GB2375538A

GB2375538A - Polypeptide constructs for FRET analysis

Info

Publication number: GB2375538A
Application number: GB0203683A
Authority: GB
Inventors: Peter Andrew Chalk; Peter Nicholas Lowe
Original assignee: Glaxo Group Ltd
Current assignee: Glaxo Group Ltd
Priority date: 2001-02-15
Filing date: 2002-02-15
Publication date: 2002-11-20
Also published as: GB0203683D0; GB0103763D0

Abstract

A polypeptide chain comprising a linker of at least 27 amino acids sandwiched between a donor and an acceptor domain, wherein FRET occurs between said donor and acceptor. The linker molecule comprises at least one binding region that may change conformation when bound by its binding partner. The linker may be the Cdc42/Rac interactive binding motif (CRIB) and the binding partner is chosen from the Ras superfamily.

Description

PAK-GFP Construct This invention relates to a novel PAK-GFP construct and its binding to Ras superfamily proteins, especially Rac and Cdc42. Moreover, it relates to a method for studying signal transduction via a Ras superfamily protein to its effectors in vivo.

Background Non-radiative transfer of energy (Fluorescence Resonance Energy TransferFRET) between fluorescent dyes is well established and has been used to study the association and disassociation of molecular complexes. The energy transfer is strongly dependent upon the distance between donor and acceptor fluorophores such that transfer is maximal at short distances. FRET is most efficient when the overlap of donor emission and acceptor excitation spectra is maximal, with high quantum yields of the donor and a high extinction coefficient of the acceptor. The green fluorescent protein (GFP) from Aquoria victoria was the first naturally fluorescent protein to be studied extensively. As a spontaneously fluorescent protein, GFP allows one to generate fluorescence, using standard molecular cloning technology, intracellularly in live, cultured cells and even in whole organisms. Proteins tagged with GFP can often be expressed in high yields. Mutagenesis and protein engineering on GFP has yielded several mutants with different emission and excitation spectra. GFP donor/acceptor pairs with high quantum yields and high extinction coefficients can give high efficiencies of FRET. Papers appeared in the literature demonstrating FRET between mutants of GFP when they have been cloned as proteins linked together by bridges of 20,25, and 18 amino acids (9,10, 11), and also when they have been linked by larger protein domains such as calmodulin and its binding peptides. Also, FRET has been demonstrated on association of GFP mutants cloned as fusion proteins with different protein binding partners (12,13, 14). It was generally believed that with increasing amino acid length FRET would diminish. One feature of GFP from Aquoria victoria that has been studied is its tendency to form dimers in vitro. Whereas fluorescent proteins isolated from luminescent species are often obligate dimers it is reported that

GFP from Aquoria victoria is predominantly a monomer but it forms dimers in vitro at concentrations above 5mg/ml (15). Similarly crystal structures of GFP and its mutants show predominantly a dimeric structure with the monomers held together by hydrophobic interactions (16).

The Ras superfamily of signal transduction proteins consists of several families of GTP binding proteins that form cell signalling networks controlling numerous cellular processes that are important in both normal and disease states. A common feature of signalling by Ras superfamily proteins is that it is mediated by the binding of the GTP activated form of the protein to downstream protein effectors, and is terminated by hydrolysis of GTP to GDP and the disassociation of the Ras protein from its downstream effector. There are over fifty Ras superfamily members, each of which are activated by one or more exchange factors and mediate downstream signals by interacting with one or more effectors. This enables the control and integration of a large number of signalling pathways. Specifically, Cdc42, Rac and Rho are members of the Rho subfamily that control processes such as morphogenesis and cytoskeletal organisation, neutrophil activation, mitogenesis and transformation, and protein kinase cascades (1,2). Upon activation they bind and activate several downstream effectors, such as p67Phox, Wiskott-Aldrich Syndrome proteins (WASPs), activated Cdc42-associated kinases (ACKs), and the p21-activated kinases (PAKs2) (3). Some, such as WASP and ACK, are specific for Cdc42, some are specific for Rac, e. g. Phox, and some appear to interact with both Rac and Cdc42, e. g. PAK. PAK was originally isolated from rat brain and recognised as a homologue of a yeast kinase STE20 that is a member of a MAP kinase cascade in Saccharomyces cerevisiae (4). PAKs contain a serine/threonine kinase domain and a conserved sequence, known as the Cdc42/Rac interactive binding motif (CRIB) which is required for binding to Cdc42 and Rac. We have shown previously that the PAK CRIB motif alone is insufficient for high affinity binding to Cdc42/Rac. However, we have now found that a 44 amino acid region of PAK (residues 75-118) containing this motif is sufficient for high affinity binding to

occur (5). Three-dimensional structures of Cdc42 complexed with this region of PAK (6) and to similar regions of ACK (7) and WASP (8), which also contain CRIB motifs, show that these effectors bind to the protein in an extended conformation with some defined secondary structure. In contrast, when uncomplexed, these regions of the effectors are either unstructured or have little defined structure.

More specifically, we have now shown that cloning the 44 amino acid Cdc42/Rac binding region of PAK between mutants of GFP allows FRET to occur between the GFP mutants in the absence of Rac, but that binding of Rac to the peptide disrupts this energy transfer. Furthermore the affinity and kinetics of the association of the PAK-GFP construct for Rac is similar to that of PAK to Rac and therefore the GFPs do not disorder the normal interaction of Rac with its effector protein. The GFP-PAK protein can be used to study binding of Rac to its effector proteins using competition assays in vitro, and also to monitor the activation state of Rac/Cdc42 in cells.

On studying the effect of unstructured protein linkers on the efficiency of FRET between tethered GFP mutants we found that, in addition to small linkers, large insertions of at least 50 amino acids can be accommodated between the mutants. We surmised that FRET between the GFP mutants derives from dimerisation of the GFPs and not from a simple tethering of the proteins in close proximity.

Summary of the Invention Accordingly, the present invention provides a polypeptide chain comprising a linker which has at least 27 amino acid residues or more.

The linker is sandwiched between a donor and an acceptor domain.

The linker is a protein domain that is inserted between a donor domain and an acceptor domain such that a novel protein is formed comprising a donor domain, a linker domain and an acceptor domain.

The donor domain, acceptor domain and linker are arranged in such a way that FRET occurs between the donor and acceptor domain.

Contrary to prior art teaching, the present inventors have found that FRET occurs with linkers which have 27 amino acid residues or more.

Accordingly, the present invention thus encompasses linkers which have 27 amino acids or more.

Preferably, the linker is an unstructured linker of 27 amino acids or more.

In fact, FRET seems to increase with increasing length of the linker. Thus the linker according to the present invention preferably has at least 30 amino acid residues or more, more preferably at least 35 amino acid residues or more, and most preferably, at least 40 amino acid residues or more. At a length of 50 amino acids residues FRET is still increasing. Thus linkers of more than 50 amino acid residues, for example, more than 60,70, 80,90, 100 or even as long as 200 amino acids or even more are also encompassed within the present invention.

The linker preferably comprises at least one binding region which is capable of binding to a binding partner and that binding partner is preferably a protein. The binding partner is preferably a partner in a cell signalling pathway. Some binding partners will only bind to the linker in either the activated or unactivated state.

Preferably, the conformation of the linker changes when it binds the binding partner. FRET changes upon the binding region of the linker binding to the

binding partner. It is thought that this may be due to the linker conformation changing in shape upon the linker binding to a binding partner.

Moreover, the present invention also encompasses a method of identifying a binding partner that is capable of binding to a linker, the method comprising : (a) providing a polypeptide according to the invention (b) providing a test substance (c) monitoring FRET activity and thereby determining whether the test substance is a binding partner capable of binding to the linker.

The present invention also includes a method of identifying a modulator that is capable of interfering with binding between a linker and a binding partner, the method comprising: (a) providing a polypeptide according to the present invention (b) providing a binding partner (c) providing a test susbstance (d) determining whether said test substance is a modulator that is capable of interfering with the binding between the linker and the binding partner.

Also, encompassed within the present invention is a method of monitoring the activation state of a binding partner comprising (a) providing a polypeptide according to the present invention (b) monitoring FRET activity and thereby determining whether the binding partner is bound to the linker.

Brief description of Figures Figure 1. Fluorescence properties of protein constructs (A) Emission spectrum of protein construct C37 before (-) and after (--) cleavage with Factor Xa. (B) Comparison of the FRET ratios obtained with different length protein constructs. FRET ratios were calculated from the emission spectra using 382 nm excitation wavelength as described in the Materials and Methods section.

Figure 2. Effect of Q61L Rac. GTP on the FRET ratio of EGFP-PAK-EBFP. The indicated concentrations of Q61L Rac. GTP were added to 0. 65, uM EGFPPAK-EBFP and the FRET ratio of the mixture determined as described in the Materials and Methods section. The solid line represents the best fit to an equation describing a single site binding isotherm.

