The present invention refers to a modified phalloidin, which may be coupled via the primary group at side chain 7 to a labelling group or a matrix, which may bind with high affinity to filamentous actin (F-actin). The compound and the matrix are suitable for purifying F-actin and compounds, e.g. proteins which specifically interact with F-actin.

Background and related art

The aim of the study was to identify the set of accessory proteins that regulate the nuclear F-actin network in oocytes. The straightforward approach is to affinity-purify nuclear F-actin and to identify the copurifying proteins. As the current protocols have drawbacks that severely reduce their usability under our requirements, we set out to develop a new method for purifying F-actin based on immobilised phalloidin, a highly F-actin specific toxin.

The techniques for F-actin affinity chromatography have seen little modification since their introduction to the scientific community. Actin binding inhibits DNAsel and DNAsel-agarose can be used to isolate G-actin and G-actin binding proteins from extracts. Some years later an F-actin affinity matrix was introduced based on the fluorescyl-coupled actin that was bound to anti-fluorescein IgG sepharose and polymerised with additional actin in the presence of phalloidin (Luna EJ. et al, J. Biol. Chem. 257 (1982), 1395-13100). The most frequently used protocol to date directly crosslinks preformed F-actin filaments to activated sepharose (Miller, K.G. and Alberts, B.M., Proc. Natl. Acad. Sci., USA, 86 (1989)). Now extracts are passed through the F-actin column and F-actin binding proteins are eluted by increasing the salt concentration. This technique has been used successfully to identify new F-actin binding proteins from diverse species (Hu, S. et al., Plant J., 24 (2000), 127-137), but there are significant drawbacks. The main problem is reproducibility of the matrix preparation. Empirical trials are necessary to find the ideal filament concentration and packing flow rate. Packing needs to be checked by applying a dye to visualise the flow path and successfully prepared columns are stable for only up to three weeks (Miller, K.G., supra). Additionally, steric hindrance due to the direct crosslinking utilized in the most popular protocol (Miller, K.G., supra) might interfere with the binding of some proteins. Another problem is the use of exogenous actin in all protocols. This makes saturation of the matrix very difficult as a massive amount of additional binding sites is presented which is especially problematic if working with small volumes of extract as mandatory in our case. Ideally, only endogenous actin is being used for the affinity chromatography.

Summary of the invention Considering these drawbacks, we focussed on developing an alternative strategy for enriching F-actin and its binding proteins from an extract. Several small molecule toxins have been known for decades to bind to F-actin with high affinity and specificity (Lengsfeld, A.M., et a/.,Proc. Natl. Acad. Sci. (1974), 2803-2807; Wieland, T., peptides of poisonous Amanita muschrooms (1986), Springer). The use of small molecules as baits for F-actin has several benefits: it minimizes the interface necessary for binding to F-actin and these molecules can be coupled to the matrix with a long, flexible spacer in between to allow unhindered access of proteins to the filament (see Figure 1). Phalloidin, a toxin produced by the fungal Amanita family, is the ideal candidate as it binds to the inside of the actin filament (Figure lb). This is a very unlikely binding site for proteins and additionally has a stabilizing effect on F- actin filaments (Dancker, P. et al, Biochim. Biophys. Acta (1975), 407-414).

The toxin is a bicyclic heptapeptide and contains an unusual thioether bridge between side chain 3 and 6 (see Figure la). Phalloidin lacks, however, a conveniently accessible coupling group (such as -COOH, -NH ₂ or -SH) and therefore needs to be chemically modified prior to immobilization. Other phallotoxins, e.g. phallacidin, do contain a carboxyl-group at side chain 2, but this position seems to be involved in contact with F-actin as judged from reduced toxicity of substances with modified amino acids at this position (see Lengsfeld, A.M., supra and Wieland, supra, page 87). Phalloidin can be purified from the fungi in reasonable amounts and is well characterized in terms of affinity to F actin (KD ~3X10 ^"8 M). Preferable amino acids for modification are also described (Faulstich, H. et cil, Physiol. Chem. (1977), 181- 184). The most versatile position is side chain 7 (γ-δ-dihydroxyleucine) at whose unique primary hydroxyl group diverse modifications have been introduced (Wieland, T. and Hollosi, M., Liebigs Ami. Chem. (1983), 1533-1540; Wieland, T., and Rehbinder, D., Liebigs Ann. Chem. (1963), 149). We used this position to introduce a thiol group attached to a flexible, hydrophilic polyethylene glycol (PEG) linker for functionalization and subsequent immobilization of the toxin to a solid support. As a proof of principle, our phalloidin-based F-actin affinity matrix was then used to map the nuclear F-actin interactome of Xenopus laevis oocytes. The matrix has been stable for several months, allowed usage of endogenous actin pools and should cause minimal interference with surface-binding interfaces on F-actin. We also coupled the functionalised phalloidin to fluorescent dyes and used it as alternative to commercial F-actin probes.

