A General Strategy for Site-Specific Double Labeling of Globular

sulfoxide; DTNB, 5,5′-dithiobis(2-nitrobenzoic acid); DTT, dithio- threitol; EDTA ..... gradient of NaCl (from 0 to 0.5 M) for 120 min at a flow. [D...
0 downloads 0 Views 136KB Size
Bioconjugate Chem. 2002, 13, 1163−1170

1163

A General Strategy for Site-Specific Double Labeling of Globular Proteins for Kinetic FRET Studies V. Ratner, E. Kahana, M. Eichler, and E. Haas* Faculty of Life Sciences, Bar-Ilan University, Ramat Gan 52900, Israel. Received April 6, 2002; Revised Manuscript Received June 5, 2002

Site-directed mutagenesis provides a straightforward means of creating specific targets for chemical modifications of proteins. This capability enhanced the applications of spectroscopic methods adapted for addressing specific structural questions such as the characterization of partially folded and transient intermediate structures of globular proteins. Some applications such as the steady state or timeresolved fluorescence resonance energy transfer (FRET) detection of the kinetics of protein folding require relatively large quantities (∼10-100 mg) of site-specific doubly labeled protein samples. Engineered cysteine residues are common targets for labeling of proteins. The challenge here is to develop methods for selective modification of one of two reactive sulfhydryl groups in a protein molecule. A general systematic procedure for selective labeling of each of two cysteine residues in a protein molecule was developed, using Escherichia coli adenylate kinase (AKe) as a model protein. Potential sites for insertion of cysteine residues were selected by examination of the crystal structure of the protein. A series of single-cysteine mutants was prepared, and the rates of the reaction of each engineered cysteine residue with a reference reagent [5,5′-dithiobis(2-nitrobenzoic acid) (DTNB)] were determined. Two-cysteine mutants were prepared by selection of pairs of sites for which the ratio of this reaction rate constant was high (>80). The conditions for the selective labeling reaction were optimized. In a first cycle of labeling, the more reactive cysteine residue was labeled with a fluorescent probe (donor). The second probe was attached to the less reactive site under unfolding conditions in the second cycle of labeling. The doubly and singly labeled mutants retained full enzymatic activity and the capacity for a reversible folding-unfolding transition. High yields (70-90%) of the preparation of the pure, site-specific doubly labeled AK mutant were obtained. The procedure described herein is a general outline of procedures, which can meet the double challenge of both site specificity and largescale preparation of doubly labeled proteins.

INTRODUCTION

The folding transition of many globular proteins involves transient intermediate states (Kuwajima, 1989; Ptitsyn, 1995; Miranker and Dobson, 1996; Fersht, 1997). This finding raises the following two research questions: what are the early intermediate steps of the folding process, and what is the order of the events leading to the final native structure? To directly respond to these questions, we propose to characterize transient intermediate structures during fast refolding transitions. The characterization of these structures presents an intricate experimental challenge, and available knowledge of such structures is limited. The main difficulties in characterizing these structures are due to their very short lifetime and the need for new methods of detection. The determination of intramolecular distances and rates of the fast changes of such distances by timeresolved fluorescence resonance energy transfer (FRET)1 measurements is a powerful method for the characterization of the structures of partially folded proteins and their * To whom correspondence should be addressed. E-mail: [email protected]. 1Abbreviations: Cl-Bi, monochlorobimane; DMSO, dimethyl sulfoxide; DTNB, 5,5′-dithiobis(2-nitrobenzoic acid); DTT, dithiothreitol; EDTA, ethylenediaminetetraacetate; ET, energy transfer; FRET, fluorescence (Fo¨rster) resonance energy transfer; I-Py, N-(1-pyrenylmethyl)iodoacetamide; TCEP, tris(2-carboxyethyl)phosphine; Tris, tris(hydroxymethyl)aminomethane; wt, wild type.

transitions. Steady state fluorescence detection of FRET is also a powerful method for monitoring fast changes of the conformations of proteins even with fairly simple instrumentation (Stryer, 1978; Van Der Meer et al., 1994; Haas, 1997). The application of this method is most effective when the dependence of distances between pairs of sites on changes of global or subdomain conformations is maximal. Such cases rarely occur naturally, and therefore, application of this method depends on the development of biochemical methods for site-specific attachment of fluorescent probes by chemical modification of proteins. The most effective design of FRET applications in the present context is achieved when pairs of fluorescent probes are attached at pairs of sites associated with secondary or tertiary structural elements. A major obstacle that hinders a wide range application of this method is the difficulty of preparing suitable protein samples. Three labeled protein species should be prepared for each well-controlled, time-resolved FRET experiment. These are two singly labeled mutants (one labeled by the donor and the other labeled by the acceptor) and the doubly labeled (by both the donor and acceptor) sample of the protein. The most general and versatile solution to the problem of site-directed labeling is based on sitedirected mutagenesis and insertion of cysteine residues at the target sites (Haran et al., 1992; Pennington, 1994; Sinev et al., 1996, 2000; DeSantis et al., 1998; Navon et al., 2001; Ortiz, 2001). Reagents for selective alkylation