Figure 3. Displacement of Q61L Rac. GTP from EGFP-PAK-EBFP by GSTPAK and GST-Phox. (A) Q61L Rac. GTP (1M) was added to 0.5 u. M EGFPPAK-EBFP, followed by GST-Phox (1-192) to final concentrations of 0.78 M (a), 2.3 11M (b), 10 zum (c), and 18 Jv) (d). The emission ratio was determined continuously at 509 and 444 nm using an excitation wavelength of 382 nm. (B) Displacement by GST-PAK (75-132) fitted to an equation describing simple competitive inhibition. (C) Displacement by GST-Phox (1-192) fitted to an equation describing simple competitive inhibition.

Figure 4. Isothermal titration calorimetry measurements of binding of Q61 L Rac. GTP to EGFP-PAK-EBFP. Q61L Rac (100 M; preliminary 2.5 J. injection followed by 19 x 10 lli injections) was titrated into 6.25 11M EGFP-PAK-EBFP, 5 11M EGFP-HCV-EBFP or buffer at 30 C. Heat changes were recorded continuously with time on a Microcal isothermal titration calorimeter as described in Materials and Methods. Upper graph-the raw data from the titration of Q61 L Rac. GTP into EGFP-PAK-EBFP and EGFP-HCV-EBFP. Lower graph-molar ratios of Q61 L Rac to EGFP-PAK-EBFP plotted against the heat output per mole of Q61L Rac. GTP. The data were analysed using MicroCal Origin software and fitted to a binding isotherm describing binding at a single site (18) to estimate the Kd and stoichiometry.

Figure 5. Stopped-flow fluorescence changes on the interaction of Q61L Rac. GTP and EGFP-PAK-EBFP. One syringe contained Q61L Rac. GTP and the other contained EGFP-PAK-EBFP. (A) Fluorescence changes observed on

mixing 0.05 M EGFP-PAK-EBFP and 2 M Q61L Rac. GTP (concentrations after mixing). The solid line is the best fit of the data to a single exponential with a first order rate constant of 1.1 s-1. (B) The experiment was repeated at several different concentrations of Q61L Rac. GTP and the observed first order rate constants obtained are plotted against Q61L Rac. GTP concentration. The solid line is the best fit to the data to a straight line with a slope of 0.56 M-1s-1 and an intercept of 0.069 s-1.

Figure 6. Fluorescence changes due to nucleotide exchange on Rac. To a solution of 0.4 M EGFP-PAK-EBFP, 2 ; nM Rac. GDP was added, followed by 10 M GTP and subsequently 2 mM MgCI2. The experiment was repeated, except that 0. zum RhoGAP was added together with the 2 mM MgCl2. Excitation was at 382 nm and emission was monitored at 509 nm.

Detailed description of the Invention The present invention provides a polypeptide chain comprising a linker of 27 amino acid residues or more which is sandwiched between a donor and acceptor domain. The donor domain, linker and acceptor domain are arranged in such a way that FRET occurs between the donor domain and the acceptor domain.

For the purpose of the present invention, Fluorescence Resonance Energy Transfer (FRET) is the non-radiative transfer of energy from a higher energy donor domain (donor fluorophore) to a lower energy acceptor domain (acceptor fluorophore) causing sensitised fluorescence of the acceptor domain and quenching of the donor domain. For efficient energy transfer the fluorescence properties of the donor and acceptor domains must be matched such that the emission wavelengths of the donor overlap with the excitation wavelengths of the acceptor. Efficiency of energy transfer has been described by the Forster equation. (Forster, T. (1948) Ann. Physik (Leipzig) 2: 55-75).

In accordance with the present invention a domain is an independent structural unit within a protein, which can be found alone, or in conjunction with other domains or repeats. Domains can be evolutionarily related. Even though the structure of a domain is not always known it is still possible to define the boundaries in many cases from sequence alone.

Specifically, a donor domain is a protein domain capable of being excited to a higher energy state and donating energy to an acceptor domain. An acceptor domain is a fluorescent protein domain that is capable of being excited to a higher energy state by the transfer of energy from a donor domain and the emission of that energy as light.

Examples of donor and acceptor domains are green fluorescent protein (GFP) donor/acceptor pairs from Aquoria victoria with high quantum yields and high extinction coefficients that can give high efficiencies of FRET. Other examples of

donor/acceptor domain pairs include mutants of GFP, such as enhanced blue fluorescent protein (EBFP) and enhanced green fluorescent protein (EGFP).

Other possible donor domains might be light emitting proteins such as luciferase, other intrinsically fluorescent proteins or proteins with fluorophores attached.

Likewise other possible acceptor domains might be intrinsically fluorescent proteins or proteins with fluorophores attached.

In accordance with the present invention a linker is a protein domain that is inserted between a donor domain and an acceptor domain using molecular cloning technology such that a novel protein is formed comprising a donor domain, a linker and an acceptor domain.

Papers appeared in the literature demonstrating FRET between mutants of GFP when they have been cloned as proteins linked together by bridges of 20,25, and 18 amino acids (9,10, 11), and also when they have been linked by larger protein domains such as calmodulin and its binding peptides. Also, FRET has been demonstrated on association of GFP mutants cloned as fusion proteins with different protein binding partners (12,13, 14). It was generally believed that with increasing amino acid length FRET would diminish.

On studying the effect of protein linkers on the efficiency of FRET between donor/acceptor domains we found that, in addition to small linkers, large insertions of at least 50 amino acids can be accommodated between the mutants. In contrast to our expectations, long linkers inserted between donor/acceptor domains allows the donor/accepor domain to come into closer association. Comparison of the FRET ratios for the different constructs shows that the values increase for the intact proteins as the length of the linker is increased. We have found that FRET increases steadily as the amino acid length of the linker increases from 10 amino acids to 50 amino acids. In fact, maximum FRET has not even been achieved at a length of 50 amino acids, so that even longer amino acid lengths of more than 50 amino acids may be inserted as the linker, maybe even as long as 60,70, 80,90, 100,200 or even

more amino acids. The linkers in accordance with the present invention are preferably unstructured linkers.

In accordance with the present invention, the linker comprises at least one binding region which is capable of binding to a binding partner. Preferably, the binding partner is a protein. More preferably, the binding partner is a partner in a cell signalling pathway. Some binding partners will only bind to the linker when the binding partner is in the activated state, whilst others will only bind in the unactivated state.

Not only does FRET change upon the binding region binding to said binding partner, but the linker conformation changes in shape upon the binding region binding to said binding partner.

A specific example of a linker according to the present invention is PAK including any isoforms. PAKs contain a serine/threonine kinase domain and a conserved sequence, known as the Cdc42/Rac interactive binding motif (CRIB) which is required for binding to Cdc42 and Rac. The CRIB domain (Cdc42/Rac Interactive Binding) is a conserved binding motif that is found in numerous target proteins for both Cdc42 and Rac GTPases. It was first recognised and described

by Burbelo et al in 1995 (Burbelo, P. D. ; Dreschesel, D. and Hall, A. (1995) J.

Biol. Chem. 270, 29071-29074).

We have shown previously that the PAK CRIB motif alone is insufficient for high affinity binding to Cdc42/Rac but a 44 amino acid region of PAK (residues 75- 118) containing this motif is sufficient for high affinity binding to occur (5). Threedimensional structures of Cdc42 complexed with this region of PAK (6) and to similar regions of ACK (7) and WASP (8), which also contain CRIB motifs, show that these effectors bind to the protein in an extended conformation with some defined secondary structure. The CRIB motif is not structured until it is bound to Cdc42/Rac.

Cdc42, Rac ACK and WASP are examples of binding partners according to the present invention which will bind to PAK and/or the PAK CRIB linker motif.

Further examples of protein domains that might be used to derive linkers either in part or as intact domains are the Kringle domain, FIZZY/CDC20 domain, phosphtyrosine interaction domain (PID), RanBP1 domain, TIR domain, TAZ finger, Dbl domain (dbl/cdc24 rhoGRF family), FRIP domain, forkhead-associated (FHA) domain, Band 4.1 family, PWWP domain, single-strand binding protein family, ubiquitin-associated domain, D111/G-patch domain, PKD domain,

ubiquitin doamin, domain found in bacterial signal proteins, combined RanBP1IWASP domain, WHEP-TRS domain, Src homology 2 (SH2) domain, citron homology domain, WW/rsp5/WWP domain, SET-domain of transcriptional regulators (TRX, EZ, ASH1 etc), Src homology 3 (SH3) domain, SOCS domain, C-terminus of STAT-inhibitors, SAM domain (Sterile alpha motif), LIM domain, RING finger, Cohesin domain, WW domain, S4 domain.

Binding partners that will bind to these specific linker are examples of binding partners.

A more detailed description of protein domains that might be used to derive linkers, their specific binding regions and binding partners that bind to the specific binding regions follows: Thus, Kringle domains are believed to play a role in binding mediators (e. g., membranes, other proteins or phospholipids), and in the regulation of proteolytic activity. Kringle domains are triple-looped, disulfide cross-linked domains with small pieces of anti-parallel beta-sheet, found in a varying number of copies, in some serine proteases and plasma proteins The FIZZY/CDC20 domain is found in proteins responsible for the activation of APC/C (Anaphase Promoting Complex/Cyclosome) at the end of mitosis. In yeast CDC20 is required for two microtubule-dependent processes, nuclear movements prior to anaphase and chromosome separation. This domain is also found in a number of, as yet, uncharacterised proteins. These include a mammalian protein, p55CDC, that is present in dividing cells and is associated with protein kinase activity.

