at a given position), and random mutagenesis (obtaining a large number of different mutations distributed over many positions of the DNA fragment to be mutated. ). Like mutagenesis directed, this technique does not incorporate a double-stranded ligation step, which is positive in the context of the invention to avoid producing false-positive clones.

In step iii), the transformation can be carried out according to various techniques and mainly by using a thermal shock or electroporation approach. It is in particular the size of the diversity to be obtained that will focus on one or the other of these two approaches: the technique by electroporation makes it possible to obtain a higher number of transformants.

Likewise, different types of bacteria can be used such as E. coli, Streptomyces and Bacillus. The choice will be made according to the culture conditions, the protein to be expressed and the quantity to be produced. In the most conventional embodiment, E. coli is most often used (Figure 3).

In step iv), the stringency concentration of kanamycin to be added will have been determined beforehand. This concentration will be determined by tests using the plasmid construct containing the non-mutant gene of interest fused to the kanamycin resistance gene. The stringing concentration will be high enough to no longer allow the bacteria transformed by this construction to grow, or to allow them only slow growth. Under these conditions, bacteria expressing a more soluble variant will have an accelerated growth rate. In a first embodiment, the cultures of the transformed mutant library will be made in a liquid medium: the selection of the most soluble mutants will then be done according to the growth rate (FIG. 4). In a second embodiment, they can also be done in solid medium: the selection will be based on the number and size of colonies (Figure 3).

In a first embodiment, the mutagenesis is carried out by random mutation, and the fragment containing the mutant genes of interest is introduced in 5 'of the kanamycin resistance gene. A DNA molecule containing the gene to be mutated is amplified under particular conditions altering the ability of the enzyme to faithfully replicate the DNA. This introduces changes in the course of the cycles, that is to say differences with respect to the initial sequence. At the end of the reaction, a large number of copies of the initial molecule is obtained, each of these molecules having different mutations. These molecules are present in the form of a bank, that is to say a mixture of molecules of different nature (differing in the nature and position of their mutations). This library of linear molecules is introduced in 5 'of the kanamycin resistance gene, taking care that the level of vector molecules religated on itself, capable of expressing the kanamycin resistance gene, is minimal.

After transformation, the bacteria are placed in the presence of a selective concentration of kanamycin to differentiate between soluble variants and insoluble variants, previously determined. Only cells expressing a soluble variant can survive.

In a second embodiment, mutagenesis is performed using Massive Mutagenesis®. A plasmid is first constructed (Figure 2), containing an origin of replication, an ampicillin resistance gene, and the gene of interest fused at 5 'and in phase with the kanamycin resistance gene. The template is prepared using a suitable plasmid DNA preparation kit.

The mutant oligonucleotides are synthesized.

The matrix is mixed with all the mutant oligonucleotides.

The matrix is denatured by heat so as to temporarily dispose of single-stranded DNA. When returning to a lower temperature, some or all of the oligonucleotides present in the mixture are fixed on the matrix at their site of homology.

- All the elements necessary to perform a replication of the matrix from the mutant oligonucleotides are added, and in particular a polymerase, the buffer allowing its activity, nucleotide triphosphates in sufficient quantity, the necessary cofactors. The replication reaction then takes place under the temperature conditions corresponding to the maximum activity of the polymerase, ie 68 ^° C. The objective is to have mutants each containing combinations of several different directed mutations. E. coli bacteria used as host cells are transformed. The clones obtained are then cultured in a medium containing ampicillin in order to select only the bacteria having integrated a plasmid.

The bacteria thus preselected are then incubated in a medium containing a stringing concentration of kanamycin, so that only the cells expressing a variant with improved solubility or higher level of expression survive.

The proteins or peptides of interest may be, for example, industrial enzymes, used in the textile, food and chemical industries in particular, or recombinant antibodies, in particular single-chain antibodies.

Description of the Figures: Figure 1: Diagram showing the preparation of a mutant library of the protein or peptide of interest. Firstly, a mutant library is prepared (mutagenesis) and then this library is cloned in phase with the kanamycin resistance gene in a vector (insertion into a plasmid).

Figure 2: Diagram showing the preparation of a mutant library of the protein or peptide of interest. Firstly, the gene coding for the protein or peptide of interest is cloned in phase with the kanamycin resistance gene in a vector (insertion into a plasmid), then a mutant library is prepared.