Description of the invention

A subject-matter of the present invention is a modified phalloidin, wherein the primary OH group of dihydroxyleucine at position 7 has been replaced by SH or SR, wherein R is a linker.

The linker is preferably a flexible linker, e.g. an alkylene or alkyleneoxy, e.g. ethyleneoxy linker which has a chain length of at least 1, 2 or 3 atoms, preferably at least 2 or 3 atoms. The chain length may be e.g. up to 20 or 30 atoms for alkylene linkers and up to 100, 200, 300 or even higher for alkyleneoxy, particularly ethyleneoxy linkers. Preferably, the linker is selected from:

-(CH ₂ ),-X and -[(CH ₂ ) _m O]„X,

wherein r is an integer > 1 , preferably > 2,

n is an integer > 1 , preferably > 2,

m is 2 or 3, preferably 2,

and X is a coupling group, e.g. SH, COOH, NH ₂ , halo etc. which may react with a complementary reactive group. The compound of the invention may be coupled via the SH or SR group to a labelling group, e.g. a fluorescent dye, or a matrix, e.g. a solid phase material including particles, particularly a chromatographic carrier material such as a silica- or sepharose-based material.

A further subject-matter of the invention is a matrix, e.g. a chromato graphic carrier material such as a silica- or sepharose-based material having coupled thereto the modified phalloidin as described above. Preferably, the modified phalloidm is coupled to the matrix via a linker, preferably a linker having a chain length of at least two atoms. More preferably, the linker is a linker as described above.

In order to enhance the specificity of F-actin binding, the matrix is preferably a passivated matrix, e.g. a matrix which has been coated with a passivating agent. A suitable passivating agent is a polyalkyleneoxy-based agent, particularly a polyethyleneoxy-based agent. More preferably, the passivating agent comprises a [(CH ₂ ) _m O] _n group,

wherein n is an integer > 1, preferably > 2,

and m is 2 or 3, preferably 2.

A further subject-matter of the invention is the synthesis of a thiol modified phalloidin as described above. The thiol modified phalloidin may be produced by activation of the primary OH group of dihydroxyleucine at position 7 of phalloidin, particularly with a sulfonate group, and introducing an SH or SR group, e.g. by reaction with an alkylene dithiol or a polyalkyleneoxy dithiol, e.g. polyethyleneoxy dithiol. The thiol modified phalloidin may be coupled to a matrix. The coupling may be effected by reaction of the SH group or the temiinal group X of the linker R with a suitable functional group on the matrix. Preferably, the coupling involves reaction of an SH group of the modified phalloidin compound with an SH reactive group on the matrix (or a fluorescent dye). Suitable SH reactive groups are e.g maleimide or iodacetamide.

The SH modified compound or a matrix having attached to the SH modified phalloidin compound may be used for F-actin affinity binding, particularly for F- actin affinity chromatography. By means of this F-actin affinity chromatography assessory proteins regulating or interacting with the F-actin network in eukaryotic cells may be identified.

Thus, the invention also refers to a method for F-actin activity binding comprising contacting a sample containing F-actin with modified phalloidin or a matrix having attached thereto a modified phalloidin under conditions wherein F-actin binds to the compound or the matrix.

Further, the present invention shall be explained by the following examples and the accompanying drawing figures. Figures

Figure 1. Structure and F-actin binding site of phalloidin

(a) Chemical structure of phalloidin. The amino acids were numbered and color- coded for clarity. The unique primary hydroxyl group (the δ-hydroxyl group of the dihydroxyleucine (side chain 7)) is labelled in bold red. (b) Model of the interaction of phalloidin with F-actin based on Xray diffraction data. The position of phalloidin is represented by the density in between the three actin subunits and is almost independent of the underlying model for F-actin used (blue density = Holmes model (Holmes, K.C. et al, Nature (1990), 44-49); green density = Lorenz model (Lorenz, M. et al, J. Mol. Biol. (1993), 826-836); magenta = based on (Tirion, M.M. et al, Biophys. J. (1995), 5-12). (b) is modified from (Oda, T. et al, Biophys. J. (2005), 2727-2736).

Figure 2. Functionalisation of phalloidin with a thiol group

Reaction a: phalloidin (1) is converted to mono-tosyl phalloidin (2) (tosyl chloride (16x molar excess) in CHC1 ₃ , pyridine; 4°C; 30 min. under argon; reaction stopped by addition of diethyl ether (Anderl, J. Synthese und Cytotoxizitat membranpermeabler Phallotoxine. Dissertation, Universitat Heidelberg, (2003))).

