WO1998001579A1 - System for detecting protein-protein interactions - Google Patents

Abstract

The invention concerns the use of a DNA sequence comprising: a nucleotide sequence coding for an indicator protein (indicator gene); a promoter; a mutated operator; the said nucleotide sequence being under the control of the said promoter comprising the said mutated operator containing two sequences and deriving from an operator containing two inverted repeat sequences (palindrome), this mutated operator being such that one of its sequences is mutated (mutation 1), while the other is for instance wild. The transcription of the indicator gene is repressed by controlling proteins in the form of heterodimers via their DNA binding sites, each monomer of these heterodimers being such that it recognises specifically one the sequence comprising mutation 1, the other the wild sequence, in particular for detecting interactions between two proteins, these interactions being revealed when there is repression of the transcription of the said indicator gene, this repression resulting from the formation of heterodimers.

Description

PROTEIN-PROTEIN INTERACTION DETECTION SYSTEM

The subject of the invention is a system for detecting protein-protein interactions, and more particularly DNA sequences for the implementation of this system, as well as cellular hosts containing said DNA sequences.

Protein-protein interactions play a fundamental role in higher living organisms, both in terms of the normal development of the organism and in certain pathological situations. The existing biochemical and biophysical methods require on the one hand that the different partners have been previously identified and, on the other hand, require large quantities of proteins (therefore their cloning and their purification). It is therefore essential to have tools enabling the initial step to be carried out, that is to say allowing the identification and cloning of proteins interacting with one another. Such systems have been recently developed using yeast as a host and activation of the transcription of a specific gene as a "probe" (see Fields and Sternglanz, 1994). However, although already well used, this system has certain limits.

The study of these protein-protein interactions can be done using biochemical and biophysical techniques in vitro. But these are only conceivable when the proteins are cloned and can be produced and purified in large quantities. On the other hand, they do not identify a new interaction partner.

An attractive alternative to biochemical methods consists in the use of cDNA expression libraries (reflecting all of the messenger RNAs) or genomic DNA. The advantage of these strategies lies in the fact that the genes of the proteins thus identified are immediately available (for a comparison of these methods see Phizicky and Fields, 1995). Among the methods based on the use of expression libraries, one can distinguish those which use in vitro selection (such as the revelation of lysis plaques by a purified and radioactively labeled protein interacting with an unknown protein) from those using in vivo selection. In this latter area, significant progress has been made thanks to the "double-hybrid" technique in yeast (Ma and Ptashne, 1988; Fields and Song, 1989; Chien et al, 1991; Gyuris et al, 1993; Durfee et al, 1993; Vojtek et al, 1993; Le Douarin et al, 1995). In the system developed by Fields and Song (1989), which makes it possible to identify interacting proteins in yeast in vivo, a known protein X is grafted onto a DNA binding domain (Gal4 or, LexA) and is looking for a second protein Y (grafted onto an activation domain or "AD-activation domain" of transcription) capable of interacting with protein X and thus causing activation of transcription. In a review, Fields S. and Sternglanz R. ("The two-hybrid system: an assay for protein-protein interaction" (1994) TIG, JLQ: 286-292) discuss the possibilities and limits of this method. One of the limits is that the two hybrid proteins must reach the nucleus of the cell. The other major difficulty consists in identifying proteins Y which interact with a protein X provided with a domain of activation of the transcription, since, in this case, the target protein (Gal4-X or LexA-X) will give a signal constitutive activation in the absence of any protein Y.

Acting on Gal4 as a DNA binding support, it was initially developed using the DNA binding domain or DBD ("DNA binding domain") of the Gal4 protein (Ma and Ptashne, 1988 ; Fields and Song, 1989). Gal4 is a modular protein of 881 amino acids with at least two functional domains. A DNA binding domain comprising amino acids 1-94 and a transcription activation domain. The activation function is in fact distributed over three different segments of the protein, namely 94-106, 148-196 and 768-881 (Marmorstein et al., 1992). The structure of Gal4j.65 involved in a complex with DNA has been determined by X-ray crystallography (Marmorstein et al., 1992) while the structure of the activation domain remains very poorly defined.

Thus, Fields and Song (1989) grafted, on the one hand the SNF1 protein (a serine / threonine kinase) on Gal4ι _j47, and on the other hand the SNF4 protein on the activation domain 768-881 of Gal4. Neither of the two hybrid proteins, Gal4ι_i47-SNF1 or SNF4-Gal47-5g_ggι alone were capable of activating the transcription of the reporter gene lacZ placed under the control of the G ALI promoter. On the other hand, the coexpression of the two hybrid proteins resulted in a notable transcription activation (180 units), although clearly less than the activation observed with the whole Gal4 protein (4000 units). This difference is attributable, on the one hand to the absence of the activation domain comprising amino acids 148-196 and, on the other hand, to the fact that the interaction between SNF1 and SNF4 does not completely replace a covalent bond between the two areas of Gal4. A direct merger between Gal4ι .j47 and Gal476 _ggι indeed only causes a reduction of a factor 2 compared to whole Gal4 (Ma and Ptashne, 1987a).

Since then, dozens of pairs of proteins X, Y have been tested. Many combinations are capable of restoring a functional activator of transcription (for a review, see Bartel et a, 1993).

In addition to the analysis of the interaction of a pair of known proteins, the double-hybrid technique also allows the identification of an unknown partner from the moment when one is in possession of the gene for the target protein X. The team by Stanley Fields was the first to exploit this possibility (Chien et a, 1991) by creating fusions between the activation domain 768-881 of Gal4 and protein sequences encoded by DNA fragments from a library of Genomic yeast DNA. This time, the activation domain was placed at the NH2-terminal end of this family of fusion proteins, in order to facilitate the construction of the hybrid proteins.

In order to avoid too large a number of false-positive clones, it is preferable to insert the library into the plasmid carrying the activation domain, since Ma and Ptashne (1987b) had shown before, that numerous protein fragments fused to The first 147 amino acids in Gal4 were able to activate transcription.

Since the article by Chien et al. (1991) the strains and plasmids used have evolved. In particular, the combination of metabolic selection associated with a lacZ screen has become widespread. While Chien et al. (1991) only worked on the lacZ screen, more recent approaches also use the acquisition of the ability to grow on a medium lacking either histidine or leucine. In principle, only cells carrying a hybrid protein from the library, capable of activating the expression of the selection gene (His3, Leu2) via an X, Y interaction will form colonies on dishes. Such a selection allows the analysis of a larger number of clones.

In addition to Gal41.147 as a support for attachment to DNA, the team of Roger Brent (Harvard Medical School) introduced the repressor LexA of Escherichia coli as a support (Golemis and Brent, 1992; Gyuris et a, 1993).

LexA is a protein of 202 amino acids, composed of two domains linked by a hinge region of approximately 30 amino acids. The first 70 amino acids constitute the DNA binding domain (Hurstel et al, 1986; Fogh et al, 1994) and amino acids 100 to 202 form the protein dimerization domain (Schnarr et al., 1985) . LexA recognizes sites palindromic DNA (SOS operators) and several adjacent operators can be occupied in a cooperative manner (Lloubès et al, 1991).

Using LexA as a DNA binding medium can have several advantages. .LexA belongs to a heterologous organism (Escherichia coli), does not affect the growth of yeast, has no residual transcriptional activity (like segment 94-106 of Gal4) and can be used in strains of yeast gal + . Therefore, one of the hybrid proteins (usually the one containing the activation domain) can be placed under the control of Gal4. Thus, the hybrid protein will be expressed only in the presence of galactose, but not in the presence of glucose. The field of activation used by Roger Brent's team also comes from Escherichia coli. It is one of the artificial activator domains isolated by Ma and Ptashne (1987b). This domain (named B42) has a lower activation potential than the activation domains of Gal4 or VP16, which makes it possible to avoid the problems of toxicity linked to a possible "squelching" (the titration of an element essential of the transcription machinery by an activation domain). A recent description of the double-hybrid system based on the use of LexA can be found in Golemis et al (1996) and Golemis and Brent (1996).

Despite the frequency of use of the various double-hybrid systems, few studies have been carried out seeking to correlate data obtained in vivo with affinity measurements in vitro. Estojak et al. (1995) sought to establish such a correlation in the case of several helix-loop-helix proteins (Myc, Max, Mxi) and in the case of mutants of the lambda repressor. Their results are somewhat disappointing.

Indeed, the double-hybrid system shows a strong polarity for the Myc / Max couple. When Myc is fixed on DBD and Max on AD, no activation of transcription is observed. On the other hand, when Max is fixed on DBD and Myc on AD, an activation is observed in agreement with biochemical data showing that the two proteins interact. When the latter geometry is used with lacZ reporters having a variable number of operators upstream of the gene, the hierarchy of apparent affinities in vivo between the Max / Myc and Max / Mxi couples is not the same depending on the reporter used.

The situation is even more confused in the case of mutants of the lambda repressor. Mutants with a greatly reduced dimerization capacity activate certain reporter genes better than the wild-type protein (Estojak et al., 1995). Furthermore, although all the variants activate the lacZ gene quite strongly, three out of four are unable to activate the leu2 gene and would therefore have fact, escaped a selection on medium deprived of leucine. In conclusion, Estojak et al. (1995) consider that no reporter gene shows a satisfactory correlation with the affinities measured in vitro.