Besides SH2, the phosphotyrosine interaction domain (PID or PI domain) is the second phosphotyrosine-binding domain found in the transforming protein Shc. Shc couples activated growth factor receptors to a signaling pathway that regulates the proliferation of mammalian cells and it might participate in the transforming activity of oncogenic tyrosine kinases. The PID of Shc specifically binds to the Asn-Pro-Xaa-Tyr (P) motif found in many tyrosine-phosphorylated proteins including growth factor receptors. The PID comprises 165 residues, 4 of which (Arg67, Arg175, Ser151 and Lys169) are responsible for binding phosphotyrosine on a IIENPQYFSDA' (NPxPY) peptide. The PI domain has a similar structure to the insulin receptor substrate-1 PTB domain, a 7-stranded beta-sandwich, capped by a C-terminal helix. However, the PI domain contains an additional short N-terminal helix and a large insertion between strands 1 and 2, which forms a helix and 2 long connecting loops. The substrate peptide fits into a surface cleft formed from the C-terminal helix and strand 5.

The RanBP1 domain is found in Ran binding proteins. Nup358 contains four Ran binding domains. The structure of the first of these is known.

In Drosophila the Toll protein is involved in establishment of dorso-ventral polarity in the embryo. In addition, members of the Toll family play a key role in innate antibacterial and antifungal immunity in insects as well as in mammals. These proteins are type-I transmembrane receptors that share an intracellular 200 residue domain with the interleukin-1 receptor (IL-1 R), the Tol !/iL-IR homologous region (TIR). The similarity between Toll-like receptors (LTRs) and IL-1 R is not restricted to sequence homology since these proteins also share a similar signaling pathway. They both induce the activation of a Rel type transcription factor via an adaptor protein and a protein kinase. Interestingly, MyD88, a cytoplasmic adaptor protein found in mammals, contains a TIR domain associated to a DEATH domain. Besides the mammalian and Drosophila proteins, a TIR domain is also found in a number of plant proteins implicated in host defense. As MyD88, these proteins are cytoplasmic. Site directed mutagenesis and deletion analysis have shown that the TIR domain is essential for Toll and IL-1 R activities. Sequence analysis have revealed the presence of

three highly conserved regions among the different members of the family: box 1 (FDAFISY), box 2 (GYKLC-RD-PG), and box 3 (a conserved W surrounded by basic residues). It has been proposed that boxes 1 and 2 are involved in the binding of proteins involved in signaling, whereas box 3 is primarily involved in directing localization of receptor, perhaps through interactions with cytoskeletal elements.

TAZ finger CBP and the related protein p300 are large nuclear molecules that interact with transcriptional activators and repressors. They belong to a class of protein containing a histone acetyltransferase activity, which suggest a role in chromatin remodeling. They are involved in biological function as diverse as cell growth, differentiation, or apoptosis. CBP/P300 proteins contain in their N and C terminal parts the so called transcriptional adaptor putative zinc finger (TAZ finger). Each TAZ domain is an around 100 amino acids domain which shows an internal triplication of a Cys-x4-Cys-x8-His-x3-Cys motif, although some of the repeats are imperfect. The binding sites for YY1, E1A and TFIIB in CBP and P300 proteins have been mapped in the region that contain the TAZ finger, suggesting a possible protein-binding function for this motif. Proteins containing this domain have been found to bind phosphorylated CREB.

The Dbl domain (dbl/cdc24 rhoGRF family) Guanine nucleotide exchange factor for Rho/Rac/Cdc42-like GTPases, is also called Db-homologous (DH) domain. It appears that domains invariably occur C-terminal to RhoGEF/DH domains The GRIP (golgin-97, RanBP2alpha, lmh1p and 230/golgin-245) domain is found in many large coiled-coil proteins. It has been shown to be sufficient for targeting to the Golgi. The GRIP domain contains a completely conserved tyrosine residue.

The forkhead-associated (FHA) domain is a putative nuclear signalling domain found in a variety of otherwise unrelated proteins. The FHA domain comprise approximately 55 to 75 amino acids and contains three highly conserved blocks separated by divergent spacer regions.

The Band 4.1 family domain is found in a number of cytoskeletal-associated proteins that associate with various proteins at the interface between the plasma membrane and the cytoskeleton. It is a conserved N-terminal domain of about 150 residues involved in the linkage of cytoplasmic proteins to the membrane.

Upon characterization of WHSC1, a gene mapping to the WolfHirschhornsyndrome critical region and at its C-terminus similar to the Drosophila ASH1/trithorax group proteins, a novel protein domain designated PWWP domain was identified. The PWWP domain is named after a conserved Pro-TrpTrp-Pro motif. It is present in proteins of nuclear origin and plays a role in cell growth and differentiation. Due to its position, the composition of amino acids close to the PWWP motif and the pattern of other domains present it has been suggested that the domain is involved in protein-protein interactions The Escherichia coli single-strand binding protein, also known as the helixdestabilizing protein, is a protein of 177 amino acids. it binds tightly, as a homotetramer, to single-stranded DNA (ss-DNA) and plays an important role in DNA replication, recombination and repair. Closely related variants of SSB are encoded in the genome of a variety of large self-transmissible plasmids. SSB has also been characterized in bacteria such as Proteus mirabilis or Serratia marcescens. Eukaryotic mitochondrial proteins that bind ss-DNA and are probably involved in mitochondrial DNA replication are structurally and evolutionary related to prokaryotic SSB The UBA-domain (ubiquitin associated domain) is a novel sequence motif found in several proteins having connections to ubiquitin and the ubiquitination

pathway. The USA domain is probably a non-covalent ubiquitin binding domain consisting of a compact three helix bundle.

The D111/G-patch domain is a short conserved region of about 40 amino acids which occurs in a number of putative RNA-binding proteins, including tumor suppressor and DNA-damage-repair proteins, suggesting that this domain may have an RNA binding function. This domain has seven highly conserved glycines.

The PKD domain was first identified in the Polycystic kidney disease protein PKD1, and contains an Ig-like fold. PKD1 is involved in adhesive protein-protein and protein-carbohydrate interactions, however it is not clear if the PKD domains mediate any of these interactions. Most of these domains are present in the extracellular parts of proteins involved in interactions with other proteins. The domain is most often found in proteins archaebacteria and some vertebrates.

Ubiquitin domain comprises ubiquitin which is a protein of seventy six amino acid residues, found in all eukaryotic cells and whose sequence is extremely well conserved from protozoan to vertebrates. It plays a key role in a variety of cellular processes, such as ATP-dependent selective degradation of cellular proteins, maintenance of chromatin structure, regulation of gene expression, stress response and ribosome biogenesis. Ubiquitin is a globular protein, the last four C-terminal residues (Leu-Arg-Gly-Gly) extending from the compact structure to form a'tail', important for its function. The latter is mediated by the covalent conjugation of ubiquitin to target proteins, by an isopeptide linkage between the C-terminal glycine and the epsilon amino group of lysine residues in the target proteins. In most species, there are many genes coding for ubiquitin. However they can be classified into two classes. The first class produces polyubiquitin molecules consisting of exact head to tail repeats of ubiquitin. The number of repeats is variable. In the majority of polyubiquitin precursors, there is a final amino-acid after the last repeat. The second class of genes produces precursor proteins consisting of a single copy of ubiquitin fused to a C-terminal extension protein (CEP). There are two types of CEP proteins and both seem to be

ribosomal proteins. There are a number of proteins which are evolutionary related to ubiquitin, including ubiquitin-like proteins from baculoviruses ; mammalian proteins GDX, FAU and RAD23-related proteins; human spliceosome associated protein 114, proteins BAT3 and CKAP1/TFCB, and ubiquitin-like proteins SMT3A, SMT3C and SMT3B; yeast proteins AD23, DK2 and SMT3; and C. elegans SMT3 and ubl-1 proteins.

DUF5 is a domain of unknown function found in bacterial sensor and chemotaxis proteins. These are usually integral membrane proteins and part of a twocomponent signal transduction pathway.

The Combined RanBP1IWASP domain is found in proteins implicated in a diverse range of signaling, nuclear transport and cytoskeletal events. This domain of around 130 amino acids is present in species ranging from yeast to mammals. It seems to be a protein protein interaction module as it has been shown that the RanBP1-WASP domain in cytoskeletal proteins bind the prolinrich motif FPPPP. Five proteins containing FPPPP sequence are yet known to bind WH1 domains: the actin cytosqueleton-related proteins ActA, Vinculin, Zyxin, the WIP protein, and the cytoplasmic domain of metalotropic lutamate receptors. Proteins of the RanBP1 family contain a WH1 domain in their N terminal region, which seems to bind a different sequence motif present in the C terminal part of RanGTP protein. Tertiary structure of the WH1 domain of the Mena protein revealed structure similarities with the pleckstrin homology (PH) domain and suggest that the WH1 domain could also be involved in phospholipid binding.