Reaction b: mono-tosyl phalloidin (2) is reacted with a large molar excess of dithiol- PEG to form thiol-PEG ₂ -phalloidin (3) (KOH (equimolar to thiol groups), MeOH; lh under argon; RT, reaction then neutralized by addition of AcOH). Figure 3. Phalloidin isolation from Amanita phalloides extract as a single, baseline-separated peak

Elution profile shows an overlay of commercial phalloidin (P2141, Sigma- Aldrich) and the fungal extract, both curves represent the absorption at 280mri in arbitrary units (mAU). A multi-step elution (10%, 25%, 100% buffer B) was used (Buffer A: lOOmM ammonium acetate; pH5, Buffer B: 60% ACN, lOOmM ammonium acetate; pH5) on a CI 8 Column (Grace- Vydac).

Figure 4. Phalloidin fraction is identified as a single, pure substance by mass spectrometry

The mass-to-charge spectrum (m/z) shows three peaks of arbitrary intensity (a.i.) for the purified substance. As only singly charged substances are detected, the m/z value directly represents the molecular mass. The three peaks all represent phalloidin in different ionisation forms ([M+H] ⁺ = 788.32 + 1 = 789.32; [M+Na] ⁺ = 788.32 + 23 = 811.32; [M+K] ⁺ = 788.32 + 39 = 827.32).

Figure 5. Mono-tosyl phalloidin is the main product of the tosylation reaction

The elution chromatogram (absorption at 280nm in arbitrary units (mAU)) shows a single main product of the tosylation reaction that is confirmed to be mono-tosyl phalloidin by mass spectrometry (see inlet). The two peaks at 50ml correspond to di- tosylated products. A multi-step elution (40%, 50%, 100% buffer B) was used (Buffer A: 10% ACN, Buffer B: 80% ACN) on a Diphenyl Column (Grace- Vydac). The mass-to-charge spectrum (m/z) shows a single main peak of arbitrary intensity (a.i.) for the purified substance.

Figure 6. Thiol-PEG ₂ -phalloidin purification shows a single product

Purification of the reduced thiol-PEG ₂ -phalloidin compound shows a single product peak (absorption at 280nm in arbitrary units (niAU)) that is confirmed to be thiol- PEG ₂ -phalloidin by mass spectrometry (see inlet). A linear elution gradient was used (Buffer A: lOOmM ammonium acetate; pH5, Buffer B: 60% ACN, lOOniM ammonium acetate; pH5) on a CI 8 Column (Grace- Vydac). The mass-to-charge spectrum (m/z) shows three peaks for the monomeric substance and a dimer peak of arbitrary intensity (a.i.) for the purified substance, indicating that the substance was oxidised during sample preparation. Please see main text for more details.

Figure 7. The phalloidin thioether-linked dye conjugate labels identical structures as commercial dye

Fluorescence microscopy analysis of formaldehyde fixed PtK2 cells stained with identical amounts of the newly designed (a) or commercial (b) phalloidin dye according to manufacturer's instruction. The scale bar represents ΙΟμιη. Note that the steric hindrance of our dye should be less than the commercial one as the linker to the fluorophore is less bulky.

Figure 8. Immobilisation of thiol-PEG ₂ -phalloidin to a silanol surface

Reaction c: thiol-PEG ₂ -phalloidin (3) is converted to a phalloidin-PEG-silane (4) by addition of equimolar amounts of maleimide-PEG-silane (Silane PEG Maleimide, MW 3400; NANOGS Inc., New York, NY) (DMF:DMSO:MeOH 1 :3.5 : 1 ; 40 min.; RT; under argon; n = ~70 PEG units). Any unreacted maleimide group is subsequently quenched by addition of a molar excess of mPEGe-SH (Polypure AS, Oslo, Norway). Reaction d: phalloidin-PEG-silane (4) is immobilised to SiMAG silanol beads to generate phalloidin beads (5) (89% toluene, 10% EtOH, 1%TEA; 24h, RT, rotating under argon).

Reaction e: phalloidin beads (5) are surface passivated by the addition of an excess of PEG _6- 9-silane (ABCR GmbH, Karlsruhe, Germany) (toluene, 1% TEA; 36h; RT; rotating under argon; m = 6-9 PEG units) to give rise to the final surface passivated phalloidin beads (6).

Figure 9. PEG-silane treatment allows complete passivation of silica surfaces (a) PEG-coated beads produced following different protocols were challenged with cytoplasmic HeLa extract to test protein repellent capacities. All proteins bound to the beads were eluted with SDS-containing sample buffer, separated by SDS-PAGE and visualized by Coomassie staining. Note: the gel has been sliced as indicated by the black line, (b) Scheme of the reactions that lead to PEG-coat deposition (modified from (Arkles, B., Chemtech (1977),766-778; Gelest, supra)).

