METHODS FOR PRODUCING ENGINEERED FLUORESCENT PROTEINS FOR ENHANCED FRET, PRODUCTS

AND USES THEREOF

FIELD OF THE INVENTION

The invention relates to the field of fluorescent proteins and their versatile applications in the field of molecular biology. More specifically, the invention relates to methods to produce genetically engineered fluorescent proteins for high efficiency FRET (Forster Resonance Energy Transfer). In particular, such methods comprise fusing a donor and acceptor fluorescent protein to a low-affinity helper interaction module. Moreover, pairs of such genetically engineered fluorescent proteins are provided as well as uses thereof in FRET-based applications. BACKGROUND

Fluorescent proteins are widely known today for their use as fluorescent markers in biomedical sciences. They are applied for a wide range of applications including the study of gene expression, protein localization, visualizing subcellular organelles in cells, visualizing protein localization and transport, as well as for detecting protein-protein interactions, or for screening purposes, amongst many others. Due to the potential for widespread usage and the evolving needs of researchers, novel fluorescent proteins have been identified with improved fluorescence intensity and maturation rates at physiological temperatures, modified excitation and emission spectra, and reduced oligomerization and aggregation properties. In addition, mutagenesis of known proteins has been undertaken to improve their chemical properties. Finally, codon usage may be optimized for high expression in the desired heterological system, for example in mammalian cells.

Among the different methods for the detection of protein-protein interactions, FRET (Forster Resonance Energy Transfer) between genetically encoded fluorescent proteins [1] has several distinct advantages: molecular interactions can be detected in real time, both in vitro or in live cells and without the need of co-factors or auxiliary reactions. FRET is fully reversible and reports both on the formation or the disruption of interactions - an important difference to other popular assays, for example, based on protein fragment complementation [2]. This, as well as the possibility to visualize interactions at sub-cellular spatial resolution, makes FRET ideally suited for the study of signaling and many other dynamic processes within cells, animals or in vitro systems.

Hypothetically, the interaction between any two proteins of interest could be monitored with the aid of two simple protein fusions: one protein is fused to a donor and the other to the acceptor fluorophore. Protein-based FRET probes are relatively bulky. Long and flexible peptide linkers between the fluorophore and the protein of interest are thus preferable in order to minimize any interference of the probe with the interaction of interest. However, energy transfer from donor to acceptor fluorophores only occurs over very short distances. As illustrated in Figure la, already a moderate separation by 10 to 12 nm - about twice the length of the fluorescent protein barrel - diminishes the signal to about 1% FRET efficiency. FRET probe engineering is therefore caught between the two opposing requirements of sufficiently long linkers versus short reach. A compromise needs to be found case by case and this has, until now, severely limited the general application of the approach.

Enhanced protein-based FRET pairs have so far been mostly a product of the ongoing efforts to improve spectral properties of individual fluorescent proteins. Photostability, higher quantum yields, single absorption peaks and better separation between donor and acceptor emission spectra all facilitate the measurement of FRET [3]. Nevertheless, these separate improvements cannot address the limitation of FRET by the strong dependence on fluorophore distances and orientations. Few studies tried to optimize a pair of donor and acceptor proteins simultaneously. The CyPet/YPet pair was developed by directed evolution and FACS screening from the cyan and yellow fluorescent proteins ECFP and EYFP [4]. The two proteins were connected through a flexible protease cleavage site and optimized for high FRET change between uncleaved and cleaved state. ECFP and EYFP are both derived from Aequorea GFP with a tendency to homodimerize (K _D ~ 0.1 M) [5]. The large improvement in FRET was later shown to be caused mainly by only two mutations that re-enforce the native homodimerization interface stemming from Aequorea GFP [6, 7] - leading to the formation of a high FRET intramolecular complex between CyPet and YPet. Such direct interactions among fluorescent proteins are traditionally considered annoyance [5] rather than virtue. They are, however, the best way to multiply FRET responses, in particular, if sensors are based on differences between high (bound, connected) and low (unbound, cleaved) local concentrations of donor and acceptor [8]. Also unimolecular FRET constructs can benefit from the weak dimerization of GFP-derived donor and acceptor domains [7, 9].

The interaction between GFP-derived donor and acceptor domains can be rationally enhanced or weakened through the introduction of point mutations among the residues that mediate the intermolecular contact [6,7 and published patent application WO 97/28261]. These residues can be inferred from the crystal structure of the GFP homodimer. Enhanced dimerization can be achieved through the introduction of additional hydrophobic contacts or the introduction of complementary electrostatic charges or both. Electrostatic charges can also be added indirectly, by engineering metal ion binding sites within the dimerization interface [48]. However, all these approaches have several disadvantages. First, as they rely on improving the native GFP homodimerization interface, these methods are limited to pairs of fluorescent proteins where both donor and acceptor are derived from Aequorea GFP. Such cyan / yellow or related FRET pairs are not without problems: the multi- exponential decay kinetics of most cyan proteins [3] complicates fluorescence lifetime measurements and the short excitation wavelength of the cyan donor provokes background from cellular autofluorescence. Secondly, the similarity of GFP-derived donor and acceptor proteins leads to a nearly symmetric interface. Owing to this symmetry, any increase in donor/acceptor heterodimerization is typically accompanied by increased donor/donor and acceptor/acceptor homodimerization. Vice versa, mutations that counter donor/donor or acceptor/acceptor homodimerization [5] also abolish the donor/acceptor heterodimerization [9]. Potentially beneficial heterodimerization thus has to be balanced against unwanted homodimerization tendencies. Thirdly, the introduction of molecular contacts within the interface stabilizes the formation of an actual donor/acceptor complex with reduced off rate or, correspondingly, a longer life time and tighter binding. Increased affinities can lead to undesired background signal of a FRET sensor which may eliminate any advantage of overall FRET increase [9]. Increased binding and life times may also affect the switching dynamics and reversibility of FRET sensors.

Thus, there is still a need for alternative FRET pairs with increased FRET efficiency and less dependency on fluorophore distances and orientations. Ideally, such pairs: (1) would be excited at longer wavelengths in order to reduce phototoxicity and autofluorescence and increase the penetrance of emission light; (2) would be composed of two un-related fluorescent proteins so that heterodimerization would not translate to homodimerization of acceptor or donor; (3) would work by creating a weakly aligned donor/acceptor encounter complex without actual binding, that means, they should show high on rates but very low half live of the donor/acceptor arrangement.

SUMMARY OF THE INVENTION

In order to overcome the current problems with existing FRET probes, different novel strategies were developed for the rational design of weak but strictly heterodimeric interactions ("helper interactions") between fluorescent proteins. We designate these methods "helper interaction FRET" (hiFRET). One of these methods is based on a modular approach for the rational design of helper interactions between, potentially, any pair of fluorescent proteins. Unmodified FRET probes were aligned through weak domain/peptide interaction modules. Using this strategy, we developed a set of red-shifted FRET pairs optimized for the robust detection of protein-protein interactions. The enhanced FRET pairs will be of immediate use both for in vitro or in vivo FRET based applications. In particular, signal enhancements through helper interactions should be of particular interest for upcoming large-scale FLIM experiments [22] or applications in thick samples. Further, it is of particularly advantage that the modular helper interactions are re-usable for other fluorescent proteins and thus uncouple issues of donor / acceptor orientation and distance from the choice of fluorophores. FRET probe developers can therefore break free from a de-facto limitation to CFP / YFP-derived variants and benefit from the full range of modern fluorescent proteins. To illustrate this point, we provide as a further non-limitative example an enhanced mTeal/mCherry FRET pair [23] with exceptional separation between donor and acceptor emission. Our helper interactions led to a large increase of FRET efficiency. Background signals in the unbound state remained near zero and could be further tuned by straightforward modifications of peptides. Thus, FRET signals between all synthetic protein constructs were characterized in vitro by donor- and acceptor-based fluorescence intensity. The most promising FRET pairs were validated by lifetime (FLIM) measurements in vitro and in live cells. To further illustrate the potential for a wide range of applications, it was demonstrated that the helper interaction FRET probes of the present invention also compared favorable to conventional sensors for the study of the transient, dynamically regulated protein-protein interaction between HRas and Rafl, and between Rafl and BRaf in live cells.

Thus, according to a first aspect, the invention relates to a method of producing a pair of engineered fluorescent proteins having an increased FRET efficiency relative to the non-engineered pair comprising the steps of: a. Providing a pair of fluorescent proteins suitable as a donor and acceptor fluorescent protein in FRET measurements,

b. Fusing said pair of fluorescent proteins to a pair of low-affinity interacting polypeptides resulting in a pair of engineered fluorescent proteins,

c. Measuring and comparing the FRET efficiency of said engineered pair of fluorescent proteins relative to the non-engineered pair in a suitable assay,

d. Selecting a pair of engineered fluorescent proteins having increased FRET efficiency relative to the non-engineered pair.

According to a specific embodiment, the step of fusing said pair of fluorescent proteins in the above described method is further characterized by a. Fusing the donor fluorescent protein of said pair of fluorescent proteins to one of the constituting polypeptides of said pair of interacting polypeptides, and

b. Fusing the acceptor fluorescent protein of said pair of fluorescent proteins to the other constituting polypeptide of said pair of interacting polypeptides. In other embodiments of any of the above described methods, said fusion may be at the C-terminal or

N-terminal end of the donor or acceptor fluorescent protein, optionally through one or more linker molecules. In a particular embodiment of any of the above described methods, said fusion may also be an internal fusion, which means an insertion at permissive sites within the donor/acceptor fluorescent protein.

In still other embodiments of any of the above described methods, said pair of interacting polypeptides can be further characterized by a dissociation constant (K _D ) of at least 50 μΜ. Preferably, said pair of interacting polypeptides comprises at least one protein domain and at least one peptide specifically interacting with said protein domain, such as for example a. a WW domain or homolog thereof, and a peptide specifically interacting with said WW domain or homolog thereof, or

b. an SH3 domain or homolog thereof, and a peptide specifically interacting with said SH3 domain or homolog thereof.

Another aspect of the invention relates to a pair of engineered fluorescent proteins comprising a donor fusion protein and an acceptor fusion protein wherein a. the donor fusion protein comprises a donor fluorescent protein, a first polypeptide, and optionally one or more linkers, and

b. the acceptor fusion protein comprises an acceptor fluorescent protein, a second polypeptide, and optionally one or more linkers, and

wherein said first and second polypeptide are characterized as a pair of low-affinity interacting polypeptides, and

wherein said pair of engineered fluorescent proteins has increased FRET efficiency relative to the non-engineered pair comprising a donor and acceptor fluorescent protein each of which is not fused to one of the interacting polypeptides.

According to specific embodiments, said donor or acceptor fluorescent protein of the above described pair of engineered fluorescent proteins are each fused to said first or second polypeptide at the C- terminal or N-terminal end of said fluorescent protein, optionally through a linker molecule. In a particular embodiment, said donor or acceptor fluorescent protein of the above described pair of engineered fluorescent proteins are each internally fused to said first or second polypeptide, which means an insertion at permissive sites within the donor/acceptor fluorescent protein.

In other embodiments of the above described pair of engineered fluorescent proteins, said pair of interacting polypeptides are characterized by a dissociation constant (K _D ) of at least 50 μΜ. Preferably, said pair of interacting polypeptides comprises at least one protein domain and at least one peptide specifically interacting with said protein domain, such as for example a. a WW domain or homolog thereof, and a peptide specifically interacting with said WW domain or homolog thereof, or

b. an SH3 domain or homolog thereof, and a peptide specifically interacting with said SH3 domain or homolog thereof.

Further, also provided in the present invention is the above described pair of engineered fluorescent proteins wherein the donor fusion protein and the acceptor fusion protein are each fused to a polypeptide of interest, optionally through one or more linker molecules.

In a preferred embodiment, the invention provides a bimolecular construct comprising: a. a donor fusion protein of the invention fused to a polypeptide of interest, optionally through one or more linker molecules, and

b. an acceptor fusion protein of the invention fused to a polypeptide of interest, optionally through one or more linker molecules.

According to another preferred embodiment, the invention provides a unimolecular construct selected from the group comprising a fusion protein construct as follows:

a. a donor fusion protein - a polypeptide of interest - an acceptor fusion protein, optionally fused through one or more linker molecules, or

b. an acceptor fusion protein - a polypeptide of interest - a donor fusion protein, optionally fused through one or more linker molecules, or

c. a polypeptide of interest - a donor fusion protein - an acceptor fusion protein - a polypeptide of interest, optionally fused through one or more linker molecules, or d. a polypeptide of interest - an acceptor fusion protein - a donor fusion protein - a polypeptide of interest, optionally fused through one or more linker molecules, or e. a donor fusion protein - a polypeptide of interest - a polypeptide of interest - an acceptor fusion protein, optionally fused through one or more linker molecules,

f. an acceptor fusion protein - a polypeptide of interest - a polypeptide of interest - a donor fusion protein, optionally fused through one or more linker molecules,

and, wherein said donor and acceptor fusion proteins are according to the invention.

Further, polynucleotides encoding any of the donor and/or acceptor fusion proteins, or any of the bimolecular/unimolecular constructs, according to the present invention, are also envisaged here, as well as expression vectors comprising suitable expression control sequences operably linked to any of the above described polynucleotides, as well as host cells comprising any of the polynucleotides or any of the expression vectors of the invention.

In a further aspect, the invention provides a kit comprising any of the polynucleotides or any of the expression vectors of the invention as described hereinbefore. Still another aspect of the invention relates to the use of the pair of engineered fluorescent proteins, or the bimolecular/unimolecular constructs, or any of the polynucleotides or any of the expression vectors, all as described hereinbefore, for in vitro and/or in vivo FRET-based applications.

According to a preferred embodiment, the present invention relates to a method of identifying a specific interaction of a first polypeptide of interest and a second polypeptide of interest, the method comprising: a. Contacting the first polypeptide of interest, which is fused to a donor fusion protein of a pair of engineered fluorescent proteins according to the invention, and the second polypeptide of interest, which is fused to the corresponding acceptor fusion protein of the pair of engineered fluorescent proteins according to the invention, under conditions that allow a specific interaction of the first and the second polypeptide of interest,

b. Exciting the donor, and

c. Detecting FRET from the donor to the acceptor, thereby identifying a specific interaction of the first and the second polypeptide of interest.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1. Principle of enhanced FRET, (a) FRET signals are highly dependent on the distance between donor and acceptor. The efficiency of FRET at various distances is compared using two Citrine structures (drawn to scale) spaced at a distance at which 90% of the signal is lost (E=10%). The Forster distance was set to 56.6 A [12]. (b, c) Schematic description of protein interaction detection by FRET between conventional (b) and enhanced (c) FRET probes. The drug-induced interaction between FRB and FKBP12 (gray) is used to quantify FRET signals in unbound and bound state.

Figure 2. Design of FRET helper interactions via a modular approach. Fluorophores were aligned by a weak interaction between (a) a WW domain and cognate peptides or (b) a SH3 domain and cognate peptides. Unstructured residues at the mCitrine and mCherry C-terminus were either retained (~) or trimmed (·). Flexible linkers are indicated by a dotted line. Average end-to-end distances (<link>) were calculated from the worm-like chain (WLC) polymer model [40] based on the number of unstructured amino acids (aa) and are drawn to scale.

Figure 3. Enhanced FRET, (a) WW - peptide helper interactions with moderate or weak affinity (shades of blue) enhance FRET over conventional probes (gray). The unstructured spacer peptides between FRET and helper modules (-...longer or «... shorter) have only minor effects. The FRB/FKBP12 reference interaction was connected N-terminally via a flexible 24-amino acid linker. Background signal in the unbound state (without rapamycin) remains negligible (at 0.5 μΜ concentration), (b) Helper interactions based on unrelated SH3 domain - peptide pairs have a similar effect. Peptide - domain affinities were stronger and weaker, respectively, than the "strong" and "weak" WW domain - peptide pairs, (c) FLIM-FRET measurements in live cells. Constructs encoding conventional or enhanced mCitrine/mCherry FRET probes were co-transfected and overexpressed in cells and visualized before (left) or after (right) the FRB/FKBP12 interaction had been induced by rapamycin. Pixel-by-pixel FRET effiency was quantified by FLIM and is color-coded from blue (no FRET) to red (highest FRET). The -WW / Wp2 helper interaction module leads to substantially increased FRET signals, (d) The WW - peptide helper module doubles FRET between mTeal (mTFPl) and mCherry. Constructs were based on the best-performing proteins described in (a) with mCitrine replaced by mTFPl. mCherry bearing a non-cognate SH3 peptide (see b) served as negative control.