A conserved domain of 46 amino acids, called WHEP-TRS domain has been shown to exist in a number of higher eukaryote aminoacyl-transfer RNA synthetases. This domain is present one to six times in the several enzymes. There are three copies in mammalian multifunctional aminoacyl-tRNA synthetase in a region that separates the N-terminal glutamyl-tRNA synthetase domain from

the C-terminal prolyl-tRNA synthetase domain, and six copies in the intercatalytic region of the Drosophila enzyme. The domain is found at the N-terminal extremity of the mammalian tryptophanyl- tRNA synthetase and histidyl-tRNA synthetase, and the mammalian, insect, nematode and plant glycyl-tRNA synthetases. This domain could contain a central alpha-helical region and may play a role in the association of tRNA-synthetases into multienzyme complexes The Src homology 2 (SH2) domain is a protein domain of about 100 amino-acid residues first identified as a conserved sequence region between the oncoproteins Src and Fps. Similar sequences were later found in many other intracellular signal-transducing proteins. SH2 domains function as regulator modules of intracellular signalling cascades by interacting with high affinity to phosphotyrosine-containing target peptides in a sequence-specific and strictly phosphorylation-dependent manner Based on sequence similarities a domain of homology the Citron homology domain has been identified in the following proteins: Citron and Citron kinase. These two proteins interact with the GTP-bound forms of the small GTPases Rho and Rac but not with Cdc42.

. Myotonic dystrophy kinase-related Cdc42-binding kinase (MRCKalpha).

This serine/threonine kinase interacts with the GTP-bound form of the small GTPase Cdc42 and to a lesser extent with that of Rac.

NCK Interacting Kinase (NIK), a serine/threonine protein kinase.

ROM-1 and ROM-2, from yeast. These proteins are GDP/GTP exchange proteins (GEPs) for the small GTP binding protein Rho1.

This domain, called the citron homology domain, is often found after cysteine rich and pleckstrin homology (PH) domains at the C-terminal end of the proteins. It acts as a regulatory domain and could be involved in macromolecular interactions

The WW domain (also known as rsp5 or WWP) is a short conserved regionin a number of unrelated proteins, among them dystrophin, responsible for Duchenne muscular dystrophy. This short domain may be repeated up to four times in some proteins. The WW domain binds to proteins with particular proline-motifs, [AP]-P-P- [AP]-Y, and having four conserved aromatic positions that are generally Trp. The name WW or WWP derives from the presence of these Trp as well as that of a conserved Pro. It is frequently associated with other domains typical for proteins in signal transduction processes. A large variety of proteins containing the WW domain are known. These include ; dystrophin, a multidomain cytoskeletal protein; utrophin, a dystrophin-like protein of unknown function; vertebrate YAP protein, substrate of an unknown serine kinase; mouse NEDD-4, involved in the embryonic development and differentiation of the central nervous system; yeast RSP5, similar to NEDD-4 in its molecular organization; rat FE65, a transcription-factor activator expressed preferentially in liver ; tobacco OB10 protein and others.

The SET-domain of transcriptional regulators (TRX, EZ, ASH1 etc) appears to be a protein-protein interaction domain. It has been demonstrated that association of SET domain and myotubularin-related proteins modulates growth control. The enhancer of zeste protein has this domain. In Drosophila this protein has a function in segment determination and in mammals it may be involved in the regulation of gene transcription and chromatin structure.

Src homology 3 (SH3) domain SH3 (src Homology-3) domains are small protein modules containing approximately 50 amino acid residues. They are found in a great variety of intracellular or membrane-associated proteins. They are found in a variety of of proteins with enzymatic activity, in adaptor proteins that lack catalytic sequences and in cytoskeletal proteins, such as fodrin and yeast actin binding protein ABP-1. The SH3 domain has a characteristic fold which consists of five or six beta-strands arranged as two tightly packed anti-

parallel beta sheets. The linker regions may contain short helices. Although usually found as single copies, two to four copies of the domain can be present in a single polypeptide, particularly in the adaptor proteins. SH3 domains are commonly found in proteins that also contain SH2 (src Homology-2) domains, suggesting that their functions are inter-related. The function of the SH3 domain is not well understood but they may mediate assembly of specific protein complexes via binding to proline-rich peptides and be involved in linking signals transmitted from the cell surface by protein tyrosine kinases to"downstream effector proteins".

The SOCS box comprising the SOCS domain, C-terminus of STAT-inhibitors was first identified in SH2-domain-containing proteins of the suppressor of cytokines signaling (SOCS) family but was latter also found in: * the WSB (WD-40-repeat-containing proteins with a SOCS box) family,

the SSB (SPRY domain-containing proteins with a SOCS box) family, 'the ASB (ankyrin-repeat-containing proteins with a SOCS box) family, * and ras and ras-like GTPases.

The SOCS box found in these proteins is an about 50 amino acid carboxy- terminal domain composed of two blocks of well-conserved residues separated by between 2 and 10 nonconserved residues. The carboxyterminal conserved region is an L/P-rich sequence of unknown function, whereas the amino-terminal conserved region is a consensus BC box, which binds to the Elongin BC complex. It has been proposed that this association could couple bound proteins to the ubiquitination or proteasomal compartments.

The sterile alpha motif (SAM) domain is a putative protein interactionmodule present in a wide variety of proteins involved in many biological processes. The SAM domain that spreads over around 70 residues is found in diverse eucaryotic organisms. SAM domains have been shown to homo-and hetero-oligomerize, nevertheless with a low affinity constant and to mediate specific protein-protein interactions. Structural analyses show that the SAM domain is arranged in a

small five-helix bundle with two large interfaces. In the case of the SAM domain of EphB2, each of these interfaces is able to form dimers. The presence of these two distinct intermonomers binding surface suggest that SAM could form extended polymeric structures.

Recently a number of proteins have been found to contain a conserved cysteinerich domain of about 60 amino-acid residues called the LIM domain. These proteins are: Caenorhabditis elegans mec-3; a protein required for the differentiation of the set of six touch receptor neurons in this nematode.

. Caenorhabditis elegans Iin-11 ; a protein required for the asymmetric division of vulval blast cells.

Vertebrate insulin gene enhancer binding protein ist-1. lsl-l binds to one of the two cis-acting protein-binding domains of the insulin gene.

. Vertebrate homeobox proteins lim-1, lim-2 (lim-5) and lim3.

. Vertebrate Imx-1, which acts as a transcriptional activator by binding to the FLAT element; a beta-cell-specific transcriptional enhancer found in the insulin gene.

Mammalian LH-2, a transcriptional regulator protein involved in the control of cell differentiation in developing lymphoid and neural cell types.

* Drosophila protein apterous, required for the normal development of the wing and halter imaginal discs.

. Vertebrate protein kinases LIMK-1 and LIMK-2.

Mammalian rhombotins. Rhombotin 1 (RBTN1 or TTG-1) and rhombotin-2 (RBTN2 or TTG-2) are proteins of about 160 amino acids whose genes are disrupted by chromosomal translocations in T-cell leukemia.

Mammalian and avian cysteine-rich protein (CRP), a 192 amino-acid protein of unknown function. Seems to interact with zyxin.

Mammalian cysteine-rich intestinal protein (CRIP), a small protein which seems to have a role in zinc absorption and may function as an intracellular zinc transport protein.

Vertebrate paxillin, a cytoskeletal focal adhesion protein.

Mouse testin.

Sunflower pollen specific protein SF3.

* Chicken zyxin. Zyxin is a low-abundance adhesion plaque protein which has been shown to interact with CRP.

. Yeast protein LRG1 Yeast rho-type GTPase activating protein RGA1/DBM1.

* Caenorhabditis elegans homeobox protein ceh-14.

'Caenorhabditis elegans homeobox protein unc-97.

. Yeast hypothetical protein YKR090w.

Caenorhabditis elegans hypothetical proteins C28H8. 6.

These proteins generally have two tandem copies of a domain, called UM (for Lin-11 Isl-1 Mec-3) in their N-terminal section. Zyxin and paxillin are exceptions in that they contains respectively three and four LIM domains at their C-terminal extremity. In apterous, isl-1, LH-2, lin-11, lim-1 to lim-3, Imx-1 and ceh-14 and mec-3 there is a homeobox domain some 50 to 95 amino acids after the LIM domains. In the LIM domain, there are seven conserved cysteine residues and a histidine. The arrangement followed by these conserved residues is C-x (2)-Cx (16, 23)-H-x (2)- [CH]-x (2)-C-x (2)-C-x (16, 21)-C-x (2, 3)- [CHD]. The LIM domain binds two zinc ions. LIM does not bind DNA, rather it seems to act as interface for protein-protein interaction.

The RING-finger is a specialized type of Zn-finger of 40 to 60 residues that binds two atoms of zinc, and is probably involved in mediating protein-protein interactions. There are two different variants, the C3HC4-type and a C3H2C3type, which is clearly related despite the different cysteine/histidine pattern. The latter type is sometimes referred to as'RING-H2 finger'. E3 ubiquitin-protein ligase activity is intrinsic to the RING domain of c-Cbl and is likely to be a general function of this domain; Various RING fingers exhibit binding to E2 ubiquitinconjugating enzymes (Ubc's). Several 3D-structures for RING-fingers are known.