Figure 10. Surface passivation under optimal conditions is crucial for minimal unspecific protein binding

(a) Phalloidin beads and the respective controls (see (b) for schematic representation) were prepared under suboptimal conditions and show high unspecific background after being challenged in HeLa cytoplasmic extract. Beads were incubated in extract, washed in parallel and bound protein was eluted with SDS-containing sample buffer (see Input lane). Four different post-processing conditions (Veiseh, M. et ai, Biomaterials (2004), 3315-3324; Gelest, Gelest Product Manual (2009)) were tested and compared (also see magnified inlet). Green arrows indicate specific bands, red arrows indicate unspecific bands that are still present in the phalloidin bead eluate after optimal surface passivation, (b) Schematic representation of the beads used in this experiment. Figure 11. Use of phalloidin PEG superparamagnetic beads to map the nuclear and cytoplasmic F-actin interactomes from Xenopus laevis oocytes

Extracts were prepared from either whole oocytes, cytoplasms or nuclei. "Input" on the left side shows SDS-PAGE analysis of those extracts. The right side shows the patterns of proteins that bound to either phalloidin beads (as described in Examples 6 and 7) or to control beads lacking the toxin. Note that the phalloidin beads retrieved not only a massive actin band, but also numerous actin-binding proteins that show a characteristic nucleo-cytoplasmic distribution.

Examples

General comment

The following examples, including the experiments conducted and results achieved, are provided for illustrative purposes only and are not to be construed as limiting upon the present invention. Identification of novel F-actin binding proteins is often done by affinity chromatography on immobilized F-actin filaments, but these approaches have drawbacks like low reproducibility and the blocking of potential binding sites on the actin filament. The nuclear F-actin network of Xenopus oocytes, which serves here as an example, stains well with phalloidin (Gard, D., Microsc. Res. Tech. (1999), 388— 414; Bohnsack, M.T. et al, Nat. Cell. Biol. (2006), 257-263), so we chose a strategy using immobilised phalloidin to purify F-actin along with its accessory proteins from Xenopus oocyte nuclear extracts. There are advantages over the currently used methods like longer matrix stability, high reproducibility, lack of steric hindrance through a long, flexible spacer between actin filament and stationary phase and minimal blocking of binding sites on the actin filament. Given the long tradition of staining F-actin with fluorescently labelled phalloidin it is surprising that such straightforward methodology for purifying F-actin and accessory proteins has never been implemented successfully. Even direct precursors for immobilizing phalloidin to a matrix, like biotinylated phalloidin that could have been bound to streptavidin matrices, existed previously (Faulstich, H. et al, J. Histochem. Cytochem. (1989), 1035-1045). Possibly, the utilised coupling chemistry resulted in a too low affinity and selectivity for F actin or other technical complications hindered a successful implementation.

Indeed, to successfully apply this method, several technical problems had to be solved. These included the introduction of an appropriate spacer with a convenient coupling group on phalloidin and the development of an immobilisation strategy for the utilised stationary phase. Additionally, a milestone for successful identification of novel F-actin binding proteins is the reduction of unspecific background binding to an absolute minimum. This is crucial because the proteins we are interested in are expected to bind in a highly substoichiometric ratio compared to actin and need to be distinguishable from non-specific binders to the stationary phase. As these processes are iterative, a large amount of starting material for the chemical synthesis, in this case phalloidin, is necessary.

All members of the Amanita family of mushrooms contain varying amounts of ama- and phallotoxins (see (Wieland, 1986, supra) for a detailed analysis of the different species). The death cap Amanita phalloides is a rather abundant representative in Germany and contains up to 1 mg phalloidin per gram of dried mushroom (Wieland, 1986, supra). Several groups of local mushroom collectors helped to supply some kilograms of the fungus. We used the phallotoxin purification scheme established by Enjalbert and Faulstich (Enj albert, F. et al , J. Chromatogr. (1992), 227-236) with several adaptations to our lab settings. The fungal material was processed to a slurry and the toxins were extracted with 50% methanol. The methanol was evaporated under reduced pressure and acetone was added to 70% final concentration to precipitate remaining proteinaceous contaminants. The soluble toxin-containing fraction was again concentrated under reduced pressure. The resulting fungal extract was loaded onto a CI 8 reversed-phase column and the constituents were fractionated by isocratic elution with steps of 10%, 25% and 100% buffer B (60% acetonitrile in lOOmM ammonium acetate at pH 5; buffer A contained no acetonitrile). Chromatography at pH 5 is important as lower pH (e.g. pH2) might cause recyclisation of phalloidin into secophalloidin, meaning ring opening and subsequent closure at another site (Wieland, T., and Sangl, I. Liebigs Ann. Chem. (1964), 160). We ran into this problem that was hard to detect because chromatography at low pH conditions (a rather mild treatment compared to the described procedure of boiling the substance in 0.2n H ₂ SO ₄ (Wieland, T., 1964, supra) seemed sufficient for the reaction causing the commercial standard we used to be modified as well. Additionally, the recyclisation resulted in a substance that was indistinguishable from the reactant in molecular mass and could be further processed without noticeable difference until the removal step of the tosylgroup.

Figure 3 shows that phalloidin could be purified as a single, baseline-separated peak that eluted after similar volume as the commercial standard (P2141, Sigma- Aldrich). The identity of the purified substance was confirmed by subsequent mass spectrometry (Figure 4).