Figure 4. Emission spectra (top) and fluorescence lifetime (bottom) of conventional and enhanced FRET pairs. Top: The best performing FRET pairs were mixed at 1:1 stoichiometry (0.5 μΜ) and their interaction induced with rapamycin. The performance of the conventional FRET pair is shown in gray (mCitrine-WW or SH3 construct paired with mCherry fused to a non-cognate peptide). Quenching of the donor emission by FRET is apparent for all pairs and is more pronounced for constructs with FRET helper interactions. Note, owing to errors in protein concentrations, exact 1:1 ratios are difficult to attain. Quantitative FRET efficiencies were therefore determined in experiments with full donor occupancy. All spectra are averages of three to six replicates and are normalized to the peak fluorescence (530 nm) of the donor-only samples. Bottom: Citrine fluorescence decay traces from the equivalent experiments on a FLIM microscope. Donor and acceptor were mixed at 0.3 and 0.5 μΜ concentrations, respectively, to ensure full donor occupancy. Raw data are shown in gray and overlaid with fits to a simple two-exponential model. Figure 5. Specificity and background activity of helper interactions, (a) The best performing FRET probe pairs from each design strategy tested at high protein concentrations (3 μΜ [donor] + 5 μΜ [acceptor]). Gains of FRET efficiency are preserved. Domain-based helper interactions can give small background signals (FRET in unbound state) that roughly scale with peptide affinity. Note: At these high concentrations, values needed to be corrected for an artificial "quenching" of Citrine fluorescence (ca. 4.5%) due to high mCherry absorption. See methods for details, (b) FRET efficiencies with mismatched helper interactions (0.3 / 0.5 μΜ standard concentration). Domain-peptide helper interactions did not cross-react and performed identical to the unmodified FRET pair. Figure 6. Correlation of in-vitro FRET efficiencies measured from changes in donor or acceptor intensities. Note that the slope of the acceptondonor correlation is directly depending on the ratio of acceptor and donor extinction coefficients which can only be measured with limited accuracy. The broken line indicates a linear least-square fit. Regression function is given.

Figure 7. In-vitro FRET efficiencies measured at varying salt concentrations. Domain-peptide binding is dominated by hydrophobic interactions and remained nearly unaffected. Error bars indicate standard deviations from 5 or 6 replicates.

Figure 8. Binding curves for peptides and helper domains estimated from the difference between donor intensity before and after induction of the FRB/FKBP12 interaction, (a) a WW domain and cognate peptides or (b) a SH3 domain and cognate peptides. No difference (Erap- = Erap+) would indicate full saturation of the donor domain with acceptor peptide already in the absence of rapamycin. The direct comparison of "background"/iininduced FRET signals at different peptide- acceptor concentrations was ruled out by the intense absorbance of mCherry samples at the required protein concentrations. Error bars indicate standard deviations from 3 replicate measurements. See Methods for details. Hypothetical binding curves for K _D values of 1 and 500 μΜ are shown as broken lines for comparison. Note that the weak peptide in (b) has close to zero affinity for the SH3 domain.

Figure 9. Comparison of the original ("wt") and enhanced mCitrine sequences. Domains added to the C-terminal of unmodified mCitrine are shown in green (WW) and blue (SH3). Unstructured peptide sequences introduced during the engineering of the original Citrine are labeled as 'spacer'.

Figure 10. Comparison of the original mCherry sequence with the four variants engineered for domain-peptide recruitment. Only the C-terminal of (the otherwise unmodified) mCherry is shown. Wpl and Wp2 denote strong and weak WW-binding peptides, respectively. Spl and Sp2 denote strong and weak SH3-binding peptides, respectively.

Figure 11. Application of conventional and enhanced FRET probes to the study of HRas : Rafl signaling in live HEK293 cells, (a) HEK293 cells expressing conventional mCherry-HRas and conventional mCitrine coupled to full-length Rafl are shown untreated (t=0) or 4 min after treatment with epidermal growth factor (EGF) (t=4min). Donor-intensity images (top) are color-coded with intensity-weighted lifetimes (middle) or intensity-weighted FRET efficiencies (bottom), (b) HEK293 cells co-expressing the same HRas / full-length Rafl sensor but now using mCitrine and mCherry enhanced through the ~WW / .Wp2 domain/peptide module, (c) HEK293 cells expressing a commonly used FRET sensor for HRas activity based on mCherry fused to HRas and mCitrine fused to a truncated fragment (the Ras Binding Domain, RBD) of Rafl. Note the nuclear de-localization of the Rafl fragment, (d) Quantification of FRET signals for selected cells without and 8 min after EGF treatment. Error bars indicate standard errors (SEM). (e) Quantification of FRET signals from membrane regions of the same cells, (f) Comparison of the dynamic range of the same FRET signals without and with EGF stimulation, evaluated for selected cells, membranes or without any selection (whole field of view). Figure 12. Comparison of the change in FRET signals after EGF stimulation, (a) Globally averaged over measurements at all-time points (4, 8, 12 and 16 or 20 min) without selection of any region of interest (whole field of view) or with cell or membrane regions selected independently at each time point; and (b) averaged, separately for each time point, over independently selected membrane regions only.

Figure 13. Application of conventional and enhanced FRET probes to the detection of the Rafl : BRaf interaction in live HeLa cells, (a) HeLa cells expressing conventional mCitrine fused to the C-terminal of full length Rafl and conventional mCherry coupled to the C-terminal of full-length BRaf are shown untreated (control) or 2 h after treatment with Raf kinase inhibitor GDC-0879. Donor-intensity images (top) are color-coded with non-weighted FRET efficiencies (middle). The distribution of FRET efficiencies over all cells in the field of view is shown in the bottom panel. No FRET signal is evident under either condition, (b) HeLa cells co-expressing the same full length Rafl / full-length BRaf sensor but now using mCitrine and mCherry enhanced through the ~WW / .Wp2 domain/peptide module. Raf inhibitor treatment now reveals a weak interaction between the two proteins.

DETAILED DESCRIPTION OF THE INVENTION

The present invention will be described with respect to particular embodiments and with reference to certain drawings but the invention is not limited thereto but only by the claims. Any reference signs in the claims shall not be construed as limiting the scope. The drawings described are only schematic and are non-limiting. In the drawings, the size of some of the elements may be exaggerated and not drawn on scale for illustrative purposes. Where the term "comprising" is used in the present description and claims, it does not exclude other elements or steps. Where an indefinite or definite article is used when referring to a singular noun e.g. "a" or "an", "the", this includes a plural of that noun unless something else is specifically stated. Furthermore, the terms first, second, third and the like in the description and in the claims, are used for distinguishing between similar elements and not necessarily for describing a sequential or chronological order. It is to be understood that the terms so used are interchangeable under appropriate circumstances and that the embodiments of the invention described herein are capable of operation in other sequences than described or illustrated herein.

Unless otherwise defined herein, scientific and technical terms and phrases used in connection with the present invention shall have the meanings that are commonly understood by those of ordinary skill in the art. Generally, nomenclatures used in connection with, and techniques of molecular and cellular biology, genetics and protein and nucleic acid chemistry and hybridization described herein are those well-known and commonly used in the art. The methods and techniques of the present invention are generally performed according to conventional methods well known in the art and as described in various general and more specific references that are cited and discussed throughout the present specification unless otherwise indicated. See, for example, Sambrook et al. Molecular Cloning: A Laboratory Manual, 2d ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. (1989); Ausubel et al., Current Protocols in Molecular Biology, Greene Publishing Associates (1992, and Supplements to 2002); Leach, Molecular Modelling: Principles and Applications, 2d ed., Prentice Hall, New Jersey (2001).

Definitions

As used herein, the terms "determining", "measuring", "assessing", "testing" and "assaying" are used interchangeably and include both quantitative and qualitative determinations.

As used herein, the terms "polypeptide", "protein", "peptide", "oligopeptide" are used interchangeably herein, and refer to a polymeric form of amino acids of any length, which can include coded and non- coded amino acids, chemically or biochemically modified or derivatized amino acids, and polypeptides having modified peptide backbones.

As used herein, the terms "nucleic acid molecule", "polynucleotide", "polynucleic acid", "nucleic acid" are used interchangeably and refer to a polymeric form of nucleotides of any length, either deoxyribonucleotides or ribonucleotides, or analogs thereof. Polynucleotides may have any three- dimensional structure, and may perform any function, known or unknown. Non-limiting examples of polynucleotides include a gene, a gene fragment, exons, introns, messenger NA (mRNA), transfer RNA, ribosomal RNA, ribozymes, cDNA, recombinant polynucleotides, branched polynucleotides, plasmids, vectors, isolated DNA of any sequence, control regions, isolated RNA of any sequence, nucleic acid probes, and primers. The nucleic acid molecule may be linear or circular.

The term "vector" as used herein is intended to refer to a nucleic acid molecule capable of transporting another nucleic acid molecule to which it has been linked. One type of vector is a "plasmid vector", which refers to a circular double stranded DNA loop into which additional DNA segments may be ligated. Other vectors include, without the purpose of being limitative, cosmids and yeast artificial chromosomes (YAC). Certain vectors are capable of autonomous replication in a host cell into which they are introduced (e.g., vectors having an origin of replication which functions in the host cell). Other vectors can be integrated into the genome of a host cell upon introduction into the host cell, and are thereby replicated along with the host genome. Moreover, certain preferred vectors are capable of directing the expression of certain genes of interest. Such vectors are referred to herein as "recombinant expression vectors" (or simply, "expression vectors"). Suitable vectors have regulatory sequences, such as promoters, enhancers, terminator sequences, and the like as desired. The term "operably linked" as used herein refers to a linkage in which the regulatory sequence is contiguous with the gene of interest to control said gene of interest, as well as regulatory sequences that act in trans, or at a distance to control the gene of interest.

The term "regulatory sequence" as used herein refers to polynucleotide sequences which are necessary to affect the expression of coding sequences to which they are operatively linked. Expression control sequences are sequences which control the transcription, post-transcriptional events and translation of nucleic acid sequences. Expression control sequences include appropriate transcription initiation, termination, promoter and enhancer sequences; efficient NA processing signals such as splicing and polyadenylation signals; sequences that stabilize cytoplasmic mRNA; sequences that enhance translation efficiency (e.g., ribosome binding sites); sequences that enhance protein stability; and when desired, sequences that enhance protein secretion. The nature of such control sequences differs depending upon the host organism. The term "control sequences" is intended to include, at a minimum, all components whose presence is essential for expression, and can also include additional components whose presence is advantageous, for example, leader sequences and fusion partner sequences. The term "recombinant host cell" ("expression host cell", "expression host system", "expression system" or simply "host cell"), as used herein, is intended to refer to a cell into which a recombinant vector has been introduced. It should be understood that such terms are intended to refer not only to the particular subject cell but to the progeny of such a cell. Because certain modifications may occur in succeeding generations due to either mutation or environmental influences, such progeny may not, in fact, be identical to the parent cell, but are still included within the scope of the term "host cell" as used herein. A recombinant host cell may be an isolated cell or cell line grown in culture or may be a cell which resides in a living tissue or organism. A "mutant" or "variant" or "derivative" (or equivalent wordings having the same meaning), as used herein, and used in the context of an amino acid or nucleotide sequence, is meant to encompass a subsequent amino acid or nucleotide sequence that has been derived from a previous amino acid or nucleotide sequence, or reference sequence, either naturally or artificially. For purposes of illustration only, a variant or mutant protein of wild type Aequorea GFP (the reference in this case) may have one or more amino acid substitutions, additions, or deletions as compared to said Aequorea wild type GFP amino acid sequence. Specific, but non-limiting, examples of fluorescent protein mutants or variants are provided in Tables 5 and 6 (examples known in the art), as well as in the Detailed Description part hereinafter (as part of the present invention). A "spectral variant" or "spectral mutant", as used herein, refers to a fluorescent protein to indicate a mutant fluorescent protein that has a different fluorescence characteristic with respect to the corresponding wild type fluorescent protein. For example, CFP, YFP, ECFP, EYFP-V68L/Q69K, and the like, are GFP spectral variants.

A "substitution", as used herein, results from the replacement of one or more amino acids or nucleotides by different amino acids or nucleotides, respectively as compared to an amino acid sequence or nucleotide sequence of a parental or reference polypeptide or nucleic acid. In certain embodiments of the present invention, the substitutions are conservative substitutions. In other embodiments, the substitutions are non-conservative substitutions. Within the context of a protein or polypeptide, conservative and non-conservative amino acid substitutions are known to those of ordinary skill in the art. It is understood that a protein or a fragment thereof may have conservative amino acid substitutions which have substantially no effect on the protein's activity. By conservative substitutions is intended combinations such as gly, ala; val, ile, leu, met; asp, glu; asn, gin; ser, thr; lys, arg; cys, met; and phe, tyr, trp.

A "deletion" is defined here as a change in either amino acid or nucleotide sequence in which one or more amino acid or nucleotide residues, respectively, are absent as compared to an amino acid sequence or nucleotide sequence of a parental or reference polypeptide or nucleic acid. Within the context of a protein, a deletion can involve deletion of about 2, about 5, about 10, up to about 20, up to about 30 or up to about 50 or more amino acids. A protein or a fragment thereof may contain more than one deletion. An "insertion" or "addition" is that change in an amino acid or nucleotide sequences which results from the insertion of addition of one or more amino acid or nucleotide residues, respectively, as compared to an amino acid sequence or nucleotide sequence of a parental or reference polypeptide or nucleic acid. "Insertion" generally refers to the insertion or addition of one or more amino acid residues within an amino acid sequence of a polypeptide, while "addition" can be an insertion or refer to amino acid residues added at an N- or C-terminus, or both termini. Within the context of a protein or a fragment thereof, an insertion or addition is usually of about 1 , about 3, about 5, about 10, up to about 20, up to about 30 or up to about 50 or more amino acids. A protein or fragment thereof may contain more than one insertion.

The terms "specifically interact" or "specifically bind" or "specific interaction or binding", as used herein, refers to the ability of one particular polypeptide to preferentially bind to another particular polypeptide, when both polypeptides are present in a homogeneous mixture of different polypeptides. Within the context of the present invention, but without being limitative, the term refers to a peptide that may specifically interact with a protein domain and not, or to a lesser degree, with other (poly)peptides in a mixture of different polypeptides. In certain embodiments, a specific binding interaction will discriminate between desirable and undesirable polypeptides in a sample, in some embodiments more than about 10 to 100-fold or more (e.g., more than about 1000- or 10,000-fold). The terms "specifically interact" or "specifically bind" and grammatical equivalents thereof are used interchangeably herein.

The term "affinity", as used herein, refers to the degree to which one particular polypeptide binds to or interacts with another particular polypeptide so as to shift the equilibrium of either polypeptide toward the presence of a complex formed by their binding. Thus, for example, where a peptide and a protein domain are combined in relatively equal concentration, a peptide of high affinity will bind to the available protein domain so as to shift the equilibrium toward high concentration of the resulting complex. Otherwise, a peptide of low affinity will usually not bind to a protein domain, unless one or both of them are present in a high concentration or if their co-recruitment or protein fusion creates a high local concentration. The dissociation constant is commonly used to describe the affinity between two polypeptides, in particular between the peptide and the protein domain. Typically, moderate to strong interactions (moderate to high affinity) imply that the dissociation constant is lower than 10 ^s M. Strong interactions (high affinity) imply a dissociation constant lower than 10 ^"6 M. Conversely, weak interactions (low affinity) typically imply that the dissociation constant is larger than 10 ^s M.