Given the complexity of assembling a transcription initiation complex in eukaryotes (see: Hori and Carey, 1994; Maldonado and Reinberg, 1995) which mobilizes at least thirty different proteins, and the functioning always misunderstood of transcription activators, these results are not so surprising.

LexA is the repressor of the SOS system at E.coli. LexA, composed of two domains, is inactivated in vivo following damage to DNA by a proteolytic cleavage located between the two domains (for a review on LexA, see Schnarr and Granger-Schnarr, 1993). The separation of the DNA binding domain from the dimerization domain then leads to a decrease in affinity for DNA of approximately 1000-fold, no longer allowing efficient transcriptional repression. On the other hand, the dimerization constant of LexA (KA = 2 x 104 Ml) is relatively low (Schnarr et al, 1985), which suggests that the transplant of a heterologous dimerization domain with modest affinity should reconstitute a repressor functional hybrid. Schmidt-Dôrr et al (1991) were able to show that the fusion of the leucine zipper ("leucine-zipper") from the Jun oncoprotein to LexAι_g7 restores transcriptional repression. On the other hand, the fusion of the leucine slide Fos ("leucine-zipper Fos"), which is unable to confer homodimerization, does not restore repression (Porte et al, 1995). On the basis of this observation, the Fos zipper part ("Fos-zipper") of the fusion protein was then mutagenized, in order to characterize the anti-determinants of Fos homodimerization (Porte et al., 1995).

The indicator bacterial strain is deficient in LexA and carries the lacZ gene under the control of an SOS operator recognized by LexA (the sulA operator). The hybrid protein, placed under the control of the lacUVS promoter inducible by 1TPTG, is produced from a plasmid so as to be able to adapt the quantity of protein to experimental needs. This system is perfectly suited to the study of a homodimeric complex, but its use becomes laborious the moment one wants to study the heterodimerization between two different proteins capable of also forming homodimers.

In addition to the LexA system, James Hu's team has developed a similar approach using the DNA binding domain of the lambda repressor (Hu, 1995). The advantage of this system is to offer the possibilities of selecting the lambda phage. Its disadvantage lies in the fact that the domain of attachment to the DNA of the lambda repressor already contains a dimerization element (helix 5), which can mask the effect of relatively weak heterologous dimerization domains.

The invention particularly aims to solve the difficulties presented by the systems of the prior art and, in particular, the invention aims to allow the study of heterodimerization.

The subject of the invention is a system allowing the detection of interaction between two grafted proteins, by visualizing the repression of transcription and not by activating transcription. Unlike the yeast system, a protein with intrinsic activating activity could be used.

The subject of the invention is a system for detecting protein-protein interactions using a bacterium (Escherichia coli) as a host organism, commonly used in all molecular biology laboratories.

The subject of the invention is a system for detecting protein-protein interactions avoiding the problem of addressing fusion proteins in the nucleus as is the case in yeast, thanks to the non-compartmentalisation of E . coli, in particular the absence of a structured nucleus.

The subject of the invention is a system suitable for studying the interactions of proteins located in the cytoplasmic membrane.

The subject of the invention is a test (repression of transcription) potentially allowing the identification of proteins interacting with the activation domains of transcription factors, among which there are many oncogenic products involved in carcinogenesis, such as Jun, Fos. or Εts family members.

The subject of the invention is a system for studying protein-protein interaction (determination of interaction forces, identification of structural or sequence elements important for this interaction, analysis of the specificity of the interaction) more reliable than the yeast system (Estojak et al. 1995).

The subject of the invention is a system for studying protein-protein interaction making it possible to detect interactions when the "hook" protein (the one for which the partner (s) is sought) has intrinsic capacities for activating transcription ( Bartel et al. 1993, Golemis and Brent 1996, Golemis et al 1996) and this occurs quite frequently (Ma and Ptashne 1988), and therefore the invention relates to a system allowing the study of the interactions of a " true "domain of transcription activation; The subject of the invention is a protein-protein interaction detection system allowing the examination of a large number of clones (> 10 ⁶ ) thanks to the transfection efficiency of E. Coli which is clearly greater than that of yeast (of the order of 1000 times) and thus the screening of cDNA or genomic DNA libraries.

The invention relates to the use of a DNA sequence comprising:

- a nucleotide sequence coding for an indicator protein (indicator gene),

- a promoter,

a mutated operator, said nucleotide sequence being under the control of the above promoter comprising the above mutated operator containing two sequences and deriving from an operator containing two inverted repeated sequences (palindrome), this operator being such that, in the non-mutated state (wild), it leads to the repression of the transcription of the indicator gene, when regulatory proteins in the form of homodimer bind to each of said inverted repeat sequences via their DNA binding site, this mutated operator being such that the one of its sequences is mutated (mutation 1), while the other is in the wild, or mutated in a silent manner or comprises a different mutation (mutation 2) from the above mutation 1, the transcription of the indicator gene being repressed by regulatory proteins in the form of heterodimers via their DNA binding site, each monomer of these heterodimers being such that it one specifically recognizes the sequence comprising the mutation 1, the other the sequence which is in the wild state, or which is mutated in a silent manner or which comprises a mutation 2, the homodimers capable of being formed from each monomers being incapable of repressing the transcription of the above-mentioned indicator gene,

- to detect or quantify interactions between two proteins or protein domains, or to screen proteins or protein domains exhibiting interactions with a specific protein or protein domain, these interactions being revealed when there is repression of the transcription of the above indicator gene, this repression resulting from the formation of heterodimers comprising:

1) a first monomer containing a first fusion protein containing the DNA binding site of a regulatory protein capable of specifically recognizing the sequence of the mutated operator containing the mutation 1, the above-mentioned binding site being fused to the '' one of the two proteins or one of the two protein domains, involved in the above interactions, 2) a second monomer containing a second fusion protein containing the DNA binding site of a regulatory protein capable of specifically recognizing the sequence not mutated, or mutated silently or comprising the mutation 2 of the mutated operator, the above-mentioned binding site being fused to the other of the two proteins or to the other of the two protein domains, involved in the above interactions.

The use according to the invention is based on the fact that a regulatory protein, for example LexA, regulates the expression of genes in vivo by binding to specific sequences located upstream of these genes (operator sequences).

The operator used in the context of the invention is a mutated operator deriving from a non-mutated operator which is such that, in the wild, its sequences are palindromic (repeated inverted) and recognized by homodimers of the regulatory protein . The regulatory protein, for example LexA, is organized into two domains, a DNA binding domain which specifically recognizes operator sequences and a dimerization domain through which the interactions between two protein monomers stabilize the interaction of dimers on DNA. Each monomer is fixed on a half-operator. By "half-operator" is meant the sequence specifically recognized by each monomer. These specific sequences are identical in the context of a wild-type or doubly mutated operator, and different in the context of a mutated operator.

This fixation is reinforced by the existence of contacts between the dimerization domains of each monomer. The DNA binding domain alone (devoid of the dimerization domain) is incapable of efficiently fixing the operator and therefore of regulating the expression of a gene. If the dimerization domain of LexA is replaced by heterologous but capable of interacting protein motifs, the regulatory activity of LexA is restored.

However, in the context of the invention, the operator sequence recognized by a regulatory protein, for example LexA, has been modified so that a half-operator remaining wild will be recognized by the wild regulatory protein, for example LexA, while the other half operator will be mutated and recognized by a mutated form of the regulatory protein.

Each of these DNA binding domains can be fused to protein domains capable of heterodimerizing and any interaction between these domains is measured by a more or less significant repression of the indicator gene (also designated by "reporter gene").

By indicator gene is meant: - Or a gene coding for a protein whose activity on the one hand is revealed by the color which it confers on the bacterial clones producing it and on the other hand can be quantitatively measured, for example the lacZ gene, the gene coding for luciferase,

- or a gene coding for a protein toxic to the bacterium so that if the protein is produced, the bacterium dies, for example the sulA gene.

We can also use a system for which:

- the toxic gene is a gene coding for resistance to an antibiotic,

- the mixed operator op408 / opWT is placed in the promoter of the gene coding for the repressor of this antibiotic resistance gene.

In the absence of interaction between LexA408-X and LexAWT-Y, the repressor of the gene in question is produced, the bacteria cannot grow on a culture medium containing this antibiotic. Otherwise, the synthesis of the repressor does not take place, the antibiotic resistance gene is produced and the bacteria give rise to colonies on a culture medium containing this antibiotic.

The term “homodimers” denotes non-covalent associations of proteins in which the protein-protein interaction domains are identical and covalently linked to heterologous or homologous protein binding domains of DNA, which are themselves identical or different. .

By "heterodimers" is meant non-covalent associations of proteins in which the protein-protein interaction domains are different and covalently linked to heterologous or homologous protein binding domains of DNA, which are themselves identical or different. .

By "silent mutation carried by one of the operator's sequences" is meant a mutation which does not affect the recognition of said sequence by the wild-type regulatory protein.

By “mutation 2 carried by the operator sequence”, is meant a mutation different from mutation 1, and which is such that the sequence comprising said mutation 2 is recognized by a regulatory protein which is not capable of recognizing the sequence carrying the mutation 1, the mutation 1 itself being as recognized by a regulatory protein which is not capable of recognizing the sequence carrying the mutation 2.

By “protein-protein interaction”, we mean the non-covalent assembly of two (or more) proteins by Van der Waals interaction (electrostatic interactions between opposite charges, hydrogen bonds, interactions hydrophobic). The area masked by such an interaction is generally from 1400 to 1600 A ° ² (for a review see: J. Janin "Principles of protein / protein recognition from structure to thermodynamics" 1995, Biochemistry 27: 497-505).

