In vivo detection of proteins interaction based on adenylate cyclase hybrid system

Abstract

The present invention relates to a method to detect the interaction between a target ligand and a moiety of interest using an adenylate cyclase enzyme (AC) and calmodulin (CaM) as interacting partners, said method comprising: i) expressing in a suitable host cell: (a) a low number of molecules of a first chimeric polypeptide containing AC, and (b) a low number of molecules of a second chimeric polypeptide containing CaM, wherein said AC in said first chimeric polypeptide and/or said CaM in said second chimeric polypeptide has decreased affinity for its interacting partner, wherein said AC in said first chimeric polypeptide is fused to a moiety of interest and said CaM in said second chimeric polypeptide is fused to a target ligand, or conversely, and wherein, when said moiety of interest and said target ligand interact, said AC is activated, and ii) detecting the activation of said AC. The present inventors herein show that only one AC/CaM complex per cell is sufficient to confer a selectable trait to the host cell. Unexpectedly, even less than one AC/CaM complex per cell can be sufficient to confer a selectable trait to the host cell. This surprising result confers a very high sensitivity, that is helpful for screening high affinity interactions, such as antigen-antibody interactions. Moreover, the low expression of the chimeric proteins that is achieved in the present invention allows to characterize toxic moieties, what was not possible before.

Claims

1. A method to detect an interaction between a target ligand and a moiety of interest using an adenylate cyclase enzyme (AC) and calmodulin (CaM) as interacting partners, said method comprising: i) expressing, in a suitable host cell: (a) between about 1 and about 10 molecules of a first chimeric polypeptide containing AC, and (b) between about 1 and about 10 molecules of a second chimeric polypeptide containing CaM, wherein said AC in said first chimeric polypeptide has decreased affinity for its interacting partner CaM and/or said CaM in said second chimeric polypeptide has decreased affinity for its interacting partner AC, wherein said AC in said first chimeric polypeptide is fused to a moiety of interest and said CaM in said second chimeric polypeptide is fused to a target ligand, or conversely, and wherein, when said moiety of interest and said target ligand interact, the AC enzyme is activated, ii) detecting the activation of said AC enzyme in said cell.

2. The method of claim 1, wherein said AC in said first chimeric polypeptide is a fragment or a mutated form of AC having 100 to 10 000 fold less affinity for CaM than a wild-type AC enzyme and said CaM in said second chimeric polypeptide is wild-type.

3. The method of claim 1, wherein said AC in said first chimeric polypeptide is wild-type and said CaM in said second chimeric polypeptide is a fragment or a mutated form of CaM having 10 to 10 000 fold less affinity for AC than a wild-type CaM.

4. The method of claim 1, wherein the activation of said AC enzyme generates a detectable signal.

5. The method of claim 1, wherein said detection is performed in a bacterial cell or in an eukaryotic cell, said cell being deficient in endogenous adenylate cyclase.

6. The method of claim 1, wherein said moiety of interest is an antibody, a toxic protein, a membrane protein, a periplasmic protein or a DNA-binding protein.

7. The method of claim 1, further comprising selecting said moiety of interest which is capable of binding said target ligand.

8. A method for screening substances, said method comprising: i) conducting the method of claim 1 in the absence of a substance to be tested, ii) conducting the method of claim 1 in the presence of a substance to be tested, wherein the substance to be tested is capable of stimulating the interaction between said target ligand and said moiety of interest when its presence substantially enhances the activation of the AC enzyme that is measured in its absence, wherein the substance to be tested is capable of inhibiting the interaction between said target ligand and said moiety of interest when its presence substantially reduces the activation of the AC enzyme that is measured in its absence.

9. A kit containing two polynucleotides, each of them being capable of expressing, per transfected cell: (a) between about 1 and about 10 molecules of a first chimeric polypeptide containing an adenylate cyclase enzyme (AC), and (b) between about 1 and about 10 molecules of a second chimeric polypeptide containing calmodulin (CaM), wherein said AC in said first chimeric polypeptide has decreased affinity for its interacting partner CaM and/or said CaM in said second chimeric polypeptide has decreased affinity for its interacting partner AC, wherein said AC in said first chimeric polypeptide is fused to a moiety of interest and said CaM in said second chimeric polypeptide is fused to a target ligand, or conversely.

10. The kit of claim 9, wherein said AC in said first chimeric polypeptide is a fragment or a mutated form of AC having 100 to 10 000 fold less affinity for CaM than a wild-type AC enzyme, and said CaM in said second chimeric polypeptide is wild-type.

11. The kit of claim 9, wherein said AC in said first chimeric polypeptide is wild-type and said CaM in said second chimeric polypeptide is a fragment or a mutated form of CaM having 10 to 10 000 fold less affinity for AC than a wild-type CaM.

12. The kit of claim 9, further containing means for detecting whether AC is activated.

13. A kit for screening substances capable of stimulating or inhibiting the interaction between a target ligand and a moiety of interest, said kit comprising: (a) between about 1 and about 10 molecules of a first chimeric polypeptide containing an adenylate cyclase enzyme (AC), and (b) between about 1 and about 10 molecules of a second chimeric polypeptide containing calmodulin (CaM), wherein said AC in said first chimeric polypeptide has decreased affinity for its interacting partner CaM and/or said CaM in said second chimeric polypeptide has decreased affinity for its interacting partner AC, wherein said AC in said first chimeric polypeptide is fused to a moiety of interest and said CaM in said second chimeric polypeptide is fused to a target ligand, or conversely.

14. A polynucleotide sequence encoding a chimeric polypeptide containing a mutated form of an adenylate cyclase enzyme (AC) having 10 to 10000 fold less affinity for calmodulin (CaM) than a wild-type AC enzyme, and a moiety of interest, wherein said moiety of interest is an antibody, a membrane protein, a toxic protein or a DNA-binding protein.