Detailed description The strong distance dependence of FRET hampers the wide-spread application of genetically encoded FRET pairs. The present invention provides strategies for the rational engineering of weak helper interactions that align donor and acceptor fluorophores leading to robust FRET signals without elaborate optimization of linker sequences or orientations.

A preferred strategy is based on the use of weak domain/peptide interaction modules and the implementation for the optimization of fluorescent protein pairs for in vitro/in vivo FRET applications. It is a particular advantageous strategy since (1) the same helper interaction module can be applied to any suitable pair of fluorescent proteins for FRET applications without further engineering, (2) the helper interaction can be easily fine-tuned to the experimental conditions like for example protein concentrations, (3) the helper interaction module can be replaced by another helper interaction module if needed for reasons of assay or host cell compatibility, (4) FRET pairs with different helper interaction modules can be employed in parallel (e.g. in multiplexed assays) without interfering with each other.

Accordingly, a first aspect of the invention relates to a method of producing a pair of engineered fluorescent proteins having an increased FRET efficiency relative to the non-engineered pair comprising the steps of: a. Providing a pair of fluorescent proteins suitable as a donor and acceptor fluorescent protein in FRET measurements,

b. Fusing said pair of fluorescent proteins to a pair of low-affinity interacting polypeptides resulting in a pair of engineered fluorescent proteins,

c. Measuring and comparing the FRET efficiency of said engineered pair of fluorescent proteins relative to the non-engineered pair in a suitable assay,

d. Selecting a pair of engineered fluorescent proteins having increased FRET efficiency relative to the non-engineered pair.

As used herein, the term "fluorescent protein" refers to any protein that can fluoresce when excited with an appropriate electromagnetic radiation. A fluorescent protein may exhibit low, medium or intense fluorescence upon irradiation with light of the appropriate excitation wavelength. The fluorescent proteins of the present invention do not include proteins that exhibit fluorescence only from residues that act by themselves as intrinsic fluorophors, i.e., tryptophan, tyrosine and phenylalanine. A fluorescent protein of the invention or for use in a method of the invention is a protein that derives its fluorescence from autocatalytically forming a chromophore. A fluorescent protein can contain amino acid sequences that are naturally occurring or that have been engineered (i.e., variants or mutants or derivatives, as defined herein). For example, a spectral variant of Aequorea GFP can be derived from the naturally occurring GFP by engineering mutations such as amino acid substitutions into the reference GFP protein. For example, ECFP is a spectral variant of GFP that contains substitutions with respect to GFP. It is to be understood that the term "fluorescent protein", as used herein, also includes variants of fluorescent proteins that have lost actual fluorescence but can still act as FRET acceptors by virtue of a "dark quenching" or "dark absorbing" chromophore. Non- limiting examples are the REACh variants of YFP [41].

Fluorescent proteins are often classified according to their spectral class. Thus, fluorescent proteins may be green fluorescent proteins which fluoresce green, or red fluorescent proteins which fluoresce red, or yellow fluorescent proteins which fluoresce yellow, or cyan fluorescent proteins which fluoresce cyan, or orange fluorescent proteins which fluoresce orange, etc. The term "green fluorescent protein" or "GFP" is used broadly herein to refer to a protein that fluoresces green light, for example, Aequorea GFP. GFPs have been isolated from the jellyfish, Aequorea victoria, the sea pansy, Renilla reniformis, and Phialidium gregarium [19, 20]. The term "red fluorescent protein", or "RFP" is used in the broadest sense and specifically covers the Discosoma RFP (DsRed), and red fluorescent proteins from any other species, such as coral and sea anemone, as well as variants thereof, as long as they retain the ability to fluoresce red light [21]. Furthermore, reference is also made to the various spectral variants and mutants that have amino acid sequences that are substantially identical to a reference fluorescent protein. Non-limiting examples of commonly known reference fluorescent proteins include, but are not limited to, A. Victoria GFP (Genebank Accession Number M62654.1), Discosoma RFP (DsRed), (Genebank Accession Number AF168419), amongst others, see [24, 25] and U.S. Patent No. 5,625,048 and International application PCT/US95/14692, now published as PCT WO96/23810, each of which is incorporated herein by reference.

Aequorea GFP-related fluorescent proteins include, for example, wild type (native) Aequorea victoria

GFP [24], allelic variants thereof, for example, a variant having a Q80R substitution [26]; and spectral variants of GFP such as CFP, YFP, and enhanced and otherwise modified forms thereof (U.S. Pat. Nos. 6,150,176; 6,124,128; 6,077,707; 6,066,476; 5,998,204; and 5,777,079, each of which is incorporated herein by reference), including GFP-related fluorescent proteins having one or more folding mutations, and fragments of the proteins that are fluorescent, for example, an A. victoria GFP from which the two

N-terminal amino acid residues have been removed. Several of these fluorescent proteins contain different aromatic amino acids within the central chromophore and fluoresce at a distinctly shorter wavelength than the wild type GFP species. For example, the engineered GFP proteins designated P4 and P4-3 contain, in addition to other mutations, the substitution Y66H; and the engineered GFP proteins designated W2 and W7 contain, in addition to other mutations, Y66W. Similarly, Discosoma

RFP (DsRed)-related fluorescent proteins include, for example, wild type (native) Discosoma RFP, allelic variants thereof, and spectral variants thereof, such as mPlum, tdTomato, mStrawberry, DsRed Monomer, mOrange [11]. To illustrate this further, several examples of wild type fluorescent proteins and variants derived thereof are listed in Tables 5 and 6, wherein further details on the properties of each listed protein are provided.

A further class of fluorescent proteins considered here are variants that are derived from other fluorescent proteins like GFP or DsRed or Citrine or mCherry by means of circular permutation. Circular permutation is the fusion of N- and C-terminal of the original protein while simultaneously creating a new N- and C-terminal within the original protein sequence, with the aim to change the orientation of the fluorescent protein within larger protein fusion constructs [29, 30].

Typically, fluorescent proteins are chosen according to the type of application and experiment, essentially based on critical factors such as emission spectra, brightness, photostability, oligomerization, for which guidance is provided in the art, see e.g. [3]. Preferably, the fluorescent proteins as used herein are suitable for Forster resonance energy transfer (FRET), also known as fluorescence energy transfer, or simply, resonance energy transfer (RET), hereinafter further referred to as "FRET". FRET is the non-radioactive transfer of excited-state energy from one molecule (the donor) to another nearby molecule (the acceptor) via a long-range dipole-dipole coupling mechanism, and is well described in the art. FRET is usually limited to distances less than ~ 10 nm, and thus provides a sensitive tool for investigating a variety of phenomena that produce changes in molecular proximity. Preferably, the fluorescent proteins as used herein are a pair of fluorescent proteins suitable as a donor and acceptor in FRET measurements, for which guidance is provided in the art, see e.g. [1, 3, and 31] as well as in the Detailed Description further herein.

There are several ways to measure or to quantify FRET which are known to the person skilled in the art. Typically, the ratio of donor and acceptor fluorescence is measured. Increasingly, instruments and microscopes allow to measure the lifetime of the donor fluorescence, which drops the more energy is transferred to the acceptor. For example, fluorescence lifetime imaging microscopy (FLIM) is now routinely used for dynamic measurements of signaling events inside living cells, including detection of protein-protein interactions. FLIM maps the spatial distribution of probe lifetimes inside living cells, and can accurately measure the shorter donor lifetimes that result from FRET (e.g. [32]). In any case, the signal-to-noise ratio depends on the FRET efficiency. The term "FRET efficiency" (£), as used herein, is the quantum yield of the energy transfer transition, i.e. the fraction of energy transfer event occurring per donor excitation event and is known to the person skilled in the art. The FRET efficiency depends on many parameters, including the distance between the donor and the acceptor, the spectral overlap of the donor emission spectrum and the acceptor absorption spectrum, the relative orientation of the donor emission dipole moment and the acceptor dipole moment. Preferably, the efficiency of FRET between the donor and acceptor is at least 10%, at least 20%, at least 30%, at least 40%, more preferably around 50%. For intensity-based FRET measurements, the efficiency of FRET is preferably even higher, at least 60%, at least 70%, and even more preferably at least 80%, at least 90%, or higher. One is referred to the Example section for an example on how FRET efficiencies can be calculated. The calculated FRET efficiencies can be compared between two pairs of fluorescent proteins, wherein one pair of fluorescent proteins may have an "increased" or "higher" FRET efficiency or may have a "reduced" or "lower" FRET efficiency relative to the other pair. Preferably, the difference in FRET efficiency is statistically significant.

The rate of energy transfer depends upon the extent of spectral overlap between the donor emission and acceptor absorption spectra, the quantum yield of the donor, the relative orientation of the donor and acceptor transition dipole moments, and the distance separating the donor and acceptor molecules. A preferred factor to be considered in choosing the donor and acceptor pair is the efficiency of fluorescence resonance energy transfer between them. The present invention provides guidance on how to improve FRET efficiency. It will be clear that spectral overlap, quantum yield and related spectroscopic properties are a direct consequence of the choice of donor and acceptor protein variant and remain valid for various implementations or applications of the same FRET pair. These were the main criteria that have been used in the past to optimize the efficiency and detectability of FRET between a donor and acceptor molecule. By contrast, distance and orientation could only be influenced by a careful choice of linkers and elaborate design and testing of custom fusion proteins. Therefore, the latter criteria could so far only be optimized on a case-by-case basis for each new FRET sensor or application individually. The present invention offers means to optimize FRET probe distance and orientation largely independent of the molecular and experimental context in which the FRET pair is applied. Distance optimized helper interaction FRET pairs can therefore be used for a variety of applications without or with little further customization. The term "a pair of engineered fluorescent proteins" or "an engineered pair of fluorescent proteins", as used herein, refers to a pair of fluorescent proteins that has been genetically engineered or modified, according to any of the above described engineering methods of the present invention; this is in contrast to a non-engineered fluorescent protein pair that has not been genetically engineered according to the invention, and that is known in the art. It should be clear that a non-engineered pair of fluorescent proteins is not necessarily restricted to a pair of wild type or reference fluorescent proteins, but may also be, for example, a homolog, a spectral variant or another mutant derived thereof. Preferably, said pair of engineered fluorescent proteins is optimized for higher FRET efficiency as compared to the non-engineered pair. It should be clear that the above described engineering method is meant to improve FRET efficiency by optimizing a pair of fluorescent proteins simultaneously (i.e. donor and acceptor protein), and not the fluorescent proteins individually.

According to a specific embodiment, the invention relates to a method of producing a pair of engineered fluorescent proteins having an increased FRET efficiency relative to the non-engineered pair comprising the steps of: a. Providing a pair of fluorescent proteins suitable as a donor and acceptor fluorescent protein in FRET measurements,

b. Fusing the donor fluorescent protein of said pair of fluorescent proteins to one of the constituting polypeptides of a pair of low-affinity interacting polypeptides,

c. Fusing the acceptor fluorescent protein of said pair of fluorescent proteins to the other constituting polypeptide of said pair of low-affinity interacting polypeptides, resulting in a pair of engineered fluorescent proteins,

d. Measuring and comparing the FRET efficiency of said engineered pair of fluorescent proteins relative to the non-engineered pair in a suitable assay,

e. Selecting a pair of engineered fluorescent proteins having increased FRET efficiency relative to the non-engineered pair.

The fusion of the donor/acceptor fluorescent protein to one of the constituting polypeptides of a pair of low-affinity interacting polypeptides may be either at the C-terminal or at the N-terminal end of said donor/acceptor fluorescent protein, optionally through one or more linker molecules. In an alternative embodiment, the fusion of the donor/acceptor fluorescent protein to one of the constituting polypeptides of a pair of low-affinity interacting polypeptides may also be an internal fusion, which means an insertion at permissive sites within the donor/acceptor fluorescent protein.

Low-affinity interacting polypeptides may thus also be fused such that they are inserted into permissive sites within the globular fluorescent protein domain. Permissive sites for the insertion of low-affinity interacting peptides are positions within the fluorescent protein where the polypeptide chain can be interrupted without severely affecting the folding or structure of the protein or the maturation of its chromophore. Permissive sites are typically surface-exposed. Circular permutation experiments have revealed many permissive sites within fluorescent proteins where the peptide chain can be opened up [49, and incorporated herein by reference]. Sequence and structure alignments can be used to infer potential permissive sites in one fluorescent protein from experimentally characterized positions in another fluorescent protein variant. Given that N- and C-terminal of fluorescent proteins are located in close proximity at one end of the beta-barrel structure, low affinity polypeptides should, preferably, be inserted at the other end of the barrel. In an optimal arrangement, the low affinity interaction and the (C or N terminally attached) protein interaction of interest can thus pull together donor and acceptor proteins from two opposing ends.

The pair of interacting polypeptides, can be freely chosen and can be any pair of polypeptides specifically interacting with each other as long as they interact with a low affinity. Thus, the term "a pair of low-affinity interacting polypeptides", as used herein, refers to a pair of polypeptides that specifically interact (as defined herein) with each other with a low affinity (as defined herein), and is also referred to as "helper interaction module". To illustrate this further, "low affinity" implies that the dissociation constant for the pair of interacting polypeptides is significantly lower than the protein concentrations used in F ET-based in vitro and/or in vivo applications, as elaborated further herein. As such, a low affinity ensures that the helper interaction module is not causing significant binding events by itself (background binding). In order to ensure less than 5%, preferably less than 3%, most preferably less than 1% background binding, the dissociation constant should be lower than the target protein concentration by a factor of 19, preferably a factor of 33, most preferably a factor of 99, respectively. Even more preferably, the dissociation constant is lower than the target protein concentration by a factor higher than 100. It should be clear that intracellular protein concentrations are usually below 0.001 mM and very rarely above 0.01 mM, even for the most strongly expressed proteins [37]. Thus, for most applications, said pair of interacting polypeptides is characterized by a dissociation constant (K _D ) in the range of 10 ^s M, preferably at least 0.05 mM, more preferably at least 0.1 mM, at least 0.2 mM, at least 0.3 mM, at least 0.5 mM, but not higher than 5 mM, preferably at most 2 mM, or more preferably at most 1 mM.

According to a more specific embodiment, said pair of low-affinity interacting polypeptides comprises at least one protein domain and at least one peptide specifically interacting with said protein domain. Domain/peptide interacting polypeptides are well-known in the art, and without the purpose of being limitative, are as described in [36]. Particular examples include protein domains such as WW, SH3, PDZ, PTB, EVH1, GYF, VHF, amongst many others, and peptides specifically binding to said domains, amongst many others.

In one preferred embodiment, said pair of interacting polypeptides in any of the above described methods is chosen from the group comprising: a. a WW domain or homolog thereof, and a peptide specifically interacting with said WW domain or homolog thereof, or b. an SH3 domain or homolog thereof, and a peptide specifically interacting with said SH3 domain or homolog thereof.

The term "homolog" of a protein domain/peptide, as used herein, encompasses polypeptides having amino acid substitutions, deletions and/or insertions, preferably by a conservative change, relative to the unmodified polypeptide in question. Homologs of domains/peptides are capable of specifically interacting with each other with different degrees of low affinity (as defined hereinbefore). Homologs may be either naturally occurring or made by man. Several different methods are known by those of skill in the art for identifying and defining these functionally homologous sequences, including phylogenetic methods, sequence similarity and hybridization methods. Percentage similarity and identity can be determined electronically. Examples of useful algorithms are PILEUP (Higgins & Sharp, CABIOS 5:151 (1989), BLAST and BLAST 2.0 (Altschul et al. J. Mol. Biol. 215: 403 (1990). Software for performing BLAST analyses is publicly available through the National Center for Biotechnology Information (http://www/ncbi.nlm. nih.gov/). Preferably, a homolog of a protein domain/peptide has a sequence identity at amino acid level of at least 30%, at least 40%, at least 50%, at least 60%, at least 70%, at least 80%, preferably at least 90%, even more preferably at least 95%, at least 98%, at least 99%, as measured in a BLASTp or any other equivalent method known in the art. Alternatively, a homolog of a protein domain/peptide can also be identified by sharing the same fold; that means, by having a three-dimensional structure that is significantly similar. Structural homologs can be identified and defined by using structure alignment programs, which are known to those skilled in the art. Examples of structure alignment algorithms are Dali and Dali server [42] as well as Tm-align [43]. Preferably, a structure homolog of a protein domain is identified by a TM-align score of at least 0.5, at least 0.6, at least 0.7, preferably higher than 0.7.