The 3D structure of the zinc ligation system is unique to the RING domain and is refered to as the'cross-brace'motif. The spacing of the cysteines in such a domain is C-x (2)-C-x (9 to 39)-C-x (1 to 3)-H-x (2 to 3)-C-x (2)-C-x (4 to 48)-C-x (2)C.

Note that in the older literature, some RING-fingers are denoted as LIM- domains. The LlM-domain Zn-finger is a fundamentally different family, albeit with similar Cys-spacing.

Cohesin domains interact with a complementary domain, termed the dockerindomain. The cohesin-dockerin interaction is the crucial interaction for complex formation in the cellulosome.

The S4 domain is a small domain consisting of 60-65 amino acid residues that was detected in the bacterial ribosomal protein S4, eukaryotic ribosomal S9, two families of pseudouridine synthases, a novel family of predicted RNA methylases, a yeast protein containing a seudouridine synthetase and a deaminase domain, bacterial tyrosyl-tRNA synthetases, and a number of uncharacterized, small proteins that may be involved in translation regulation. The S4 domain probably mediates binding to RNA.

According to the present invention inserting a linker of 27 amino acids or more, between a donor domain and an acceptor domain allows FRET to occur between the donor/acceptor domain. Moreover, the linker within the polypeptide chain can bind to a binding partner. Thus, the donor/acceptor domains do not disrupt the

normal interaction of the region with the binding partner. On binding of the binding partner to the binding region of the linker energy transfer between the donor and acceptor domains is disrupted. This provides a rapid, sensitive and simple method to measure the binding of binding partners to the linker.

Furthermore by using competition assays to displace the bound linker the binding of other ligands to binding partners can be detected in vitro.

Furthermore, small molecule inhibitors which bind to either the linker or the binding partner may be discovered using such assays. Such inhibitors might be pharmaceutical agents.

The present invention therefore provides a method to measure the binding of binding partners either directly or using competition assays and also to detect agents that interfere with binding in vitro.

More specifically, inserting a 44 amino acid region, containing the CRIB binding region of PAK, between mutants of GFP allows FRET to occur between the GFP mutants. The PAK moiety in the GFP protein construct binds to Rac with similar affinity as to an uncomplexed PAK fragment. Thus, GFPs do not disrupt the normal interaction of PAK with Rac. On binding of Rac to the EGFP-PAK-EBFP construct energy transfer between the GFP mutants is disrupted. This provides a rapid, sensitive and simple method to measure the binding of effector proteins to Rac/Cdc42 in vitro. Furthermore by using competition assays to displace the bound EGFP-PAK-EBFP the binding of other ligands to Rac or Cdc42 can be detected in vitro.

PAK and Rac are known to be involved in cell proliferation and thus any ligands that bind to either PAK or Rac are potential pharmaceuticals, in particular for the treatment of diseases associated with cell proliferation and more particularly the treatment of cancer. Similarly, molecule inhibitors which bind to either PAK or Rac or interfere with the binding of Rac to PAK may be useful drugs, particularly anti-cancer agents.

The present invention therefore also encompasses a method of identifying agents that might be useful in the treatment of diseases connected with cell proliferation, in particular cancer.

There has been speculation in the literature as to what extent the dimerisation of GFP might adversely affect its utility for FRET based assays (15). The tendency of GFP to dimerise might be expected to distort the binding affinities of binding partners if they were each tagged separately with one of the GFP FRET partners. However, in the situation described herein where one binding partner (linker region) is tagged with both GFPs, the results suggest that tethering the GFPs promotes their dimerisation and that the weak affinity of GFP for itself may be an important advantage for the development of successful GFP based biosensors. The binding data obtained using the protein construct is in close agreement with data obtained by more traditional methods (Table 1). An advantage of the approach used in this study is that only one binding partner (linker region) needs to be tagged with GFP and its binding to multiple other binding partners can be observed. Also a significant advantage is that the interpretation of results is much simplified because the ratio of GFP FRET partners is always 1: 1. This eliminates differences in measured fluorescence ratio that would otherwise arise with different mixtures of GFPs because of direct excitation of the acceptor GFP mutant on excitation of the donor.

Clearly since moieties tagged with donor/acceptor domains can often be expressed in high yields in cells the method according to the present invention also has the potential to be used to monitor the binding of binding partners to the linker in cells. Furthermore by using competition assays to displace the bound linker the binding of other ligands to binding partners can be detected in cells. Furthermore, small molecule inhibitors which bind to either the linker or the binding partner may be discovered using such assays in cells. Such inhibitors might be pharmaceutical agents.

Moreoversince moieties tagged with donor/acceptor domains can often be expressed in high yields in cells the method according to the present invention also has the potential to be used to monitor the activation state of binding partners, especially when the binding partner is a partner in a cell signalling pathway. Moreover, the present invention also provides a method for visualising signal transduction events in live cells and subcellular compartments in real time.

It has advantages over prior art methods where both the linker and the binding partner are labelled and thus requires the introduction of both the binding partner and the linker region into cells. Using the method according to the present invention only the polypeptide chain comprising the "donor domain-linkeracceptor domain"needs to be tranfected into cells and can be used to monitor the activation state of endogenous binding partners within the cell. Thus, this system can be used in vitro to measure kinetics and binding constants as described and also has the potential to be used in vivo to monitor signal transduction events associated with the activation state of the binding partner.

By chosing the appropriate linker region it is possible to monitor any desired intracellular interactions.

More specifically, proteins tagged with GFP can often be expressed in high yields in cells the method according to the present invention also has the potential to be used to monitor the activation state of Rac/Cdc42 and to visualise signal transduction events in live cells and subcellular compartments in real time. It has advantages over the method described in Kraynov et al. (22) where Rac and PAK are both labelled and the method required transfection of Rac and microinjection of PAK into cells. Using this method only EGFP-PAK-EBFP needs to be tranfected into cells and can be used to monitor the activation state of endogenous Rac within the cell. Thus, this system can be used in vitro to measure kinetics and binding constants as described and also has the potential to be used in vivo to monitor signal transduction events associated with the

activation state or Rac/Cdc42. Similar constructs containing a different linker region could be used to monitor other intracellular interactions.

Examples Example 1 Vector Construction The genes for EBFP and EGFP were amplified from their respective Clontech vectors using PCR (Pfu DNA polymerase, Stratagene). The EBFP gene was inserted into a Novagen pET30a vector between BamH1 and Xhol restriction sites using a 5'primer incorporating an EcoR1 restriction site. This was followed by insertion of the EGFP gene between Bglll and EcoR1 restriction sites using a 3'primer incorporating a BamH1 restriction site, a short linker and an EcoR1 restriction site. The resultant vector (C9) had EGFP and EBFP linked by the amino acid sequence SGSSSGEFS containing BamH1 and EcoR1 restriction sites. In addition C9 carries N terminal Hise/thrombin and C terminal His6 sequences derived from the pET30a vector. Other constructs were derived from C9 by excision of the BamH1/EcoR1 region of C9 and ligation of oligonucleotides (Life Technologies) coding for the desired linkage sequence. Oligonucleotides encoding the forward and reverse sequences were mixed at 10 I- ! M, heated at 95 C for 5 min and ligated into the cut C9 vector. Manipulation of vectors were carried out in E. coli DH5a and the identity of all constructs were confirmed by sequencing using forward and reverse sequencing primers designed to the insert region. Linkers also incorporated a factor Xa protease cleavage sequence (GIEGR) and had the following linker sequences: C15, SGSSSGIEGRSSEFS ; C20, SGSGASSGIEGRSSSGAEFS ; C25, SGSGAGSGSSGIEGRSSSGAGTEFS ; C30, SGSGAGSGAGSSGIEGRSSSGAGTGAGEFS ; C37, SGSGAGSGAGSGAGSSGIEGRSSSGAGTGAGSGAEFS ; C50, SGSGAGSGAGSGAGSGAGSGASSGIEGRSSSGAGTGAGSGAGSGAGSEFS.

A control construct EGFP-HCV-EBFP was made with the linkage sequence SGSGAGSGAGSGAGEDVVPCSMSYGAGSGAGSGAEFS. The PAK insert was made by PCR amplification of residues 75-118 of human aPAK with primers incorporating BamH1 and EcoR1 restriction sites and the restriction enzyme cut product was ligated into cut C9 vector as above. The final construct, EGFP-PAKEBFP, had the linkage sequence SGSISLPSDFEHTIHVGFDAVTGEFTGMPEQWARLLQTSNITKSEQKEFS.

Example 2 Protein Expression Vectors encoding His-tagged proteins were transformed into E. coli BL21 (DE3).

Stationary cultures were diluted 1 in 20 in L Broth containing 30 flg. mr1 Kanamycin, grown at 25 C to an A600 of 0.8 and induced with 1 mM IPTG for 24 h. Cells were harvested, resuspended in 20 ml 20 mM Tris/HCI pH 8,100 mM NaCI, disrupted by sonication, and centrifuged for 1 hour at 130,000g at 4 C.