15. The polynucleotide sequence of claim 14, wherein said mutated form has the sequence SEQ ID NO:3.

16. A recombinant vector containing: a) a polynucleotide sequence encoding either: a mutated form or fragment of an adenylate cyclase enzyme (AC) that has decreased affinity for calmodulin (CaM), or a mutated form or fragment of CaM that has decreased affinity for AC, and b) at least one restriction site enabling to insert a moiety of interest, in frame with said AC or said CaM, said vector being characterized in that transcriptional and translational control sequences upstream of the Open Reading Frame of the polynucleotide sequence a) have been mutated so as to produce only between about 1 and about 10 polypeptides, when transfected in a cell.

17. A polynucleotide sequence encoding a chimeric polypeptide containing a mutated form of calmodulin (CaM) having 10 to 10 000 fold less affinity for an adenylate cyclase enzyme (AC) than a wild-type CaM, and a moiety of interest, wherein said moiety of interest is an antibody, a membrane protein, a toxic protein or a DNA-binding protein.

Description

FIGURE LEGENDS

(1) FIG. 1 discloses the principle of the high sensitive adenylate cyclase hybrid (HSACH) system of the invention. (A) The two boxes represent CaM and the catalytic domain of B. pertussis adenylate cyclase, modified to decrease its affinity for CaM (ACM). When expressed at low level in E. coli cyaA, CaM cannot activate ACM and there is no cAMP synthesis. (B) When ACM and CaM are fused to two interacting proteins, X and Y, they are brought into close proximity and CaM can activate ACM to produce cAMP. Cyclic AMP then binds to the catabolite gene activator protein, (CAP) and the cAMP/CAP complex can stimulate the transcription of the catabolite genes, such as the lactose operon or the maltose regulon.

(2) FIG. 2 discloses a schematic representation of the HSACH plasmids. The colored boxes represent the ORFs of different genes, with the arrow indicating the direction of transcription/translation. The hatched boxes correspond to the multicloning site sequences (MCS) fused to the Cter of ACM or N-ter of CaM. The origins of replication of the plasmids are indicated by shaded boxes. cI corresponds to the thermosensitive repressor cI.sup.857 that strongly repress the promoter at low temperature (30 C. or below), pT7 to the T7 promoter and RBS to the ribosome Binding site. For each plasmid, the relative expression level of the ACM or CaM fusion proteins, expressed as number of molecules per bacterial cell, and estimated by western blot analysis, is given on the right.

(3) FIG. 3 discloses the expression levels of hybrid proteins in vivo.

(4) (A) Western blot analysis of the expression of the ACM-GFP hybrid protein in DHM1. Lines 1-5: 10, 3, 1, 0.3, and 0.1 ng respectively of the purified ACM335-GFP hybrid protein (molecular weight of 73 kDa; 0.1 ng of ACM335-GFP fusion correspond to 810.sup.8 protein molecules) were separated by electrophoresis, electro-transferred to nitrocellulose and detected with a 3D1 monoclonal antibody. Lines 6-9: One OD600 of DHM1 cells, corresponding to 10.sup.9 bacteria, harboring the following combinations of plasmids were probed in parallel by Western blot line 6: pCm-ACGFP/pTr-CaM; line 7: pCm-ACM247-GFP/pTr-3G9A-CaM; line 8: pCm-ACM335-GFP/pK1-3G9A-CaM; line 9: pCm-ACM335-GFP/pK2-3G9A-CaM.

(5) (B) Western blot analysis of the expression of the CaM fusion proteins in DHM1. Lines 10-13: 3, 1, 0.3, and 0.1 ng, respectively of the purified 3G9A-CaM-3D1-zip protein (molecular weight of 42 kDa; 0.1 ng of ACM335-GFP fusion correspond to 1.410.sup.9

(6) molecules) were separated by electrophoresis, electro-transferred to nitrocellulose and

(7) detected with 3D1 monoclonal antibody. Lines 14 and 15: One OD600 of DHM1 cells (corresponding to 10.sup.9 bacteria) harboring the following combinations of plasmids were probed in parallel by WB; line 14: pCm-ACM335-zip/pK2-3G9A-CaM-3D1-zip; line 9: pCm-ACM335-zip/pK1-3G9A-CaM-3D1-zip.

(8) FIG. 4 discloses the in vivo detection of rapamycin-induced ACM-FKBP/FRB-CaM interaction.

(9) DHM1 cells transformed with the indicated plasmids were grown overnight at 30 C. in LB medium containing appropriate antibiotics then diluted 1:100 in LB medium containing plus antibiotics and IPTG (100 M) and incubated until early exponential phase at 30 C. Rapamycin (5 M) was added at time 0 and cells were imaged at the indicated time on a Nikon epi-fluorescence microscope (4A: 0, 1 h, 2 h, 3 h and 4B: 5 h, 7 h). Bottom left of 4B: cell images after 24 hr incubation in LB medium plus antibiotics IPTG and rapamycin (5 M). Bottom right of 4B: cell images of after 1 hr incubation in LB medium plus antibiotics IPTG and 2 mM cAMP.

(10) FIG. 5: Schematic representation of plasmids expressing CaM variants.

(11) The boxes represent the ORFs of the different genes, with the arrow indicating the direction of transcription/translation. CaM is the black bar (with the amino acid residues indicated below), the multicloning site sequences (mcs) is the white arrow, the 3K1K VhH is the hatched arrow, and the beta-lactamase (AmpR) is the light grey arrow. The ColE1 origin of replication is indicated by the dotted bar. The white bar in CaM indicates the position of the 3 Glu residues modified to Lys residues in CaM.sub.VU8.

(12) FIG. 6: Schematic representation of plasmids expressing membrane-associated ACM & CaM hybrid proteins.

(13) The boxes represent the ORFs of the different genes, with the arrow indicating the direction of transcription/translation. ACM3335: light grey bar; leucine zipper motif (Zip): vertical stripped bar; OppB transmembrane segment (TM): black bar; 3G9A VHH: grey bar, CaM: black bar; 3D1 (horizontal stripped bar) in CaM plasmids correspond to the very C-terminal segment of AC containing the 3D1 epitope (inserted during cloning of the leucine zipper motif). The chloramphenicol resistant marker (CmR) and the beta-lactamase (AmpR) are indicated by grey arrows while the p15A and ColE1 origins of replication are indicated by dotted rectangles.