Any naturally or engineered fluorescent protein, including a variant of a fluorescent protein, can be chosen as a donor or acceptor fluorescent protein, as long as together they are suitable for FRET measurements, as was described hereinbefore. Other examples are provided in Table 5 and 6. For the sake of clarity, the term "variant" as defined hereinbefore, encompasses homologs of the donor or acceptor fluorescent proteins. Homologs of a protein encompass polypeptides having amino acid substitutions, deletions and/or insertions, preferably by a conservative change, relative to the unmodified protein in question and having similar biological and functional activity as the unmodified protein from which they are derived; or in other words, without significant loss of function or activity, (for example, homologous fluorescent proteins derived from one species, e.g. Aequorea GFP, either naturally occurring or made by man). Several different methods are known by those of skill in the art for identifying and defining these functionally homologues sequences, and are described hereinbefore. Preferably, a homolog of a donor or acceptor fluorescent protein has a sequence identity at protein level of at least 30%, at least 40%, at least 50%, at least 60%, at least 70%, at least 80%, preferably at least 90%, even more preferably at least 95%, at least 98%, at least 99%, as measured in a BLASTp or any other equivalent method known in the art. Preferably, the donor and acceptor fluorescent protein of the pair of fluorescent proteins do not show initial intrinsic interaction, such as by choosing fluorescent protein variants that have been made monomeric, such as for example, without the purpose of being limitative, mCitrine, mCherry, mTFPl.

A second aspect of the present invention relates to a pair of engineered fluorescent proteins comprising a donor fusion protein and an acceptor fusion protein wherein a. the donor fusion protein comprises a donor fluorescent protein, a first polypeptide, and optionally one or more linkers, and

b. the acceptor fusion protein comprises an acceptor fluorescent protein, a second polypeptide, and optionally one or more linkers, and

wherein said first and second polypeptide are characterized as a pair of low-affinity interacting polypeptides, and

wherein said pair of engineered fluorescent proteins has increased FRET efficiency relative to the non-engineered pair comprising a donor and acceptor fluorescent protein each of which is not fused to one of the interacting polypeptides.

According to specific embodiments, the donor or acceptor fluorescent protein may be fused to said first or second polypeptide at the C-terminal or N-terminal end of said fluorescent protein, optionally through a linker molecule. In an alternative embodiment, the fusion of the donor or acceptor fluorescent protein to said first or second polypeptide may also be an internal fusion, which means an insertion at permissive sites within the donor or acceptor fluorescent protein. Further, specific embodiments related to the pair of interacting polypeptides and the donor/acceptor fluorescent proteins are as described hereinbefore.

In a preferred embodiment, the donor fusion protein and the acceptor fusion comprised in the pair of engineered fluorescent proteins are each fused to a molecule of interest, preferably a polypeptide of interest or a target polypeptide, optionally through one or more linker molecule. A "polypeptide of interest" or a "target protein" or grammatically equivalents thereof, as used herein, can be any polypeptide, including, for example, a sensor protein such as calmoduline, or a cellular polypeptide such as an enzyme, a G-protein, a growth factor receptor, or a transcription factor, amongst others, and can be one or two or more proteins that can interact or associate to form a complex.

The engineered donor and acceptor fusion proteins of the present invention may each be linked to a molecule of interest either directly or indirectly, using any linkage that is stable under the conditions to which the protein-molecule complex is to be exposed. Where the molecule of interest is a polypeptide, a convenient means for linking or fusing an engineered fluorescent protein and the molecule is by expressing them as a fusion protein from a recombinant nucleic acid molecule, which comprises a polynucleotide encoding, for example, a fluorescent protein operably linked to a polynucleotide encoding the polypeptide molecule. Preferred "linker molecules" or "linkers" are peptides of 1 to 50 amino acids length and are typically chosen or designed to be unstructured and flexible. These include, but are not limited to, synthetic peptides rich in Gly, Ser, Thr, Gin, Glu or further amino acids that are frequently associated with unstructured regions in natural proteins [33]. Examples used herein are (GS)s (SEQ ID NO: 69) or (GS)i ₀ (SEQ ID NO: 70) or a 50 amino acid randomized sequence of Gly, Serjhr, Gin, Glu (SEQ ID NO: 71, SEQ ID NO: 98). Further examples include, but are not limited to, a protease cleavage site such as Factor Xa cleavage site having the sequence IEG (SEQ ID NO: 72), the thrombin cleavage site having the sequence LVPR (SEQ ID NO: 73), the enterokinase cleaving site having the sequence DDDDK (SEQ ID NO: 74), or the PreScission cleavage site LEVLFQGP (SEQ ID NO: 75). For the sake of clarity, a polypeptide of interest may be fused to an engineered donor or acceptor fusion protein, either at the free N- or C-terminal end of the donor or acceptor fluorescent protein or otherwise at the free N- or C-terminal end of the low-affinity interacting polypeptides. I will be understood that this will depend on how the donor and acceptor fusion proteins are constructed themselves; more specifically, whether the donor/acceptor fluorescent protein are each fused to one of the constituting interacting polypeptides either at the C-terminal or at the N-terminal end of said donor/acceptor fluorescent protein.

Preferably, the amino acid linker sequence is relatively short, but long enough to allow the contact of enhanced donor and acceptor fluorescent proteins, and does not interfere with the biological activity of the proteins. Note that the terms "linker" and spacer" are used interchangeably herein. Non-limiting examples of suitable linker sequences are also described in the Example section. Thus, the invention also provides a fusion protein comprising a polypeptide of interest fused to an engineered donor fusion protein or to an engineered acceptor fusion protein, as described hereinbefore. More specifically, the present invention also encompasses a "bimolecular construct" of two such fusion proteins, wherein one fusion protein comprises a polypeptide of interest fused to an engineered donor fusion protein and one other fusion protein comprises a polypeptide of interest fused to an engineered acceptor fusion protein. It should be understood that bimolecular constructs refer to two separate fusion proteins. On the other hand, the bimolecular constructs as described herein may be expressed from a single recombinant nucleic acid molecule or from two separate recombinant nucleic acid molecules, as described further herein. Such a bimolecular construct is particularly useful for detection of protein-protein interactions, which, in turn, can serve as indicator of changes in protein signaling, protein modifications or protein localization.

Thus, according to a more specific embodiment, the invention envisages a bimolecular construct comprising: a. a donor fusion protein fused to a polypeptide of interest, optionally through one or more linker molecules, and

b. an acceptor fusion protein fused to a polypeptide of interest, optionally through one or more linker molecules, and

wherein said donor and acceptor fusion proteins are comprised in the pair of engineered fluorescent proteins as described hereinbefore.

Alternatively, a fusion protein is provided comprising one or more polypeptide(s) of interest, an engineered donor and an acceptor fusion protein as described hereinbefore, which is then refered to as a "unimolecular construct" or "tandem construct". The different polypeptides can be fused to each other in any order and either directly or indirectly through one or more linker molecules.

Such a unimolecular construct can be particularly useful for the detection of conformational changes or intra-molecular binding within a polypeptide of interest which, in turn, can serve as proxy for changes in cellular signaling, ion concentrations or enzymatic activities. Unimolecular constructs are also used for the detection of peptide cleavage events, in particular, due to protease activities. Thus, in such an alternative embodiment, the invention envisages a unimolecular construct selected from the group comprising a fusion protein construct as follows:

a. a donor fusion protein - a polypeptide of interest - an acceptor fusion protein, optionally fused through one or more linker molecules, or

b. an acceptor fusion protein - a polypeptide of interest - a donor fusion protein, optionally fused through one or more linker molecules, or

c. a polypeptide of interest - a donor fusion protein - an acceptor fusion protein - a polypeptide of interest, optionally fused through one or more linker molecules, or d. a polypeptide of interest - an acceptor fusion protein - a donor fusion protein - a polypeptide of interest, optionally fused through one or more linker molecules, or e. a donor fusion protein - a polypeptide of interest - a polypeptide of interest - an acceptor fusion protein, optionally fused through one or more linker molecules,

f. an acceptor fusion protein - a polypeptide of interest - a polypeptide of interest - a donor fusion protein, optionally fused through one or more linker molecules,

and, wherein said donor and acceptor fusion proteins are comprised in the pair of engineered fluorescent proteins as described hereinbefore.

The polypeptides comprised in any of the above described fusion protein constructs can be linked through peptide bonds, as described hereinbefore. The fusion proteins may be expressed from a recombinant nucleic acid molecule containing a polynucleotide encoding an engineered fluorescent protein of the invention operatively linked to one or more polynucleotides encoding one or more polypeptides of interest. In another embodiment, the invention relates to one, two, or more polynucleotides encoding the engineered donor and/or acceptor fusion proteins of the pair of engineered fluorescent proteins, or the bimolecular or unimolecular constructs, or the fusion proteins, all as described hereinbefore. Non- limiting examples of polynucleotide sequences encoding engineered fluorescent proteins and synthetic protein constructs according to the invention, as well as the corresponding amino acid sequences, are listed in Table 7.

The invention further concerns expression vectors containing such polynucleotides and host cells containing such polynucleotides or vectors. The vector generally contains elements required for replication in a prokaryotic or eukaryotic host system or both, as desired. Such vectors, which include plasmid vectors and viral vectors such as bacteriophage, baculovirus, retrovirus, lentivirus, adenovirus, vaccinia virus, semliki forest virus and adeno-associated virus vectors, are well known and can be purchased from a commercial source (e.g. Promega, Madison Wl; Stratagene, La Jolla CA; GIBCO/B L, Gaithersburg MD) or can be constructed by one skilled in the art. Where the vector is a viral vector, it can be selected based on its ability to infect one or few specific cell types with relatively high efficiency. For example, the viral vector also can be derived from a virus that infects particular cells of an organism of interest, for example, vertebrate host cells such as mammalian host cells. Viral vectors have been developed for use in particular host systems, particularly mammalian systems and include, for example, retroviral vectors, other lentivirus vectors such as those based on the human immunodeficiency virus (HIV), adenovirus vectors, adeno-associated virus vectors, herpesvirus vectors, vaccinia virus vectors, and the like. The construction of expression vectors and the expression of a polynucleotide in transfected cells involves the use of molecular cloning techniques also well known in the art (see Sambrook et al., In "Molecular Cloning: A Laboratory Manual" (Cold Spring Harbor Laboratory Press 1989); "Current Protocols in Molecular Biology" (eds., Ausubel et al.; Greene Publishing Associates, Inc., and John Wiley & Sons, Inc. 1990 and supplements).

A further aspect of the invention is drawn to a kit used for making and using a pair of engineered fluorescent proteins according to the invention in laboratory methods or other applicable uses, including, for example, to construct a fluorescent fusion protein comprising an engineered fluorescent protein of the invention that can be expressed in living cells, tissues, and organisms. Preferably, the invention provides kits comprising any of the above described polynucleotides or any of the above described expression vectors. Alternatively, or in addition, the kits can provide any of the above described (pair of) engineered fluorescent proteins or fusion proteins themselves. In addition, kits of the present invention can also include, for example but not limited to, apparatus and reagents for sample collection and/or purification, apparatus and reagents for product collection and/or purification, reagents for bacterial cell transformation, reagents for eukaryotic cell transfection, previously transformed or transfected host cells, sample tubes, holders, trays, racks, dishes, plates, instructions to the kit user, solutions, buffers or other chemical reagents, suitable samples to be used for standardization, normalization, and/or control samples. Also present in the kits may be antibodies specific to the provided proteins. The components of the kit may be packaged in combination or alone in the same or in separate containers, depending on, for example, cross-reactivity or stability, and can also be supplied in solid, liquid, lyophilized, or other applicable form. A container means of the kits may generally include, for example, at least one vial, test tube, flask, bottle, syringe or other container means, into which a component may be placed, and preferably, suitably aliquoted. Such a kit may be useful for any of the applications of the present invention as described further herein. Still another aspect of the invention relates to the use of any of the engineered fluorescent proteins (whether as a pair or individual, and whether in a bimolecular or unimolecular construct) or any polynucleotide encoding such protein in any method that employs a fluorescent protein. In a preferred embodiment, the engineered fluorescent proteins are useful as a FRET pair for in vitro and/or in vivo FRET-based applications, such as detection of protein-protein interactions, conformational changes of a protein, protease activity, protein modifications, changes in concentrations of metabolites, ions or signaling molecules. In a specific embodiment, the aforementioned FRET-based assay is part of a screening process, for example, in order to discover or characterize compounds, conditions or processes that trigger or disrupt protein-protein interactions or that lead to other changes in the state of cellular or in-vitro systems. Thus, according to one embodiment, engineered fluorescent proteins having features of the invention are useful as a FRET pair in a method of identifying a specific interaction of a first molecule and a second molecule, for example a specific interaction between proteins, or between a protein and a nucleic acid, or between nucleic acids. The first and second molecules can be cellular proteins that are being investigated to determine whether the proteins specifically interact, or to confirm such an interaction. Such first and second cellular proteins can be the same, where they are being examined, for example, for an ability to oligomerize, or they can be different where the proteins are being examined as specific binding partners involved, for example, in an intracellular pathway. The first and second molecules also can be a polynucleotide and a polypeptide, for example, a polynucleotide known or to be examined for transcription regulatory element activity and a polypeptide known or being tested for transcription factor activity. For example, the first molecule can comprise a plurality of nucleotide sequences, which can be random or can be variants of a known sequence, that are to be tested for transcription regulatory element activity, and the second molecule can be a transcription factor, such a method being useful for identifying novel transcription regulatory elements having desirable activities. The conditions for such an interaction can be any conditions under which is expected or suspected that the molecules specifically interact. In particular, where the molecules to be examined are cellular molecules, the conditions generally are physiological conditions. As such, the method can be performed in vitro using conditions of buffer, pH, ionic strength, and the like, that mimic physiological conditions, or the method can be performed in a cell or using a cell extract.

Accordingly, in a preferred embodiment, the invention envisages a method of identifying a specific interaction of a first polypeptide of interest and a second polypeptide of interest, the method comprising: a. Contacting the first polypeptide of interest, which is fused to a donor fusion protein of the pair of engineered fluorescent proteins of the invention, and the second polypeptide of interest, which is fused to the corresponding acceptor fusion protein of the pair of engineered fluorescent proteins of the invention, under conditions that allow a specific interaction of the first and the second polypeptide of interest,

b. Exciting the donor, and

c. Detecting FRET from the donor to the acceptor, thereby identifying a specific interaction of the first and the second polypeptide of interest.

According to specific embodiments of the above method, the first polypeptide is a first cellular protein or fragment thereof and the second polypeptide is a second cellular protein or fragment thereof. The use of the engineered fluorescent proteins of the present invention for such a purpose provides a substantial advantage in that (1) the FRET signal is increased over noise and background, (2) FRET signals can be obtained without optimizing the distance and orientation between the interacting proteins of interest and the donor and acceptor probes which, (3) allows to use standardized long (rather than customized short) molecular linkers between proteins of interest and fluorescent probes, which (4) diminishes the risk of artifacts or modification of the biological activity of interest.

The above processes can be miniaturized and automated to enable screening many thousands of molecules in a high throughput format.

