Undecagold Cluster Labeling of Proteins at Reactive Cysteine Residues

Journal of Structural Biology 127, 101–105 (1999) Article ID jsbi.1999.4110, available online at http://www.idealibrary.com on

Undecagold Cluster Labeling of Proteins at Reactive Cysteine Residues Daniel Safer Pennsylvania Muscle Institute, University of Pennsylvania, Philadelphia, Pennsylvania 19104-6085 Received December 1, 1998, and in revised form March 4, 1999

METHODS

Undecagold cluster labeling of reactive cysteine residues in numerous proteins has allowed the labeled sites to be identified by electron microscopy, providing high-resolution information on the location and orientation of subunits in oligomeric enzymes, virus capsids, crystalline sheets of membrane proteins, and muscle thin filaments. The range of applications of undecagold cluster labeling has been greatly extended by the availability of sitedirected mutagenesis to introduce cysteine residues at sites of interest. In this paper I discuss factors that can influence the extent and specificity of labeling, methods for biochemical analysis of undecagold-labeled proteins, and the effects of undecagold cluster labeling on biological activity. r 1999

Undecagold clusters. Undecagold clusters were prepared as previously described and activated by conversion of primary amines to maleimides (Safer et al., 1986). Most of the experiments described here were performed with gold clusters bearing an average of one aminopropylamidophenyl group per cluster. More recently, undecagold clusters were synthesized using tris(4-Nmethylcarboxamidophenyl)phosphine as the sole ligand (Safer et al., 1986), and primary amine ligands were introduced into the purified cluster by ligand exchange (Jahn, 1989): the undecagold cluster was dissolved to 5–10 mM in dry ethanol, 2 mol bis(aminopropyl)phenyl phosphine/mol undecagold cluster was added under an argon atmosphere, and the mixture was allowed to equilibrate for at least 48 h at ambient temperature. Undecagold clusters were then separated from excess phosphine ligand by gel filtration. Ligand exchange was confirmed by reverse-phase highperformance liquid chromatography (HPLC). Undecagold clusters prepared by this procedure have been found to label proteins as effectively as those prepared by the previous method, but have the advantage that the spacer arm is five atoms shorter, which may allow the label to be localized with higher resolution. Following ligand exchange, the undecagold cluster preparation consists of a mixture of species bearing zero, one, two, or more bis(aminopropyl)phenyl phosphine ligands. Neither the heterogeneity of the gold cluster preparation nor the presence of two amino groups on the reactive ligand appears to interfere with the specificity of labeling: we have not observed any detectable level of protein crosslinking or of labeling at multiple sites using these unfractionated preparations. The gold cluster preparation can be fractionated into homogeneous components by cation-exchange chromatography (Safer et al., 1986); however, the undecagold clusters in these initially homogeneous fractions slowly equilibrate by ligand exchange into mixtures of species with different numbers of reactive ligands, as long as the material is stored in solution. Ligand exchange may be prevented by storing undecagold cluster preparations in dry form, but this must be done in vacuo or under an inert atmosphere, since dry films of undecagold clusters decompose slowly due to air oxidation. For labeling purposes, we have found no obvious benefit in using homogeneous fractions rather than the mixture of undecagold clusters with different numbers of reactive groups. Proteins and reagents. Actin, myosin and its subfragments, and tropomyosin were prepared from rabbit skeletal muscle by published methods (Pardee and Spudich, 1982; Margossian and Lowey, 1982; Cummins and Perry, 1973). Histones and spinach calmodulin were obtained from Sigma. Bis(aminopropyl)phenyl phosphine was obtained from Alfa Aesar. Undecagold cluster labeling. Labeling was performed at 4°C for 16–24 h, generally using 4 mol undecagold cluster/mol protein. Actin was labeled in 10 mM imidazole–Cl, 0.2 mM ATP, 0.2 mM CaCl2, pH 8.0, and was dialyzed against the same buffer during labeling to eliminate residual salt in the undecagold cluster

Academic Press

Key Words: actin; chromatography; cysteine; electrophoresis; gold cluster; HPLC; myosin; thiol; undecagold.