(14) FIG. 7: Diverse topology of ACM & CaM hybrid proteins

(15) (A) The interaction is detected between the cytosolic hybrid proteins ACM335-zip and 3G9A-CaM-Zip. (B) The interaction is not detected for ACM335-TM-Zip (with leucine zipper in the periplasm) and 3G9A-CaM-Zip (with leucine zipper in the cytosol), neither for (C) ACM335-Zip (Zip in cytosol) and 3G9A-CaM-TM-Zip (Zip in periplasm). (D) The membrane associated ACM335-TM-Zip and 3G9A-CaM-TM-Zip hybrids can efficiently interact through the dimerization of their leucine zipper motifs located in the periplasm.

(16) This invention will be described in greater detail with reference to the following examples.

EXAMPLES

(17) 1. Materials and Methods

(18) General Methods

(19) Bacteria were routinely grown at 30 C. in LB broth (0.5% yeast extract, 1% tryptone) containing 0.5% NaCl (Miller J. H. et al., 1992, Cold Spring Harbor Laboratory Press). Unless stated otherwise, antibiotics were added at the following concentrations: ampicillin (100 g/ml), chloramphenicol (30 g/ml), kanamycin (50 g/ml). Standard protocols for molecular cloning, PCR, DNA analysis, transformation and P1 transduction were used (Miller J. H. et al., 1992). The E. coli strain XL1-Blue (Agilent Technologies Stratagene) was used for all routine cloning experiments. PCR primer's synthesis and DNA sequencing were carried out by the company Eurofins MWG Operon (Ebersberg, Germany). The synthetic genes coding for the 3G9A and 3K1K VHH, barnase and barstar were obtained from Geneart (Life-technologies, France). Plasmids encoding FKBP and FRB (J Am Chem Soc. 2005, 127: 4715-4721) were kindly provided by Drs Yves Jacob. Plasmids coding for AC wild-type, ACM247, ACM335 and CaM were described in Ladant et al. (J. Biol. Chem, 1992, 267(4):2244-2250) and Vougier et al. (J. Biol. Chem., 2004, 279(29):30210-30218). pCm-AC and pCm-ACM are derived from the pT25 plasmid (Karimova G. et al, PNAS. 1998, 95(10):5752-5756) upon removal of transcriptional and translational sequences in front of AC(M), and appending a multicloning at the 3 end of AC(M) open-reading frame. The genes encoding GFP, the FKBP polypeptide, barnase or the GCN4 leucine zipper (Zip) were inserted in frame into the multicloning of pCm-ACM (or pCm-AC). The pTr-CaM plasmid is a derivative of pDLTCaM41 (Vougier et al. J. Biol. Chem., 2004, 279(29):30210-30218) containing a 6His tag and a multicloning site at its N-terminus. The genes encoding the 3G9A or 3K1K V.sub.HH, barstar, or the FRB polypeptide were inserted in frame into the multicloning of pTr-CaM. The ACM-GFP fusions were expressed in E. coli after subcloning of the ACM-GFP genes into appropriate sites of the pTRAC expression plasmid, and purified as described in Vougier et al. (J. Biol. Chem., 2004, 279(29):30210-30218). The 3G9A-CaM-FLAG and 3G9A-CaM-3D1-Zip proteins were expressed in E. coli after subcloning into the pTr-CaM plasmid derivative and purified as described in Vougier et al. (J. Biol. Chem., 2004, 279(29):30210-30218). Protein purity was monitored by SDS-PAGE analysis and the protein concentration was determined by absorption at 280 using molecular extinction coefficients calculated form the amino acid sequence. For Western Blot analysis, the proteins were separated by 10% SDS-polyacrylamide gel electrophoresis, electrotransferred onto a polyvinylidene difluoride membrane (Millipore), incubated with the anti-cyaA monoclonal antibody 3D1 (Santa Cruz Biotechnology) revealed with a horseradish peroxidase-conjugated mouse secondary antiserum (Amersham Bio-sciences) and detected by enhanced chemiluminescence (ECL-Plus kit; Amersham Biosciences).

(20) HSACH Complementation Assays

(21) HSACH complementation assays were carried out in the E. coli cya strain DHM1 (Karimova G. et al, J. Bacteriol 2005, 187(7):2233-2243). After transformation with appropriate plasmids, cells were plated on LB agar containing X-Gal, IPTG plus antibiotics and incubated at 30 C. for 24-36 hours. Efficiency of interaction between hybrid proteins was quantified by measuring -galactosidase (-Gal) activity in liquid cultures in 96-well format assay (Karimova G. et al, J. Bacteriol. 2012, 194(20):5576-5588). For each set of transformation, the -Gal assay was performed on eight overnight cultures that were grown at 30 C. in 300 L LB broth in the presence of 0.5 mM IPTG and appropriate antibiotics in a 96-well microtiter plate (2.2 ml 96-well storage plate, Thermo Fisher Scientific). For screening experiments, the DHM1 cells, after electroporation with appropriate plasmids, were incubated in LB broth at 30 C. for 90 min, then washed several times with M63 synthetic medium, and spread on M63 solid medium supplemented with maltose (0.2%), 5-bromo-4-chloro-3-indolyl-b-D-galactoside (XGal, 40 g/ml), isopropyl--D-galactopyronoside (IPTG, 0.5 mM), kanamycin (25 g/ml) and chloramphenicol (20 ng/ml). Plates were incubated at 30 C. for 2-3 days until appearance of blue cya.sup.+ (Mal.sup.+ and Lac.sup.+) colonies. Cyclic AMP was measured on boiled liquid culture with an ELISA assay as previously described (Karimova G. et al, PNAS. 1998, 95(10):5752-5756).