10.1021/bc025537b CCC: $22.00 © 2002 American Chemical Society Published on Web 08/20/2002

1164 Bioconjugate Chem., Vol. 13, No. 5, 2002

Ratner et al. THEORETICAL BACKGROUND

Calculation of the Maximal Yield of Site-Specific Labeling of Two-Cysteine Mutants. Let [X] be the concentration of the SH groups at the first site, [Y] the concentration of SH groups at the second site, and [D] the concentration of the modifying reagent. All three concentrations decrease during a reaction that modifies cysteine residues in accordance with eqs 1a-c.

d[X]/dt ) -kx[X][D] -

∑kix[Ai][X]

(1a)

d[Y]/dt ) -ky[Y][D] -

∑kiy[Ai][Y]

(1b)

d[D]/dt ) -kx[X][D] - ky[Y][D] - kiD[D]

Figure 1. Backbone structure of wt E. coli adenylate kinase (AKe) (Schultz et al., 1991). The residues which serve as targets for chemical modifications are marked.

of sulfhydryl residues are available. The challenge here is to develop methods for selective modification of one of the two reactive sulfhydryl groups in the same protein molecule. Sub-milligram amounts of labeled proteins can suffice for a full set of FRET experiments under equilibrium conditions. Such minute amounts of site-specific doubly labeled protein samples can be prepared, based on highresolution chromatographic separations. Hundred-fold larger preparations of labeled proteins are needed for flow experiments. In this case, it is desirable to develop methods in which the selectivity is achieved by differential reactivity of selected sites on the protein molecule. The development of a new general systematic procedure for selective labeling of each one of two cysteine residues in a protein molecule is reported here, using Escherichia coli adenylate kinase (AKe) as a model protein. AKe is a ubiquitous enzyme, which catalyzes the phosphoryl transfer reaction ATP + AMP T 2ADP (Noda, 1973). The E. coli AKe molecule (Figure 1) consists of a single chain of 214 residues folded in three domains: the large CORE domain (∼154 residues) and two smaller domains (∼30 residues each) (Schultz, 1990; Mu¨ller and Schultz, 1992; Schultz et al., 1997). Residues 161-188 of AKe form a long stretch of helical structure. Ratner et al. (2000) studied the refolding of fluorescently labeled mutants of AKe by means of fast mixing and time-resolved FRET detection. The starting mutant of AKe used by Ratner et al. (2000) (AK) contained the C77S mutation. The protein was labeled at residue 73 with N-(1-pyrenylmethyl)iodoacetamide (I-Py) and residue 203 with lucifer yellow, with a yield of 80%. This experiment showed at least two steps in the refolding kinetics of AK (Ratner et al., 2000). A transient intermediate, formed within 2 ms from the initiation of refolding, was found. The characterization of this intermediate, and the search for possible additional transient intermediate states of the refolding of this protein using FRET detection, requires labeling of additional pairs of sites in 50-100 mg quantities. Residues 169 and 188 were labeled with a donor and an acceptor of excitation energy, respectively. Tens of milligrams of each specifically labeled (>95%) mutant were prepared and characterized.

(1c)

where kiX and kiY are the miscellaneous reaction rate constants for various competing reactions of the SH groups, which make the SH groups nonreactive with the modifying reagent. Ai terms represent various solvent additives that can be involved in such reactions. These reactions can be eliminated by the addition of the reducing agent such as tris(2-carboxyethyl)phosphine (TCEP) to the reaction mixture. This reagent maintains the reduced state of the cysteine side chains, and it does not react with alkylating reagents. Thus, the last term in eqs 1a and 1b can be neglected. kiD is the rate constant for reaction of D with solvent components resulting in the deactivation of the reagent. The division of eq 1a by eq 1b (the last terms are ignored) yields

d[X]/d[Y] ) (1/R)([X]/[Y])

(2)

where R ) ky/kx. The division of [X] and [Y] by [P], the protein concentration, gives the corresponding fractions of the reactive SH groups, at sites 1 and 2, respectively; x ) [X]/[P], and y ) [X]/[P]. Integration of eq 2 then yields x ) yR. The following four species are found in a reaction mixture of any modifying reagent with a protein that carries two reactive cysteine groups at sites 1 and 2: C1-C2, D1-C2, C1-D2, and D1-D2 (where Ci and Di represent the unmodified and modified sites, respectively, in each species). The corresponding fractions are xy, (1 - x)y, x(1 - y), and (1 - x)(1 - y). In a case where ky, the rate constant for the reaction of the sulfhydryl group at site 2, is greater than the rate of reaction of the first site, kx, i.e., R > 1, the goal is to maximize the fraction of the species (C1-D2) in the product mixture. Combining the expressions x ) yR and the fraction of that species, x(1 - y), one obtains

[C1,D2] ) yR(1-y)

(3)

This goal can be achieved when y corresponds to the condition d[C1-D2]/dy ) 0. The solution yields

yopt ) R/(1 + R)

(4)

At this magnitude of y, the concentration of the desired product is maximal and is provided by eq 5.