Proteins were purified from the supernatant using Clontech TALONspin metal affinity columns according to manufacturer's instructions. Cultures generally yielded about 15 mg. i' of His-tagged protein. The purified proteins were visible as a single band on SDS PAGE and their molecular weights were verified by Electrospray Mass Spectrometry. Q61L Rac, GST-PAK (75-132), GST-PAK (75- 118) and GST-Phox (1-192) were prepared as described previously (5). Protein concentrations were calculated from their Asso using extinction coefficients calculated from the amino acid composition of the proteins and the extinction coefficients of tyrosine, tryptophan (14) and guanine nucleotide.

Example 3 Evaluation of FRET Fluorescence measurements were carried out using a Perkin-Elmer LS50B spectrofluorimeter. Unless stated otherwise fluorescence measurements were carried out using 10 zu of protein in a volume of 0.5 ml. For calculation of FRET

ratios samples were excited at 382 nm and emission spectra were taken between 400 and 600 nm. Background corrected fluorescence measurements were taken at 444 nm and 509 nm and the FRET ratio was calculated according to the formula FRET ratio = (1509-0. 21x ! 444)/'444 [The factor 0.21 corrects for the proportion of the fluorescence emission at 509 nm due to direct emission from the donor (EBFP) calculated from its optimum emission at 444 nm]. Protease cleavage of protein constructs was carried out in 50 mM Tris/HCI pH 7.9, 100 mM NaCI and 1 mM CaCI2 overnight at 30 C in 0.5 ml using 20 - 100 g of protein construct and 0.13 units of Factor Xa (Boehringer). Equilibrium binding experiments of PAK-GFP2 to Q61L Rac were carried out in a 0.5 ml stirred cuvette with 0.65 11M PAK-GFP2 in 50 mM Tris/HCI buffer pH 7.9. Q61L Rac was added to final concentrations between 0.05 and 3 uM and fluorescence measurements performed as above. Similarly measurements of the displacement of Rac from PAK-GFP2 by GST-Phox and GST-PAK were determined with the spectrometer in ratiometric mode using 1 11M Rac, 0.4 or 0.5 M PAK-GFP2 and and GST-PAK or GST-Phox additions where indicated.

Example 4 Isothermal Titration Calorimetry ITC was performed in a Microcal MCS calorimeter (15) in 50 mM Tris/HCI pH 7.9.

Q61L Rac (1 00 M ; preliminary 2. 5 ! injection followed by up to 10 x 10 0 I injections) was injected into 6. 25 M PAK-GFP2, 5 u. M HCV-GF ? 2 or buffer (for control) at 30 C. The heat output was analysed using MicroCal Origin software and the resultant data fitted to a binding isotherm describing binding at a single site (15) to estimate Kd, and stoichiometry.

Example 5 Stopped-Flow Kinetics Stopped-flow experiments were performed using an Applied Photophysics SX. 17MV stopped-flow fluorimeter operated in the single push mode. Excitation at 382 nm was obtained from a mercury-xenon arc lamp and a monochromator, and emitted light was measured through an APL 515 nm bandpass filter. All measurements were carried out in 50 mM Tris/HCI pH 7.5, 2 mM MgCI2, 1 mM DTT at 25 C. Reagents were loaded via two syringes: one contained PAK-GFP2 (0.1 M) and the other contained Q61L Rac. GTP (0.5-6. 5 M). Data were collected using SpectraKinetic Workstation software (Applied Photophysics) and analysed using Grafit software (Erithacus) to give observed rate constants (kobs) for the fluorescence intensity changes seen. Multiple traces were collected for each reaction and observed rate constants were averaged.

Example 6 Scintillation Proximity Assays Scintillation proximity assays were similar to those described previously (5).

Affinities of either PAK (75-118) or PAK-GFP2 for Q61L Rac. [H] GTP were measured using a competition assay consisting of 1.25 mg of Protein A SPA PVT beads (Amersham Cat No. RPNQ 0019), 4. 5 Ilg of anti-GST antibody (Molecular Probes), 0.025 M Q61L Rac. [3H]GTP, 0. 02 M GST-PAK (75-132), 50 mM Tris/HCI pH7.5, 2 mM MgCI2, 1 mM DTT, 0.2 mg. Ml-1 of BSA and competing protein (up to 3. 2 M) in a final volume of 200 Ill. Apparent KdS for competing proteins were determined as in Ref. (5), taking the Kd for Q61L

RacfH] GTP binding to GST-PAK (75-132) to be 0. 035 M. The affinity of Q61 L Rac. [3H]GTP and GST-PAK (75-132) was measured by a direct titration SPA with 0. 02 M GST-PAK (75-132) as in Ref. (5).

Example 7 Initial Evaluation of FRETand Comparison of the Constructs In initial experiments we investigated to what extent the FRET signal between concatenated GFP mutants is affected by inserting different length unstructured amino acid chains between the GFPs. In this way we hoped to gain insights into how the nature of the amino acid linker between GFP mutants might effect the FRET obtained. A recurring motif of GSGA was chosen for this on the assumption that it would give an unstructured and flexible linker. Within each linker there was a Factor Xa cleavage site. With all the constructs the fluorescence emission spectra on excitation of the EBFP donor at 382 nm showed two distinct peaks of emission at 444 and 509 nm corresponding to emission from the donor and acceptor GFPs respectively (Figure 1A). Protease cleavage of the constructs with factor Xa resulted in a large change in their fluorescence properties and after cleavage the FRET ratio of all constructs tested decreased to a ratio of 1.0. For example, the FRET ratio of construct C37 decreases from 4.5 to 1.0 (Figure 1A). That this represents complete cleavage of the proteins was verified by SDS PAGE and by extrapolation from reaction progress curves (results not shown). The final ratio of 1.0 derives from the overlap of EGFP and EBFP fluorescence spectra since mixing equal quantities of the two proteins gives the same ratio (results not shown). It is unlikely that EGFP and EBFP interact at this concentration since a FRET ratio of 1.0 can be derived either if the proteins are measured as a mixture or measured separately and the spectra added together electronically. Therefore it appears that, with these GFP mutants, the minimum achievable FRET ratio as measured here is 1.0. FRET between mutants of GFP has previously been reported using linker lengths of about 18-25 amino acids (9,10, 11). Since the efficiency of FRET falls off markedly with separation distance of the fluorophores we expected that increasing the length of unstructured linkers above this would give rise to a marked decline in FRET ratio to a value of 1.0. Comparison of the FRET ratios for the different constructs shows that the values increase for the intact proteins

as the length of the unstructured linker is increased (Figure 1 B). It has been calculated that FRET between EGFP and EBFP would be negligible at separation distances greater than about 100 A (13) whereas, allowing 3 A per peptide bond, 50 amino acids in an extended form represents a distance of 150 A . Therefore it is not plausible that FRET in this instance derives simply from a direct tethering of the two proteins in close proximity. This suggests that, in contrast to our expectations, long linkers inserted between the GFP mutants allows them to come into closer association. A likely explanation, in view of the weak dimerisation observed with untethered GFPs, is that tethering promotes intramolecular dimerisation and hence FRET. The structure of GFP monomers from X-ray crystallographic studies is a cylinder of about 30 AO diameter and 40 A long (16). The N and C peptide termini are closely grouped at one end of the barrel and GFP dimers are formed by head to tail adjacent alignment of the barrels. Therefore, for tethered dimers to form a similar conformation the linker would need to be around 40 AO long, i. e. in the region of 10 to 13 amino acids.

We observed a maximum FRET ratio of around 5 with these protein constructs (Figure 1b). Since flexible linkers of 20 to 30 amino acids do not give optimum FRET between the mutants it seems that flexible linkers even as long as this restrict the ability of the tethered GFPs to form intramolecular dimers. From our results it is not possible to predict what the maximum achievable FRET ratio would be for tethered GFP protein constructs. Similarly it is not clear to what extent the EGFP and EBFP are dimerised in the protein constructs. When measurements are performed in buffers containing high concentrations of sodium chloride (3 M), which has been reported to promote the dimerisation of GFP (19), we obtained a FRET ratio of 11 with construct C20 (results not shown). Possibly this effect derives, at least in part, from an increased dimerisation of GFP. However, this is not straightforward since at this concentration of salt the fluorescence of EGFP is enhanced and that of EBFP is diminished, an effect that would also give an increase in apparent FRET ratio.

Under these high salt conditions, the calculated FRET ratio of an equimolar

mixture of EBFP and EGFP, due to overlap of emission spectra on excitation at 382 nm, would be approximately 4. DeAngelis et al have reported that when two GFP molecules are brought into proximity there are marked changes in the fluorescence properties of the protein (20). This effect was utilised by them to observe the dimerisation of GFP-tagged proteins in vivo and evidence for the formation of more than one form of GFP dimer was obtained (20). It is difficult to relate the changes in excitation spectra of wild type GFP on dimerisation observed by DeAngelis et al. to the differences in emission spectra we have detected on dimerisation of GFP mutants in our constructs. However, it may be that the spectral changes we have observed derive from a similar process. Therefore, it is possible that the changes in fluorescence we find on the formation of EBFP/EGFP dimers are not due to FRET per se but rather are due to direct changes in the fluorescence properties of the GFPs because of structural changes of the proteins on direct contact of two GFP molecules.