In another embodiment, the engineered fluorescent proteins are useful as a FRET pair to detect cleavage of a substrate having the donor and acceptor coupled to the substrate on opposite sides of the cleavage site. Upon cleavage of the substrate, the donor/acceptor pair physically separate, eliminating FRET. Such an assay can be performed, for example, by contacting the substrate with a sample, and determining a qualitative or quantitative change in FRET. A fluorescent protein variant donor/acceptor pair also can be part of a fusion protein coupled by a peptide having a proteolytic cleavage site. In other embodiments, useful applications include FRET-based sensors for protein kinase and phosphatase activities or indicators for small ions and molecules such as Ca ²⁺ , Zn ²⁺ , cyclic 3', 5'- adenosine monophosphate, and cyclic 3', 5' -guanosine monophosphate.

Further, the fluorescent protein variants are useful as fluorescent markers in the many ways fluorescent markers already are used, including, for example, coupling fluorescent protein variants to antibodies, polynucleotides or other receptors for use in detection assays such as immunoassays or hybridization assays, or to track movement of proteins in cells. Examples wherein the fluorescent protein variants can be used as labeling substance, include, without the purpose of being limitative, biological and/or medicinal imaging, fluorescent microscopy, FRET-based assays or screening or imaging (in vitro, in cells or in vivo). Fluorescence in a sample generally is measured using a fluorimeter and methods of performing assays on fluorescent materials are well known in the art (see, for example, Lakowicz, "Principles of Fluorescence Spectroscopy" (Plenum Press 1983); Herman, "Resonance energy transfer microscopy" In "Fluorescence Microscopy of Living Cells in Culture" Part B, Meth. Cell Biol. 30:219-243 (ed. Taylor and Wang; Academic Press 1989); Turro, "Modern Molecular Photochemistry" (Benjamin/ Cummings Publ. CoJ, jfric. 1978), pp. 296-361, each of which is incorporated herein by reference).

The sample to be examined can be any sample, including a biological sample, an environmental sample, or any other sample for which it is desired to determine whether a particular molecule is present therein. Preferably, the sample includes a cell or an extract thereof. The cell can be obtained from a vertebrate, including a mammal such as a human, or from an invertebrate, and can be a cell from a plant or an animal. The cell can be obtained from a culture of such cells, for example, a cell line, or can be isolated from an organism. As such, the cell can be contained in a tissue sample, which can be obtained from an organism by any means commonly used to obtain a tissue sample, for example, by biopsy of a human. Where the method is performed using an intact living cell or a freshly isolated tissue or organ sample, the presence of a molecule of interest in living cells can be identified, thus providing a means to determine, for example, the intracellular compartmentalization of the molecule.

The following examples are intended to promote a further understanding of the present invention. While the present invention is described herein with reference to illustrated embodiments, it should be understood that the invention is not limited hereto. Those having ordinary skill in the art and access to the teachings herein will recognize additional modifications and embodiments within the scope thereof. Therefore, the present invention is limited only by the claims attached herein.

EXAMPLES EXAMPLE 1. Model protein-protein interaction FRET sensor

We chose the widely used fluorescent protein variants mCitrine [10] and mCherry [11] as our reference donor and acceptor fluorophores for the comparison of conventional and enhanced FRET. The yellow FP Citrine [10] is derived from Aequorea victoria GFP. It is assumed to retain the weak homodimerization tendency of GFP but can be converted into a strict monomer (mCitrine) by the interface-breaking mutation A206K [5]. mCherry [11] is a monomeric and improved variant of Discosoma red fluorescent protein. As the two fluorophores originate from different species (jelly fish and coral), they show no intrinsic interaction. (m)Citrine / mCherry are an excellent long-wavelength FRET pair combining good spectral separation and single exponential donor decay kinetics with among the longest Forster distance reported for any genetically encoded pair [12]. We used the chemically induced protein-protein interaction between FRB and FKBP12(T2089L) [13, 14] as a test system for conventional and enhanced FRET probes. As outlined in Figure lb, addition of rapamycin triggers a high-affinity binding between the two domains yet there is no interaction in absence of the drug [14]. This allowed us to measure FRET signals both in bound (active, induced) and unbound state.

FRET donor constructs all shared the same primary structure architecture and were composed (from N- to C-terminal) of: (1) a FKBP12 domain preceded by a Thr-Gly spacer, (2) a 20 amino acid linker consisting of 10 repeats of a Gly-Ser dipeptide and preceded and followed by a Thr-Gly spacer (amounting to an overall linker length of 24 amino acids), (3) the mCitrine reference or modified sequence followed by a Ser-Gly spacer, (4) a recognition and cleavage site for the PreScission protease followed by a Thr-Gly spacer, (5) a hexa -Histidine tag for purification purposes followed by a Thr-Gly spacer and STOP codon.

FRET acceptor constructs all shared the same primary structure architecture and were composed (from N- to C-terminal) of: (1) a FRB domain preceded by a Thr-Gly spacer, (2) a 20 amino acid linker consisting of 10 repeats of a Gly-Ser dipeptide and preceded and followed by a Thr-Gly spacer (amounting to an overall linker length of 24 amino acids), (3) the mCherry reference or modified sequence followed by a Ser-Gly or a Gly-Ser spacer, (4) a recognition and cleavage site for the PreScission protease followed by a Thr-Gly spacer, (5) a hexa -Histidine tag for purification purposes followed by a Thr-Gly spacer and STOP codon.

Table 3 lists all proteins constructed for this study.

We improved our previous protocol for in vitro FRET measurements [14]: (1) Purification by gel filtration was crucial for consistent results. (2) Extinction coefficients (see Table 4) were determined by comparison with absorbance at 280 nm and averaged over various protein constructs containing (m)Citrine or mCherry and are somewhat different from published values [3]. (3) To account for inevitable inaccuracies in protein concentrations, we ensured full donor or acceptor occupancy by providing acceptor protein in excess during the measurement of donor-based FRET and, vice-versa, an excess of donor for sensitized emission measurements. In vitro FRET efficiencies measured with the modified protocol are typically reproduced to within 1% E also between different preparations of equivalent proteins. A good agreement between donor- and acceptor-based FRET measurements (Figure 6) indicates that mCitrine and mCherry populations were generally homogeneous and intact.

EXAMPLE 2. FRET enhancement by domain/peptide interaction modules We tested a strategy relying on weak interaction modules that can be reused and added to any pair of fluorescent proteins.

We tested this strategy by modifying the model interaction sensor described above; For donor constructs, the mCitrine sequence (3 above) was replaced by the mCitrine sequence (full length or C- terminally trimmed) followed by a Ser-Gly spacer, followed by the enhancement domain (WW or SH3) followed by a Thr-Gly spacer. For acceptor constructs, the mCherry sequence (3 above) was replaced by the mCherry sequence (full length or C-terminally trimmed) followed by a Ser-Gly or a Gly-Ser spacer, followed by the enhancement peptide of 6 to 11 amino acids length with varying affinities for either WW or SH2 domain followed by a Thr-Gly spacer.

WW domain / peptide module. In a first implementation, illustrated in Figure 2a, we fused a stabilized (L30K) [15] variant of the WW domain of human YAP65 to the C terminal of mCitrine. Affinities of cognate peptides reported for this domain [15] range from (K _D ) 40 to 700 μΜ. We fused either the strongest or a weakly binding peptide to the C terminal of mCherry. Peptide sequences and affinities are listed in Table 1. The N- and C terminal of all peptide sequences were flanked by Gly-Ser (GS) and Thr-Gly (TG), respectively, in order to prevent spurious interactions with neighboring residues. The conventional mCitrine as well as mCherry contain an additional short unstructured C terminal spacer sequence (GMDELYK) originally introduced for better compatibility with protein fusions. Judging from structural models, the WW / peptide interaction appeared compatible with a tight coupling and the deletion of this spacer sequence should bring donor and acceptor into even closer proximity (see Figure 2a for distances). We thus prepared two variants of mCitrine-WW and each mCherry-peptide fusion - one with the traditional spacer sequence (GMDELYKGS) and one without (GSG). As shown in Figure 3a, the addition of the domain - peptide interaction module nearly doubled FRET efficiencies from 23.1±1.0 % to 43.8±0.8 % (stronger peptide) or 41.3±1.3 % (weak peptide). The spacer between fluorescent protein and WW domain or peptide had only a minor effect. Nevertheless, the combination of loosely linked WW domain (~WW) and tightly fused peptide ("peptide) gave a slight advantage over all other pairs. Example spectra and lifetime traces are shown in Figure 4. The domain - peptide helper interactions remained active and unaffected over a wide range of ionic strength (Figure 7).

SH3 domain / peptide module. The enhancement of FRET probes through domain-based helper interactions is a general design concept. To prove this point, we replicated the same design using an unrelated SH3 domain / peptide interaction module (Figure 2b). We paired the SH3 domain from Saccharomyces cerevisiae protein Shol [16, 17] with two different peptides, chosen to have a stronger or weak affinity similar to the two WW / peptide pairs (Table 1). The SH3 domain was fused to mCitrine and peptides were fused to mCherry. As before, each construct was prepared with and without the MDELYK spacer sequence. Results from in vitro measurements are given in Figure 3b. SH3 / peptide helper interactions doubled FRET efficiencies to up to 48.6±1.1 % when the domain was paired with the higher affinity peptide. The, for practical purposes more important, weak peptide gave less pronounced improvements to only 32.0±1.1 % (a 50% gain in signal). Helper affinities

Our original design was aiming for peptide-domain affinities of about 500 μΜ Kd which should ensure strong intra-complex helper binding without incurring serious background interactions even in concentrated protein solutions. Nevertheless, affinities had been less well characterized for the SH3 than for the WW interaction module and we had to rely on theoretical estimates. The different behavior of SH3 and WW weak peptides might therefore have been due to larger than expected deviations from target affinities. Indeed, our F ET-based measurements of peptide / domain binding (Figure 8, see methods for details) revealed substantial disparities. K _D values are listed in Table 1. The strong WW and SH3 peptides showed similar affinities for their target domains. However, the weak WW peptide exhibited a threefold stronger binding than expected from literature, presumably due to the stabilizing effect of the WW L30K mutation [15]. Conversely, there was close to no signal detectable for the weak SH3 / peptide interaction. The actual K _D of this pair appears to be well above the 1 mM mark.

EXAMPLE 3. Specificity and background of enhanced FRET probes None of the various helper interactions did cause any significant background signal (FRET signal in unbound state) at the standard 0.5 μΜ protein concentrations of our in vitro experiments. At 10-fold higher concentrations (Figure 5a), an appreciable background signal appeared for FRET probes connected to the strong peptide / domain interactions - which displayed up to 5 and 7% FRET efficiency (for the best WW and SH3 modules, respectively) already before the addition of rapamycin. Note that these values are corrected for significant unspecific quenching (4-6% E) that occurred at the elevated protein concentrations even in the absence of helper interactions. After this correction, a low background was also recorded for the weaker WW peptide (2.5% E). No background signal was observed from the much weaker SH3 / peptide.

The peptide - domain helper interactions were highly specific, despite low affinities and despite the fact that both SH3 and WW domain recognize proline-rich motifs. As shown in Figure 5b, we did not observe any FRET increase for any off-target combination of WW and SH3 domain peptides.

EXAMPLE 4. Lifetime and intracellular measurements

Table 2 compares FRET efficiencies of the most important FRET pairs characterized with various protocols in bulk samples as well as in a microscopy setup. Absolute FRET efficiencies are consistent across a variety of measurement methods. In particular, efficiencies obtained from the decrease of amplitude-weighted average lifetimes (x _amp ) confirm the results of intensity-based experiments. Donor fluorescence decay remained monoexponential before the induction of protein-protein interactions, testifying to the absence of background FRET. Post induction, the enhancement of FRET was evident from the shortening of average lifetimes. Interestingly, a simple double exponential was only a poor model for the excited state time course of both conventional and enhanced FRET pairs. The assumption of only two states, FRET-on or -off, would indeed be at odds with the free tumbling of fluorophores attached to highly flexible linkers. Peptide-mediated helper interactions should reduce average distances between donor and acceptor but were expected to leave substantial freedom for the sampling of different relative orientations and distances. A stretched exponential model [18] accounts for the dynamic averaging over disordered ensembles and resulted in quantitative agreement between intensity and lifetime-based FRET efficiencies.

Figure 3c demonstrates the application of the domain / peptide enhanced FRET probes in live cells. We selected the WW / weak peptide pair (Cit(K)~WW + Che.Wp2) as the enhancement module with the most promising properties in vitro. Plasmids encoding the model proteins as introduced above, with rapamycin-inducible interaction domains and enhanced or conventional mCitrine/mCherry FRET probes, were co-transfected into HeLa cells. In-line with in vitro results, WW - peptide co-recruitment doubled the FLIM-FRET signal. Notably, despite strong overexpression of the two partner proteins, no background signal was apparent before the addition of rapamycin.

EXAMPLE 5. Application to another FRET pair (mTeal/mCherry)

In order to further test the general applicability of our approach, we designed an alternative model FRET donor construct with a primary architecture (sequence) identical to the one described in Example 1 except that the mCitrine (module 3 in Example 1) was replaced by full-length mTFPl [23], also known as mTeal, as alternative donor fluorescent protein unrelated to mCitrine. This donor protein was paired with the enhanced mCherry acceptor protein described in Example 2, bearing mCherry (C- terminally trimmed) and the peptides Wpl and Wp2 with stronger or weak affinity for the WW domain. Without a matching WW domain in the donor, this protein pair corresponds to a conventional un-enhanced FRET pair. Upon induction of the model interaction with Rapamycin, a FRET signal was detected. Owing to the low overlap of mTFPl emission and mCherry absorbance spectra, the signal was comparatively low and only reached 15% FRET efficiency. Results of this and additional control experiments are shown in Figure 3d. These results reinforce the notion that, based on their spectroscopic properties, mTFPl : mCherry would generally be considered a poor FRET pair.

An enhanced version of this donor protein was prepared with a primary architecture identical to the one described in Example 2, except that the mCitrine-WW module was replaced by full-length mTFPl followed by a Ser-Gly spacer followed by the WW enhancement domain, followed by a Thr-Gly spacer. This enhanced donor protein was again paired with the enhanced acceptor protein described above. Upon induction of the model interaction with rapamycin, a much higher FRET signal was detected. Similar to the effect described in Example 2, the helper interaction between donor and acceptor doubled the signal to give 31% FRET efficiency. Figure 3d compares FRET efficiencies measured for the enhanced and non-enhanced versions of this FRET pair. This example demonstrated that the enhancement of FRET signals through modular helper interactions is not limited to the mCitrine / mCherry FRET pair but can be easily applied to other, entirely unrelated, pairs of fluorescent proteins.

EXAMPLE 6. Direct detection of H-Ras / Rafl interactions As a further example, we applied conventional and enhanced FRET probes for studying the interaction between HRas and Rafl. Activation of the small GTPase HRas leads to the binding of Rafl (mediated by the RasGTP binding domain, RBD, contained within Rafl) and the recruitment of the complex to the membrane where further signalling events are triggered [38].

Existing bimolecular FRET sensors consisting of mCherry-HRas and RaflRBD-mCitrine fusion proteins display a high FRET efficiency in the activated state and thus have an excellent dynamic range distinguishing between unbound and bound state [38]. This performance is explained by the structure of the complex between HRas and RaflRBD which places the N-terminal of HRas into close proximity of the C-terminal of the RBD fragment. FRET probes can thus be attached through very short linker peptides and are in optimal position for FRET. However, the RaflRBD fragment (81 amino acids) may not be a suitable model for the physiological interaction between full-length Rafl (comprising 648 amino acids) and HRas. It lacks two additional domains as well as several functionally important sites of post-translational modification, protein interaction and localization signals. RBD is located near the N- terminal of Rafl which rules out C-terminal attachment of FRET probes. An unstructured region of 50 amino acids length separates the RasGTP binding domain also from the N-terminal of the protein. Furthermore, already the N-terminal of the binding domain itself is pointing away from the interacting HRas. Any protein fusion will thus place the fluorescent probe remote from the interacting HRas. It is thus not surprising that no high performance bi-molecular FLIM/FRET probe has so far been described for the physiologically relevant interaction between HRas and full-length Rafl.