(22) For fluorescence microscopy studies, overnight cultures of DHM1 cells harboring appropriate plasmids were diluted 1:100 in LB medium containing IPTG (100 M), and appropriate antibiotics and incubated until early exponential phase at 30 C. Rapamycin (5 M) was added to induce association of ACM-FKBP with FRB-CaM. Images of living, nonfixed cells were acquired on a Nikon epi-fluorescence microscope Eclipse 801 equipped with a 100 Plan-Apo oil immersion objective and a 100 W mercury lamp. Images were captured with a 5-megapixel colour CCD DS-SMc device camera and processed using Adobe Photoshop software (Karimova G. et al., J. Bacteriol 2012, 194(20):5576-5588).

(23) 2. Results

(24) 2.1. Design of a High Sensitive Adenylate Cyclase Hybrid (HSACH) System

(25) In the novel system of the invention, two proteins of interest are separately fused to AC and CaM and co-expressed in an E. coli cya strain (FIG. 1). To render the AC activation dependent upon the association of the hybrid proteins, AC was engineered to disable its high affinity for CaM (concentration for half-maximal activation, K.sub.1/20.1 nM in the presence of calcium) by introducing appropriate mutations, so that when both the modified AC and CaM would be expressed alone at low level in a E. coli cya strain, they could not spontaneously interact. Among the various modifications known to decrease CaM affinity, two were chosen, ACM247 and ACM335, consisting in two-amino acids insertions within the T18 moiety of AC, which were previously shown to decrease CaM affinity by more than a 5,000 and 500 fold, respectively (Ladant et al., J. Biol. Chem. 267(4):2244-2250). These two-codon insertion mutations (Leu-Gln and Cys-Ser for ACM247 and ACM335 respectively) are expected to be less prone to reversion toward a wild-type, high-affinity phenotype than a single point mutation replacing a critical residue involved in CaM-binding.

(26) As a model system of high affinity interacting proteins, the Inventors used an antigen-binding fragment (V.sub.HH #3G9A) from a camelidae heavy chain antibody that interacts with high affinity (K.sub.D0.5 nM) with the green fluorescent protein (GFP) as reported by Kirchhofer et al (Nat. Struct. Mol Biol. 2010; 17(1):133-138), who determined its structure in complex with GFP.

(27) 2.2. In Vivo Detection of Active Hybrid AC/CaM Complexes

(28) Different expression systems were explored in order to express AC in E. coli at the minimal possible level yet enough to confer a selectable cya.sup.+ phenotype to an E. coli cya strain. Among them, the Inventors selected an expression vector (pCm-AC) derived from the low-copy plasmid pACYC184 (Chloramphenicol resistant), in which all transcriptional and translational control sequences upstream of the AC open reading frame (residues 1 to 399 from B. pertussis CyaA) were deleted (FIG. 2). Expression of the wild-type AC as a fusion with GFP from this vector (pCm-AC-GFP) was able to restore a cya+ phenotype (as assessed by blue colonies on LB X-gal, -galactosidase assays in liquid cultures and cAMP measurements on total bacterial extract; Table 1) to the E. coli cya strain DHM1 provided the host cells also harbored a compatible plasmid expressing CaM, pTr-CaM (ColE1 origin; ampicillin resistant). When the ACM247 variant was similarly fused to GFP (encoded by plasmid pCm-ACM247-GFP) and co-expressed with CaM in DHM1, it failed to restore a cya.sup.+ phenotype. However, when the plasmid pCm-ACM247-GFP was co-transformed in DHM1 with a pTr-CaM derivative (pTr-3G9A-CaM, FIG. 2) that expresses CaM as a fusion with the 3G9A VHH, the transformants exhibited a cya.sup.+ phenotype, although the cAMP and -galactosidase expression levels were lower than in DHM1 co-expressing the wild-type AC-GFP and CaM (Table 1). It was concluded that 3G9A-CaM, but not CaM alone, could activate in vivo the ACM247-GFP variant as a result of the specific interaction between the 3G9A V.sub.HH and the GFP moieties.

(29) The second AC variant, ACM335, similarly expressed as a GFP fusion (from plasmid pCm-ACM335-GFP) also conferred a robust cya.sup.+ phenotype to DHM1 when co-transformed with pTr-3G9A-CaM as expected, but also with plasmid pTr-CaM that expresses CaM alone (i.e. not fused to the 3G9A VHH). It was hypothesized that the CaM expression level achieved with the pTr-CaM plasmid via the residual transcription from the cI857-repressed promoter used to drive CaM expression (FIG. 2), was high enough to spontaneously activate the ACM335 variant that has a higher affinity for CaM than ACM247 (Ladant D. et al, 1992, J. Biol. Chem. 267(4):2244-2250). The Inventors therefore tested alternative expression systems in order to reduce the level of the CaM-fusion. Two plasmids, pK1-3G9A-CaM and pK2-3G9A-CaM, were constructed, both harboring a ColE1 origin and a kanamycin resistant gene, in which the 3G9A-CaM fusion was expressed under the control of a T7 promoter with or without an RBS sequence, respectively (FIG. 2). The synthetic CaM gene in these plasmids was also fused at its C-terminus to a FLAG epitope. These plasmids, when co-transformed with pCm-AC-GFP conferred a robust Cya.sup.+ phenotype to DHM1 cells, as revealed by cAMP production and -galactosidase expression (Table 1). DHM1 cells cotransformed with pCm-ACM335-GFP and either pK1-3G9A-CaM or pK2-3G9A-CaM, also synthesized high levels of cAMP and expressed high -galactosidase activity (although the cAMP and -galactosidase levels were significantly lower with pK2 as compared to pK1). Co-transformation of either pK1-3G9A-CaM or pK2-3G9A-CaM into DHM1 together with pCm-ACM247-GFP yielded only a barely detectable cya.sup.+ phenotype, likely due to the lower specific activity of the ACM247 variant as compared to ACM335 (Ladant D. et al, 1992, J. Biol. Chem. 267(4):2244-2250), while a cya.sup. phenotype was obtained upon co-transformation with pCm-ACM247 as expected. Moreover, when pK1-3G9A-CaM or pK2-3G9A-CaM were co-transformed into DHM1 with a plasmid expressing ACM335 as a fusion to the FK506-binding protein, FKBP (pCm-ACM335-FKBP), the cells exhibited a cya phenotype (data not shown). Hence the 3G9A-CaM fusion produced from the pK1 or pK2-3G9A-CaM plasmids could efficiently activate the ACM335-GFP hybrid but not the ACM335-FKBP one (Table 1).