This procedure yielded pure phalloidin with the expected molecular weight of about 788 Da. As expected, the measured molecular weight (Figure 4) was approximately one Dalton larger than the sum of the chemical elements, because the substance was singly charged, in our case protonated (often described as [M+H] ⁺ ). Only charged molecules can be measured in MALDI-TOF mass spectrometry. The two additional peaks that can be seen in Figure 4 resulted from the addition of a sodium [M+Na] ⁺ or potassium [M+K] ⁺ ion instead of a hydrogen and thus resulted in measured masses that are about 23 or 39 Da higher than the actual mass.

Phalloidin lacks a conveniently accessible coupling group, but has a unique primary, two secondary and one tertiary hydroxyl group. We used the ability of tosylchloride to specifically convert the unique primary hydroxyl group (dihydroxyleucine (side chain 7)) into a suitable leaving group (Wieland, T. and Hollosi, M., 1983, supra). Figure 2 provides an overview of the coupling scheme. We subsequently introduced a dithiol-polyethylene glycol (dithiol-PEG ₂ ) at this position that gives rise to unique thiol group, an excellent functional group that can be reacted very efficiently e.g. with a maleimide for further modification of the toxin. The thiol is attached to the toxin via a short flexible linker. The resulting group at the site of attachment to the toxin (thioether) is probably less bulky than the resulting group of the classical coupling reaction (dithiolane), in which periodate oxidation at the dihydroxyleucine sidechain allows for a subsequent formation of an aminomethyldithiolane derivative by reaction with 2,3-dimercaptopropylamine (Faulstich, H. et al. , J. Muscle Res. Cell. Motil. (1988), 370-383). This procedure is still used for commercial phalloidin dyes. Structures of the different results can be compared in Figure 7.

To functionalise phalloidin it needs to be tosyl-activated and subsequently reacted with the dithiol-PEG ₂ (Figure 2). Pyridine acts as base to deprotonate the hydroxyl groups of the toxin and the resulting alkoxide reacts in an S 2 reaction with tosylchloride (Figure 2a). This nucleophilic substitution kinetically favours primary hydroxyl groups over secondary and tertiary groups, which is the reason for the selectivity of the reaction for side chain 7 (Anderl, J. Synthese und Cytotoxizitat membranpermeabler Phallotoxine. Dissertation, Universitat Heidelberg (2003)). The chromatogram of the subsequent purification on a Diphenyl reversed-phase column showed that mono-tosyl phalloidin was the main product, which was confirmed by mass spectrometry (Figure 5).

One of the thiol groups of the dithiol-PEG ₂ subsequently replaces the tosylate (Figure 2b). This reaction leads to a thiol-PEG ₂ -phalloidin and introduces an excellent coupling group attached to the toxin by a flexible, hydrophilic linker of approximately 10 A length. The dithiol- linker had to be used in great molar excess in order not to cross-link two activated phalloidins molecules. It should be noted that the solubility of the dithiol-linker in H ₂ 0 had to be increased by equimolar addition of a base (either KOH or NaOH) and that the reaction had to be kept under argon to avoid oxidization and disulfide bond formation of the free thiols. Any disulfide bonds that might still have formed during the reaction could be reduced by addition of DTT prior to the purification over a CI 8 column. The single product peak of the reversed-phase purification corresponded to thiol-PEG ₂ -phalloidin (Figure 6). This was supported by mass spectrometric analysis of the peak fraction (see inlet Figure 6), yet the use of KOH for increasing solubility of the dithiol-linker could be traced to the m/z spectrum. The [M+H] ⁺ ionization state of the product (952.3 + 1 = 953.3 Da) was rather underrepresented and the [M+K] ⁺ ionization (952.3 + 39 = 991.3 Da) dominated the spectrum. During preparation for mass spectrometry the sample partially oxidized and formed a dimer, connected through a disulfide bond. This resulted in the peak at 1941.7 Da which represents [2M -2H + K] ⁺ or [(2 x 952.3) - (2 x l) + 39] - 1941.6 Da. The thiol-PEG ₂ -phalloidin we produced can now be efficiently coupled to a variety of substances via maleimide or iodacetamide chemistry. As a proof of principle we linked our thiol-phalloidin to maleimide- ATT0565 and compared the staining patterns on formaldehyde-fixed PtK2 cells side-by-side with a commercial phalloidin dye following the manufacturer's instructions. The resulting F-actin probe perfonned without notable difference to the commercial substance at identical concentration and stained stress fibres in the cytoplasm of the cells. This led to the conclusion that our functionalization chemistry did no interfere with the F-actin binding ability of the toxin in a significant way. In order to purify F-actin and accessory proteins from an extract, the thiol-PEG?- phalloidin needs to be immobilised to a stationary phase (also called matrix). The thiol group on the toxin and the numerous variations of homo- and hetero- bifunctional crosslinkers on the market allow for diverse combinations of matrices and coupling chemistry. We tested various different materials like Sepharose, porous and non-porous silica particles and paramagnetic particles with amino- and silanol- functions on their surface. Best results in terms of F-actin and associated protein enrichment in combination with minimal unspecific background binding were achieved using paramagnetic particles with a non-porous silanol surface and PEG surface coating for reduction of unspecific binding. An overview of the immobilisation and passivation chemistry starting from thiol-PEG ₂ -phalloidin (substance 3) is shown in Figure 8.