(Mutagenesis).

Figure 3: Diagram showing the transformation of competent bacteria by the mutant library of the protein or peptide of interest followed by selective solid medium culture containing kanamycin.

Figure 4: Diagram showing the transformation of competent bacteria by the mutant library of the protein or peptide of interest followed by selective liquid culture containing kanamycin. Figure 5: Diagram representing the negative control.

Figure 6: Graph showing the resistance to kanamycin bacteria transformed with ContK, Pet23, TA2K or TB2K6 (results of Example 1).

Figure 7: Chart showing the chloramphenicol resistance of the bacteria transformed with ContCam, Pet + Cam, TA2Cam or TB2Cam (results of the comparative example of Example 1).

Figure 8: Table showing the results of Example 2. Figure 9: Diagram of the construction pETl 1-KanK. Figure 10: Growth of cells transformed with a plasmid encoding wild type or mutant kanamycin kinase II (Ql 55W) as a function of kanamycin concentration.

Figure 11: Growth of cells transformed with a plasmid encoding wild type or mutant kanamycin kinase II (Q155W, S114H and Q155W + S114H) as a function of kanamycin concentration.

Figure 12: Kanamycin resistance efficacy of Kat and KK as a fusion protein.

Figure 13: Growth of cells transformed with a plasmid encoding a mutant Q1 55 W kanamycin kinase II and two TA or TB fusion proteins with mutant kanamycin kinase II Q155W as a function of kanamycin concentration. TB and TA respectively mean a soluble and insoluble variant of the FE65 protein.

The following examples illustrate the present application.

Examples

Example 1 Fe65-PTB2 Protein

Genes encoding a soluble variant and an insoluble variant of the protein

FE65-PTB2 were introduced into plasmid petl 1. The construct was made to fuse said protein to a kanamycin resistance gene in the absence of a peptide spacer. The vector also has a resistance gene at

Ampicillin.

We obtained different constructions:

- Pet: plasmid petl 1 which is at the base of the different constructions; it does not contain a kanamycin resistance gene and therefore serves as a negative control (Figure 5).

ContK: plasmid petl 1 on which the kanamycin resistance gene has been cloned (SEQ ID Nos. 1 and 2); this construction serves as a positive control. TA2K: plasmid petII on which was cloned a TA2 fusion protein-kanamycin resistance gene (SEQ ID Nos. 3 and 4); TA2 protein is an insoluble variant of the protein taken as a model. - TB2K: plasmid petl 1 on which was cloned a fusion protein TB2-kanamycin resistance gene (SEQ ID Nos 5 and 6); the TB2 protein is a soluble variant of the protein taken as a model.

E. coli BL21 DE3 bacteria were then transformed by heat shock and then incubated in a liquid medium containing ampicillin at 100 μl / ml at 37 ° C. for 12 h.

After 12h of culture, 10 ^"9 UOD were seeded into cultures containing a kanamycin variable concentration.

A range of kanamycin antibiotics ranging from 100 μg / ml to 0.75 μg / ml were tested to find the stringency concentration for this set of mutants, i.e., the concentration to differentiate mutants with greater solubility.

The cultures were incubated and the OD measured with a spectrophotometer at 600 nm to measure bacterial growth.

A difference of 2.7-fold in resistance was then observed between the soluble variant (half-strength = 90 μg / ml) and the insoluble variant (half-strength = 33 μg / ml) (Figure 6).

In order to make a comparison, the same experiment was performed with the chloramphenicol resistance gene.

Contcam: plasmid petl I on which the chloramphenicol resistance gene has been cloned (SEQ ID Nos. 7 and 8); this construction serves as a positive control. TA2cam: plasmid petl I on which was cloned a TA2 fusion protein-chloramphenicol resistance gene (SEQ ID Nos. 9 and 10); TA2 protein is an insoluble variant of the protein taken as a model.

- TB2cam: plasmid petl I on which was cloned a TB2-chloramphenicol resistance gene fusion protein (SEQ ID Nos 11 and 12); the TB2 protein is a soluble variant of the protein taken as a model.

The experiment was carried out following the same steps as for the antibiotic kanamycin.