(30) Additional pK1 and pK2 derivatives (pK1 and pK2-FRBCaM, respectively) were constructed to express CaM as a fusion with the FKBP-rapamycin binding domain, FRB, that binds with high affinity to FKBP in the presence of rapamycine (Banaszynski L. A., 2005, J. Am. Chem. Soc., 127(13):4715-4721). As shown in Table 1, the FRB-CaM fusion was able to specifically activate in vivo the ACM335-FKBP only when the cells were grown in the presence of rapamycine (Table 1).

(31) Altogether these data indicate that the CaM fusions produced by pK1 or pK2 plasmids could activate the ACM335 hybrids in a highly selective manner, dictated by the specific association between the protein modules appended to CaM and ACM335.

(32) TABLE-US-00001 TABLE 1 ACM-CaM complementation in DHM1 strain cAMP Phenotype gal nmol/mg dry CaM plasmids AC plasmids on LB/Xgal Rel. Units weigth 1 pDL1312 pCm-AC-GFP White 1 <0.1 (- no CaM) 2 pTr-CaM pCm-AC-GFP Blue 106 >250 3 pCm-ACM247-GFP White 1 <0.1 4 pCm-ACM335-GFP Blue 80 NT 5 pTr-3G9A-CaM pCm-AC-GFP Blue >100 >250 6 pCm-ACM247-GFP Blue 26 >50 7 pCm-ACM335-GFP Blue 78 >60 1 pK1-3G9A-CaM pCm-AC-GFP Blue 102 230 25 2 pCm-ACM247 White 1 <0.1 3 pCm-ACM247-GFP Pale Blue 4 1.4 0.5 4 pCm-ACM335-GFP Blue 57 60 10 5 pCm-ACM335-FKBP White 2 <0.1 6 pK2-3G9A-CaM pCm-AC-GFP Blue 76 160 15 7 pCm-ACM247 White 1 <0.1 8 pCm-ACM247-GFP Pale Blue 4 0.25 0.1 9 pCm-ACM335-GFP Blue 23 11 2 10 pCm-ACM335-FKBP White 2 <0.1 11 pK1-FRB-CaM pCm-ACM335-GFP White 1 NT 12 pCm-ACM335-FKBP White 2 NT 13 pCm-ACM335-FKBP NT 48 NT (+Rapa) 14 pK2-FRB-CaM pCm-ACM335-GFP White 1 NT 15 pCm-ACM335-FKBP White 2 NT 16 pCm-ACM335-FKBP NT 40 NT (+Rapa)
2.3. Expression Levels of Hybrid Proteins.

(33) The Inventors then attempted to determine the level of expression of the ACM and CaM hybrid proteins in the bacterial cells by western blot (WB). The ACM proteins could be detected with a monoclonal antibody (Mab) 3D1 that recognizes an epitope located between residues 373 and 400 of AC (Lee S J. et al. 1999, Infect. Immun. 67(5): 2090-2095). To quantify the amount of protein per cells the ACM335-GFP hybrid protein was over-expressed in E. coli (Material and Methods) and purified to homogeneity to serve as a standard. The Mab 3D1 was able to detect 0.1 ng of ACM335-GFP fusion (FIG. 3), which correspond to about 810.sup.8 protein molecules (molecular weight of 73 kDa). One OD600 of bacterial extracts, corresponding to 10.sup.9 bacteria, of DHM1 cells harboring different combinations of plasmids were probed in parallel by WB. As shown in FIG. 3, no signal could be detected by WB in these extracts, indicating that the bacteria harboring the pCm-ACM335-GFP expressed, as a mean, less than one ACM335-GFP (or AC-GFP or ACM247-GFP) molecule per cell (i.e. below 0.1 ng of fusion per 10.sup.9 cells).

(34) The Inventors similarly aimed to quantify the level of expression of the CaM fusions achieved with the pK1 and pK2 plasmids by WB with an anti-FLAG Mab. The 3G9A-CaM protein (with an appended FLAG tag) was overexpressed in E. coli and purified to homogenenity to serve as a standard in WB calibration. Unfortunately, the anti-FLAG Mab could not detect less than 10 ng of the 3G9A-CaM fusion protein (which corresponds to 17010.sup.9 molecules of this 35 kDa polypeptide). No WB signals were detected in the bacterial extracts of 10.sup.9 cells harboring pK1 or pK2-3G9A-CaM (data not shown), indicating an upper limit of about 200 3G9A-CaM molecules per bacterial cell.

(35) To determine more precisely the level of 3G9A-CaM, the Inventors replaced the Flag epitope by the AC 3D1 epitope. For this, the Inventors fused in frame to the Cter of CaM, the AC residues 373 to 400 followed by a leucine zipper motif (from GCN4) and classically used as positive interaction control in standard BACTH (Karimova G. et al, 1998, PNAS 95(10):5752-5756). The Inventors checked that the 3G9A-CaM-3D1-zip fusion protein, expressed from pK1 or pK2 plasmids, could interact specifically both with ACM335-GFP via the 3G9A VHH as well as with an ACM335-zip hybrid (encoded by pCm-ACM335-zip plasmid) via their leucine zipper motifs. About 0.1 ng of the 3G9A-CaM-3D1-zip fusion could be detected by Mab 3D1 in WB (FIG. 3), corresponding to about 1410.sup.8 protein molecules (molecular weight of 42 kDa). A similar signal was detected in extracts of 10.sup.9 DHM1 cells harboring pK2-3G9A-CaM-3D1-zip indicating a ratio of about 1-2 molecules per cell while DHM1 cells harboring pK1-3G9A-CaM-3D1-zip expressed about 5-10 CaM hybrids per bacteria.