Therefore, we applied conventional and enhanced FRET probes for studying the interaction between HRas and Rafl. For control purposes, we fused conventional mCitrine (SEQ ID NO: 1) to the flexible N- terminal of full-length Rafl (SEQ ID NO: 26). A control FRET acceptor was constructed from conventional mCherry fused to the N-terminal of HRas (SEQ ID NO: 28) through a randomized synthetic 50 amino acid linker designed to be unstructured and flexible (SEQ ID NO: 71). Figure 11a shows that low though significant FLIM/FRET signals were recorded and that the signals increased after stimulation of cells with EGF. Introduction of the ~WW / Wp2 helper interaction led to increased FLIM/FRET signals (Figure lib) as well as a sharper delineation of membrane-anchored signalling. Notably, the background FLIM/FRET signal before EGF stimulation was lower than what was observed with the conventional FRET pair. Thus, the weak helper interaction module appears not to cause any spurious interaction. For comparison and as a positive control, the performance of the classic HRas/RaflRBD truncated FRET sensor was also tested. As expected, FLIM/FRET signals were strong (Figure 11c). However, RaflRBD appeared de-localized throughout the cell. By contrast, neither of the full-length Rafl constructs was entering the nucleus, in-line with the physiological function of the protein.

This qualitative picture is confirmed by quantitative analysis of mean apparent FRET efficiencies measured over regions of interest, such as cells (Figure lid, between 3 and 14 cells were analyzed for each field of view) or, functionally more relevant, membrane boundaries of these cells (Figure lie). The WW enhanced FRET pair outperformed the conventional full-length Rafl probe both in terms of intensity-weighted FRET efficiencies of whole cells as well as non-weighted FRET efficiencies of membrane regions. In relative terms, the WW helper module improved the dynamic range to an approximately 4-fold signal increase upon stimulation, compared to a only 2.2 fold increase in case of the conventional FRET pair (Figure llf). These quantitative differences persist regardless of selection scheme and regions of interest. In fact, differences are even still evident if fields of view are analyzed as a whole, without the filtering of non-expressing cells or cell debris (Figure llf), or if data are moreover averaged over multiple measurements of different cells at different time points (Figure 12a).

The classic truncated HRas/RaflRBD FRET sensor is close to optimal in terms of the geometry for FRET

- very short linkers (about 4 amino acids) connect donor and acceptor to the respective target protein. Moreover, the interaction between HRas and Ras binding domain positions the two fluorophores in very close proximity to each other. By contrast, the sensor based on HRas and full length FRET has a very unfavorable geometry for FRET - donor and acceptor fluorophores are removed from each other through a combined 100 amino acids of flexible linkers (comprising 50 amino acids flexible N-terminal of Rafl and 50 amino acids synthetic linker on the HRas side) which are, furthermore, connecting to opposing ends of the interaction partners. Remarkably, the enhancement with helper interactions improves the HRas/HRafl FRET sensor to approximately the same level of performance as observed for the classic truncated HRas/RaflRBD construct. Clearly, the enhanced HRas/HRafl FRET sensor is a biologically more relevant model for the interaction, as the lack of regulatory sequences and the effective de-localization of the RaflRBD protein fragment may lead to non-physiological artifacts in more detailed aspects of signaling. In fact, the approximate time course of FRET signals (Figure 12b) suggests a delayed activation of HRas/RaflRBD sensors which could be tentatively explained by a slow exchange of RaflRBD fragments between cytosolic and non-physiological nuclear fractions. In conclusion, this example demonstrates the straightforward recovery of intracellular high-performance FRET signals from a transient and dynamically regulated protein-protein interaction with a molecular geometry that would be unfavorable for conventional FRET probe design.

EXAMPLE 7. Direct detection of Rafl / BRaf interactions

As a further example of a biological application, we applied conventional and enhanced FRET probes for studying the interaction between the two related signaling proteins Rafl and Braf. Both Rafl (also described as CRaf) and BRaf are part of the Ras-ERK MAPK signaling pathway (see Example 6). This pathway is often involved in the malignant proliferation of human cancer cells, for example, 43% of melanomas are characterized by oncogenic mutations activating BRaf [46]. BRaf is a downstream effector of activated H-Ras. Inhibitors of Raf kinase activity were therefore considered promising antitumor agents as they should block proliferation signals both from oncogenic BRaf as well as from mutated H-Ras. By contrast, Raf kinase inhibitors failed to act against tumors driven by oncogenic Ras. Furthermore, therapeutic success against oncogenic BRaf is often limited by the rapid development of drug resistance. Both failures have recently been traced back to activating interactions between Rafl and BRaf [45-47]. In the presence of activated Ras, Raf kinase inhibitors turned out to promote the heterodimerization of BRaf with Rafl leading to the activation of downstream proliferation signals by Rafl kinase thus circumventing the block of BRaf kinase. Depending on cellular context (oncogenic Ras instead of oncogenic BRaf), these drugs can thus promote rather than inhibit cancer growth.

Despite its clinical importance, the Rafl / BRaf interaction has until now not been visualized in live cells. Existing assays are biochemical (immunoblots of phosphorylated substrate proteins or co- immunoprecipitation of Rafl and BRaf), and are performed in cell lysates. We have tested whether the interaction of full length Rafl and BRaf can be monitored by FRET in live cells. We fused conventional mCitrine (SEQ ID NO: 1) to the flexible C-terminal of full-length Rafl (SEQ ID NO: 26) through a randomized synthetic 50 amino acid linker designed to be unstructured and flexible (SEQ ID NO: 98). We then fused conventional mCherry (SEQ ID NO: 2) to the C-terminal of BRaf (SEQ ID NO: 76) through the same randomized 50 amino acid linker. Both constructs were introduced into HeLa cells and FLIM measurements were performed with or without the drug candidate GDC-0897. GDC-0897 has previously been shown to stimulate Rafl / BRaf interaction in biochemical assays [47]. Nevertheless, as shown in Figure 13a, no FRET signal was observed in either stimulated or unstimulated cells. We then replaced the conventional FRET probes by mCitrine and mCherry enhanced with the ~WW/Wp2 helper interaction module (see Table 3 and 7 for synthetic protein constructs). As shown in Figure 13b, stimulation with the drug now leads to a significant FRET signal, distributed throughout the cytosol of cells. As before, unstimulated cells exhibit a sharp distribution of fluorescence life times around 3.06 ns indicating no FRET. By contrast, life times in stimulated cells show a broad bi-modal distribution. Average FRET signals are weaker than in Example 6 but, nevertheless, distinct from controls and unstimulated cells. We note that, as before, this result was achieved without optimization of linker lengths or protein truncation. The interaction is mediated by the Raf kinase domains which is remote from both N- and C terminal of the protein. This would explain the failure of conventional FRET probes. Truncated sensors based on only those domains would likely provide a better signal but could be prone to artifacts as described in Example 6. In conclusion, this example demonstrates the utility of hiFRET probes in visualizing a weak and transient protein interaction that cannot be detected by conventional FRET probes.

Example 8. Unimolecular Caspase sensors

Activation of executioner Caspase 3 is a crucial event in programmed cell death (apoptosis). Caspase 3 is a tightly regulated protease that recognizes a specific cleavage sequence (DEVD). Several well- known FRET sensors allow the detection of Caspase 3 activity, and thus initiation of apoptosis in life cells [50]. Caspase sensors are simple unimolecular construct where a donor fluorescent protein is fused via a short linker to an acceptor fluorescent protein resulting in a strong FRET signal between the two. The short linker contains the specific Caspase cleavage sequence. Upon activation of Caspase 3, this linker is cleaved and the FRET signal is rapidly lost. The cleavage can be followed by simply measuring the ratio of acceptor (sensitized emission) versus donor fluorescence since the stochiometry of both is fixed. Known Caspase sensors perform well and are difficult to improve on. However, simple proteolytic sensors such as this have often been used to evaluate or optimize new donor / acceptor FRET pairs [4]. We constructed a new Caspase sensor as an example for the application of helper interactions in unimolecular settings, involving one of the most recent and highly optimized fluorescent proteins, mTurquoise2 [51], and to test different arrangements of helper interaction and fluorescent proteins. For comparison, we first constructed a conventional Caspase 3 sensor (SEQ ID NO: 88) consisting, from N- to C-terminal, of mTurquoise2 fused to mCitrine through a relatively short 18 amino acid linker containing the DEVD cleavage site recognized by Caspase 3 (see Table 3 and 7 for synthetic protein constructs). The sensor was tested to detect initiation of apoptosis in HeLa cells. Upon stimulation with Stauroporin, the acceptor / donor emission ratio switched from 1.72±0.23 to 0.84±0.12, as expected (all values given are averages over about 30 single cells). The introduction of a helper interaction at either end of the short linker was unlikely to much reduce the distance between donor and acceptor. However, we tested an alternative arrangement where helper interactions were fused to the mTurquoise2 N-terminal (WW domain) and Citrine C-terminal (Wp2 peptide) (SEQ ID NO: 89) and measured an improved emission ratio of 1.99±0.28 switching to 0.79±0.07 after apoptosis. This experiment demonstrated that the helper interaction principle can be applied to mTurquoise2 and does not have any significant effect on switching dynamics or proteolytic separation.

We then modified the conventional Caspase sensor by inserting the helper interaction module within a surface exposed loop at the end of the fluorescent protein barrel that is opposing the juxtaposed N and C terminal (SEQ ID NO: 91). We introduced a truncated version of the WW domain into position 174 (Gly) of mTurquoise2 and inserted the weak affinity peptide Wp2 into the equivalent position 174 (Gly) of mCitrine. A flexible asymmetric linker of 18 amino acids was added to one side of the Wp2 peptide in order to ensure that the helper interaction would orient itself outside of rather than in between the donor and acceptor barrel. For control purposes, we constructed a third Caspase sensor identical to the former, except that Wp2 was replaced by a flexible non-binding sequence of same length (SEQ ID NO: 90) (see Table 3 and 7 for synthetic protein constructs). In HeLa cells, this control construct had a lower FRET efficiency than the conventional sensor (ratio before apoptosis 1.28±0.16) but switched to a comparable ratio after cleavage (0.72±0.05). This could indicate side effects of the domain insertion such as slower maturation or less efficient folding. Yet, the construct with helper interaction (complete with cognate peptide) more than compensates for this effect and improves FRET to a ratio of 1.78±0.13 switching to 0.77±0.05, which is equivalent or better than the conventional sensor. Follow up experiments with purified Caspase sensors in vitro reproduced the result with the conventional sensor protein (ratio of 1.72 switching to 0.48) but show an even higher ratio for the inserted helper interaction (ratio 2.21 switching to 0.49) corresponding to an apparent FRET efficiency of 75%. Overall, this set of experiments demonstrated that helper interaction modules can also be inserted within the fluorescent protein structure. Preferred positions of insertion are sites such as Gly 174 which had previously been shown to tolerate circular permutation [49]. Further optimization of insertion site and surrounding sequences may help to mitigate effects on fluorophore maturation and folding.

In conclusion, experiments with unimolecular Caspase sensors confirmed that helper interaction modules can be robustly attached at either N- or C-terminal or even inside of different fluorescent proteins and can lead to significant improvements of ratiometric FRET measurements.

METHODS TO THE EXAMPLES

METHODS Construct design and cloning

Protein-coding constructs were obtained by gene synthesis from DNA 2.0 (CA, USA) and delivered in expression plasmid pJExpress411, codon-optimized for E.coli. For expression in HeLa cells, selected constructs were transferred from the pJExpress411 into a pcDNA3.1 vector backbone using In-Fusion (Clonetech) or isothermal assembly [39] recombining. The appropriate PC reactions were performed with Phusion Hotstart polymerase (NEB). All constructs were verified by DNA sequencing. Constructs containing H-Ras, Rafl, Rafl RBD, BRaf or Caspase3 sensors were again obtained by gene synthesis from DNA 2.0 and delivered in pJ603-neo (mCitrine and Caspase constructs) or pJ609-puro (mCherry constructs) mammalian expression vectors and codon-optimized for Homo sapiens. For expression in E. coli, sequences encoding Caspase sensors were transferred from pJ603-neo into the pJExpress411 vector backbone, including C terminal fusion to PreScission recognition site and hexa-Histidine tag, using PCR and isothermal assembly [39]. Protein expression and purification

Expression plasmids were transformed into E. coli BL21 (DE3) (Invitrogen). Starter cultures (LB, 50 g/ml kanamycin) were inoculated from single colonies, grown over night at 37°C and then used for 1:100 inoculation of 0.5 I production cultures (2xTY, 50 g/ml kanamycin). Production cultures were grown shaking to an O.D. of around 0.6, induced with 0.5 mM IPTG and incubated over night at 20°C . Cells were harvested by centrifugation for 15 min at 6000 g and 4°C, washed once in 15 ml PBS, weighed and stored at -80°C. Pellets were resuspended in 5 ml/(g pellet) BugBuster lysis buffer (Novagen), supplemented with Complete protease inhibitor (Roche) at 1 tablet/50 ml. The lysis mix was incubated for 20 min slowly shaking at room temperature. Cell debris was removed by 5 min centrifugation at 1500 g at 4°C, followed by 30 min centrifugation at 20,000 g and 4°C to remove insoluble protein. The supernatant was mixed with 4 ml Ni-NTA Agarose resin (Qiagen, washed twice), diluted to 40 ml with binding buffer (25 mM Tris-Hcl, 20 mM imidazole, 0.5 M NaCI, 10% glycerol, pH 7.4) and incubated rotating for 30 min at 4°C . The resin was washed (1 min centrifugation at 2000 g) twice with 40 ml washing buffer (25 mM Tris-Hcl, 40 mM imidazole, 1 M NaCI, 10% glycerol, 0.1% Tween 20, pH 7.4), transferred to gravity flow columns (BioRad), settled with 30 ml binding buffer and protein was then eluted by gravity flow with 2 x 1 ml elution buffer (25 mM Tris-Hcl, 0.5 M imidazole, 0.5 M NaCI, 10% glycerol, pH 7.4). Samples were clarified again and then subjected to gel filtration on a Superdex75 16/300 column (GE Healthcare) using modified HBSP+ running buffer (10 mM HEPES, 300 mM NaCI, 50 μΜ EDTA, 0.05% P20, 10% glycerol, pH 7.4). Peak fractions were pooled and concentrated into HBSP+ storage buffer (as above but 150 mM NaCI and 1 mM DTT) with centrifugal spin concentrators (GE Healthcare) with 10 kD size exclusion limit. Protein concentrations were determined from absorbance at 280 nm of native protein samples using sequence-based extinction coefficients [27] as calculated by the ProtParam server [28].