SiMAG-Silanol superparamagnetic beads (chemicell GmbH, Berlin, Germany) possess an unmodified non-porous silica surface. The free silanol groups on the bead surface readily react with silane containing compounds (e.g. trimethoxy-silanes), fomiing a covalent linkage. Thiol-phalloidin can be coupled to a maleimide-PEG- silane (Silane PEG Maleimide, MW 3400; NANOGS Inc., New York, NY) to generate a surface-reactive phalloidin-PEG-silane (Figure 8c). The PEG spacer has a length of approx. 70 PEG units, which corresponds to 300 A. This provides enough conformational freedom for flexible binding of F-actin filaments to the beads. After deposition of phalloidin-PEG-silane to the beads (Figure 8d), any remaining silanol groups on the surface need to be quenched (Figure 8e). Otherwise these groups would be negatively charged at neutral pH (Towns, J., Anal. Chem. (1991),1126- 1132) and thereby act as a cation-exchangers. This would generate high levels of unspecific binding of proteins, which would make meaningful interpretation of the elution fraction difficult (Figure 10).

The use of F-actin chromatography in order to find novel F-actin interactors has one important difference compared to other approaches used to find interaction partners like e.g. coimmunoprecipitation. The expected ratio in which actin and the accessory proteins interact, and therefore will be purified, is highly substoichiometric and not nearly equimolar as in most other protein-protein complexes. This sets totally different requirements for the amount of tolerable unspecific background binding in the experiment, which has to be reduced to an absolute minimum in order to make valid predictions for the specificity of the co-eluting proteins. Polyethylene glycol (PEG) has been known for decades as a protein repellent agent on surfaces (Lee, J.H. et al, J. Biomed. Mater. Res. (1989), 351-368). There is a vast body of literature on the use of polyethylene glycol coats on surfaces (Harris, J.M. (1992). Poly(Ethylene Glycol) Chemistry: Biotechnical and Biomedical Applications (Topics in Applied Chemistry) (Springer)) and diverse examples of successful application (Papra, A. et al, Langmuir (2001), 1457-1460; Kannan, B. et al , Biosensors and Bioelectronics (2006), 1960-1967; Kamisetty, N.K. et al, Analytical and Bioanalytical Chemistry (2006), 1649-1655). The underlying molecular mechanism is not fully clarified (Banerjee, I. et al, Adv. Mater. (2010), 690-718), but is most probably based on effects that can be explained by themiodynamics. To get to the minimum free energy state, the ether groups of the PEG bind water (an exothermic reaction) and take on a random, i.e. high entropy conformation. Additionally, there might be a layer of water molecules on top of the PEG coat (Jeon, S. and Andrade, J., Journal of Colloid and Interface Science (1991), 159-166). The approach of a protein leads to compression and orientation of the PEG molecules that will generate an elastic repulsion force and loss in entropy. Water molecules have to be removed from the hydrated PEG chains at the cost of energy as well. These effects sum up to generate a thermodynamically unfavourable penalty that underlies the repulsive properties of PEG coated surfaces.

We used a PEG ₆ -9silane (3-[Methoxy(polyethyleneoxy)propyl]trimethoxysilane; ABCR GmbH) that contains 6-9 PEG groups corresponding to approx. 21 - 31 A for surface passivation (Figure 9). The reactions that occur during covalent surface binding of the substance are summarized in Figure 9b. Hydrolysis of the three methoxy groups is followed by a condensation step of the newly formed silanol groups during which oligomers of PEG-silane appear. Upon contact with the silanol groups on the silica surface, hydrogen bonds form and condense to a covalent bond over time. The water needed for the initial hydrolysis may be added, but atmospheric amounts are also sufficient to start the reaction (All les, B., Chemtech (1977), 766- 778).