The different bacterial cultures were seeded at 10 ^-4 UOD / ml (previously tested as minimum seeding). A range of chloramphenicol ranging from 100 μg / ml to 12.3 μg / ml was tested in order to find the stringering concentration for this mutant mixture, that is to say the concentration to differentiate the mutants with greater solubility. Cultures were incubated and OD measured spectrophotometer at 100 nm to measure bacterial growth.

In this experiment, a chloramphenicol resistance increased by a factor of 1.4 of the soluble variant of the protein (half-strength = 35 μg / ml) compared to the insoluble variant (half-strength = 25 μg / ml) ( Figure 7).

These results demonstrated that the kanamycin resistance gene, in particular kanamycin kinase II, can be used as a reporter for the solubility of a protein fused to it. The kanamycin resistance gene is much better than the chloramphenicol resistance gene, for which it is found that the window of variation of resistance of the two variants is not very extensive and thus does not allow the selection of mutants within 'a bank.

Example 2: In order to apply the technique to the screening of a mutant library, the same experiment was performed as above but with a mixture combining the two variants: soluble and non-soluble.

The stringency concentration of kanamycin determined in Example 1 is used to select only the mutants expressing the soluble protein. The cultures were carried out in solid medium in order to be able to count the colonies.

Several types of cultures were performed: a culture with only the variant expressing the soluble protein, a culture with only the variant expressing the insoluble protein and several cultures with mixtures of the two variants in different proportions (Figure 8).

A colony number was obtained proportional to the protein mixture: for an inoculation containing only clones expressing the soluble protein, 17,000 colonies were obtained; for inoculation containing only clones expressing the insoluble protein, 4 colonies were obtained; for seeding together 10% of clones expressing the soluble protein and 90% of clones expressing the insoluble protein, 4000 colonies were obtained.

This technique has made it possible to demonstrate the presence of mutants expressing a soluble protein in a mixture of variants at a proportion that can be as low as 0.1%.

Example 3;

In order to create soluble mutants of a protein of interest, a mutant library of a plasmid constructed in such a way that the kanamycin resistance gene is fused in phase and C-terminal to that of to express the protein of interest.

Mutagenesis is performed using Massive Mutagenesis® technology:

The plasmid is mixed with a set of mutant oligonucleotides.

- The mixture is denatured by heat. All the elements necessary to perform a replication of the matrix from the mutant oligonucleotides are added.

E. coli bacteria used as host cells are transformed by electroporation. The clones obtained are then cultured in a liquid medium in the presence of ampicillin, so as to preselect the bacteria that have actually integrated a plasmid. The culture is then used to inoculate a liquid medium containing a predetermined kanamycin stringing concentration so as to select the cells expressing a soluble variant.

EXAMPLE 4 In order to identify soluble mutants of a protein of interest, diversity is first generated by random mutagenesis: a DNA molecule containing the gene to be mutated is amplified by error-prone PCR ". The molecules obtained are cloned into a plasmid constructed so that the mutant genes are fused in phase and 5 'of the kanamycin resistance gene. The mutant library is then screened as set forth in the previous example, or using a solid medium to differentiate on the number or size criterion the colonies corresponding to a more soluble variant of the protein of interest. Example 5;

In order to create soluble mutants of a protein of interest, a mutant library of a plasmid constructed such that the kanamycin resistance gene is fused to the C-terminal fused with that for expressing is made. the protein of interest.

Mutations are carried out by recombination of DNA on the gene of interest:

Several homologous genes are isolated from each other, corresponding to the desired activity.

These genes are amplified using flanking oligonucleotides, using very short amplification cycles (STEP).

The DNA fragment obtained is cloned in such a way that the mutant genes are fused in phase and in 5 'to the kanamycin resistance gene present on the vector.

The mutant bank is then screened as set forth in the previous example.

Example 6

The gene coding for kanamycin kinase II (APH (3 ') -IIa) was cloned into the NcoI and NotI sites of the vector pET1 1 (Novagen) behind the T7 promoter to give the construct called pET1 1 -KanK (FIG. 9). .

A kanamycin kinase mutant library was then constructed using Massive Mutagenesis® technology (FR2813314): total diversity was introduced at each position 3 rounds of mutagenesis were performed to obtain a library with a final diversity of 5.10 ⁶ clones. The strain used was E. coli DH10B. Sequencing of 24 randomly selected clones revealed an average number of mutations per molecule of 2.5. Moreover, there was no major bias in the sequences.