By "DNA binding domain of a protein" is defined the part of the protein whose integrity is essential for its non-covalent attachment to DNA.

By "interaction detection" is meant a method for visualizing an interaction between two proteins.

By "quantification of the interaction", we designate a method allowing to quantify the strength of this interaction.

The expression "protein capable of recognizing the sequence of an operator" designates a protein capable of making non-covalent bonds with a particular DNA sequence.

The expression “binding domain fused to a protein” designates a protein domain linked by a peptide bond to another protein or protein domain.

According to an advantageous embodiment of the invention, the regulatory protein is devoid of its dimerization domain, which can however be tolerated if the interactions resulting from the presence of the dimerization domain are weaker than those which it is desired to bring into play. evidence.

The invention also relates to a DNA sequence comprising:

- a nucleotide sequence coding for an indicator protein (indicator gene),

- a promoter,

a mutated operator, said nucleotide sequence being under the control of a promoter comprising a mutated operator containing two sequences and deriving from an operator containing two inverted repeat sequences (palindrome), this operator being such that, in the non-mutated state (wild), it leads to the repression of the transcription of the indicator gene, when regulatory proteins in the form of homodimer bind to each of said inverted repeat sequences via their DNA binding site, this mutated operator being such that the one of its sequences is mutated (mutation 1), while the other is in the wild, or mutated in a silent manner or comprises a different mutation (mutation 2) from the above mutation 1, the transcription of the indicator gene being repressed by regulatory proteins in the form of heterodimers via their DNA binding site, each monomer of these heterodimers being such that it recognizes s one specifically the sequence with mutation 1, the other the sequence which is in the wild, or which is mutated in a silent manner or which contains a mutation 2, the homodimers capable of being formed from each of the monomers being incapable of repressing the transcription of the above-mentioned indicator gene, said mutation 1 being further as it is specifically recognized by a regulatory protein comprising a DNA binding site forming part of a first fusion protein, which binding site is fused with a protein or protein domain capable of interact with another protein or another protein domain forming part of a second fusion protein containing the DNA binding site of a regulatory protein capable of specifically recognizing the non-mutated sequence, or mutated silently or with mutation 2.

The invention relates to a DNA sequence characterized in that it comprises a mutated operator containing two sequences and derived from an operator containing two inverted repeat sequences (palindrome), one of these two sequences being mutated, and the another being in the wild.

The mutated sequence can be upstream or downstream of the sequence in the wild.

The invention also relates to a DNA sequence characterized in that the mutated operator contains the following sequences:

CCGT and ACAG

spaced apart by a sequence of approximately 3 to approximately 20 nucleotides, and in particular from 8 to 12, and advantageously 8 nucleotides, and in particular characterized in that the mutated operator contains the following sequence:

5'-CGGAAπCAATAGGGπGATCTTTGTTGTCACTGGATGTACCGTACATCCATACAGTAACTCACAGGGGC-3 '

The CCGT and ACAG sequences correspond to the CTGT and ACAG sequences of the non-mutated operator respectively.

The present invention relates to a DNA sequence characterized in that the nucleotide sequence coding for the indicator protein is the lacZ gene in E. coli, coding for β-galactosidase, the mutated operator contains two sequences and is derived from an operator containing two inverted repeat sequences (palindrome), one of these two sequences being mutated, and the other being in the wild, the mutated sequence preferably being that located downstream of the sequence in the wild, the wild sequence being recognized specifically by the wild LexA protein or by a fragment of LexA containing the DNA binding site, in particular a fragment defined by positions 1 to x of LexA, x varying from 69 to 87, and in particular by fragments 1 to 69 or 1 to 87 of .LexA, and the mutated sequence being recognized by the mutated LexA protein or by a mutated fragment of LexA containing the DNA binding site, in particular a mutated fragment of LexA corresponding to that defined by positions 1 to x of LexA, x varying from 69 to 87, and in particular by fragments 1 to 69 or 1 to 87, and the mutations of LexA being advantageously P40 → A, N41 → S, A42 - »S, the dimerization domain of .LexA being absent from both LexA or the fragment of LexA and mutated LexA or the mutated fragment of LexA.

The three simultaneous mutations P40 - »A, N41 → S, A42 → S in LexA give rise to a mutant designated by LexA408.

By "dimerization domain" is meant the part of the protein whose integrity is essential for non-covalent interaction with itself (homodimer) or with another dimerization domain (heterodimer).

It is necessary that the dimerization domain of .LexA and of LexA408 or of the fragment used of LexA and of LexA408 are absent otherwise when the two proteins (LexA and LexA408) are coexpressed in the cellular host implicated in the invention, the gene indicator would be suppressed.

The present invention also relates to a complex between - a DNA sequence comprising a nucleotide sequence coding for an indicator protein, said nucleotide sequence being under the control of a promoter comprising a mutated operator containing two sequences and derived from an operator containing two inverted repeat sequences (palindrome), this operator being such that, in the non-mutated (wild) state, it results in the repression of the transcription of the indicator gene, when regulatory proteins in the form of homodimer bind to each of the said inverted repeat sequences via their DNA binding site, this mutated operator being such that one of its sequences is mutated (mutation 1), while the other is in the wild, or mutated silently or contains a different mutation (mutation 2) from the above mutation 1, the transcription of the indicator gene being repressed by regulatory proteins in the form of heterodimers via their DNA binding site, each monomer of these heterodimers being such that it specifically recognizes one the sequence comprising the mutation 1, the other the sequence which is in the wild state , or which is transferred in a silent way or which includes a mutation 2, the homodimers capable of being supplied from each of the monomers being incapable of repressing the transcription of the above-mentioned indicator gene,

- and heterodimers whose monomers constituting them (respectively first monomer and second monomer) are involved between them in a protein / protein interaction:

1) the first monomer being such that it contains a first fusion protein containing the site of attachment to the DNA of a regulatory protein capable of specifically recognizing the sequence of the mutated operator containing the mutation 1, the aforementioned site of fixation being fused to one of the two proteins or to one of the two protein domains, involved in the above interactions,

2) the second monomer being such that it contains a second fusion protein containing the DNA binding site of a regulatory protein capable of specifically recognizing the sequence not mutated, or mutated silently or comprising the mutation 2 of the mutated operator, the above-mentioned binding site being fused to the other of the two proteins or to the other of the two protein domains, involved in the above interactions.

The present invention also relates to a complex characterized in that:

- the DNA sequence is as defined above and in that

- the heterodimers contain:

1) a monomer containing a first fusion protein containing the DNA binding site of mutated LexA or of a mutated fragment of LexA recognizing the sequence of the mutated operator containing mutation 1, which binding site is fused with a specific protein or a specific protein fragment capable of being involved in an interaction as described above, which protein or which protein domain contains, for example, an activation or dimerization domain,

2) a monomer containing a second fusion protein containing the binding site to DNA of wild LexA, or carrying a silent mutation or a mutation 2 or a fragment of LexA recognizing the non-mutated sequence of LexA or mutated in a silent manner or comprising a mutation 2, which binding site is fused with a specific protein or a specific protein fragment, containing for example a protein or a protein domain capable of interacting with the protein or the fragment protein defined in 1) or with the above domains of activation or dimerization.

The invention also relates to a cell host containing a DNA sequence as described above.

The invention also relates to a method for detecting and quantifying interactions between two proteins or protein fragments, characterized

- in that the nucleotide sequences coding for the first and the second monomer respectively containing a first and a second fusion protein as defined above are brought into contact with a DNA sequence as described above, and

- in that the possible repression of the transcription of the indicator gene as defined above is detected, in particular by transfection of a cellular host as described above, with vectors containing, on the one hand, the above nucleotide sequences coding respectively for the first and the second monomer respectively containing the first and the second fusion protein and, on the other hand, the elements necessary for the expression of the said monomers.

The invention relates to a method for identifying a protein interacting with a specific protein, characterized

- in the presence of the following elements:

. a nucleotide sequence coding for a first monomer containing a first fusion protein between, on the one hand, a regulatory protein according to the invention and, on the other hand, the determined protein,

. a genomic library, containing in particular DNA fragments, and in particular cDNA, each capable of coding for a protein interacting with the determined protein, each of the different elements of the genomic library being respectively fused to the sequence coding for a protein regulatory as defined above,

. a DNA sequence according to the invention, and

- in that the possible repression of the transcription of the nucleotide sequence coding for an indicator protein as described above is detected, in particular by transfection of a cellular host according to the present invention, on the one hand, with a vector containing the nucleotide sequence coding for the above first monomer containing the first fusion protein and, on the other hand, a set of vectors containing a nucleotide sequence coding for a second monomer containing a second fusion protein, each nucleotide sequence coding for every second monomer containing the nucleotide sequence coding for the regulatory protein according to the present invention, and downstream of this one of the DNA elements of the above-mentioned library, all the vectors further containing the elements necessary for the expression of the aforesaid monomers.

This process is the subject of the example below. In this example, the indicator gene is lacZ and has been placed under the control of a promoter / operator sequence characterized in that the operator has been mutated as indicated below. This gave rise to the construction of an indicator strain (SU202). This strain makes it possible to identify, on the basis of the color of the bacterial clones, the existence of protein-protein interactions between two proteins or protein domains fused respectively to the DNA binding domain of wild LexA and LexA408 as described below. This strain makes it possible to quantify and analyze this interaction by mutagenesis.