Several published protocols for surface passivation using PEG-silane were compared (Gelest, Gelest Product Manual (2009), 1-60; Sharma, S. et al, Langmuir (2004), 348-356; Sui, G. et al, Anal. Chem. (2006), 5543-5551). The beads were incubated in the respective reaction mixes (Figure 9a) and the resulting PEG-coated bead surfaces were analysed for protein repellent capacity as follows: the beads were incubated in cytoplasmic HeLa extract, washed in parallel and any bound material was eluted with SDS-loading buffer. The eluates were then analysed by SDS-PAGE and Coomassie staining. The untreated matrix has charged silanol groups on the surface, which caused a high degree of protein binding (Figure 9a). The key factor for effective PEG coat deposition to the bead surface was the presence of an inert base, here triethylamine (TEA), as catalyst for the condensation reaction (see Figure 9a). The addition of water to the reaction mix did not significantly improve coat formation, while deposition from toluene worked slightly better than from ethanol. The importance of optimal surface passivation becomes particularly evident in the context of identifying substoichiometric binding partners (Figure 10). Three kinds of beads with different surface modifications (see Figure 10b) were produced. The initial attempt to passivate the surface failed (see Figure 10a Input), which led to a high unspecific background binding when challenged in cytoplasmic HeLa extract. The two control beads (quenched and PEG only) of the Input preparation show an almost identical pattern of bands as the phalloidin beads. The only obvious exception was the actin band (about 43 kDa) seen in the phalloidin bead eluate. Yet, identification of proteins that specifically co-enriched with actin would be very difficult from such an experiment. Contaminating proteins would outcompete the specific actin binders for detection in mass spectrometry. Therefore, the suboptimally coated beads were subjected to different post-processing conditions based on the successful toluene protocol seen in Figure 9. Under optimal conditions (toluene; 3% PEG-silane, 1% TEA; 25°C or 60°C) the control bead eluates showed dramatically reduced unspecific background binding. This became strikingly evident when comparing the phalloidin bead eluates between optimal and suboptimal postprocessing conditions at greater detail (see magnified inlet Figure 10a). All seemingly specifically co-enriched bands were still present under optimal conditions (represented by green arrow) while the majority of the contaminating bands vanished. Some seemingly unspecific bands were still present (indicated by red arrows), but the complexity of the sample was greatly reduced. This allows mass spectrometric analysis and confident identification of even highly substoichiometric binding partners of F-actin. Example 1. Toxin-containing extract from Amanita phalloides

The following protocol is based on a method published by (Enj albert, F. et al , J. Chromatogr. (1992), 227-236) and was adapted to the settings in our lab. Fresh fruiting bodies were chopped into small pieces and stored at -80 °C. To prepare the extract, 280g fungi were weighted in frozen state and two volumes of 50% MeOH in H ₂ 0 (560ml) were added. The thawed fungi were homogenized using a grinder (Ultra-Turrax, IKA, Staufen) until milkshake consistency was reached (1-2 min.). Toxins were extracted o/n at 4°C. The homogenate was centrifuged at 6,000 g for 15 min at 4°C and the combined supematants were filtered and measured for volume. Two equivalent volumes of acetone were added to the supernatant and incubated at -20 °C for lh. Precipitates, mostly proteins, were removed by centrifugation (6,000g, 15min, 4°C) and subsequently filtered (use solvent approved filter, e.g. MN QF 10 Filter (Macherey-Nagel GmbH, Diiren)). Organic solvents were removed from the extract by vacuum rotary evaporation at 15mbar and 30°C resulting in approx. 140 ml of aqueous toxin-containing fungal extract that was further concentrated to 70 ml final. In a separatory funnel, one volume of diethyl ether (70 ml) was added to extract any remaining lipids. Shaking for 2 min. mixed the two phases, while care was taken to release the increasing pressure inside the funnel, deriving from the evaporating ether, in short intervals. Subsequently, the two phases were allowed to build again and the coloured, aqueous phase was collected. Residual ether in the aqueous phase was removed by vacuum rotary evaporation. The solvent-free extract was filtered again through MN QF 10 and frozen at -80°C or processed directly.

Example 2. Isolation of phalloidin

Phalloidin was isolated from the other toxins by chromatography on a CI 8 reversed- phase column (218TP1010, GraceVyday, Deerfield, IL) using the Akta Purifier System (Pharmacia, Upsala, Sweden). All buffers were filtered and degassed prior to use. Chromatography was performed at RT in an ammonium acetate based buffer system at pH 5. The extract was chromatographed using a multi-step elution (10%, 25% and 100% buffer B; Buffer A: lOOmM ammonium acetate; pH5, Buffer B: 60% acetonitrile (ACN) in lOOmM ammonium acetate; pH5) at flow rate of lml/min. The phalloidin elution time was determined using a commercial phalloidin as standard (P2141, Sigma) and the gradient was adapted to yield a baseline-separated peak for the substance. Peak fractions were pooled, snap-frozen and lyophilised (Alpha 1-2 LD plus, Christ Martin GmbH, Osterode). Subsequently the sample was desalted by dissolving it in 20% ACN in H ₂ 0, rebinding it to the CI 8 column and washing with seven column volumes at 15% ACN in H ₂ 0. Phalloidin was eluted by shifting to 70% ACN in H ₂ 0. Peak fractions were pooled, snap-frozen and lyophilised again and identity of the purified substance was checked by mass spectrometry. Example 3. Mono-tosylation of phalloidin