INTRODUCTION

Undecagold cluster labeling has been used to locate specific sites by electron microscopy in a variety of macromolecules and assemblies. While a few studies have employed noncovalent, highaffinity binding of undecagold-labeled ligands (Safer et al., 1982; Steinmetz et al., 1998), most applications have relied on the covalent labeling of thiols by undecagold clusters bearing maleimide groups (Safer et al., 1986; Hainfeld, 1987; Milligan et al., 1990; Hainfeld et al., 1991; Crum et al., 1994; Schnyder et al., 1995; Zlotnick et al., 1997). In order for the site of interest to be located by electron microscopy, it must be labeled with high occupancy and specificity, while preserving native structure. In the hope of making the undecagold-labeling method more widely accessible, I describe methods for analyzing labeled proteins to determine the extent and site of labeling, conditions that affect the efficiency of labeling, and the effects of labeling on the biological activity of several proteins. 101

1047-8477/99 $30.00 Copyright r 1999 by Academic Press All rights of reproduction in any form reserved.

102

DANIEL SAFER

preparation, which tended to inhibit labeling by causing polymerization. Myosin subfragment 1 was labeled in 30 mM KCl, 20 mM triethanolamine–Cl, 1 mM NaN3, pH 8.0; tropomyosin was labeled in 50 mM triethanolamine–Cl, 1 mM NaN3, 2 M urea, pH 8.0; histones were labeled in 0.2 M NaCl, 20 mM triethanolamine– Cl, pH 8.0; and spinach calmodulin was labeled in 25 mM sodium Hepes, 1 mM EGTA, 1 mM NaN3, pH 8.1, with 7 M urea. Chloride ion has some tendency to react with and inactivate the maleimide group of the maleimide–undecagold cluster (Hainfeld, personal communication, confirmed in our laboratory); this is particularly evident at pH ⬍8. Where high-ionic-strength buffers are required, other salts such as acetate salts are preferable. Analytical methods. The concentration of undecagold cluster, both free and protein-bound, was determined spectrophotometrically using E420 ⫽ 47 100 M⫺1 cm⫺1 (Safer et al., 1986). Protein concentrations were determined using the Lowry, Bradford, or bicinchonic acid colorimetric assays (Lowry et al., 1951; Bradford, 1976; Smith et al., 1985). Control experiments, in which the undecagold cluster was added to known quantities of protein, showed that the undecagold cluster did not interfere significantly with any of these assays. Polyacrylamide gel electrophoresis (PAGE) was performed both in the presence of sodium dodecyl sulfate (SDS; Laemmli, 1970) and under nondenaturing conditions (Safer, 1989). Samples were prepared for nondenaturing PAGE by the addition of 10% glycerol and bromophenol blue to the native protein and were kept at 4°C. HPLC was performed using an Isco dual-pump chromatograph with variable-wavelength detector. For tryptic fingerprinting, actin was S-alkylated with iodoacetamide and pyrene-iodoacetamide as described under Results and then succinylated with 0.1 M succinic anhydride (added as a 1 M solution in dimethylformamide) in 0.2 M triethanolamine–HCl, pH 8.4, to block lysine ⑀-amino groups and thus restrict cleavage to arginine residues (Elzinga and Phelan, 1984). After 2 h, the S-alkyl, succinyl actins were dialyzed against 500 vol of 50 mM ammonium bicarbonate. To facilitate subsequent peptide analysis, aliquots of the S-alkyl, succinyl actins were incubated in the cold for 72 h with 10% ␤-mercaptoethanol containing approximately 5% sodium borohydride. This procedure causes decomposition of the gold clusters without apparent degradation of the S-alkyl, denatured proteins. The reactions were then dialyzed against two changes of 20 mM ammonium bicarbonate, lyophilized, dissolved to 4 mg actin/ml in 0.5% ammonium bicarbonate– 0.1% n-butanol, and digested overnight at room temperature with trypsin at a ratio of 1/20 (wt/wt) (Elzinga and Phelan, 1984). RESULTS AND DISCUSSION

The Reactivity of Cysteine Residues Small thiol compounds react avidly with alkylating reagents over a broad range of conditions. Reaction rates increase with pH, but competing reactions, such as alkylation of primary amines or hydrolysis of the alkylating reagent, are also promoted by alkaline pH. In our experience, undecagold cluster labeling at pH 7–8, using a fairly small excess of undecagold cluster/protein, has given efficient labeling of cysteine residues without detectable labeling of amino groups. Other alkylating agents generally show similar specificity when used in low excess over cysteine residues (Means and Feeney, 1971). For proteins that contain unusually reactive lysine residues, it may be useful to label at pH 6.5 or lower in order to