However, screening a DNA library requires the analysis of at least 10 ^ clones. This implies obtaining on an appropriate solid indicator medium (as indicated in FIG. 5) and the experiment of 10 ^ bacterial colonies whose color should be checked (white or pink in the event of interaction and red in the absence of interaction) . Although theoretically feasible, this is very costly not only in time, but also in disposable equipment.

To identify new interactions, the following selection system is advantageously used. In this type of application, the protein-protein interactions are no longer visualized on the basis of the color of the bacterial clones, but on the simple fact of their existence. Indeed, if the indicator gene is replaced by a gene coding for a toxic protein:

the absence of protein-protein interaction between parts Y and X of LexAι_ 7W -Y and of LexAι_g7408-X (as described in FIG. 1) leads to the production of the toxic protein and therefore the death of the bacteria,

- the existence of these interactions leads to the repression of the toxic gene, the bacteria can then divide and give rise to a colony on solid culture medium.

Such a system based on the survival of bacteria is obviously easier to implement. However, it is obvious that, as in all selection systems, there are so-called "false positives", that is to say clones which survive for reasons other than those we identify, for example, spontaneous mutations that inactivate the toxic gene product. To remove this ambiguity, use the indicator strain SU202. Indeed, only the clones growing white or pink are considered "true positive", the surviving clones due to a mutation in the toxic gene being unable to repress LacZ and growing red.

The toxic gene used is the sulA gene. It codes for a protein, SULA, an inhibitor of cell division. In its presence, the bacteria no longer divides and dies.

So far, the construction of putting sulA under the control of a transferred operator has been completed. Using the model system described below with the hybrid proteins of LexAι_g7408-cJunZip and LexAι_ 7WT-cFosZip, it has been shown that the system works. Indeed, the constniction (sulA under the control of a mutated operator) can only be propagated in a bacterial host if it also carries the genes coding for the hybrid proteins of LexAι_g7408-cJunZip and LexAι_g7WT-cFosZip.

The discovery of new protein-protein interactions requires the constitution of a DNA library ligated in fusion following the binding domain. For this, a DNA library was formed which is ligated in fusion following the DNA binding domain of wild LexA or of a LexA mutant recognizing the wild part of the operator but with a lower affinity. such as those carrying one of the following mutations G - »D or G -> S in position 23 (ref. 21). The purpose of this alternative is to limit the appearance of "false positive" clones which may be due to too strong homodimeric interaction of a protein or protein domain present in the library.

Description of the figures:

- Figure 1: Figure 1 relates to the principle of operation of the double-hybrid system in E. coli

The system is based on the fact that a mutated operator containing 1/2 wild operator (WT) and 1/2 mutated operator (408) is only recognized by a mixed LexAWT / .LexA408 dimer. An indicator gene (lacZ represented by an empty rectangle) was put under the control of this mixed operator (represented by a black rectangle for the non-mutated and annotated WT part and by a hatched rectangle for the mutated and annotated part 408). Proteins or subdomains of proteins X and Y capable of interacting are fused to the DNA binding domains of LexAi-87WT (represented by a black sphere) and LexAi-g7408 (represented by a hatched sphere). .LexAi-g7WT- X (represented by a hatched rectangle connected to a hatched sphere) and LexA 1-87408- Y (represented by a black rectangle connected to a black sphere) are cloned on compatible plasmids and their expression is controlled by a inducible promoter (lacUV5, Oertel-Buchheit P., Lamerichs RMJN, Schnarr M. and Granger-Schnarr M. 1990 Mol. Gen. Broom. 222: 40-48 Genetic analysis of the LexA repressor: isolation and characterization of LexA (Def) mutant proteins.). The plasmids are represented by circles, a thickened part of which is separated into two, one represents the gene coding for the proteins, LexAι_87WT-X or LexA 1-87408-Y and the other for their promoter sequence, lacUV5). These proteins, LexAi-87WT-X and LexAi-87408-Y can exist either in free form in the bacteria, or in form linked to DNA (represented by a black rectangle linked to a gray sphere interlaced with a hatched rectangle linked to a hatched sphere themselves connected to the black and hatched rectangles representing the mixed operator described above.). The LacUV5 promoter is itself under the control of a regulatory protein, the lactose repressor (represented by a square with rounded corners and segmented into boxes). The gene coding for this repressor (lacl gene) is cloned on an episome, factor F '(Willetts N. and Skurray R. in E. coli and Salmonella Typhimurium, Cellular and Molecular Biology Ed. Neidhardt FC 1987, pp 1110-1133 and Oertel-Buccheit et al 1990) itself represented by an arc of a circle. The lactose repressor can exist either in free form in the bacterium, or in form linked to DNA (represented by a square with rounded corners and segmented in cartons linked to the promoter sequences, lacUV5 described above). The term LexA (Def) means that the strain is devoid of endogenous LexA.

- Figure 2: Figure 2 shows the construction of the indicator strain for the implementation of a double hybrid system in E. coli.

The plasmid pRS415 (represented by the circle containing inside the indication "pRS415" (SIMONS RW, HOUMAN F. & KLECKNER N. (1987). Improved single and multicopy lac-based cloning vectors for protein and operon fusions. Gene 52: 85-96) carries the lacZ gene, lacking a promoter / operator region (represented by a thickening of the line) and two unique cleavage sites for two restriction enzymes EcoRI and BamHI. "Lac" means that this plasmid is incapable of produce the lacZ gene product since it lacks a promoter / operator sequence. This plasmid is cut by these two enzymes and then ligated to an oligonucleotide of 70 base pairs (70 mer and represented by a black rectangle) whose sequence is the next one :

GAATTCAATAGGGTTGATCTTTGTTGTCACTGGATGTACCGTACA TCCATACAGTAACTCACAGGGGCACCCTAGG-3 'this oligonucleotide thus introduces before the lacZ gene a promoter / operator sequence allowing the lacZ gene to be transcribed and regulated by the mixed operator as defined in the legend for figure 1. This ligation by an arrow between the plasmid and the 70mer creates the plasmid pGC202 (represented by the circle

-I- containing inside the indication "pGC202"). "Lac" means that this plasmid is now capable of producing the lacZ gene product since it is provided with a promoter / operator sequence. The indicator gene thus formed (represented by a black rectangle for the promoter / operator sequence and by a white rectangle for the lacZ gene) is then transferred to the DNA of a lambda phage (λ) by a double recombination event between l Plasma and phage DNA (double recombination event is understood to mean a complex molecular mechanism occurring in vivo and resulting in the exchange of DNA fragments between partially homologous molecules, this is shown in the figure by two crosses between the plasmid pGC 202 and a line symbolizing the phage DNA) giving rise to the phage λGC 202. This phage has the capacity to integrate into the bacterial chromosome by a complex mechanism taking place in vivo (Elledge et al. 1991). The advantage of this system lies in the fact that the indicator gene will be on the bacterial chromosome and no longer on a plasmid, which makes it possible to use the plasmid vectors as a vehicle for fusion proteins.

- Figure 3: Figure 3 concerns the characterization of the SU202 strain as a function of the quantity of proteins present in the cell.

On the abscissa are the concentrations of IPTG (M) and on the ordinate, the relative expression of β-galactosidase.

The variation in the quantity of proteins is obtained by cultivating the bacteria in the presence of variable concentrations of IPTG, the genes coding for the chimeric proteins being under the control of the inducible promoter lacUV5. The β-galactosidase units are measured according to the method described by Miller (1972). The relative expression of β-galactosidase is calculated by making the ratio between the β-galactosidase units measured in the presence of the expression vectors and those measured in their absence, the latter correspond to the non-repressed state of the lacZ gene. The white squares correspond to the values measured when the strain contains a plasmid expressing LexA 1-87408-cJunZip (pDL606, see Figures 6 and 7). The white circles correspond to the values measured when the strain contains a plasmid expressing the DNA binding domain alone (JWL96, Little JW and Hill SA (1985) Deletion within the hinge region of a specifies DNA-binding protein Proc. Natl. Acad. Sci USA, 22: 2301-2305). The white triangles correspond to the values measured when the strain contains a plasmid expressing LexAi-g7WT-cFosZip (pMS404, see Figures 8 and 9). The black squares correspond to the values measured when the strain contains both the plasmid expressing the domain DNA binding alone (JWL96) and that expressing LexAi-87408-cJunZiρ (pDL606, see Figures 6 and 7). The black circles correspond to the values measured when the strain contains both the plasmid expressing LexAi-87408-cJunZip (pDL606, see Figures 6 and 7) and that expressing LexAi-87WT-cFosZip (pMS404, see Figures 8 and 9).

- Figure 4: Figure 4 represents the bacterial strain SU202 which makes it possible to selectively measure the heterodimerization of fusion proteins capable of homodimerizing.

On the abscissa are the concentrations of IPTG (M) and on the ordinate, the relative expression of β-galactosidase.