[C1-D2]max ) RR/[(1 + R)R+1]

(5)

For a given value of R, the control of the yield of the desired product depends on the initial concentrations of the mutant protein ([P]) and the reagent ([Do]) and the ratio ky/kiD. Practically, two modes of control of the extent

Bioconjugate Chem., Vol. 13, No. 5, 2002 1165

of the reaction are available: either the limitation of the initial concentration of the modifying reagent, [Do], or the limitation of the time course of the reaction. In a case where kiD , ky, the composition of the product mixture can be controlled by the duration of the reaction. In this case, the link between [Do] and y is as expected when the calculated extent of labeling of site 2, (1 - y), reaches the optimal value (eq 5). The expected optimal duration of the modification reaction can be determined using the standard kinetics rate law for bimolecular reactions and the rate constants for the reactions of the pairs of single-cysteine mutants and their ratios (R). It was found that for the reaction of iodoacetamidofluorescein, monochlorobimane (Cl-Bi), I-Py, and 4-iodoacetamidosalicylic acid with AK mutants, the competing deactivation reactions can be ignored; i.e., kiD , ky. In the case where kiD g ky, i.e., the rate of deactivation of the modifying reagent is greater than the rate of the modification reaction, both modes of control can be used. When the extent of the reaction is controlled by the initial concentration of D ([Do]), the reaction is allowed to proceed to completion, without any time limitation. Under these conditions, the link between [Do] and y is given by

[Do] ) [P](y + y1/R) - (kiD/ky) ln(1 - y)

(6)

The rate constants (kx, ky, and kiD) can be determined for each modification reagent in preliminary experiments using minute amounts of the single-cysteine mutants. Given these rate constants, eq 6 can be applied to calculate the optimal [Do], which corresponds to the optimal final concentration of the product (C1-D2). The commonly used ratio [Do]/[P] is not valid in this case. A typical example of the fluorescent reagent for which kiD . ky dye is the iodoacetamido derivative of lucifer yellow (Molecular Probes). In an aqueous solution at pH 7.2 and room temperature, this reagent is hydrolyzed in the dark within 15 min. EXPERIMENTAL PROCEDURES

Procedure for the Systematic Search for Potential Sites for Mutagenesis and Labeling. The task is to prepare a mutant protein with two different organic groups attached to two different sites. These sites were selected according to the needs of a biophysical experiment with minimal perturbation of the native structure and function of the protein. A general procedure for the selection of suitable pairs of sites includes several steps in which the set of residues considered for chemical modification is reduced. (1) Screening for all the residues that could be labeled. The available three-dimensional native structure of the target protein is screened insearch for all the solvent-exposed residues, i.e., those with g30% solvent accessibility (algorithms such as the Swiss PDB viewer can be used). The product of this search is a set of residues expected to be accessible to soluble reagents in solution and whose chemical modification is expected to cause minimal perturbation of the global and local structure of the protein. Exposed charged residues are not considered for modification and are eliminated from this set. (2) Producing single-cysteine mutants. Residues included in the reduced set are mutated, one at a time, and single-cysteine mutants are produced and purified. (3) Determining the rate constants for the reaction of the single-cysteine mutants with a reference reagent. A systematic search will be conducted for reaction conditions. These conditions should enhance