maximize the specificity of labeling for cysteine residues. Cysteine residues in proteins vary greatly in reactivity, and reactivity toward the maleimide–undecagold cluster has generally been consistent with published or preliminary experiments using fluorescent alkylating agents. Thus, G-actin is readily labeled at cysteine-374, while F-actin labels poorly; the sterically shielded cysteines of tropomyosin and spinach calmodulin are labeled efficiently only under denaturing conditions (2 M urea for tropomyosin, 7 M urea for calmodulin); and myosin subfragment 1 labels at several reactive cysteines, with the distribution of label depending strongly on solvent conditions. Cysteine residues are rarely exposed on the surface of native proteins, but can be introduced by site-directed mutagenesis (Zlotnick et al., 1997). In preliminary experiments, we have found that the major difficulty in labeling highly exposed cysteine residues is their tendency to oxidize rapidly in the absence of reducing agents. In some cases, high levels of labeling have been obtained by rapidly separating the protein from reducing agent, e.g., by adsorption onto a small ion-exchange column and then rapidly washing, releasing, and adding label. Separation of Proteins from Free Undecagold Cluster In most separation methods, free undecagold cluster behaves like a small, weakly basic protein. Sedimentable proteins can be separated from free undecagold cluster by centrifugation (Crum et al., 1994); precipitation with ammonium sulfate, however, is not recommended, since the undecagold cluster precipitates in the same range as many proteins. Soluble proteins that bind to anion exchangers can conveniently be recovered at high concentration by adsorption to DEAE-cellulose or -agarose. In labeling experiments using myosin subfragment 1 or spinach calmodulin, the labeled protein was applied to a small bed (⬍0.5 ml/mg protein) of the anion exchanger, and free gold cluster was washed through at low ionic strength. The labeled protein, visible as an orange band at the top of the column bed, was released with high-salt buffer and collected in a few drops. Soluble proteins larger than ⬃20 kDa can be separated from undecagold cluster by gelfiltration (Hainfeld, 1987; Zlotnick et al., 1997). Gel filtration was advantageous in the case of G-actin, since labeling resulted in a small amount of aggregated material, and gel filtration on Sephacryl 200 allowed the labeled actin monomers to be separated from both aggregate and free undecagold clusters. Separation of labeled histones from free undecagold cluster posed unusual problems, because histones are similar to undecagold cluster in both size and charge.

UNDECAGOLD LABELING OF PROTEIN THIOLS

103

Separation was obtained by taking advantage of the hydrophobicity of undecagold cluster: gel filtration was performed on Toyopearl HW40, a somewhat hydrophobic polymeric matrix, so that the free undecagold cluster was retarded by adsorption as well as sieving. Undecagold cluster labeling is frequently substoichiometric, and most attempts to separate labeled from unlabeled protein have been unsuccessful. In the case of spinach calmodulin, however, labeled and unlabeled protein could be separated by hydrophobic interaction chromatography on Butyl Toyopearl in 25 mM Tris–Cl buffer at pH 8.0. Unlabeled calmodulin was released from the column at 0.7 M (NH4 ) 2SO4, while the undecagold-labeled calmodulin remained bound and was released with Tris–Cl buffer. Analysis of Undecagold-Labeled Proteins If sufficient material is available, the stoichiometry of labeling can be determined by measuring the concentration of undecagold cluster in the labeled preparation spectrophotometrically and using a colorimetric assay for protein. Labeling can also be assessed by gel electrophoresis, provided that the sample preparation does not require exhaustive reduction of the protein, since high concentrations of mercaptoethanol or other reducing agents destroy the gold cluster. Nondenaturing PAGE has proved useful for the analysis of soluble, acidic proteins, since samples need not be reduced. Undecagold cluster labeling results in a significantly lower electrophoretic mobility for G-actin (Fig. 1) or spinach calmodulin (not shown). SDS–PAGE has been useful for proteins that can be analyzed under nonreducing conditions (Crum et al., 1994). For proteins that contain only a single cysteine residue, specificity of labeling is not an issue. Where more than one cysteine is present, but one residue is known to be particularly reactive, specificity can be tested by determining whether alkylation of the reactive cysteine residue abolishes incorporation of undecagold cluster. For multisubunit proteins with cysteine residues on more than one subunit, the labeled subunit can be identified by SDS–PAGE, provided that samples can be prepared satisfactorily with little or no reducing agent. For proteins that are prone to aggregate in the absence of reducing agent, reverse-phase HPLC has been a useful analytical method: samples are prepared and analyzed at pH ⱕ2, which strongly inhibits formation of disulfides and does not disrupt the undecagold cluster (Fig. 2). The labeled subunit can be distinguished by comparing elution profiles at two different wavelengths, e.g., 220 nm to detect all subunits and 310 nm to detect the undecagold cluster.