Fos leucine zipper mutants were constructed in the laboratory, these mutants acquired the ability to homodimerize (unlike Fos leucine zipper, which is incapable of this). Four of these mutants are used here and are described in: Porte D., Oertel-Buccheit P., Granger-Schnarr M. and Schnarr M. 1995 J. Biol. Chem. 22Q: 22721-22730 Fos leucine zipper variants with increased association capacity, These are:

* a mutant for which the five positions "a" of the slide ("zipper") (defined in Porte et al 1995) are occupied by leucine residues (LexAi-87WT-cFosZipLLLLL) and which is represented by white circles in Figures 4 A, B and C,

* of a mutant for which the five positions "a" of the slide ("zipper") (defined in Porte et al. 1995) are successively occupied by two valines, an isoleucine, a phenylalanine and a valine (LexAi-87WT- cFosZipWIFV) and which is represented by black circles in FIGS. 4A, B and C,

* a mutant for which the five positions "a" of the slide ("zipper") (defined in Porte et al. 1995) are successively occupied by two isoleucines, a leucine and two isoleucines (LexAi-87WT-cFosZipIILII) and which is represented by white squares in Figures 4A, B and C,

* a mutant for which the five positions "a" of the slide ("zipper") (defined in Porte et al. 1995) are successively occupied by an isoleucine, a leucine, a phenylalanine, a leucine and a phenylalanine (LexAi -87WT-cFosZipILFLF) and which is represented by black squares in FIGS. 4A, B and C,

* the black triangles represent wild LexAi-87WT-cFosZip,

* the dotted curves represent the measurements made in the absence of plasmid. Each of these constructions is either introduced separately into the indicator bacterial strains (FIG. 4A for the strain SU202 and 4B for the strain SU101) or co-introduced with the fusion protein LexAi-87408-cJunZip (FIG. 4C). The β-galactosidase activity is then measured under the conditions described by Miller et al 1972 as a function of variable concentrations of IPTG. In the absence of protein, the number of units measured remains constant, indicating that the gene coding for β-galactosidase is completely expressed, when the number of units measured decreases, this means that the gene coding for β-galactosidase is repressed , the intensity of the repression being a function both of the amount of protein present in the cell (it increases when the concentration of IPTG increases) and of the interaction strength of the proteins with each other.

A- The strain SU202 (as described in this claim) does not "respond" to homodimerization (lacZ controlled by a mixed operator WT / 408).

B- The homodimerization of the various proteins is demonstrated in a strain (SU101) in which lacZ is controlled by a WTC operator -The strain SU202 selectively measures their heterodimerization with LexA408JunZip

- Figure 5: in Figure 5, we see that the SU 202 strain is transformed either with each of the plasmids carrying the genes coding for the fusion proteins LexAi-87408-cJunZip and LexA _j _87WT-cFosZip is cotransformed with the two plasmids. The transformed cells are left overnight at room temperature in LB supplemented with either kanamycin, chloramphenicol, ampicillin or kanamycin, chloramphenicol, tetracycline or the four antibiotics: kanamycin, chloramphenicol, ampicillin and tetracycline, before being spread on solid medium. McConkey containing the same antibiotics. Cells transformed by only one of the two plasmids grow red. This staining means that the β-galactosidase gene is not repressed by the fusion proteins and that the enzyme produced ferments the lactose in the medium. The bacteria cotransformed by the two plasmids grow white, the β-galactosidase gene is suppressed.

- Figure 6: Figure 6 shows the map of plasmid pDL606 (for cloning the "hook"). The restriction sites indicated are unique sites. Ap represents the gene coding for resistance to ampicillin, p15, at the origin of replication of the plasmid. The vector is of pACYC type (New England Biolabs, Inc): pACYC177. - Figure 7: Figure 7 shows the sequence of part of the plasmid pDL606 coding for the fusion protein LexAi-87408-cJunZip. The indicated sites make it possible to locate oneself on the map of figure 6.

- Figure 8: Figure 8 shows the map of plasmid pMS404. The restriction sites indicated are unique sites. Tet represents the gene coding for resistance to tetracycline, Ori the origin of replication of the plasmid. The vector is of the pUC type (New England Biolabs, Inc).

- Figure 9: Figure 9 shows the sequence of part of the plasmid pMS404 encoding the fusion protein LexAi-87WT-cFosZip. The indicated sites make it possible to locate oneself on the map of figure 8.

- Figure 10: Figure 10 shows the map of plasmid pL23B (plasmid vector for bank cloning for SU202). The restriction sites indicated are unique sites. Tet represents the gene coding for resistance to tetracycline, Ori the origin of replication of the plasmid. This plasmid carries the DNA binding domain of a LexA mutant, LexA 1-823 (G-> S mutation in position 23). This mutation does not affect recognition specificity (LexA23 recognizes the same sequence as LexAWT) but does affect the strength with which this protein / DNA interaction occurs. In fact, to improve the system, we found that by decreasing the force of interaction between LexA and DNA, we gained in sensitivity. This plasmid is intended for the cloning of a library.

- Figure 11: Figure 11 shows the sequence of part of the plasmid pL23B coding for LexA 1-8723. The bank can be cloned behind LexAi-8723 using one of the following unique sites: Xhol, EcoRI, Sfil. Downstream of these sites, three stop codons (TAA) were inserted into the three reading frames in order to limit the fusion protein. The indicated sites make it possible to locate oneself on the map of figure 10.

The invention will be supplemented by the detailed description and the examples which follow, given by way of illustration and not limitation.

Principle of the double-hybrid system in E. coli The double-hybrid system in E. coli is based on the use of LexA, a bacterial repressor. Wild LexA specifically recognizes and in the form of dimers a palindromic sequence of 16 base pairs, CTGT (N) 8 ACAG, called operator. The 4 external base pairs are highly conserved, while the central motif is more variable. Each LexA monomer is composed of a DNA binding domain which recognizes the CTGT sequence and a dimerization domain essential for binding to DNA since the DNA binding domain lacking a dimerization domain, only fixes DNA very weakly.

The literature describes mutants of .LexA having a new specificity for binding to DNA. The one that was used, LexA408, carries a triple mutation (P40 → A, N41- »S, A42 → S) and recognizes the following sequence CCGT (N) 8 ACGG.

The system according to the present invention is based on the fact that a mutated operator containing 1/2 wild-type operator and 1/2 mutated operator is recognized only by a mixed LexAWT / LexA408 dimer. An indicator gene (lacZ) also designated by reporter gene was put under the control of this operator. Proteins or subdomains of proteins X and Y capable of interacting are then fused to the DNA binding domains, LexA _87WT and LexAι_ 87408. FIG. 1 illustrates this system. LexAι_87WT-X and LexAι_87408-Y are cloned on compatible plasmids and their expression is controlled by an inducible promoter (lacUVS). The indicator strain overexpressing the repressor of the lactose promoter (since lacF \), one can then modulate the expression of LexAι_87WT-X and LexAι_87408-Y by adding variable concentrations of the inducer IPTG.

This system has two main uses:

I) - X and Y can be proteins or protein domains characterized and interacting with each other, they then give the hybrid proteins the ability to bind stably to DNA and thus repress the lacZ gene. Such an interaction can be visualized on an appropriate indicator medium (for example Mac Conkey and described below) and can also be quantified by assaying the lacZ gene product, β-galactosidase according to the method described by Miller JH, 1972 "Experiments in Molecular Genetics, Cold Spring Harbor Lab. Cold Spring Harbor, NY); it is thus possible:

1) quantify the strength of this interaction;

2) to identify the determinants of this interaction (by mutagenesis for example);

3) to identify molecules capable of interfering with this interaction and capable of being used in vivo; small molecules that can be put in the culture medium and pass through the bacterial wall.

II) - X (or Y) is a protein or a protein domain capable of interacting with a protein (s) still unknown that we want to identify. A bank is then grafted onto Y (or X), this bank being cointroduced into the indicator strain with the gene coding for the hybrid protein for which the partner or partners are sought. Any element of the coding bank for a protein or a protein domain interacting with X (or Y) gives rise to a bacterial clone whose color on an appropriate medium is different from those for which there is no interaction. This or these elements are then identified by sequencing.

EXAMPLE 1:

Construction of the indicator strain for the implementation of a double hybrid system in E. coli

This construction requires:

1) cloning of lacZ under the control of a promoter regulated by an operator of the CCGT (N) 8 ACAG type;

2) to pass this construction on the chromose of a bacterial host deleted from its lactose operon, deficient in chromosomal LexA and expressing the lactose repressor, strain JL1434 (Little JW & Hill SA 1985 Deletions within a hinge region of a specifies DNA- binding protein, Proc Natl, Acad Sci, USA £ 2: 2301-2305 and this via a lambda phage (Figure 2).

1. Construction of plasmids carrying the lacZ gene placed under the control of a mutated operator WT / 408

The plasmid pRS415 carrying the lacZ gene, devoid of promoter / operator region (Simons RW, Houman F. & Kleckner N. (1987). Improved single and multicopy lac-based cloning vectors for protein and operon fusions. Gene 5_2: 85-96 .) is open to EcoRI / BamHI restriction sites. Between these two unique restriction sites, the following double-stranded oligonucleotide has been inserted:

S ^, -GAATTC> VATAGGGTTGATCTTTGTTGTCACTGGATGTACCGTACATCCATACAGT ^ ACTCACAGGGC.CACCCTAGG-3 '

(bases in bold = promoter elements, underlined bases = operator) thus creating the plasmid pGC202.

2. Isolation of Recombinant Lac ⁺ Bacteriophages

The strain JL1434 (LexA (Def), Lac "), transformed by the plasmids pGC202, was infected by the bacteriophage λφd (Lac ^~ <Hochschild A. & Ptashne M. (1988). Interaction at a distance between λ repressors disrupts gene activation Nature 226: 353-357) whose genome carries immediately to the left of a lacZ gene devoid of promoter and initiation codon (ATG) of translation, a sequence derived from the vector pBR322 and which is also found to the left of the EcoRI cloning site of the plasmid pRS415 (see figure 2). Thanks to the homologous sequences located on either side of the "promoter / operator / initiator" sequence and following a double recombination event between the plasma and phage DNA, the lacZ gene is found under the control of the promoter / clone operator in pGC202 giving rise to phage λGC 202.