In-vitro FRET

Measurements were performed in 150 μΙ volumes in black flat-bottom 96-well plates with 0.3 μΜ final donor and 0.5 μΜ final acceptor concentration (for donor FRET efficiency) or, vice versa, 0.5 μΜ final donor and 0.3 μΜ acceptor concentration (for sensitized emission) in HBSP+ buffer (see above) at pH

7.4 and 25 to 27 °C. All proteins were pre-diluted to 15 μΜ and then diluted to working solutions of 0.45 μΜ donor 9 and 1.5 μΜ acceptor (donor FRET) or 0.75 μΜ donor and 0.9 μΜ acceptor (sensitized emission). Six replicas each of three samples were prepared for each protein pair on a single plate: (D) 100 μΙ donor + 50 μΙ buffer, (A) 100 μΙ buffer + 50 μΙ acceptor, (AD) 100 μΙ donor + 50 μΙ acceptor. The FRB / FKBP12 interaction was triggered by adding 2 μΙ 112.5 μΜτβρΒΓηγάη to a final concentration of

1.5 μΜ. After 15 s shaking, donor fluorescence was measured at 495 nm excitation and 530 nm emission wavelengths. Sensitized emission was determined at 516 nm (Citrine) excitation and 610 nm (mCherry) emission. Measurements were performed before and after addition of rapamycin and, for control of full binding, repeated after an additional dose of 2 μΙ rapamycin. The donor-based FRET efficiency E was calculated as: where F _D is the fluorescence intensity of the donor-only sample (D), and F _AD is the intensity of the donor+acceptor sample (AD) corrected for (in practice negligible) acceptor "bleed-through" F ⁵³⁰ _A . For control purposes, efficiency of FRET was also calculated from measurements before and after addition of rapamycin:

Efficiency of FRET was also calculated, indirectly, from the sensitized emission of the acceptor after donor excitation at 516 nm:

Equation 3 still needs correction for f ⁶¹ ^ , the (non-negligible) donor fluorescence at Cherry emission wavelengths. F ₆₁ o° is determined from the donor-only measurements (D) but is subject to FRET-based quenching in the mixed (AD) sample:

Where / = F _AD /F _D is the correction factor for apparent donor quenching which was determined from 495 nm excitation / 516 nm emission measurements in the same experiment. Note, / = 1-E _app is derived from the apparent donor-based FRET efficiency E _app which was determined according to equation 1 but is lower than E _D above, owing to the fact that the donor was not fully occupied during the sensitized emission measurements. All fluorescent intensities F were averages of five to six replicates. Overall standard deviations for E were determined by error propagation. Standard deviations reported for E _A do not include the error arising from the determination of extinction coefficients ε ⁵¹⁶ _Λ and s ⁵¹⁶ _D (Table 4). Owing to these additional uncertainties, E _A is unlikely to perfectly agree with E _D but serves as a consistency check. Large deviations of E _A would indicate errors in protein concentrations or systematic defects in binding or fluorescent domains.

In vitro Caspase3 assay with purified Caspase sensors

Recombinant human Caspase 3 was purchased from R&D Systems, MN. Caspase reactions were performed in 150 μΙ volumes in black flat-bottom 96-well plates with 0.5 μΜ FRET sensor concentration in the buffer recommended by the manufacturer manufacturer (25 mM HEPES pH 7.5, 10 mM DTT, 0.1 % CHAPS) at 30 °C . Donor fluorescence (excitation 434 nm, emission 474 nm) and sensitized emission (excitation 434, emission 529 nm) were measured on a Tecan M1000 plate reader with 5 nm bandwidth for excitation and emission. 5 μΙ Caspase 3 containing a total of 50 ng enzyme (or buffer as control) were then added to each well simultaneously. After short initial shaking, donor and sensitized emission signals were recorded every 30 s for 40 min. ln-vitro FLIM

FLIM (fluorescence lifetime imaging)-FRET was measured by time-correlated single-photon counting (TCSPC) with an inverted multiphoton laser scanning microscope (Leica TCS SP5) using a 63x water immersion N.A 1.2 Plan-Apochromat objective, and equipped with a single molecule detection platform and single-photon counting electronics (PicoHarp 300) from PicoQuant GmbH (Berlin). Donor (mCitrine) two-photon excitation was performed at 950nm from a Mai Tai ThSapphire laser (Spectra Physics) with a repetition rate of 80 MHz. Photons were detected by a SPAD set up (PicoQuant). A fluorescence bandpass filter (500-550 nm) limited the detection to the donor fluorescence only.

In vitro FLIM measurements were performed in 500μΙ volumes in Lab-Tek 8-well coverglass dishes with 0.3 μΜ final donor and 0.5 μΜ final acceptor concentration in HBSP+ buffer at pH 7.4. Three samples were prepared for each protein pair on a single dish: (D) 250μΙ donor + 250μΙ buffer, (DA) 250μΙ donor + 250μΙ acceptor, (DA) 250μΙ donor + 250μΙ acceptor. The FRB - FKBP12 interaction was triggered by adding 5μΙ rapamycin to a final concentration of 1.5μΜ. Donor fluorescence was measured before and after addition of rapamycin. Fluorescent decay curves were analyzed in IGOR Pro (WaveMetrics, Portland, Oregon). Mean FRET efficiency values, E, were calculated from: where r _DA is the amplitude-weighted mean fluorescence lifetime of the donor (mCitrine) in the presence of both acceptor (mCherry) and rapamycin. r _D is the mean fluorescence lifetime of the donor (mCitrine) in the presence of acceptor (mCherry) without rapamycin. For non-FRET conditions, T _D of the donor in the presence of the acceptor but without rapamycin was calculated from a mono- exponential fit to the fluorescence lifetime decays. Under FRET conditions, experimental decay curves were fit to a stretched bi-exponential model [18]. The non-interacting protein's lifetime was fixed to T _D and the value of T _DA and stretching factor β were estimated.

ln-vivo FLIM

FRB/ FKBP12 interaction

HeLa cells were cultured at 37°C and 5% C02 in DMEM(Gibco), supplemented with 10% FBS, 1% glutamate, and 1% pen/strep. After 48 h, 1x10 ^s cells were seeded on glass-bottomed 6-cm cell culture plates (MatTek Corp.) and grown overnight. Cells were transfected with 4 μg of donor only (control) or double transfected with 4 μg of donor- and 4 μg of acceptor-expressing vectors using Lipofectamine 2000 (Invitrogen) in OptiMEM media for 24 hours at 37°C according to the manufacturer's instructions. Media was changed to imaging buffer containing no Phenol Red and Rapamycin was added at a concentration of 1 μΜ in 1.5 ml imaging buffer 30 min prior to FLIM experiments. mCitrine and mCherry were visualized on the same Leica TCS SP5 confocal microscope with a 40x 1.25NA objective at a zoom factor of 3x (1024x1024 pixels, 0.126 μιτι/pixel). FLIM was performed on 5 fields per sample and analyzed as described above. FRET efficiency images in Figure 11 were calculated with the FLIM- FRET script from SymphoTime software. IRF (instrument response function) and MLE (maximum likelihood estimator) were applied in the reconvolution fit for the calculation of the lifetimes. Ras / Raf interaction

HEK293 cells were cultured as described for HeLa cells above and 1x10 ^s cells were seeded on MatTek 6 cm cell culture plates 24 h before transfection. Cells were transfected or double transfected as before and the OptiMem media was changed 6 h after transfection to DMEM with 10% FBS, 1% Pen/Strep, and 1% glutamate for 12 h before FRET measurements. EGF was added at a concentration of 0.05 μg/ml in 1.5 ml imaging buffer 4 minutes prior to FLIM experiments. mCitrine and mCherry were visualized with a 63x 1.25NA objective at a zoom factor of 4x. FLIM was performed on 5 fields per sample (4 minutes between FLIM readings). Regions of interest were selected in ImageJ [44] and cell or membrane areas within were defined automatically using custom ImageJ macros. Fluorescent decay data were analyzed in IGOR Pro (WaveMetrics, Portland, Oregon) using the pFLIM software module [31] following the authors' instructions. The pFLIM module was also used for the generation of intensity, intensity-weighted lifetime and FRET efficiency images. Rafl / BRaf interaction

HeLa cells were cultured as described above. 2.5x105 cells were seeded on MatTek 6 cm cell culture plates 24 hours before transfection. Cells were transfected with 4 μg of donor only (control) or double transfected with 4 μg of donor- and 4 μg of acceptor-expressing vectors using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions; media was changed 6 hours after transfection to DMEM with 10% FBS, 1% Pen/Strep, and 1% glutamate for 12 h before FRET measurements. mCitrine lifetime was measured before and 2h after treatment with GDC-0879 lOuM or DMSO (as a control). FLIM images were recorded with a 40x 1.25NA Oil PlanApo objective at a zoom factor of 1.7x (256x256 pixel). FLIM was performed on 3 fields per sample. Fluorescent decay data were analyzed with IGOR Pro / pFLIM as above.

Caspase3 assay in HeLa cells

HeLa cells were cultured at 37 and 5% C02 in DMEM (Gibco), supplemented with 10% FBS, 1% glutamate, and 1% pen/strep. 2.5x105 cells were seeded on MatTek 6 cm cell culture plates 24 hours before transfection. Cells were transfected with 4 μg of caspase3 sensor expressing vectors using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions; media was changed 6 hours after transfection to DMEM with 10% FBS, 1% Pen/Strep, and 1% glutamate for 12 h before FRET measurements. Media was changed to imaging buffer (DMEM without Phenol Red). Staurosporine was added at a concentration of 2 μΜ in 2 ml imaging buffer 10 min after ratiometric FRET experiment started. Images were acquired with a Leica TCS SP5 confocal microscope using a 63x 1.4 NA PL APO objective (Leica Microsystems GmbH). For each Caspase sensor, images were obtained every 5 minutes for 5 hours in 12 different positions of the same culture dish. Cells were excited with a 458 nm laser and donor and FRET channels were imaged with 2 hybrid detectors (HyD) with a 458/514 double dichroic beamsplitter and acquisition bandpass filters set between 469 and 507 nm and between 523 and 576 nm, respectively. Image size was 512x512 pixels and laser scanning speed was set to 400 Hz bidirectional. Laser AOTF and HyD voltage settings were kept constant for all the tested sensors for comparative purposes. Image processing and analysis was performed with FIJI software. Briefly, images were median filtered for noise reduction and background was subtracted from ROIs. After this processing ratio images were calculated dividing the FRET channel by the Donor channel. Ratiometric values of at least 30 cells were measured for each Caspase sensor.

Reverse FRET peptide binding assay

Binding affinities of selected domain-peptide helper interactions were estimated from FRET signals at varying concentrations of FRB-mCherry-peptide proteins. However, at higher acceptor concentrations, FRET cannot any longer be directly quantified from the comparison of donor-only and donor + acceptor samples. The intense coloration of already low μΜ mCherry samples obscures the donor fluorescence with an additional absorbance component that is difficult to quantify. While we are thus lacking a clear reference sample for the unbound state, we can, conversely, use the rapamycin-induced binding as a reference for the fully bound (Y=l) state of the donor-acceptor complex. Without rapamycin, peptide- domain binding leads to a reduction of the fluorescence

(6)

Where a is the degree of binding between donor (D) and acceptor (A): a =Y = DA/(DA+D), F0 is the (here unknown) donor fluorescence in absence of any FRET and E is the molecular FRET efficiency. We can assume that rapamycin induces 100 % binding between FRB and FKBP12 domains (K _D < 1 nM[14]). The donor emission with rapamycin thus becomes

It ί ^ 1 ILuit ί ^ ^

(7)

The absolute FRET efficiency can be determined as described above at lower donor and acceptor concentrations (0.5 μΜ) where we do not observed peptide - domain interactions. The unknown F0 is eliminated by combining equations 6 and 7 to F

(8)

From which a can be determined as

Peptide binding assays were performed on 384 well plates to reduce sample amounts. Per well, ΙΟμΙ of 1.5 μΜ donor (mCitrine) solution were mixed with 20 μΙ from a dilution series of peptide-bearing acceptor protein. Donor fluorescence was measured as before on a plate reader with 495 nm excitation and 530 nm emission wavelength. 2 μΙ 112.5 μΜ rapamycin were then added and the measurement repeated. All dilution series and measurements were performed in triplicates. E = 1-F+/F- was determined at low acceptor concentrations between 0.6 μΜ and 1 μΜ. The degree of binding Y was calculated according to equation 8 from the mean of fluorescence readings. Standard deviations for each Y were determined by error propagation:

where / = F-/F+. KD was determined from a numeric least-square fit of a 1:1 binding model Y = A/(K _D +A) to Y and errors using Graph Pad Prism (GraphPad Software Inc, La Jolla).

Table 1: Domain / peptide helper interactions

Pair helper interaction module K| in μΜ mCitrine mCherry expected measured

WW/ Wpl Hs Yap65 WW (L30K) (GS)GTPPPPYTVG(TG) 40 ^{l h}

WW/ Wpl Hs Yap65 WW (L30K) ίΠ^ ΡΙ ΡΥΤίΤΠ^ 500 ¹⁶ ' ^e 170±6.8

S H 31 > l ^{, 7} ' ^ft 30± 1.2

SH 1 Sj 2i Sc Sho 1 3H3 (GS)IVNKPLAPLPV(TG) ~500 ¹⁸ ' ^c > 1000

" measured with less stable wt (L30) WW domain; * not measured, modified from strongest binder (IRSKPLP- PLPV); estimated from a position-weighted matrix (PWM) model;

Table 2: Selected FRET efficiency results

FRET pair in-vitro intensity in-vitro FLIM cell FLJM

Citrine constructs mCherry constructs helper donor ⁰ sens, emission* ^amp count'' tamp ¹

44 FK-20-Cit(A) 41 FR-20-Che - 0.237 ±0.006 0.220 ±0.008 0.222 0.250 0.08 ±0.06

64 KK-20-Citi Ki W W 41 FR-20-Che - 0.232 ±0.008 0.212 ±0.012 0.193 0.246

70 FR-20-Che.Wpl strong 0.438 ±0.008 0.410 ±0.005 0.530 0.467

71 FR-20-Che.Wp2 weak 0.413 ±0.013 0.393 ±0.004 0.410 0.392 0.20 ±0.08

51 FK-20-€it(K).SH3 41 FR-20-Che - 0.202 ±0.002 n.d. 0.224 0.228

52 FR-20-Che Spl strong 0.486 ±0.01 1 n.d. 0.428 0.541

53 FR-20-Che Sp2 weak 0.320 ±0.01 1 n.d. 0.254 0.288

Selected absolute FRET efficiencies (between 0.0 and 3.0 ± standard deviation) determined with different methods. Enhanced FRET pairs recommended for application are highlighted in bold face, " FRET efficiency <E) based on donor quenching measured in vitro (plate reader); ^b E based on acceptor sensitized emission measured in vitro (plate reader); ^c E based on decrease of amplitude- weigh ted average lifetime (single measurement); ^d E based on decrease of photon count during same experiment (single measurement); Note, IDs refer to the full list of protein constructs given in a separate table.

Table 3: Synthetic proteins constructed in this study.

ID name composition comment

bdg ^a link ^fo FP ^C helper^ reference proteins without enhancement

41 FR-20-Che FRB 24 mCherry

44 FK-20-Cit(A) FKBP 24 Citrine

46 FK-20-Cit(K) FKBP 24 mCitrine

93 FK-20-TFP FKBP 24 mTFPl (Teal)

SH3 / peptide helper module

50 FK-20-Cit(K)~SH3 FKBP 24 mCitrine SH3 full-length Citrine

51 FK-20-Cit(K).SH3 FKBP 24 mCitrine SI 13 Citrine C trimmed

52 FR-20-Che~Spl FRB 24 mCherry strong SH3 bdg full-length Cherry

53 FR-20-Che~Sp2 FRB 24 mCherry weak SH3 bdg full-length Cherry

54 FR-20-Che.Spl FRB 24 mCherry strong SH3 bdg Cherry C trimmed

55 FR-20-Che.Sp2 FRB 24 mCherry weak SH3 bdg Cherry C trimmed

WW / peptide helper module

64 FK-20-Cit(K)~WW FKBP 24 mCitrine WW full-length Citrine

65 FK-20-Cit(K).WW FKBP 24 mCitrine WW Citrine C trimmed

68 FR-20-Che~Wpl FRB 24 mCherry strong WW bdg full-length Cherry

69 FR-20-Che~Wp2 FRB 24 mCherry weak WW bdg full-length Cherry

70 FR-20-Che.Wpl FRB 24 mCherry strong WW bdg Cherry C trimmed

71 FR-20-Che.Wp2 FRB 24 mCherry weak WW bdg Cherry C trimmed mTFPl / mCherry with WW / peptide helper module

90 FK-20-TFP~WW FKBP 24 mTeal WW

see separate table for Ras/Raf sensor constructs

" binding domain; ^h length of flexible linker in amino acids (sequence TG[GS] _X TG with x=10 or x=20); ^c fluorescent protein domain; ^d helper domain or peptide,

Table 3: Synthetic proteins constructed in this study (continued).