(36) All together these results highlight the exquisite sensitivity of the AC/CaM signaling cascade that could detect in E. coli interactions between hybrid proteins expressed at a minimal level of few molecules per cell in the case of the CaM fusions, or even and more strikingly, at less than one molecule per cellas a meanin the case of the ACM fusions.

(37) 2.4. Characterization of Interactions Involving Toxic Proteins

(38) To further establish that the ACM fusions are expressed in vivo at an extremely low level, the Inventors explored the interaction of the toxic enzyme barnase, a ribonuclease secreted by the bacterium Bacillus amyloliquefaciens, with barstar a specific inhibitor that binds with high affinity to barnase and blocks its RNAse activity. Barnase is lethal to the cell when expressed without its inhibitor Barstar (Frisch C. et al, J. Mol Biol., 1997, 267(3):696-706; Jucovic M. et al, 1996, PNAS, 93(6):2343-2347). Synthetic Barnase and Barstar genes were cloned into the pCm-ACM335 and pK1-CaM plasmids respectively. DHM1 cells cotransformed with the two resulting plasmids pCm-ACM335-Barnase and pK1-Barstar-CaM exhibited a strong cya.sup.+ phenotype, while control co-transformations with various pCm-ACM335 and pK1-CaM derivatives demonstrated the selectivity of interaction between the Barnase and Barstar modules (not shown). Noticeably, the pCm-ACM335-Barnase plasmid could be transformed into DHM1 cells that did not expressed any Barstar fusions (i.e. harboring pK1-FRB-CaM, PK1-3G9A-CaM or no additional plasmid) and the transformed cells did not exhibit any detectable growth problem. This confirms that the ACM335-Barnase hybrid protein was expressed at a level low enough not to affect the bacterial physiology, yet sufficient to allow detection of Barnase-Barstar interaction. Hence the HSACH system may be useful to characterize the interaction properties of many toxic proteins in bacteria, including a wide variety of toxin-antitoxin systems.

(39) 2.5. Direct Screening of Antigen-Antibody Interactions in Bacteria.

(40) The remarkable ability of the HSACH system to detect down to a single complex of hybrid proteins per host cell, suggests that it should be particularly adapted for direct in vivo screening of high affinity antibodies or other binders to antigens of interest. The Inventors showed above the successful detection of the specific association of the 3G9A camelidae V.sub.HH with GFP. Another camelidae V.sub.HH, 3K1K (Kirchhofer A. et al, 2010, Nat. Struct. Mol. Biol. 17(1):133-138), also exhibiting a high affinity for GFP, was similarly tested. Again, the Inventors found that the 3K1K-CaM fusion selectively activated in vivo the ACM335-GFP fusion but not other ACM335 hybrid proteins (i.e. fused to FKBP or Zip moiety) (data not shown).

(41) To demonstrate that the HSACH system could be applied for in vivo selection of high affinity binders, the Inventors randomly PCR-mutagenized the 3G9A V.sub.HH at amino acid residue 107, located at the interface with GFP in the crystal structure, and screened for variants that did not interact anymore with GFP. The Inventors picked up one 3G9A variant (variant 1, harboring a Tyr to Asn modification that fully abolished the interaction with GFP (as revealed by the cya.sup. phenotype of DHM1/pCm-ACM335-GFP/pK2-3G9A.sub.Y107N-CaM transformants) for a second round of mutagenesis. The pK2-3G9A.sub.Y107N-CaM plasmid was then randomly mutagenized at the three positions 105, 106 or 107 of the V.sub.HH. The mutagenized plasmid pool was then co-transformed into DHM1/pCm-ACM335-GFP and plated on an indicator plate (LB-Xgal). Both blue (i.e. lac+/cya+ bacteria expressing interacting hybrid proteins) and white (lac/cya bacteria expressing non-interacting hybrids) colonies were randomly picked for plasmid purification and the pK2 plasmids in each clone were sequenced. All blue colonies contain an aromatic residue (Tyr or Phe, and in one case, Trp) at position 107, and a glycine residue at position 106, indicating that these residues were mostly critical for interaction of 3G9A with GFP. In contrast, a variety of residues could be found in cya+ clones at codon 105, indicating that this position is less important for interaction, in good agreement with the 3D structure of the GFP-3G9A complex (not shown). The mutagenized plasmid pool was also co-transformed into DHM1/pCm-ACM335-GFP and plated on a selective medium, that is a minimal medium that has maltose as a unique carbon source: as the maltose regulon is under a very stringent cAMP/CAP control, only cya+ bacteria can grow on this medium (Xgal and IPTG were also added to better visualize the cya+ colonies). The Inventors randomly picked up a number of cya+ colonies that grew on this selective medium and sequenced the 3G9A-CaM fusions. All colonies had either a Tyr or Phe residue at position 107, and all had a glycine residue at position 106, confirming the importance of these residues for the 3G9A/GFP interaction. As a control, the mutagenized plasmid pool was also co-transformed into DHM1/pCm-ACM335-FKBP and plated on the same selective medium. No colonies could be detected in these conditions highlighting the stringency of the in vivo selection.

(42) All together these experiments indicate that the HSACH system could allow for direct in vivo selection of bacteria expressing V.sub.HH specific for a given protein target.

(43) 2.6. Visualization of Active Hybrid AC/CaM Complexes In Vivo

(44) Finally, to further document the counter intuitive observation that cells could express less than one ACM/CaM active complex and yet display a selectable cya+ phenotype, the Inventors attempted to visualize the complementation between hybrid proteins in vivo, on individual bacteria through a fluorescent reporter.