3. Isolation of Lysogenic Bacteria for Recombinant Lac ⁺ Bacteriophages

The purpose of isolating lysogenic bacteria is to obtain stable integration of the recombinant phage into the bacterial chromosome, which is then in the prophage state. This is done from a second infection of the bacteria JL1434 with mini-preparations of the phage previously obtained in order to get rid of the initial plasmid and thus to ensure that the Lac character "* ^" is well carried by the phage.

After infection of the JL1434 strain, the recombinant λ phages begin a lysogenic cycle which leads to stable integration by site-specific recombination of their genomes (and therefore of the lacZ gene controlled by the promoter / operator element described above) in the bacterial chromosome, at the att b site between the gay and organic operons. The lysogenic Lac ^" * ^* SU202 strain was thus isolated. This indicator strain carries the genes for resistance to Kanamycin and Chloramphenicol.

Description of the plasmids carrying the hybrid proteins

The genes coding for the hybrid proteins are carried by two compatible plasmid vectors carrying different origins of replication (pMBl or pl5A) as well as the selection for two different markers (tet ^ or amp ^) thus making it possible to maintain them simultaneously in the same bacterium. host.

Use of the SU202 strain to identify and quantify protein / protein interactions

The indicator strain SU202 was tested with the following chimeric proteins LexAi-87408-cJunZip and LexAi-87WT-cFosZip, they consist of the DNA binding domain (mutant 408 or wild-type (WT)) of the repressor LexA (aa 1 -87) and of the dimerization domain (leucine slide or "leucine zipper") of the oncoproteins cJun (aa 277-315) and cFos (aa 102-200).

The gene coding for LexAi-87408-cJunZiρ is carried by the plasmid pDL606 (see Figures 6 and 7). FIG. 6 represents a map of the entire plasmid accompanied by the unique restriction sites, the vector is a pACYC177 (New England Biolabs, Inc.); it carries resistance to ampicillin and pi 5 A as the origin of replication. FIG. 7 represents the part of the sequence of this plasmid corresponding to the fusion protein, this sequence being identifiable on the general map using the restriction sites.

The gene coding for LexAι_87WT-cFosZip is carried by the plasmid pMS404 (see Figures 8 and 9). FIG. 8 represents a map of the whole of the plasmid accompanied by the unique restriction sites, the vector is a plasmid related to a pBR322 (New England Biolabs, Inc.); it carries resistance to tetracycline and pMB1 as the origin of replication. FIG. 9 represents the part of the sequence of this plasmid corresponding to the fusion protein, this sequence being identifiable on the general map using the restriction sites.

These dimerization domains are such that the Jun Jun homodimers are much less stable than the Jun Fos heterodimers, but nevertheless more than the Fos / Fos homodimers which do not exist in vivo (Sassone- Corsi P., Ransone LJ, Lamph WW & Verma IM (1988). Direct interaction between fos and jun nuclear oncoproteins: role of the "leucine-zipper" domain. Nature (London) 2: 692-695.).

The genes encoding these two fusion proteins are controlled by the inducible promoter lacUV5. The indicator strains overexpressing this repressor (since lacl'ï), make it possible to modulate the expression of LexAi- 87408-cJunZip and LexAi-87WT-cFosZip in the presence of different concentrations of inducer (IPTG).

Qualitative test

The SU 202 strain is transformed either with each of the plasmids carrying the genes coding for the fusion proteins LexAι_87408-cJunZip and LexAi- 87WT-cFosZip or cotransformed with the two plasmids. The transformed cells are left overnight at room temperature in LB (Maniatis et al.) Added respectively with either kanamycin, chloramphenicol, ampicillin or kanamycin, chloramphenicol, tetracycline or the four antibiotics: kanamycin, chloramphenicol, ampicillin and tetracycline, before d 'be spread on McConkey medium containing the same antibiotics. Figure 5 shows, after transplanting onto the same box, the results obtained.

This type of culture medium contains inter alia lactose which is hydrolyzed by β-galactosidase to allolactose. As allolactose is the natural inducer of the lactose promoter, when the bacteria having one or both genes coding for the chimeric proteins are cultured on this medium, these are partially induced. It is estimated that the induction corresponds to that which one observed in LB medium supplemented with 10 ^" 5 M in IPTG. A significant difference was observed in the coloration of the bacterial clones indicating that the indicator strain constructed met the defined criteria, in fact:

- cells transformed by only one of the two plasmids grow red; this coloration means that the β-galactosidase gene is not repressed by the chimeric proteins and that the enzyme produced ferments the lactose in the medium; in the case LexAi-87408-cJunZip, the homodimers which are capable of forming are incapable in vivo of recognizing the mutated operator and therefore of repressing the lacZ gene. Regarding the leucine zipper ("leucine zipper") from Fos, the answer was expected since it only very poorly moderates;

- the bacteria cotransformed by the two plasmids grow white, the β-galactosidase is therefore not produced by the cells; this double transformation allows the cell to simultaneously produce the two chimeric proteins LexAi-87408-cJunZip and .LexAi-87WT-cFosZip; the absence of production of β-galactosidase indicates that the heterodimer LexA i -87408- cJunZip / LexAι_87WT-cFosZip, is unlike homodimers, capable of repressing the lacZ gene.

This indicator strain is therefore very functional, it makes it possible to highlight protein / protein interactions on the simple basis of the color of the clones, it is very specific and makes it possible to discriminate between homo- and heterodimerization.

Quantitative test

The efficiency with which the indicator gene is suppressed can be quantitatively measured by in vitro assay of β-galactosidase activity. This test is carried out under the conditions described by Miller J.H. (1972) in Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

The values of the dosages described represent the average of at least three independent experiments and are carried out in the presence of variable concentrations of IPTG.

Figure 3 shows that:

* whatever the IPTG concentration, the unprocessed strain produces the same number of β-galactosidase units (on average 2,500 units);

* when it is transformed with the plasmid coding for LexAi-87WT-cFosZip and whatever the IPTG concentration (that is to say even by overexpressing the chimeric protein), the number of units does not vary; * when it carries the two chimeric proteins, a very strong repression of the reporter gene is observed (the number of β-galactosidase units drops to 20 for the 10 "and 10 ^" 3M concentrations in IPTG);

* This repression is due to the specific interaction between the leucine zipper parts ("leucine zippers") of the chimeric proteins leading to the formation of active heterodimers, and not to the simple coexistence of two chimeric proteins including the binding domains. to DNA would be able to attach independently to each half site of the mutated operator; in fact, when the strain carries the gene coding for the WT DNA binding domain of LexA devoid of its dimerization domain and that coding for LexAi-87408-cJunZip, no repression of the reporter gene is observed.

We therefore have a test which makes it possible to demonstrate protein / protein interactions which can be used as a screen but also with the possibility of measuring and comparing the strength of these interactions.

Search and identification of mutants affected in these interactions

This strain can also be used to identify the determinants of these interactions either by random mutagenesis or by site-directed mutagenesis. For example, it was hypothesized that the specificity of heterodimerization between the zippers of Jun and Fos was due to the non-homodimerization of the zipper of Fos. This strain made it possible to show that this was not the case since Fos mutants capable of homodimerizing did not lose this specificity (Figure 4).

The cells transformed only by the plasmids coding for the Fos mutants grow red on McConkey medium. Homodimerization is therefore not detected in this strain.

In the case where the bacteria also produce LexAi-87408-cJunZip, they grow white if the Fos mutants are improved in their protein / protein interaction compared to wild Fos, and pink if the Fos mutants are less good. The intensity of the color is always inversely proportional to the capacity for heterodimerization (β-galactosidase assay).

Use of the strain SU202 to identify the partner (s) of a protein or a characterized protein domain

Two plasmids are used:

1- the plasmid carrying the protein or the protein domain characterized for which it is desired to identify the partner (s) fused to the DNA binding domain of LexA408, pDL606: constituting the "hook". 2- the plasmid carrying the bank, pLE23B (see FIGS. 10 and 11). FIG. 10 represents a map of the whole of the plasmid accompanied by the unique restriction sites, the vector is a plasmid related to a pBR322 (New England Biolabs, Inc.) it carries resistance to tetracycline and pMB1 as the origin of replication. FIG. 11 represents the part of the sequence of this plasmid corresponding to the fusion protein, this sequence being identifiable on the general map using the restriction sites.

The two plasmids are simultaneously introduced into the SU202 strain and the transformed bacteria are:

- either directly selected on solid medium LB + antibiotics (kanamycin, chloramphenicol, ampicillin and tetracycline), the positive clones then being revealed by replicating the dishes on McConkey indicator medium + the same antibiotics,

- either selected in LB liquid medium + antibiotics (kanamycin, chloramphenicol, ampicillin and tetracycline) overnight, then revealed by spreading the cultures on McConkey indicator medium + the same antibiotics.

Each clone carries a different library element coding for a different protein or protein domain, when this protein or protein domain is capable of establishing protein / protein interactions with the "hook", it gives rise to a clone white or pink among a carpet of red clones, which represent all the elements of the library coding for proteins or protein domains which do not interact with the hook.