Example 8 Fluorescence properties of EGFP-PAK-EBFP and it's use in an assay to detect binding to Rac and binding of Rac to its effector proteins.

When a linker corresponding to the 44 amino acid Cdc42/Rac-binding domain of p21 activated kinase (residues 75-118), was inserted between the GFPs (EGFP-PAK-EBFP) a FRET ratio of 4.2 was observed for the protein construct. This ratio is slightly less than the ratio one would predict for a flexible linker of this length from Figure 1B but clearly shows significant FRET between the GFPs. In order to investigate the effect of Rac binding to PAK on FRET, Q61L Rac. GTP was titrated into EGFP-PAK-EBFP. Q61 L Rac. GTP was used as it has a low intrinsic hydrolysis rate, compared to wild-type Rac, allowing the binding interaction between Rac and PAK to be studied in isolation from the GTP cleavage step. The addition of increasing concentrations of Q61 L Rac. GTP resulted in a clear decline in FRET ratio to 2.2 (Figure 2) showing that the PAK fragment retained the ability to bind to Rac. Assuming a value of 1.0 for the

measured fluorescence ratio in the absence of FRET (see above), this binding disrupts approximately 60% of the FRET compared to the uncomplexed material. The Kd of this association calculated from the single site binding isotherm in Figure 2 is 0.27 IlM. Furthermore, we found that this system could form the basis of a simple and rapid competition method to measure the binding of effector proteins to Rac and Cdc42. Thus titration of GST-Phox (1-192) (Figure 3A and Figure 3B) or GST-PAK (75-132) (Figure 3C) into assays with fixed concentrations of Q61L Rac. GTP and EGFP-PAK-EBFP resulted in the restoration of FRET on EGFP-PAK-EBFP. Binding affinities of 0. 66 (Jv ! and 0. 231lM were calculated for the binding of Q61L Rac. GTP to GST-Phox and GST-PAK, respectively, from these studies (Table 1). To examine whether the presence of the GFP proteins in the EGFP-PAK-EBFP construct affect the normal interaction between Rac and PAK, equilibrium binding measurements of Q61L Rac. GTP to either EGFP-PAK-EBFP or untagged PAK (75-118) were made using scintillation proximity assays (SPAs). In these assays, a signal is obtained when Q61L Rac. [3H]GTP binds to GST-PAK (75-132) linked to SPA beads. From experiments in which the concentration of Q61 L Rac. H] GTP was varied, the apparent Kd of Q61 L Rac. GTP for GST-PAK (75-132) was found to be 0.035 IlM (Table 1). Then, SPAs were performed in which EGFP-PAK-EBFP or untagged PAK were titrated into assays with fixed concentrations of Q61L Rac. [3H]GTP and GST-PAK (75-132), thereby competing for radiolabelled Rac and causing a decrease in signal. Using this procedure, the apparent Kds for Q61 L Rac. GTP binding to EGFP-PAK-EBFP and PAK (75-118) were found to be 0. 11 ils and 0.12 M (Table 1). It appears, therefore, that the presence of the GFP moieties on either side of PAK has no effect on the affinity of Q61L Rac. GTP for this fragment of PAK. In order to investigate this further the binding affinity of EGFP-PAK-EBFP for Q61 L Rac. GTP was measured directly by isothermal titration calorimetry (ITC). This method detects binding of components directly by monitoring the heat change obtained during titration of one binding component into another. In this way we obtained a clear binding

isotherm for the binding of Q61 L Rac. GTP to EGFP-PAK-EBFP with a Kd of 0.41 u. M and a stoichiometry of binding of 1.2 Q61 L Rac to EGFP-PAK-EBFP (Table 1). With control titrations using a linker unrelated to PAK (HCV-GFPz) no binding was detect by ITC confirming that EGFP-PAK-EBFP binds to Q61L Rac. GTP in a 1: 1 ratio via the PAK (75-118) moiety.

Example 9 Kinetics of binding of EGFP-PAK-EBFP to Rac Since the decrease in FRET signal obtained on binding of Rac to EGFP-PAKEBFP might depend upon the dissociation of the GFP moieties it was important to investigate whether this gives rise to a two stage kinetic process, with a rapid binding of EGFP-PAK-EBFP to Rac followed by a slower dissociation of the GFPs. We investigated, therefore, the kinetics of the interaction between EGFPPAK-EBFP and Q61 L Rac. GTP in a stopped-flow fluorimeter. The EBFP donor was excited at 382 nm and emission from the EGFP acceptor was detected using a 515 nm bandpass filter. On mixing EGFP-PAK-EBFP with a large molar excess of Q61L Rac. GTP, a single exponential decrease in fluorescence signal was observed (Figure 5A). The timecourse of the exponential decrease ranged from 5 to 20 sec using 0.1-6. 4 M 061 L Rac. GTP. The data were fitted to a single exponential and the observed rate constants ranged from 0.2 to 1.9 s-1. The rate constants were linearly dependent on the concentration of Q61L Rac. GTP (Figure 5B). It appears therefore that binding of Q61L Rac. GTP to EGFP-PAK-EBFP occurs as a single step process and we obtained no evidence for the dissociation of GFPs in EGFP-PAK-EBFP markedly changing the kinetics of the association of Q61 L Rac. GTP to EGFP-PAK-EBFP. From these data, the second order association rate constant (ki) was found to be 0.56 x 106 M-1s-\ and the first order dissociation rate constant (k-i) was found to be 0.069 s-1. The Kd value for the interaction of Q61 L Rac. GTP and EGFP-PAK-EBFP, calculated from Kd = k. i/ki, was 0.12 M, in close agreement to the values obtained with the equilibrium binding experiments. The amplitudes of the fluorescence changes ranged from 0.24-0. 55 and showed a hyperbolic dependence on the Q61L

Rac. GTP concentration (data not shown). The Kd value derived from these data was 0. 34 uM (Table 1).

Example 10 Interaction of EGFP. PAK-EBFP with Rac. GDP versus Rac. GTP If it can be shown that the interactions between EGFP-PAK-EBFP and Rac. GDP or Rac. GTP can be differentiated using this method, it is possible that EGFPPAK-EBFP could be used to monitor the activation state of Rac in live cells.

Therefore, the FRET-based fluorescence changes associated with the exchange of nucleotide on Rac when bound to EGFP-PAK-EBFP were investigated.

Initially, 0. 4 lM EGFP-PAK-EBFP was incubated at 37 C, the EBFP donor was excited at 382 nm and emission from the EGFP acceptor was monitored at 509 nm. On adding 2 p. ! v) Rac. GDP no significant changes in fluorescence occurred (Figure 6). Addition of 10 pu GTP caused a decrease in fluorescence, due to exchange of the nucleotide on Rac from GDP to GTP. That this fluorescence decrease represents exchange of nucleotide on Rac was verified by the addition of 0.8 mM EDTA at the same time as the GTP (data not shown). This caused a rapid increase in the rate of the fluorescence change, consistent with the known ability of EDTA to cause an increase in the dissociation rate of nucleotide from Rac (21). In the presence of EDTA, Rac GTPase activity is inhibited. On addition of excess (2 mM) MgCI2 a slow increase in the fluorescence was observed, corresponding to hydrolysis of the GTP bound to Rac to GDP. The rate of this increase in fluorescence is enhanced by the addition of 0. 05 lM RhoGAP, which acts to catalyse GTP hydrolysis on Rac. These data show that the interactions of EGFP-PAK-EBFP with Rac. GDP or Rac. GTP can be distinguished using this method.

Further experiments, in the presence of 2 mM MgCI2, confirmed that Rac. GDP up to 16. 5 Jvt did not give any FRET change on adding to EGFP-PAK-EBFP,

whereas addition of 211M Rac. GTP or 111M 061 L Rac. GTP gave approximately maximal FRET changes. Thus, EGFP-PAK-EBFP binds to Rac. GDP with significantly lower affinity than to Rac. GTP.

Example 11 Interaction of EGFP-PAK-EBFP with Rac1 in mammalian cells In order to demonstrate that the FRET binding pair has the capability to work in mammalian cells, EGFP-PAK-EBFP was cloned into the mammalian expression vector, pcDNA-4 (Invitrogen). HEK293T cells were transfected with this vector.

24-86h later cells were harvested and the lysate analysed by Western Blot analysis using an antibody against the His-Tag and by recording fluorescence emission spectra. The blots showed a single band of the expected mobility for the construct and no indication of any degradation. The spectra showed a similar FRET ratio (4.64) as for bacterially expressed protein.

Bacterial expressed 061 L Rac1 was then titrated into the cell lysate, recording fluorescence emission spectra after each addition. Rac caused a reduction in FRET, and the dose-response curve and calculated Kd were very similar to those observed using purified bacterial-expressed proteins.

teferences 1. Van Aelst, L. , and D'Souza-Schorey, C. (1997) Rho GTPases and signaling networks. Genes Dev. 11,2295-2322 2. Hall, A. (1998) Rho GTPases and the actin cytoskeleton. Science 279, 509-514 3. Bishop, A. L., and Hall, A. (2000) Rho GTPases and their effector proteins. Biochem. J. 348,241-255 4. Manser, E. , Leung, T., Salihuddin, H. , Zhao, Z. , and Lim, L. (1994) A brain serine/threonine protein kinase activated by Cdc42 and Rac1.