(45) For this, the gene coding for the ZsGreen fluorescent protein (Clonetech Laboratories) was placed under the transcriptional control of a cAMP/CAP dependent lac promoter and inserted on the pK1-FRB-CaM plasmid. The resulting plasmid pK1-FRB-CaM-placZs was co-transformed into DHM1 cells with either pCm-ACM335-FKBP or pCm-ACM335-GFP. The co-transformants displayed a low background fluorescence signal when grown in LB medium but were highly fluorescent when grown in LB medium supplemented with cAMP, which can diffuse inside the cells to stimulate cAMP-dependent gene transcription. As expected, DHM1 cells cotransformed with pK1-FRB-CaM-placZs and pCm-ACM335-FKBP also displayed high fluorescence when grown in LB medium supplemented with rapamycin, while those co-transformed with pK1-FRB-CaM-placZs and pCm-ACM335-GFP remained nonfluorescent (data not shown). This indicated that the placZsgreen fluorescent reporter could detect in vivo the rapamycin-induced interaction between ACM335-FKBP and FRB-CaM and the resulting activation of AC enzymatic activity.

(46) The Inventors then explored the kinetics of rapamycin-induced activation of ACM335-FKBP and FRB-CaM grown in vivo in LB medium. As shown in FIG. 4, within the first hours of growth in the presence of rapamycin, only about 20-25% of the cells became fluorescent. This fraction progressively increased to more than 90% of total population after an overnight culture. In contrast, all the cells became highly fluorescent within 0.5-1 hr after addition of cAMP in the medium, and as expected no fluorescent cells were detected when rapamycin was added to DHM1/pK1-FRB-CaMplacZs/pCm-ACM335-GFP (FIG. 4).

(47) These results can be interpreted as follows: in DHM1 harboring the pCm-ACM335-FKBP, the ACM335-FKBP fusion protein is stochastically expressed and present only in about 20-25% of the cells. Upon addition of rapamycin, the hybrid enzyme interacts with the co-expressed FRB-CaM fusion and these bacteria start to express the ZsGreen fluorescent reporter. The other cells, not expressing ACM335-FKBP, obviously cannot produce cAMP and thus remain non-fluorescent. However, upon prolonged exposure to rapamycin, the progeny of these ZsGreen-cells will progressively become fluorescent as a result of stochastic expression of the ACM335-FKBP that should occur statistically once every 2-3 cell cycles (if present statistically in 20-25% of the cells). The progeny of the ZsGreen.sup.+ cells should retain the fluorescence of the mother cell due to the diffusion of the ZsGreen protein as well as that of the cAMP/CAP complex, which can trigger de novo ZsGreen expression in daughter cells. In addition, one of the 2 daughter cells inherits the active ACM/CaM complex and thus continues to produce high amounts of cAMP.

(48) All together these data support the view that the exquisite sensitivity of the HSACH system allows colonies to be selected for their cya+ phenotype, even though at any given time only a fraction of all bacteria may harbor an active ACM/CaM hybrid.

(49) 2.7. CaM variant and fragment as AC activator in vivo

(50) The inventors also explored the possibility of using a wild-type AC in the first chimeric polypeptide and a modified CaM with decreased affinity for AC as the second chimeric polypeptide. The VU-8 calmodulin of SEQ ID NO:15 (in which 3 glutamic acid residues at position 82 to 84 of CaM are substituted with 3 lysine residues) was used as mutated CaM. This mutated CaM, CaM.sub.VU8, has a 1000-fold lower affinity for wild-type AC than the native CaM (Haiech, et al. J. Biol. Chem. 1988 (263, 4259)).

(51) CaM.sub.VU8 was expressed from plasmid pCaM.sub.VU8 (harboring a ColE1 origin and ampicillin resistant gene), under the control of a T7 promoter and an RBS sequence, respectively. A multicloning site and an HA tag was also appended to the C-terminus. As positive control for interaction, the camelidae V.sub.HH 3K1K was tested. As explain above, this chain interacts with high affinity (K.sub.D0.5 nM) with GFP. The 3K1K gene was cloned into the MCS of pCaM.sub.VU8 to yield plasmid pCaM.sub.VU8-3K1K (see SEQ ID NO:16 and FIG. 5).

(52) The inventors also analyzed the possibility of using a fragment of CaM instead of the full-length CaM protein as a potential activating partner of AC. Here, they tested a fragment of CaM, CaM.sub.Cter, encompassing residues 77 to 148 of mammalian calmodulin (illustrated on SEQ ID NO:8) and corresponding to the C-terminal half of CaM. This domain is able to activate wild-type AC with a 10-100 fold less affinity than that of full-length CaM (Wolff et al., Biochemistry, 1986; 25:7950).

(53) CaM.sub.Cter was expressed from plasmid pCaM.sub.Cter (harboring a ColE1 origin and ampicillin resistant gene), under the control of a T7 promoter and an RBS sequence, respectively and with a multicloning site and a HA tag at its C-terminus. The 3K1K V.sub.HH gene was then cloned into the MCS of CaM.sub.Cter to yield plasmid pCaM.sub.Cter-3K1K (see FIG. 5).

(54) DHM1 bacteria were transformed with the indicated plasmids and plated on LB agar supplemented with appropriate antibiotics, IPTG, and X-gal and grown at 30 C. for 36 hrs in the presence of 0.5 mM IPTG plus appropriate antibiotics. The -galactosidase activities (expressed in relative units) were determined on liquid cultures grown overnight at 30 C. in LB plus appropriate antibiotics and IPTG. For each transformant, the values are the average obtained on eight independent colonies (SD when not indicated were below 20%).