The constitution of the bank takes place via a phage shuttle Elledge S.J. et al, 1991, Proc. Natl. Acad. Sci. USA, .88: 1731-1735. "λYes: multifunctional cDNA expression vector for the isolation of genes by complementation of yeast and E. coli mutations".

EXAMPLE 2:

Construction of the selection system using an indicator gene coding for a toxic protein:

Modification of the operator sequence of the sulA gene so as to make its expression dependent on heterodimeric interactions between proteins or protein domains X and Y fused respectively to the DNA binding domains of .LexA408 and .LexAWT.

This modification consists in introducing a point mutation in the natural operator sequence of sulA:

CTGTACATCCATACAG = WT operator, CCGTACATCCATACAG = mixed operator 408 / WT.

The sulA gene is in vivo repressed by LexA, the introduction of the op408 mutation prevents this repression, the gene product is then synthesized and the bacteria die, making the selection of such a construction impossible. It is therefore necessary to find a means other than suppression to control the production of SULA.

Prior to the establishment of this mutation, two to four STOP codons were introduced into the coding phase of sulA (a STOP codon is a specific sequence of three base pairs serving as a stop signal for the translational machinery). Thus, only an inactive truncated protein is produced and this allows the establishment of the op408 mutation.

When the system is implemented, it suffices to use a bacterial host carrying a suppressor tRNA of this STOP codon (a suppressor tRNA is tRNA capable of recognizing a STOP codon and of allowing further translation). The bacterial host (SU202) which is used being carrying the suppressor tRNA SupE which, in place of the STOP codons (UAG), introduces a Glutamine residue, two to four glutamine residues have been replaced by TAG codons .

The sulA gene nailed into a plasmid was given by S. Cole (Institut Pasteur, Paris). The STOP TAG codons were introduced at the level of Glutamines 42,47 on the one hand, and 42,47 55 and 56 on the other hand, giving rise to constructions called respectively sulA 2STOP and sulA 4STOP. This was carried out by site-directed mutagenesis according to the method described by Eckstein (Nakamaye, K. and Eckstein, F. Nucleic Acids Res. 1986, 14: 9679-9698) and using a kit sold by the company Amersham.

The 408 mutation was then introduced by the same mutagenesis process and identical results were obtained with sulA 2STOP and sulA 4STOP. The constructions obtained were called op408 / opWT sulA 2STOP and op408 / opWT sulA 4STOP.

Implementation of the selection system;

The system was tested in the SU202 strain carrying a suppressor tRNA, SupE. The following results have been observed. i) op408 / opWT sulA 2STOP or op408 / opWT sulA 4STOP cannot be introduced into the strain. Indeed, whatever the transformation method used (calcium chloride or electroporation), no clone was obtained, while an identical plasmid but not carrying the construction gives rise to 200 or 2000 clones per nanogram of plasmid depending on the transformation method used; ii) on the other hand, op408 / opWT sulA 2STOP or op408 / opWT sulA 4STOP can be introduced into the strain SU202 if they are co-transformed with plasmids carrying the genes coding for the fusion proteins LexA _ 7408- cJunZip and .LexAι_g7WT -cFosZip.

In the first experiment, the sulA operation is not busy and the STOP codons are "read" by the suppressor tRNA, SupE. The SULA protein is produced and causes the death of the bacteria.

In the second experiment, the heterodimers LexAι_87408cJunZip / LexA -87 WT-cFosZip are able to bind to the operator of sulA and thus prevent the production of the protein SULA, the bacteria can divide and give rise to clones.

REFERENCES

1. Bartel, P.L., Chien, C.-T. , Sternglanz, R. & Fields, S. (1993) "Using the two-hybrid system to detect protein-protein interactions", In: Cellular Interactions in Development: A practical approach (Oxford University Press, Oxford) pp. 153-179.

2. Chien, C.-T., Bartel, PL, Sternglanz, R. & Fields, S. (1991), "The two-hybrid system: A method to identify and clone genes for proteins that interact with a protein of interest ", Proc. Natl Acad. Salt USA. 88: 9578-9582.

3. Durfee, T., Becherer, K., Chen, P.-L. , Yeh, S. -H. , Yang, Y., Kilburn, A.E., Lee, W.-H. & Elledge, S.J. (1993), "The retinoblastoma protein associâtes with the protein phosphatase type 1 catalytic subunit", Genes and Dev. 2: 555-569.

4. Estojak, J., Brent, R. & Golemis, E.A. (1995), "Correlation of two- hybrid affinity data with in vitro measurements", Mol. Cell Biol. JL5_: 5820-5829.

5. Fields, S. & Song, O.-K. (1989) "A novel genetic system to detect protein-protein interactions", Nature 2fiû: 245-246.

6. Fogh, RH, Ottleben, G., Ruterjans, H., Schnarr, M., Boelens, R. & Kaptein, R. (1994), "Solution structure of the LexA repressor DNA binding domain: determination by 1H NMR spectroscopy ", EMBO J.. 0: 3936-3944.

7. Golemis, E.A. & Brent, R. (1992), "Fused protein domains inhibit DNA binding by LexA", Mol. Cell Biol. 12: 3006-3014.

8. Golemis, E.A. & Brent, R. (1996), "The interaction trap", In: Two- hybrid Systems: A practical approach (Oxford University Press, Oxford) Chapter 4.

9. Golemis, EA, Gyuris, J. & Brent, R. (1996), "Interaction trap / two- hybrid system to identify interacting proteins", In: Current Protocols in Molecular Biology, John Wiley & Sons, New York (Eds : Ausubel, F., Brent, R., Kingston, R., Moore, D., Seidmann, J., Smith, JA & Struhl, K.) Unit 20.1.

10. Gyuris, J., Golemis, E.A., Chertkov, H. & Brent, R. (1993) "Cdil, a human Gl and S phase protein phosphatase that associâtes with Cdk2", Cell 25: 791-803.

11. Hori, R. & Carey, M. (1994) "The role of activators in assembly of RNA polymerase II transcription complexes", Current Opinion in Genetics and Dev. 4: 236-244.

12. Hu, J.C. (1995) "Repressor fusions as a tool to study protein-protein interactions", Structure 2: 431-433.

13. Hurstel, S., Granger-Schnarr, M., Daune, M. & Schnarr, M. (1986), "In vitro binding of LexA repressor to DNA: evidence for the involvement of the amino-terminal domain", EMBO J. 5: 793-798.

14. Le Douarin, B., Pierrat, B., vom Baur, E., Chambon, P. & Losson, R. (1995) "A new version of the two-hybrid assay for detection of protein- protein interactions", Nucl. Acids Res. .22: 876-878.

15. Lloubès, R., Granger-Schnarr, M., Lazdunski, C. & Schnarr, M. (1991), "Interaction of a regulatory protein with a DNA target containing two overlapping binding sites", J. Biol Chem. 2 ^: 2303-2312.

16. Ma, J. & Ptashne, M. (1987a), "Deletion analysis of Gal4 defines two transcrptional activating segments", Cell 48: 847-853.

17. Ma, J. & Ptashne, M. (1987b), "A new class of yeast transcriptional activators", Cell 51: 113-119.

18. Ma, J. & Ptashne, M. (1988), "Converting a eukaryotic transcriptional inhibitor into an activator", Cell .55: 443-446.

19. Maldonado, E. & Reinberg, D. (1995), "News on initiation and elongation of transcription by RNA polymerase II", Current Opinion in Cell Biol 2: 352-361. 20. Marmorstein, R., Carey, M., Ptashne, M. & Harrison, SC (1992), "DNA recognition by Gal4: structure of a protein-DNA complex", Nature 256: 408-414.

21. Oertel-Buchheit, P., Lamerichs, R., Schnarr, M. & Granger-Schnarr, M. (1990), "Genetic analysis of the LexA repressor: isolation and characterization of LexA (Def) mutant proteins", Mol Gen. Broom. 223: 40-48.

22. Phizicky, E.M. & Fields, S. (1995), "Protein-protein interactions: methods for detection and analysis", Microbiol. Rev. 55: 94-123.

23. Porte, D., Oertel-Buchheit, P., Granger-Schnarr, M. & Schnarr, M. (1995), "Fos leucine zipper variants with increased association capacity", J. Biol. Chem. 22Q: 22721-22730.

24. Schmidt-Dôrr, T., Oertel-Buchheit, P., Pernelle, C, Bracco, L., Schnarr, M. & Granger-Schnarr, M. (1991), "Construction, purification and characterization of a hybrid protein comprising the DNA binding domain of the LexA repressor and the Jun leucine zipper ", Biochemistry 3_Q: 9657-9664.

25. Schnarr, M. & Granger-Schnarr, M. (1993), "LexA, the self-cleaving transcriptional repressor of the SOS System", Nucleic Acids and Mol. Biol 7: 170-189 (Eds. F. Eckstein & D.M.J. Lilley)

26. Schnarr, M., Pouyet, J., Granger-Schnarr, M. & Daune, M. (1985), "Large-scale purification, oligomerization equilibria, and specifies interaction of the LexA repressor of Escherichia coli", Biochemistry 24 : 2812-2818.

27. Vojtek, A.B., Hollenberg, S.M. & Cooper, J. A. (1993), "Mammalian Ras interacts directly with the serine / threonine kinase Raf", Cell 24: 205-214.