The library was then transformed into E. coli strain BL21 (DE3) to allow the expression of kanamycin kinase and spread on different media: non-selective media: ampicillin 100 μg / ml and kanamycin 25 μg / ml to estimate transformation efficiency of the bank

selection medium: kanamycin 300 μg / ml to allow the selection of mutants resistant to high concentrations of kanamycin The number of clones after transformation was estimated at 1.10 ⁶ by counting on ampicillin medium. The medium containing 300 μg / ml of kanamycin made it possible to select 170 clones. After minipreparation of the plasmids isolated from 12 of them, a sequencing step was performed using oligos T7pro and T7ter. The sequence of the 12 clones revealed a single mutation at position 155 where a glutamine residue is replaced by a tryptophan residue.

CAG → TGG Gin (Q) → Trp (W) This Q155W mutant was then reconstructed on the pET11-KanK template by site-directed mutagenesis. The growth in liquid medium of this mutant in the presence of different concentrations of kanamycin was then compared to that of WT. Briefly, increasing concentrations of kanamycin were added to cultures ά'Ecoli BL21 (DE3) transformed with pETll, pETll-Kank and pETll-KanK_Q155W diluted to a D0 ₆ oo _nm of 10 ^"4. After 15h of growth at 37 ° C, the ODs of the different cultures were measured.

The result is shown in Figure 10.

Since the structure of kanamycin kinase II is known (Nurizzo et al., 2003), the mutation can be placed in the structure of the enzyme. The residue Ql 55 is placed on a flexible loop (residues 151-166) which borders the fixing cavity of the kanamycin substrate. On this loop, there are also crucial residues for the activity of the enzyme such as D159 and E160. The proximity of this residue to the active site may therefore explain an increase in the activity of the enzyme.

A new library was then constructed with the mutant pET11-KanK_Q155W as the starting template. The protocol followed is the same as during the first step of mutagenesis (3 successive rounds, sequencing 24 clones for quality control). After transformation into BL21 (DE3) ₅ the library was selected on 1 mg / ml kanamycin to isolate still resistant variants kanamycin as Q155W. After sequencing a few clones transplanted on selective medium, a new mutation appears: the serine 114 is mutated into Histidine.

TCC → CAC Ser (S) → His (H) This new mutant was constructed by directed mutagenesis on a pET11-KanK (single mutant Sl 14H) matrix and on a ρET11-KanK_Q155W matrix (double mutant Q155W, Sl14H). The result of the liquid growth test of the different mutants and WT is shown in Figure 11.

LC50 for each clone was defined as the concentration of kanamycin for which the measured OD corresponds to 50% of the maximum OD.

Under these conditions (dilution 10 ^'6 for the starting pinoculum), the calculation of the EC50s is as follows:

The improvement factor is important. Indeed, it is between 3 and 7 for single mutants and is greater than 14 for the double mutant. It should be noted that the effect of the two mutations is more than additive, it constitutes a synergy between the two mutations.

Example 7

The kanamycin resistance efficacy of two types of kanamycin resistance genes, namely Kat (kanamycin nucleotidyltransferase) and KK (kanamycin kinase II), in the form of fusion proteins with reporter genes (GFP or interferon gamma) was tested. The result is shown in Figure 12. It can thus be seen that, surprisingly, the resistance efficiency of Kat as a fusion protein is seriously reduced, while the efficiency of KK is generally preserved.

Example 8 The results presented here are based on the use of a protein encoded by the antibiotic resistance gene kanamycin (kanamycin kinase II gene) as a reporter for the solubility of a protein-target (Fe65) fused to the latter . This system is based on the assumption of transmitting the soluble or insoluble nature of the protein of interest to the protein conferring resistance to kanamycin. The purpose of the experiments presented here was to generate a mutant bank of the Fe65 protein of interest using Massive Mutagenesis® and to verify that the selection system through kanamycin resistance makes it possible to directly select variants that provide an improvement. of solubility.

constructions

Several constructs are used to carry out the tests: pET11: plasmid which is at the base of the different constructions; it's the negative control. pETHK: plasmid pET11 in which the kanamycin resistance gene has been cloned. This construction serves as a positive control.