FIG. 1. Nondenaturing polyacrylamide gel electrophoresis of G-actin before and after labeling. Lane 1, unlabeled; lane 2, gold cluster-labeled; lane 3, pyrene-labeled under conditions that restrict labeling to cysteine-374 (Kouyama and Mihashi, 1981); lane 4, pyrene-labeled followed by incubation with maleimide– gold cluster. Pyrene labeling completely abolishes formation of the low-mobility, undecagold-labeled derivative.

FIG. 2. Analysis of undecagold-labeled myosin subfragment-1 by reverse-phase HPLC, using a Polymer Laboratories PLRP-S 1000 column; solvent A is 1% H3PO4 in water, solvent B is acetonitrile, resolving gradient from 30 to 60% B in 15 min at 2 ml/min. Undecagold-labeled subfragment-1 was sedimented with F-actin in the absence of nucleotide and then released with MgATP. Detection at 220 nm (upper trace) shows the alkali light chain and heavy chain eluting at 14 and 18 min, respectively. Detection at 310 nm (lower trace) shows that the undecagold cluster is predominantly on the light chain.

104

DANIEL SAFER

When a polypeptide contains more than one potentially reactive cysteine, it may be necessary to determine which residue is labeled by digesting the protein and identifying the labeled peptide. Direct identification of an undecagold-labeled peptide has not been feasible, because proteins with multiple cysteines must be exhaustively reduced and alkylated before digestion, which causes breakdown of the undecagold cluster into heterogeneous products; the labeled cysteine residue then bears only a phosphine or phosphine oxide, which does not make it readily identifiable. To confirm which of the five cysteine residues in actin was labeled by undecagold cluster, a ‘‘difference method’’ was used (Fig. 3): both unlabeled actin and undecagold-labeled actin were exhaustively alkylated with pyrene-iodoacetamide, a chromophoric alkylating agent. The labeled peptide was identifiable by the absence of the corresponding pyrene-labeled peak from the digest of undecagoldlabeled actin.

Effects of Undecagold Cluster Labeling on Biological Activity Alkylation of cysteine residues with undecagold cluster can have greater effects on protein function than alkylation with smaller reagents, particularly when the labeled site is at a protein–protein interface. Filaments polymerized from undecagold-labeled actin were identical in structure to unlabeled filaments, but were less stable, and phalloidin was required to produce specimens suitable for electron microscopy (Milligan et al., 1990). Labeling of the intrinsic cysteine of skeletal myosin light chain A2 disrupted its function even more severely, completely preventing its exchange into myosin subfragment 1, while labeling of intact subfragment 1 gave a mixture of products, with the label distributed between the alkali light chain and one or more sites on the heavy chain. Subfragment 1 that labeled only on the light chain, as determined by HPLC (Fig. 2), showed ATP-sensitive actin binding and retained most of its native ATPase activity, while most of the material labeled on the heavy chain no longer bound actin and retained ⬍30% of native ATPase activity. Thus, the maleimide–undecagold cluster was less selective for the reactive cysteines of myosin than smaller alkylating agents (Reisler, 1982), a result that precluded structural studies of the labeled protein. Where intersubunit contacts are not affected, undecagold-labeled proteins can assemble normally: ␤-tropomyosin labeled with undecagold cluster gave Mg2⫹ paracrystals of normal periodicity (results not shown), and undecagold-labeled hepatitis B virus capsid protein assembled into capsids indistinguishable from wild-type (Zlotnick et al., 1997). This work was supported by NIH Grants AM31984, AR38976, and AR40840.

FIG. 3. Analysis of undecagold-labeled actin by tryptic digestion and HPLC, using a Vydac C18 column; solvent A is 0.1% trifluoroacetic acid in water, solvent B is 0.08% trifluoroacetic acid in acetonitrile, resolving gradient from 18 to 50% B in 16 min at 2 ml/min. Detection at 340 nm shows only the pyrene-labeled peptides. When native actin was labeled with pyrene at cysteine374, and the remaining cysteine residues were alkylated with iodoacetamide in 8 M urea, tryptic digestion produced one major pyrene-labeled peak and one minor pyrene-labeled peak, probably due to incomplete cleavage at Lys-373 (lower trace: single arrow denotes minor peak, double arrow denotes major peak). When actin was labeled with pyrene in 8 M urea so that all cysteine residues were labeled, four additional major pyrene-labeled fragments were detected (center trace). When native actin was labeled with undecagold cluster, and the remaining cysteine residues were labeled with pyrene in 8 M urea, the peaks corresponding to pyrene labeling at cysteine-374 were virtually absent, while other pyrene-labeled peaks were unaffected, indicating that cysteine374 was specifically alkylated by maleimide–undecagold cluster in native G-actin.