Claims

1- Use of a DNA sequence comprising:

- a nucleotide sequence coding for an indicator protein (indicator gene),

- a promoter,

a mutated operator, said nucleotide sequence being under the control of the above promoter comprising the above mutated operator containing two sequences and deriving from an operator containing two inverted repeated sequences (palindrome), this operator being such that, in the non-mutated state (wild), it leads to the repression of the transcription of the indicator gene, when regulatory proteins in the form of homodimer bind to each of said inverted repeat sequences via their DNA binding site, this mutated operator being such that the one of its sequences is mutated (mutation 1), while the other is in the wild, or mutated in a silent manner or comprises a different mutation (mutation 2) from the above mutation 1, the transcription of the indicator gene being repressed by regulatory proteins in the form of heterodimers via their DNA binding site, each monomer of these heterodimers being such that it one specifically recognizes the sequence comprising the mutation 1, the other the sequence which is in the wild state, or which is mutated in a silent manner or which comprises a mutation 2, the homodimers capable of being formed from each monomers being incapable of repressing the transcription of the above-mentioned indicator gene,

- to detect or quantify interactions between two proteins or protein domains, or to screen proteins or protein domains exhibiting interactions with a specific protein or protein domain, these interactions being revealed when there is repression of the transcription of the above indicator gene, this repression resulting from the formation of heterodimers comprising:

1) a first monomer containing a first fusion protein containing the DNA binding site of a regulatory protein capable of specifically recognizing the sequence of the mutated operator containing the mutation 1, the above-mentioned binding site being fused to the '' one of the two proteins or one of the two protein domains, involved in the above interactions,

2) a second monomer containing a second fusion protein containing the DNA binding site of a regulatory protein capable of specifically recognizing the sequence not mutated, or mutated so silent or comprising the mutation 2 of the mutated operator, the aforementioned binding site being fused to the other of the two proteins or to the other of the two protein domains, involved in the above interactions.

2. DNA sequence comprising:

- a nucleotide sequence coding for an indicator protein (indicator gene),

- a promoter,

a mutated operator, said nucleotide sequence being under the control of a promoter comprising a mutated operator containing two sequences and deriving from an operator containing two inverted repeat sequences (palindrome), this operator being such that, in the non-mutated state (wild), it leads to the repression of the transcription of the indicator gene, when regulatory proteins in the form of homodimer bind to each of said inverted repeat sequences via their DNA binding site, this mutated operator being such that the one of its sequences is mutated (mutation 1), while the other is in the wild, or mutated in a silent manner or comprises a different mutation (mutation 2) from the above mutation 1, the transcription of the indicator gene being repressed by regulatory proteins in the form of heterodimers via their DNA binding site, each monomer of these heterodimers being such that it recognizes specifically one the sequence comprising the mutation 1, the other the sequence which is in the wild state, or which is mutated in a silent manner or which comprises a mutation 2, the homodimers capable of being formed from each of the monomers being incapable of repressing the transcription of the above-mentioned indicator gene, said mutation 1 being further such that it is specifically recognized by a regulatory protein comprising a site for attachment to the DNA entering into the constitution of a first fusion protein, which binding site is fused with a protein or protein domain capable of interacting with another protein or another protein domain forming part of a second fusion protein containing the DNA binding site of a protein regulatory capable of specifically recognizing the sequence not mutated, or mutated silently or comprising the mutation 2.

3. DNA sequence according to claim 2, characterized in that it comprises a mutated operator containing two sequences and deriving from an operator containing two reverse repeated sequences (palindrome), one of these two sequences being mutated, and the other being in the wild.

4. DNA sequence according to one of claims 2 or 3, characterized in that the mutated operator contains the following sequences:

CCGT and ACAG

spaced apart by a sequence of approximately 3 to approximately 20 nucleotides, and in particular from 8 to 12, and advantageously 8 nucleotides, and in particular characterized in that the mutated operator contains the following sequence:

5 '-CGGAATTC TO ATAGGGTTGATCTTTGTTGTCACTGGATGT ACCGTACATCCATACAGTAACTCACAGGGGC-3'

5. DNA sequence according to one of claims 3 to 4, characterized in that the nucleotide sequence coding for the indicator protein is the lacZ gene in E. coli, coding for β-galactosidase, the mutated operator contains two sequences and derivative of an operator containing two inverted repeat sequences (palindrome), one of these two sequences being mutated, and the other being in the wild state, the mutated sequence preferably being that located downstream of the sequence in the wild, the wild-type sequence being recognized specifically by the wild .LexA protein or by a LexA fragment containing the DNA binding site, in particular a fragment defined by positions 1 to x of LexA, x varying from 69 to 87, and in particular by fragments 1 to 69 or 1 to 87 of LexA, and the mutated sequence being recognized by the mutated LexA protein or by a mutated fragment of LexA containing the DNA binding site, in particular a mutated fragment from Le xA corresponding to that defined by positions 1 to x of LexA, x varying from 69 to 87, and in particular by fragments 1 to 69 or 1 to 87, and the mutations of LexA being advantageously P40 - → A, N41 → S, A42 → S, the dimerization domain of LexA being absent from both LexA or from the fragment of LexA and from mutated LexA or from the mutated fragment from LexA.

6. Complex between

a DNA sequence comprising a nucleotide sequence coding for an indicator protein, said nucleotide sequence being under the control of a promoter comprising a mutated operator containing two sequences and deriving from an operator containing two inverted repeat sequences (palindrome), this operator being such that, in the non-mutated (wild) state, it causes the repression of the transcription of the indicator gene, when regulatory proteins in the form of homodimer bind to each of said inverted repeat sequences via their binding site to the DNA, this mutated operator being such that one of its sequences is mutated (mutation 1), while the other is in the wild, or mutated silently or contains a different mutation (mutation 2) from the said mutation 1, the transcription of the indicator gene being repressed by regulatory proteins in the form of heterodimers via their DNA binding site, each monomer of these heterodimers being such that it specifically recognizes one of the sequences comprising the mutation 1, the other the sequence which is in the wild state, or which is mutated in a silent manner or which comprises a mutation 2, the homodimers capable of being supplied from each of the monomers being incapable of repressing the transcription of the above-mentioned indicator gene,

- and heterodimers whose monomers constituting them (respectively first monomer and second monomer) are involved between them in a protein / protein interaction:

1) the first monomer being such that it contains a first fusion protein containing the site of attachment to the DNA of a regulatory protein capable of specifically recognizing the sequence of the mutated operator containing the mutation 1, the aforementioned site of fixation being fused to one of the two proteins or to one of the two protein domains, involved in the above interactions,

2) the second monomer being such that it contains a second fusion protein containing the DNA binding site of a regulatory protein capable of specifically recognizing the sequence not mutated, or mutated silently or comprising the mutation 2 of the mutated operator, the above-mentioned binding site being fused to the other of the two proteins or to the other of the two protein domains, involved in the above interactions.

7. Complex according to claim 6, characterized in that:

- the DNA sequence is as defined according to any one of claims 3 to 5, and in that

- the heterodimers contain:

1) a monomer containing a first fusion protein containing the DNA binding site of mutated LexA or of a mutated fragment of LexA recognizing the sequence of the mutated operator containing mutation 1, which binding site is fused with a specific protein or fragment determined protein capable of being involved in an interaction according to claim 1, which protein or which protein domain contains, for example, an activation or dimerization domain,

2) a monomer containing a second fusion protein containing the site of attachment to DNA of wild LexA, or carrying a silent mutation or a mutation 2 or a fragment of LexA recognizing the non-mutated sequence of. LexA or mutated in a silent manner or comprising a mutation 2, which binding site is fused with a specific protein or a specific protein fragment, containing for example a protein or a protein domain capable of interacting with the protein or the protein fragment defined in 1) or with the above fields of activation or dimerization.

8. Cell host containing a DNA sequence according to claim 2.

9. Method for detecting and quantifying interactions between two proteins or protein fragments, characterized

- in that one puts in the presence of the nucleotide sequences coding respectively for the first and the second monomer respectively containing a first and a second fusion protein as defined in claim 1, with a DNA sequence according to one any of claims 2 to 5, and

- in that the possible repression of the transcription of the indicator gene as defined in claim 1 is detected, in particular by transfection of a cellular host according to claim 8, with vectors containing, on the one hand, the the above nucleotide sequences coding respectively for the first and the second monomer containing respectively the first and the second fusion protein and, on the other hand, the elements necessary for the expression of the said monomers.

10. Method for identifying a protein interacting with a determined protein, characterized

- in the presence of the following elements:

. a nucleotide sequence coding for a first monomer containing a first fusion protein between, on the one hand, a regulatory protein according to claim 1 and, on the other hand, the determined protein,

. a genomic library, containing in particular DNA fragments, and in particular cDNA, each capable of coding for a protein interacting with the determined protein, each of the different elements of the genomic library being respectively fused to the sequence coding for a protein regulator according to claim 1,

. a DNA sequence according to any one of claims 2 to 5, and

- in that the possible repression of the transcription of the nucleotide sequence coding for an indicator protein according to claim 1 is detected, in particular by transfection of a cellular host according to claim 8, on the one hand, with a vector containing the nucleotide sequence coding for the above first monomer containing the first fusion protein and, on the other hand, a set of vectors containing a nucleotide sequence coding for a second monomer containing a second fusion protein, each nucleotide sequence coding for each second monomer containing the nucleotide sequence coding for the regulatory protein according to claim 1, and downstream thereof one of the DNA elements of the aforementioned library, all the vectors further containing the elements necessary for the expression of the aforesaid monomers.