TA2K: plasmid pET11 in which was cloned the hybrid gene Fe65TA2-kanamycin resistance gene. TA2 protein is the insoluble variant of the model Fe65 protein. TB2K: plasmid pET11 in which was cloned the hybrid gene Fe65TB2-kanamycin resistance gene. The TB2 protein is the soluble variant of the Fe65 protein.

Conditions previously determined Concentration of kanamycin selective to discriminate between soluble and insoluble proteins (in liquid and solid media): 50 μg / ml

Experiences

Screening of a mini-bank The purpose of the experiments presented here was to make a mini-bank of variants on the gene coding for the insoluble Fe65 model protein (TA2) using Massive Mutagenesis® and to verify that the kanamycin system makes it possible to select directly the soluble mutant. The latter was contained in the oligonucleotides chosen for mutagenesis.

13 oligonucleotides introducing mutations were drawn, one of them corresponding to the mutation TB2 (SolPhe). Oligonucleotides used for the construction of the library: SoINOID GGCAGCGCCTAAGGATGAGTTGGTC SEQ ID No. 19

SolQ02E GAATGAGTTGGTCGAGAAGTTCCAAGTC SEQ ID NO 20 SolN03V TATTACCTGGGGGTAGTACCTGTTGCTAA SEQ ID NO 21 SolN04D TAGATGTGATTGATGGGGCCCTCGAGT SEQ ID No. 22 SolQ05E AGCAGCCGTGAAGAATGGACCCCAAG SEQ ID NO 23 S0IQO6E ACCATCTTGCACGAGCAGACAGAGGCAGT SEQ ID NO 24 SolQ07E ACCATCTTGCACCAGGAGACAGAGGCAGT SEQ ID NO 25 S0INO8D TTCTGGTGCGAGCCCGATGCTGCCAGC SEQ ID NO 26 SolQ09E TCAGAGGCTGTGGAGGCTGCGTGCAT SEQ ID No. 27

SoIQlOE ATGCTTCGCTACGAGAAGTGTGCGG SEQ ID NO 28

SoIQl IE TCAGAGGCTGTGGAGGCTGCGTGCAT SEQ ID NO 29

SolPhe AGTGTCGGGTGCGTTNNCTCTCCTTCCTGG SEQ ID NO 30

SolPro ATGTCAGTGTGGCCCNNGCTACCCTCACCA SEQ ID No 31

One round of Massive Mutagenesis® (PLCR) was carried out on the pET1 1TA2 construct with the oligonucleotides described above; the result of the reaction was transformed into the strain of Escherichia coli DH1OB and spread on a box containing ampicillin. Counting the number of colonies obtained made it possible to verify that all the diversity of the bank was represented.

The colonies obtained were scraped off and a DNA preparation was made from these colonies.

The DNA preparation corresponding to the library was then transformed into the E. coli strain. coli BL21 (DE3) to allow the expression of fusion proteins.

The transformation result was spread on medium containing ampicillin 100 μg / ml (as a control of the transformation efficiency) and on a selective medium containing kanamycin at 50 μg / ml (which should make it possible to select only soluble mutants).

Results: On ampicillin medium, 1500 clones were counted. The same amount of DNA transformed on kanamycin yielded only 12 clones. DNA minipreparations were performed on each of the 12 clones and the hybrid genes were sequenced in their entirety to determine which mutations were retained.

Result: Of the 12 clones sequences, 11 had the mutation corresponding to variant TB2.

• 7 have only TB2 mutation (TTC → TBI: oligo SolPhe)

• 3 have TB2 + other mutation (+ S0INO8D, SolQ09E, SoIQlOE)

• One mutation other than TB2 (SolQ09E)

Example 9

The genes encoding a soluble variant (TB) and an insoluble variant (TA) of the FE65 protein were introduced into the plasmid pET1 1. The construct was made to fuse said protein with kanamycin kinase having the Q1 mutation. W. The following constructions were obtained:

pET11_KanK: Q155W: plasmid pET11 in which the sequence of the kanamycin kinase kinase at position 155 (Q 155W) is cloned

pET11_TA-KanK: Q155W: plasmid pET11 in which is cloned the insoluble variant of FE65 fused to the sequence of kanamycin kinase mutated on position 155 (Q155W)

- pETl l_TB-KanK: Q155W: plasmid pET11 in which is cloned the soluble variant of FE65 fused to the sequence of the kanamycin kinase kinase at position 155 (Q155W)

E. coli BL21 DE3 bacteria were then transformed by heat shock and then incubated in a liquid medium containing ampicillin at 100 μl / ml at 37 ^° C. for 12 h.