10 mg of lyophilized phalloidin (M _r 788.3 g/mol) was dissolved in ΙΟΟμΙ anhydrous pyridine. 40mg of 4-Toluenesulfonyl chloride (tosylchloride; M _r 190.6 g/mol; Fluka) was dissolved in 50μ1 CHCI ₃ (a 16-fold molar excess over the phalloidin). The phalloidin was cooled on ice and the tosylchloride was added slowly during constant mixing. The sample was set under argon and rotated for 30min at 4°C (Anderl, 2003). The reaction was stopped by addition of 1 ml diethyl ether that led to precipitation of the product. The precipitates were pelleted in a tabletop centrifuge at 14,000 rpm for 2min and ether was removed under reduced pressure. The dried precipitate was resuspended in 50%> ACN in H ₂ 0 and chromato graphed on a Diphenyl reversed-phase column (219TPDiphenyl lOu, GraceVydac, Deerfields, IL) using multi-step elution (40%, 50%, 100%, buffer B; Buffer A: 10% ACN, Buffer B: 80%) ACN). Peak fractions were pooled, snap-frozen and lyophilised and stored at - 80°C. The main product of the reaction was mono-tosylated phalloidin as confirmed by mass spectrometry.

Example 4. Mono-thiolation of phalloidin

6mg of mono-tosyl phalloidin (M _r 942.3 g/mol) were resuspended in 1.2ml MeOH. 500μ1 of 3,6dioxal,8octanedithiol (dithiol PEG linker; 6.11M, Sigma) were mixed with 200μ1 MeOH and solubilized by deprotonating thiol groups with 400μ1 triethylamine (7.17 M, Roth GmbH) or 380μ1 KOH (10M). Phalloidin was added slowly under constant mixing, the vast molar excess of dithiol PEG linker precluded crosslinking of two activated phalloidins. Reaction was allowed to proceed for lh at RT under argon. Thiol-PEG ₂ -phalloidin was extracted from the solution by the addition of 1.8ml lOOmM ammonium acetate and the pH was set to 5 with ~150μ1 of glacial acetic acid. This caused phase separation between the surplus dithiol PEG linker and the toxin-containing aqueous phase that was collected for clean up of the reaction on CI 8 reversed-phase column (linear gradient; Buffer A: lOOmM ammonium acetate; pH5, Buffer B: 60% ACN in lOOmM ammonium acetate; pH5; flow rate lml/min). Peak fractions of the single reaction product were pooled, snap- frozen and lyophilised and stored at -80°C under argon.

Example 5. Custom Fluorescent phalloidins

Thiol-PEG ₂ -phalloidin (M _r 952.3 g/mol) was reacted with an equimolar amount of the respective maleimide-functionalized ATTO dye (ATTO-Tec GmbH, Berlin) in PBS for lh at RT (according to manufacturer's instructions). The resulting product was purified by C18-reversed phase chromatography (linear gradient; Buffer A: lOOmM ammonium acetate; pH5, Buffer B: 60% ACN in lOOmM ammonium acetate; pH5; flow rate lml/min) and identified by mass spectrometry.

Example 6. Phalloidin - Beads

ΙΟΟμΙ Silane PEG Maleimide (1.5mM in DMF:DMSO:MeOH 1 :3.5 : 1 ; M _r 3400 g/mol; NANOGS Inc., New York, NY) was mixed with 250μ1 thiol-PEG ₂ -phalloidm (0.5mg/ml) and incubated for 40min. at RT under argon. Addition of 25 μΐ mPEG _f ,- SH (250mM in DMF; Polypure, Oslo, Norway) quenched any remaining functional maleimide group. A control sample lacking the phalloidin was prepared in parallel. SiMAG-Silanol superparamagnetic beads (chemicell GmbH, Berlin) were resuspended and three samples of 500μ1 slurry each were washed in 100%o EtOH and then brought into 90% toluene / 10% EtOH (5ml final volume). The phalloidin-PEG- silane mix and the control mix were added to one sample each, the third one remained untreated at this stage. Deposition of silanes to the surface of the beads was started by the addition of 50μ1 triethylamine (7.17 M, Roth GmbH) and was allowed to proceed for 36h at RT (rotating under argon). Subsequently, all three samples were washed in toluene. Example 7. PEG-coating the beads

All three bead samples ("phalloidin", "quenched" and "PEG only") were resuspended in 5ml toluene and 150μ1 PEG ₆ -9-silane (3- [Methoxy(polyethyleneoxy)propyl]trimethoxysilane; 90%; ABCR GmbH, Karlsruhe) was added. Again, deposition was started by the addition of 50μ1 triethylamine (7.17 M, Roth GmbH) and was allowed to proceed for 36h at RT (rotating under argon). Subsequently, all three samples were washed three times in toluene and then in three times in EtOH. Finally, the beads were stored in 500μ1 EtOH at 4°C.