(55) TABLE-US-00002 TABLE 2 Modified or truncated CaM as a activating partners of AC in in vivo interaction assays. -galactosidase (relative units) pCaM.sub.VU8 pCaM.sub.VU8-3K1K pCaM.sub.Cter-3K1K pCm-AC 6 5 29 (5) pCm-AC-GFP 5 111 (10) 187 (13) pCm-ACM335 7 5 4 pCm-ACM335-GFP 6 5 141 (15) pCm-ACM335-FKBP 6 5 4
Results:

(56) As shown in Table 2, the CaM.sub.VU8 was unable to activate in vivo AC or the AC-GFP fusion, while the CaM.sub.VU8-3K1K fusion was able to efficiently activate in vivo the AC-GFP fusion but not AC alone. Altogether these data indicate that in vivo, activation of the wild-type AC by the modified CaM.sub.VU8 only occurs when the two proteins are fused to interacting modules (here, GFP and 3K1K). Interestingly the CaM.sub.VU8-3K1K fusion was unable to activate in vivo the ACM335-GFP fusion, likely because of the too-low affinity of the modified CaM.sub.VU8 for the ACM335 variant.

(57) As shown in Table 2, the CaM.sub.Cter-3K1K fusion expressed from pCaM.sub.Cter-3K1K was able to activate in vivo the AC alone. This indicates that the CaM.sub.Cter moiety is able by itself to stimulate AC in bacteria, as shown previously with the full-length CaM.

(58) As shown in Table 2, the CaM.sub.Cter-3K1K fusion expressed from pCaM.sub.Cter-3K1K was able to activate in vivo the AC-GFP fusion but also the AC alone. This indicates that, as found with the full-length CaM, the CaM.sub.Cter moiety is enough to stimulate AC in bacteria in these conditions. More importantly the CaM.sub.Cter-3K1K fusion efficiently activated the ACM335-GFP fusion but not ACM335 or the ACM335-FKBP fusion. This indicates that CaM.sub.Cter fragment can be used together with the ACM335 variant for high sensitive detection of interactions in vivo in bacteria. As compared to full-length CaM, CaM.sub.Cter is only 70 amino-acid long and thus a smaller fusion moiety.

(59) 2.8. In Vivo Detection of Interaction Between Membrane Proteins or Occurring in the Periplasm

(60) The Inventors finally explored the capacity of their highly sensitive approach to detect interaction between membrane proteins or between periplasmic proteins. For this, they constructed:

(61) i) the plasmid pCm-ACM335-TM-zip of SEQ ID NO:17 that expresses ACM335 fused to a short peptide encoding the first trans-membrane (TM) segment of the E. coli OppB protein, an oligopeptide transporter subunit (Ouellette et al., 2014, Env. Micro. Rep., 6:259) and the leucine zipper dimerization domain of GAL4.

(62) Plasmid pCm-ACM335-TM-zip was constructed by subcloning between the BamHI and XhoI sites of pCm-ACM335-GFP (SEQ ID NO:9), a PCR-amplified DNA fragment (with appropriate primers introducing a BglII sitecompatible with BamHIand a XhoI site) that codes for the linker region, the E. coli OppB first TM segment and the GAL4 leucine zipper (zip) domain from plasmid pUT18C-TM-zip (Ouellette et al 2014, Env. Micro. Rep., 6:259).

(63) ii) the plasmid pAR-3G9A-CaM-TM-Zip of SEQ ID NO: 18 that expresses CaM fused at its N-terminus to the 3G9A V.sub.HH and at its C-terminus to the OppB TM segment and the GAL4 leucine zipper (see FIG. 6).

(64) Plasmid pAR-3G9A-CaM-TM-Zip is a derivative of pUC18 (without promoter but with a RBS sequence) that expresses CaM fused at its N-terminus to the 3G9A V.sub.HH and at its C-terminus to the OppB TM segment and the GAL4 leucine zipper (see FIG. 6).

(65) Control plasmids pAR-3G9A-CaM expressing the 3G9A-CaM fusion and pAR-3G9A-CaM-Zip expressing the 3G9A-CaM-Zip fusion (i.e., with the leucine zipper but without the TM segment) were also constructed and tested for complementation in two-hybrid assay as above (see Table 3).

(66) In vivo complementation assays between various AC and CaM fusions. DHM1 bacteria were transformed with the indicated plasmids and plated on LB agar supplemented with appropriate antibiotics, IPTG, and X-gal and grown at 30 C. for 36 hrs in the presence of 0.5 mM IPTG plus appropriate antibiotics. The -galactosidase activities (expressed in relative units) were determined on liquid cultures grown overnight at 30 C. in LB plus appropriate antibiotics and IPTG. For each transformant, the values are the average obtained on six to eight independent colonies (SD when not indicated were below 20%).

(67) TABLE-US-00003 TABLE 3 Interaction assays of membrane associated proteins -galactosidase pCm- pCm- (relative units) ACM335-Zip ACM335-TM-Zip pAR-3G9A-CaM 6 7 pAR-3G9A-CaM-Zip 160 (10) 9 pAR-3G9A-CaM-TM-Zip 7 157 (17)

(68) As shown in Table 3 and illustrated in FIG. 7D, the membrane associated ACM335-TM-Zip and 3G9A-CaM-TM-Zip hybrids can efficiently interact through the dimerization of their leucine zipper motifs located in the periplasm. The interaction was similar to that detected between the cytosolic hybrid proteins ACM335-zip and 3G9A-CaM-Zip (FIG. 7A).

(69) Interestingly, ACM335-TM-Zip (with leucine zipper in the periplasm) did not interact with 3G9A-CaM-Zip (with leucine zipper in the cytosol) (FIG. 7B), neither ACM335-Zip (Zip in cytosol) with 3G9A-CaM-TM-Zip (Zip in periplasm) (FIG. 7C).

(70) Hence these results indicate that the system of the invention may be used to characterize the topology of integral membrane proteins and also to probe the biological functionality of in silico predicted TM segments (i.e., their capacity to insert properly into a membrane as well as the orientation of their insertion).

(71) The HS-ACH system may also be exploited to explore the subcellular localization of given proteins in E. coli. As the ACM and CaM hybrids in the system of the invention are expressed at extremely low level, if they are spatially addressed at distant subcellular location in the cell, they will not be able to encounter and there will be no activation. Conversely if the hybrid proteins are colocalized in the cell they could then associate and be activated.

(72) In summary these data indicate that the system of the invention can efficiently report interactions between integral membrane proteins or occurring in the periplasm.