After 12 hours of culture, 10 ⁶ UOD were used to seed cultures containing a variable concentration of kanamycin (range of antibiotic kanamycin ranging from 400 .mu.g / ml to 12.5 .mu.g / ml). The cultures were incubated and the OD measured with a spectrophotometer at 600 nm to measure bacterial growth.

A difference in resistance of a factor of 4 between the soluble variant (half-resistance = 156 μg / ml) and the insoluble variant (half-strength = 39 μg / ml) was then observed (Figure 13). These results demonstrated that the kanamycin resistance gene mutated at position 155 is useful as a reporter for the solubility of a protein fused to it. The difference of a factor 4 between the soluble and insoluble variant shows that the system is of great sensitivity.

Claims

1) Process for obtaining soluble or better expressed variants of a protein or peptide of interest, characterized by the generation of a mutant library of the gene encoding said protein or said peptide of interest fused to the polynucleotide encoding a kanamycin phosphotransferase in an expression vector, and by selecting mutant genes corresponding to more soluble or better expressed proteins or peptides by culturing the bacteria transformed by said expression vector library into a culture medium containing kanamycin. 2) Method according to claim 1, characterized in that kanamycin phosphotransferase is kanamycin kinase II, a fragment or a functional variant thereof. 3) Method according to claim 2, characterized in that kanamycin phosphotransferase is an improved kanamycin kinase II, a fragment or a functional variant thereof comprising at least one substitution selected from Q155W and S114H. 4) Method according to one of claims 1-3, characterized in that the method comprises: i) generating mutants of a gene encoding the protein or peptide of interest; ii) Preparation of a mutant library by cloning the mutants obtained in i) in phase with the polynucleotide encoding a kanamycin phosphotransferase in an expression vector; iii) transformation of competent bacteria by the mutant bank obtained in ii); and iv) selecting, using a kanamycin-containing culture medium, bacteria transformed by a vector comprising a mutant corresponding to a protein or peptide of mutant interest that is more soluble or better expressed than the starting protein of interest. 5) Method according to claim 4, characterized in that the mutants are generated in step i) by random mutagenesis. 6) Method according to claim 4, characterized in that the mutants are generated in step i) by recombination of genes. 7) Method according to one of claims 1-3, characterized in that the method comprises: i) manufacturing an expression vector containing the gene encoding the protein or peptide of interest fused in phase with the polynucleotide encoding kanamycin phosphotransferase; ii) generating a mutant library of the gene coding for the protein or peptide of interest on the expression vector obtained in i); iii) Transformation of competent bacteria by the mutant bank obtained in ii). iv) Selection, using a culture medium containing kanamycin, of bacteria transformed by a vector comprising a mutant corresponding to a protein or peptide of mutant interest that is more soluble or better expressed than the protein of starting interest. 8) Method according to claim 7, characterized in that the mutants are generated in step ii) by massive mutagenesis. 9) Process according to any one of claims 1 to 8, characterized in that the expression vector is constituted such that the gene encoding the protein or peptide of interest is located in 5 'of the resistance gene to kanamycin. 10) Method according to any one of claims 1 to 9, characterized in that the gene coding for the protein or peptide of interest and the kanamycin resistance gene are separated by a spacer that does not change the reading phase. . 11) Process according to any one of claims 1 to 10, characterized in that the bacteria are transformed by electroporation or by thermal shock. 12) Process according to any one of claims 1 to 11, characterized in that the culture medium containing kanamycin is solid or liquid. 13) Method according to any one of claims 1 to 12, characterized in that the culture medium contains a quantity of stringy kanamycin. 14) Use of an expression vector encoding a fusion protein between a mutant of a protein or peptide of interest and the polynucleotide encoding kanamycin phosphotransferase to select a variant of this protein or peptide having a improved solubility or increased expression. 15) Use according to claim 14, characterized in that kanamycin phosphotransferase is kanamycin kinase II, a fragment or a functional variant thereof. 16) Use according to claim 14 or 15, characterized in that kanamycin phosphotransferase is an improved kanamycin kinase II, a fragment or a functional variant thereof comprising at least one substitution selected from Q155W and S114H.