Nature 367,40-46 5. Thompson, G. , Owen, D., Chalk, P. A., and Lowe, P. N. (1998) Delineation of the Cdc42/Rac-binding domain of p21-activated kinase.

Biochem. 37,7885-7891

6. Morreale, A., Venkatesan, M., Mott, H. R., Owen, D., Nietlispach, D., Lowe, P. N. , and Laue, E. D. (2000) Structure of Cdc42 bound to the GTPase binding domain of PAK. Nat. Struct. Biol. 7,384-388 7. Mott, H. R., Owen, D., Nietlispach, D. , Lowe, P. N. , Manser, E., Lim, L. , and Laue, E. D. (1999) Structure of the small G protein Cdc42 bound to the GTPase-binding domain of ACK. Nature 299,384-388 8. Abdul-Manan, N. , Aghazadeh, B. , Liu, G. A., Majumdar, A., Ouerfelli, 0., Siminivitch, K. A. , and Rosen, M. K. (1999) Structure of Cdc42 in complex with the GTPase-binding domain of the'Wiskott-Aldrich syndrome' protein. Nature 399,379-383

9. Mitra, R. D., Silva, C. M. , and Youvan, D. C. (1996) Fluorescence resonance energy transfer between blue-emitting and red-shifted excitation derivatives of the green fluorescent protein. Gene 173,13-17 10. Heim, R. , and Tsien R. Y. (1996) Engineering green fluorescent protein for improved brightness, longer wavelengths and fluorescence resonance energy transfer. Current Biology 6: 178-182 11. Xu, X., Gerard, A. L. V. , Huang, B. C. B. , Anderson, D. C. , Payan, D. G. , and Ying, L (1998) Detection of programmed cell death using fluorescence energy transfer. Nucleic Acids Research 26,2034-2035 12. Romoser, V. A., Hinkle P. M. , and Persechini A. (1997) Detection on living cells of Ca 2±dependent changes in the fluorescence emission of an indicator composed of two green fluorescent protein variants linked by a calmodulin-binding sequence. J. Biol. Chem. 272,13270-13274 13. Miyawaki, A. , Liopis J. , Heim, R. , McCaffery, J. M. , Adams, J. A., Ikura, M. , and Tsein R. Y. (1997) Fluorescent indicators for Ca2+ based on green fluorescent proteins and calmodulin. Nature 388,882-887 14. Mahajan, N. P. , Harrison-Shostak, D. C. , Michaux, J. , and Herman, B.

(1998) Novel mutant green fluorescent protein protease substrates reveal the activation of specific caspases during apoptosis. Nature Biotechnology 16, 547-552 15. Ward W. W. (1998) in Green Fluorescent Protein (Chalfie, M. , and Kain, S. Eds) pp 45-76, Academic Press, San Diego 16. Philips G. N. (1998) in Green Fluorescent Protein (Chalfie, M. , and Kain, S. Eds) pp 77-96, Academic Press, San Diego

17. Mach, H. , Middaugh, C. R. , and Lewis, R. V. (1992) Statistical determination of the average values of the extinction coefficients of tryptophan and tyrosine in native proteins. Anal. Biochem. 200,74-80 18. Wiseman, T., Williston, S. , Brandts, J. F. , and Lin, L. N. (1989) Rapid measurement of binding constants and heats of binding using a new titration calorimeter. Anal. Biochem. 179,131-137 19. Cutler, M. W. , and Ward, W. W. (1997) in Bioluminescence and Chemiluminescence ; Molecular Reporting with Photons (Hastings, J. W., Kricka, L. J. , and Stanley, P. E. Eds) pp 403-406, John Wiley 20. DeAngelis, D. A. , Miesenbock, G., Zemelman, B. V. , and Rothman, J. E.

(1998) PRIM : Proximity imaging of green fluorescent protein-tagged polypeptides. Proc. Natl. Acad. Sci. USA 95,12312-12316 21. Zhang, B., Zhang, Y. , Wang, Z, and Zheng, Y. (2000) The role of Mg2+ cofactor in the guanine nucleotide exchange and GTP hydrolysis reactions of Rho family GTP-binding proteins. J. Biol. Chem. 275,25299- 25307 22. Kraynov, V. S., Chamberlain, C. , Bokoch, G. M. , Schwartz, M. A., Slabaugh, S. , and Hahn, K. M. (2000) Localized Rac activation dynamics visualized during cytoskeletal rearrangements in living cells. Science 290,333-337

Table 1. Equilibrium dissociation constants for binding to Q61 L Rac. GTPa

Binding Partner Method Kd (M) EGFP-PAK-ITC 0.41 EBFP (75-118) SPA 0.11 FRET 0.27 Stopped-0. 12 Flow Stopped- 0. 34c Flow PAK (75-118) SPA 0.12 GST-PAK (75- SPA 0.04 132) FRET 0.23 GST-Phox FRET 0.66 a Catalytic and equilibrium dissociation constants were measured using the indicated method as described in the Materials and Methods Section and in the Legends to Figures 1-6. b Calculated from dependence of rate constants on Q61 L Rac. GTP concentration. c Calculated from dependence of amplitude of fluorescence change on Q61 L Rac. GTP concentration.

Claims

Claims 1. A polypeptide chain comprising a linker which has at least 27 amino acid residues or more, said linker being sandwiched between a donor and an acceptor domain, wherein FRET occurs between said donor and acceptor domain.
2. A polypeptide chain according to claim 1, wherein the linker has at least 30 amino acid residues or more.
3. A polypeptide chain according to claim 1 or 2, wherein the linker has at least 35 amino acid residues or more.
4. A polypeptide chain according to any of the above claims, wherein the linker has at least 40 amino acid residues or more.
5. A polypeptide chain according to any of the above claims, wherein the linker comprises at least one binding region.
6. A polypeptide chain according to claim 5 above, wherein the binding region is capable of binding to a binding partner.
7. A polypeptide chain according to claim 6 above, wherein the binding partner is a protein.
8. A polypeptide chain according to claim 6 and 7, wherein FRET changes upon the binding region binding to said binding partner.
9. A polypeptide chain according to claim 6,7 or 8, wherein the linker conformation changes in shape upon the binding region binding to said binding partner.
10. A polypeptide chain according to claim 6,7, 8 or 9, wherein said binding partner will only bind to the linker in the activated state.

<Desc/Clms Page number 42>
11. A polypeptide chain according to claim 6,7, 8 or 9, wherein said binding partner will only bind to the linker in the unactivated state.
12. A polypeptide chain according to any of claims 6 to 11 wherein said binding partner is a partner in a cell signalling pathway.
13. A polypeptide chain according to any of claims 1 to 9, wherein the linker is the Cdc42/Rac interactive binding motif (CRIB).
14. A polypeptide chain according to any of claims 1 to 9, wherein the linker may be chosen from any of Pak, the Kringle domain, FIZZY/CDC20 domain, phosphtyrosine interaction domain (PID), RanBP1 domain, TIR domain, TAZ finger, Dbl domain (dbl/cdc24 rhoGRF family), FRIP domain, forkhead-associated (FHA) domain, Band 4.1 family, PWWP domain, single-strand binding protein family, ubiquitin-associated domain, D111/G- patch domain, PKD domain, ubiquitin domain, domain found in bacterial signal proteins, combined RanBPI/WASP domain, WHEP-TRS domain, Src homology 2 (SH2) domain, citron homology domain, WW/rsp5NVWP domain, SET-domain of transcriptional regulators (TRX, EZ, ASH1 etc), Src homology 3 (SH3) domain, SOCS domain, C-terminus of STAT- inhibitors, SAM domain (Sterile alpha motif), LIM domain, RING finger, Cohesin domain, WW domain, S4 domain.
15. A polypeptide chain wherein the binding partner is one that will bind to a linker region as disclosed in claim 13 or 14.
16. A polypeptide chain according to claim 7, wherein the binding partner is chosen from a Ras superfamily protein.
17. A polypeptide chain according to claim 7, wherein the binding partner is chosen from Rac & Cdc42.
18. A method of identifying a binding partner that is capable of binding to a linker, the method comprising: (a) providing a polypeptide according to any of claims claims 6 to 15

<Desc/Clms Page number 43>

(b) providing a test substance (c) monitoring FRET activity and thereby determining whether the test substance is a binding partner capable of binding to the linker.
19. A method of identifying a modulator that is capable of interfering with binding between a linker and a binding partner, the method comprising : (a) providing a polypeptide according to any of claims 6 to 15, (b) providing a binding partner (c) providing a test susbstance (d) determining whether said test substance is a modulator that is capable of interfering with the binding between the linker and the binding partner.
20. A method of monitoring the activation state of a binding partner comprising (a) providing a polypeptide according to any of claims 6 to 12 (b) monitoring FRET activity and thereby determining whether the binding partner is bound to the linker.
21. A method according to claim 17 or 19, wherein said method is carried out in vitro.
22. A method according to claim 17 or 20, wherein said method is carried out in vivo.