REFERENCES Bradford, M. M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248–254. Crum, J., Gruys, K. J., and Frey, T. G. (1994) Electron microscopy of cytochrome c oxidase crystals: Labeling of subunit III with a monomaleimide undecagold cluster compound. Biochemistry 33, 13719–13726. Cummins, P., and Perry, S. V. (1973) The subunits and biological activity of polymorphic forms of tropomyosin. Biochem. J. 133, 765–777. Elzinga, M., and Phelan, J. J. (1984) F-actin is intermolecularly crosslinked by N,N8-p-phenylene-dimaleimide through lysine191 and cysteine-374. Proc. Natl. Acad. Sci. USA 81, 6599– 6602. Hainfeld, J. F. (1987) A small gold-conjugated antibody label: Improved resolution for electron microscopy. Science 236, 450– 453. Hainfeld, J. F., Sprinzl, M., Mandiyan, V., Tumminia, S. J., and

UNDECAGOLD LABELING OF PROTEIN THIOLS Boublik, M. (1991) Localization of a specific nucleotide in yeast tRNA by scanning transmission electron microscopy using an undecagold cluster. J. Struct. Biol. 107, 1–5. Jahn, W. (1989) Synthesis of water soluble undecagold clusters for specific labeling of proteins. Z. Naturforsch. 44b, 1313–1322. Kouyama, T., and Mihashi, K. (1981) Fluorimetry study of N-(1pyrenyl)iodoacetamide labelled F-actin. Eur. J. Biochem. 114, 33–38. Laemmli, U. K. (1970) Cleavage of structural proteins during assembly of the head of bacteriophage T4. Nature 227, 680–685. Lowry, O. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) Protein measurement with the Folin phenol reagent. J. Biol. Chem. 193, 265–275. Margossian, S. S., and Lowey, S. (1982) Preparation of myosin and its subfragments from rabbit skeletal muscle. Methods Enzymol. 85, 55–71. Means, G. E., and Feeney, R. E. (1971) Chemical Modification of Proteins. Holden-Day, San Francisco. Milligan, R. A., Whittaker, M., and Safer, D. (1990) Molecular structure of F-actin and location of surface binding sites. Nature 348, 217–221. Pardee, J., and Spudich, J. A. (1982) Purification of muscle actin. Methods Enzymol. 85, 164–181. Reisler, E. (1982) Sulfhydryl modification and labeling of myosin. Methods Enzymol. 85, 84–93.

105

Safer, D. (1989) An electrophoretic method for detecting proteins that bind actin monomers. Anal. Biochem. 178, 32–37. Safer, D., Bolinger, L., and Leigh, J. S., Jr. (1986) Undecagold clusters for site-specific labeling of biological macromolecules: Simplified preparation and model applications. J. Inorg. Biochem. 26, 77–91. Safer, D., Hainfeld, J., Wall, J. S., and Reardon, J. E. (1982) Biospecific labeling with undecagold: Visualization of the biotinbinding site on avidin. Science 218, 290–291. Schnyder, T., Tittmann, P., Winkler, H., Gross, H., and Wallimann, T. (1995) Localization of reactive cysteine residues by maleidoyl undecagold in the mitochondrial creatine kinase octamer. J. Struct. Biol. 114, 209–217. Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallia, A. K., Gartner, F. H., Provenzano, M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J., and Klenk, D. C. (1985) Measurement of protein using bicinchonic acid. Anal. Biochem. 150, 76–85. Steinmetz, M. O., Stoffler, D., Muller, S. A., Jahn, W., Wolpensinger, B., Goldie, K. N., Engel, A., Faulstich, H., and Aebi, U. (1998) Evaluating atomic models of F-actin with an undecagoldtagged phalloidin derivative. J. Mol. Biol. 276, 1–6. Zlotnick, A., Cheng, N., Stahl, S. J., Conway, J. F., Steven, A. C., and Wingfield, P. T. (1997) Localization of the C terminus of the assembly domain of hepatitis B virus capsid protein: Implications for morphogenesis and organization of encapsidated RNA. Proc. Natl. Acad. Sci. USA 94, 9556–9561.