ID composition ⁰ helper ¹ ' comment

100 mCitrine- 14aa-cRaf _ full-length Rafl (c-Raf) without enhancement

101 mCitrine~W W- 14aa-cRaf WW full-length Rafl with WW domain

102 RBD-mCitrine _ classic sensor; Ras-binding-domain of Rafl

103 mCherry.X-50aa-hRas _ mCherry with nonbinding peptide; 50aa random linker; H-Ras

104 mCherry. Wp2-50aa-hRas Wp2 mCherry with weak WW peptide; 50aa random linker; H-Ras

105 mCherry-6aa-hRas classic sensor; short-linker; to be paired with RBD

" domain composition from N- to C-terminal; 14aa ... linker TG[GS]\oTG; 50aa ... randomized flexible linker (see text for details); 6aa ... short linker SGGSGT. * helper module if any Table 3: Synthetic proteins constructed in this study (contin

a domain composition from N- to C-terminal; 50 aa: randomized flexible linker (see text for details); helper module if any

Table 3: Synthetic proteins constructed in this study (contin

³ domain composition from N- to C-terminal; 50 aa: randomized flexible linker (see text for details); ^b helper module if any

Table 4: Extinction coefficients in M ^~1 cm

ε (468 nm) ε (495 nni) ε (516 nm) ε (587 nm) mTFP ^a 41900 ±3650

(m)Citrme* 23500 ±850 89600 ±1950

mCherry ^c 2600 ±200 7200 ±390 14400 ±1000 60000 ±1200

Exerimental extinction coefficients (peak values in bold face).

a determined from 2 different proteins in 4 measurements; * determined from 10 different protein; 0 determined from 7 different proteins

Table 5: Properties of common fluorescent protein variants per spectral class (1)

) References: [3, 34, 35 US patent application US2006/0275827; www.eyrogen.com; www.clontech.com Table 6: Mutations of common fluorescent protein variants [3, 11]

such as K26R, Q80R, N146H, H231L, etc.variants

Many GFP variants contain V inserted after Metl so that the mRNA should contain an ideal translational start sequence. We number such a V as la to preserve wild-type numbering for the rest of the sequence.

Table 7: Overview amino acid and nucleotide sequences used in this study.

Non-limiting examples of reference and engineered polypeptides.

SEQ IDs SEQ IDs

(AA sequences) (NT sequences)

Synthetic protein constructs ^{(2) (3)}

FKBP-20AA-mCitrine~WW (=ID 64) 29

F B-20AA-mCherry.Wp2 (=ID 71) 30

FKBP-20AA-mTFPl~WW (=ID 90) 31

FKBP-20AA-mCitrine (=ID 46) 32

FRB-20AA-mCherry (=ID 41) 33

FKBP-20AA-mTFPl (=ID 93) 34 mCitrine-lOAA-cRaf (=ID 100) 35

mCitrine~WW -lOAA-cRaf (=ID 101) 36

RBD-mCitrine (=ID 102) 37

mCherry.X-50AA-hRas (=ID103) 38

mCherry.Wp2-50AA-hRas (=ID 104) 39

mCherry-6AA-hRas (=ID 105) 40 cRaf-50aa-mCitrine 77

cRaf-50aa-mCitrine~WW 78

bRaf-50aa-mCherry 79

bRaf-50aa-mCherry.Wp2 80 rg2130 (conventional caspase sensor) 88

rg2131 (enhanced caspase sensor) 89

rg2132 (control for domain insertion caspase sensor) 90

rg2133 (caspase sensor enhanced through domain 91

insertion)

For nomenclature, see also Table 1

For nomenclature, see also Table 3

Nucleotide sequences of synthetic protein constructs can be easily derived by combining the ucleotide sequences as provided in SEQ ID NOs: 41-68, 81, 92-97.

REFERENCES

[I] Miyawaki, A. Development of probes for cellular functions using fluorescent proteins and fluorescence resonance energy transfer. Annual review of biochemistry 80, 357-73, June (2011).

[2] Michnick, S. W., Ear, P. H., Landry, C, Malleshaiah, M. K., and Messier, V. A toolkit of protein- fragment complementation assays for studying and dissecting large-scale and dynamic protein- protein interactions in living cells. Methods in Enzymology 470, 335-368 (2010).

[3] Shaner, N. C, Steinbach, P. A., and Tsien, R. Y. A guide to choosing fluorescent proteins. Nature Methods 2(12), 905-909, December (2005).

[4] Nguyen, A. W. and Daugherty, P. S. Evolutionary optimization of fluorescent proteins for intracellular FRET. Nat Biotech 23(3), 355-360, March (2005).

[5] Zacharias, D. A. Partitioning of Lipid-Modified Monomeric GFPs into Membrane Microdomains of Live Cells. Science 296(5569), 913-916 (2002).

[6] Vinkenborg, J. L., Evers, T. H., Reulen, S. W. A., Meijer, E. W., and Merkx, M. Enhanced Sensitivity of FRET-Based Protease Sensors by Redesign of the GFP Dimerization Interface. ChemBioChem 8(10), 1119-1121 (2007).

[7] Ohashi, T., Galiacy, S. D., Briscoe, G., and Erickson, H. P. An experimental study of GFP-based FRET, with application to intrinsically unstructured proteins. Protein Science 16(7), 1429-1438 (2007).

[8] Merkx, M. Rational Design of FRET-Based Sensor Proteins. In Reviews in Fluorescence 2008,

Geddes, C. D., editor, 69-87. Springer, New York (2010).

[9] Kotera, I., Iwasaki, T., Imamura, H., Noji, H., and Nagai, T. Reversible Dimerization of Aequorea victoria Fluorescent Proteins Increases the Dynamic Range of FRET-Based Indicators. ACS Chemical

Biology 5(2), 215-222, February (2010).

[10] Griesbeck, O., Baird, G. S., Campbell, R. E., Zacharias, D. A., and Tsien, R. Y. Reducing the environmental sensitivity of yellow fluorescent protein. Mechanism and applications. The Journal of Biological Chemistry 276(31), 29188-29194, August (2001).

[II] Shaner, N. C, Campbell, R. E., Steinbach, P. A., Giepmans, B. N. G., Palmer, A. E., and Tsien, R. Y.

Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nature Biotechnology 22(12), 1567-1572, December (2004).

[12] Akrap, N., Seidel, T., and Barisas, B. G. Forster distances for fluorescence resonant energy transfer between mCherry and other visible fluorescent proteins. Analytical Biochemistry 402(1), 105-106, July (2010).

[13] Belshaw, P. J., Ho, S. N., Crabtree, G. R., and Schreiber, S. L. Controlling protein association and subcellular localization with a synthetic ligand that induces heterodimerization of proteins. Proceedings of the National Academy of Sciences of the United States of America 93(10), 4604- 4607, May (1996). [14] Grunberg, R., Ferrar, T., and Sloot, A. V. D. Building blocks for protein interaction devices. Nucleic acids 38(8), 2645-2662, May (2010).

[15] Pires, J., Taha-Nejad, F., Toepert, F., Ast, T., Hoffmuller, U., Schneider-Mergener, J., Kuhne, R., Macias, M., and Oschkinat, H. Solution structures of the YAP65 WW domain and the variant L30 K in complex with the peptides GTPPPPYTVG, N-(n-octyl)-GPPPY and PLPPY and the application of peptide libraries reveal a minimal binding epitopel. Journal of molecular biology 314(5), 1147- 1156, December (2001).

[16] Zarrinpar, A., Park, S.-H., and Lim, W. A. Optimization of specificity in a cellular protein interaction network by negative selection. Nature 426(6967), 676-680, December (2003).

[17] Tonikian, R., Xin, X., Toret, C. P., Gfeller, D., Landgraf, C, Panni, S., Paoluzi, S., Castagnoli, L,

Currell, B., Seshagiri, S., Yu, H., Winsor, B., Vidal, M., Gerstein, M. B., Bader, G. D., Volkmer, R.,

Cesareni, G., Drubin, D. G., Kim, P. M., Sidhu, S. S., and Boone, C. Bayesian Modeling of the Yeast

SH3 Domain Interactome Predicts Spatiotemporal Dynamics of Endocytosis Proteins. PLoS Biol

7(10), el000218, October (2009).

[18] Lee, K. C, Siegel, J., Webb, S. E., Leveque-Fort, S., Cole, M. J., Jones, R., Dowling, K., Lever, M. J., and French, P. M. Application of the stretched exponential function to fluorescence lifetime imaging. Biophysical journal 81(3), 1265-74, September (2001).

[19] Ward et al., Photochem. Photobiol. 35:803-808, 1982

[20] Levine et al., Comp. Biochem. Phvsiol. 72B:77-85, 1982

[21] Matz et al, Nature Biotechnology 17:969-973, 1999

[22] Grecco, H. E., Roda-Navarro, P., Girod, A., Hou, J., Frahm, T., Truxius, D. C, Pepperkok, R., Squire,

A., and Bastiaens, P. I. H. In situ analysis of tyrosine phosphorylation networks by FLIM on cell arrays. Nature methods 7(6), 467-72, June (2010).

[23] Ai, H.-wang, Henderson, J. N., Remington, S. J., & Campbell, R. E. (2006). Directed evolution of a monomeric, bright and photostable version of Clavularia cyan fluorescent protein: structural characterization and applications in fluorescence imaging. The Biochemical journal, 400(3), 531-

40. Portland Press Ltd.

[24] Prasher et al, Gene 111 :229-233, 1992

[25] Heim et al, Proc. Natl Acad. ScL USA 91:12501-12504, 1994

[26] Chalfie et al., Science 263:802-805, 1994,

[27] Pace, C. N., Vajdos, F., Fee, L., Grimsley, G., and Gray, T. How to measure and predict the molar absorption coefficient of a protein. Protein Science 4(11), 2411-23, November (1995).

[28] Gasteiger, E., Hoogland, C, Gattiker, A., Duvaud, S., Wilkins, M. R., Appel, R. D., and Bairoch, A.

Protein identification and analysis tools in the ExPASy server. In The proteomics protocols handbook, Walker, J. M., editor, volume 112, 571-607. Humana Press, Totowa, NJ, January (2005).

[29] Baird, G. S., Zacharias, D. A., & Tsien, . Y. (1999). Circular permutation and receptor insertion within green fluorescent proteins. PNAS, 96(20), 11241-11246. doi:10.1073/pnas.96.20.11241;

[30] Nagai, T., Yamada, S., Tominaga, T., Ichikawa, M., & Miyawaki, A. (2004). Expanded dynamic range of fluorescent indicators for Ca(2+) by circularly permuted yellow fluorescent proteins. PNAS, 101(29), 10554-9.

[31] Walther, K. A., Papke, B., Sinn, M. B., Michel, K., & Kinkhabwala, A. (2011). Precise measurement of protein interacting fractions with fluorescence lifetime imaging microscopy. Mol. BioSyst, 7, 322-336. The Royal Society of Chemistry.

[32] Sun et al. 2011 Nature protocols, 6, 1324-1340

[33] Dosztanyi, Z., Csizmok, V., Tompa, P., & Simon, I. (2005). lUPred: web server for the prediction of intrinsically unstructured regions of proteins based on estimated energy content. Bioinformatics (Oxford, England), 21(16), 3433-4.

[34] Shcherbo, D., Merzlyak, E. M., Chepurnykh, T. V., Fradkov, A. F., Ermakova, G. V., Solovieva, E. A., Lukyanov, K. A., et al. (2007). Bright far-red fluorescent protein for whole-body imaging. Nature methods, 4(9), 741-6.

[35] Shcherbo, D., Murphy, C. S., Ermakova, G. V., Solovieva, E. A., Chepurnykh, T. V., Shcheglov, A. S., Verkhusha, V. V., et al. (2009). Far-red fluorescent tags for protein imaging in living tissues. The Biochemical journal, 418(3), 567-74.

[36] Pawson, T., & Nash, P. (2003). Assembly of cell regulatory systems through protein interaction domains. Science, 300(5618), 445-52.

[37] Ghaemmaghami, S., Huh, W.-K., Bower, K., Howson, R. W., Belle, A., Dephoure, N., O'Shea, E. K., et al. (2003). Global analysis of protein expression in yeast. Nature, 425(6959), 737-41.

[38] Rocks, O., Peyker, A., Kahms, M., Verveer, P. J., Koerner, C, Lumbierres, M., Kuhlmann, J., et al.

(2005). An Acylation Cycle Regulates Localization and Activity of Palmitoylated Ras Isoforms. Science, 307(5716), 1746 -1752.

[39] Gibson, D. G., Young, L, Chuang, R.-Y., Venter, J. C, Hutchison, C. A., & Smith, H. O. (2009).

Enzymatic assembly of DNA molecules up to several hundred kilobases. Nature methods, 6(5), 343-5.

[40] Evers, T. H., van Dongen, E. M. W. M., Faesen, A. C, Meijer, E. W., & Merkx, M. (2006).

Quantitative understanding of the energy transfer between fluorescent proteins connected via flexible peptide linkers. Biochemistry, 45(44), 13183-92.

[41] Ganesan, S., Ameer-Beg, S. M., Ng, T. T. C, Vojnovic, B., & Wouters, F. S. (2006). A dark yellow fluorescent protein (YFP)-based Resonance Energy-Accepting Chromoprotein (REACh) for Forster resonance energy transfer with GFP. Proceedings of the National Academy of Sciences of the United States of America, 103(11), 4089-94.

[42] Holm L, Rosenstrom P (2010) Dali server: conservation mapping in 3D. Nucl. Acids Res. 38, W545- 549) as well as Tm-align.

[43] Y. Zhang, J. Skolnick (2005) TM-align: A protein structure alignment algorithm based on TM-score ,

Nucl. Acids Res. 33: 2302-2309).

[44] Abramoff, M.D., Magalhaes, P.J., Ram, S.J. (2004) Image Processing with ImageJ. Biophotonics

International 11, issue 7, pp. 36-42.

[45] Poulikakos, P. I., Zhang, C, Bollag, G., Shokat, K. M., & Rosen, N. (2010). RAF inhibitors transactivate RAF dimers and ERK signalling in cells with wild-type BRAF. Nature, 464(7287), 427-

30. doi:10.1038/nature08902

[46] Heidorn, S. J., Milagre, C, Whittaker, S., Nourry, A., Niculescu-Duvas, I., Dhomen, N., Hussain, J., et al. (2010). Kinase-dead BRAF and oncogenic RAS cooperate to drive tumor progression through

CRAF. Cell, 140(2), 209-21. doi:10.1016/j. cell.2009.12.040

[47] Hatzivassiliou, G., Song, K., Yen, I., Brandhuber, B. J., Anderson, D. J., Alvarado, R., Ludlam, M. J. C, et al. (2010). RAF inhibitors prime wild-type RAF to activate the MAPK pathway and enhance growth. Nature, 464(7287), 431-5. doi:10.1038/nature08833

[48] Jensen KK, Martini L, Schwartz TW. Enhanced fluorescence resonance energy transfer between spectral variants of green fluorescent protein through zinc-site engineering. Biochemistry. 2001

40:938-45.

[49] Baird, G. S., Zacharias, D. A., & Tsien, R. Y. (1999). Circular permutation and receptor insertion within green fluorescent proteins. Proceedings of the National Academy of Sciences, 96(20), 11241-11246.

[50] Luo, K. Q., Yu, V. C, Pu, Y., & Chang, D. C. (2001). Application of the fluorescence resonance energy transfer method for studying the dynamics of caspase-3 activation during UV-induced apoptosis in living HeLa cells. Biochemical and biophysical research communications, 283(5), 1054-60.

[51] Goedhart, J., Von Stetten, D., Noirclerc-Savoye, M., Lelimousin, M., Joosen, L, Hink, M. a., Van Weeren, L, et al. (2012). Structure-guided evolution of cyan fluorescent proteins towards a quantum yield of 93%. Nature Communications, 3, 751.