differences in the reactivity of the selected and purified cysteine mutants with sulfhydryl modification reagents. (4) On the basis of the results of this search, it is possible to select pairs of sites, which show large differences in the reaction rates with the reference reagent, reducing the set by the selection of pairs whose modification is of interest for specific biophysical experiments. These can be pairs of sites where probes can detect changes of conformation according to the goals of the experiment, e.g., secondary or tertiary structural elements, loops, etc. (5) Measurement of the spectroscopic properties (fluorescence lifetimes, spectral shifts, and parameters affecting the Fo¨rster critical distance Ro) of the potential probes in situ (at the selected sites) is performed to select the most appropriate probes for each experiment. (6) Optimizing the methods for separation of the protein mutants labeled by a single probe at different sites. (7) Measuring of the kinetics of the reaction of the two singlecysteine mutants with the modification reagents under the optimized conditions. (8) Preparing two-cysteine mutants (cysteine residues inserted at the two sites selected in step 5). (9) Carrying out the reaction of the two-cysteine protein with a chemical modification reagent under the optimized conditions. This reaction produces a mixture of the unlabeled protein and three products: two singly labeled species and the doubly labeled species. Enrichment of the fraction of a selected singly labeled species in the reaction mixture is possible on the basis of knowledge of the ratio of the reaction rate constants of these sites with a modifying reagent. Expression and Purification of AKe. The expression of cysteine-mutated E. coli adenylate kinase and protein purification were performed as described previously (Sinev et al., 1996). The protein was stored in 50% glycerol in 40 mM sodium phosphate buffer (pH 6.5) containing 0.1 M NaCl, 2 mM EDTA, and 2 mM DTT at -18 °C. The starting mutant of E. coli adenylate kinase used in this study (AK) contained the C77S mutation. Small-Scale Labeling of AK Mutants. Reactivation of Stored Protein Samples. Samples of the stored protein (3 mg in 0.2 mL) were diluted 5-fold with 100 mM TrisHCl buffer (pH 8.0) containing 10 mM DTT and 2 mM EDTA (the reactivation buffer). After 30 min, the protein solution was applied to a Sephadex G15 column (1.5 cm × 5 cm for 1 mL protein samples) equilibrated and eluted with 20 mM HEPES buffer (pH 7.2) containing 2 mM EDTA. Labeling of Single-Cysteine Mutants. The single-cysteine mutants, AK(V169C) and AK(A188C), were labeled with the fluorescent reagent N-(1-pyrenylmethyl)iodoacetamide (I-Py) (Molecular Probes). The labeling was performed in 20 mM HEPES (pH 7.2) which contained 2 mM EDTA and 30 µM reducing agent tris(2-carboxyethyl)phosphine (TCEP) (Molecular Probes) (folding buffer). The reaction mixture contained 20 µM protein in this buffer and 75 µM I-Py, which was added in DMSO [the final concentration of DMSO was 10% (v/v)]. The reaction was allowed to proceed at room temperature. Aliquots of the reaction mixture were taken at several time points between 4 and 24 h and analyzed. Excess insoluble pyrene was filtered using a 0.2 µm sterile filter, and the soluble excess of the reagent was removed by gel filtration on a Sephadex G15 column equilibrated with 20 mM Tris-HCl (pH 8). The separation of the labeled protein was performed by anion exchange chromatography. Samples of ∼0.5 mg were applied to a MonoQ column (0.5 cm × 5 cm) (Pharmacia) equilibrated with 20 mM Tris-HCl (pH 8). Proteins were eluted with a linear gradient of NaCl (from 0 to 0.5 M) for 120 min at a flow

1166 Bioconjugate Chem., Vol. 13, No. 5, 2002

Ratner et al.

Figure 2. Absorbance and fluorescence spectra of AK(V189CPy) (donor) (- - -) and AK(A188C-Bi) (acceptor) (s). Emission spectra were recorded by excitation at 345 and 390 nm for Py and Bi, respectively (bandwidth of 3 nm).

Figure 3. Elution profile (s) of the preparative reaction mixture applied to a Phenyl Sepharose 6 Fast Flow column. Four peaks were resolved. The fractions were as follows: I, unlabeled AK; II, AK(V169C/A188C-Py); III, AK(V169C-Py/ A188C); and IV, AK(V169C-Py/A188C-Py) The salt gradient is shown (- - -). The reference preparations for identification of these fractions were the base mutant, AK(V169C-Py), AK(A188C-Py), and AK(V169C-Py/A188C-Py) labeled in unfolding buffer with excess I-Py.

rate of 0.5 mL/min. The free protein eluted first, at 100111 mM NaCl; AK(V169C-Py) eluted at 112-119 mM NaCl, and AK(A188C-Py) eluted at 180-190 mM NaCl. The extent of protein labeling was evaluated by determining the area under the peaks, divided by molar extinction coefficients of the unlabeled protein at 280 nm (11 800 M-1 cm-1) or the labeled protein at 280 nm (49 200 M-1 cm-1). AK(A188C) (14 µM) was also labeled with monochlorobimane (Molecular Probes) (final concentration of 530 µM), 20 mM HEPES (pH 7.8), 1.8 M GuHCl, 30 µM TCEP, and 2 mM EDTA (unfolding buffer). The concentration of the soluble reagent in the reaction was determined by the absorbance at 390 nm, using a molar extinction coefficients of bimane of 5560 M-1 cm-1 in HEPES buffer (pH 7.2). Large-Scale Double Labeling. A quantity of 65 mg of reactivated AK(V169C/A188C) was dialyzed in 20 mM HEPES (pH 7.2) and 2 mM EDTA. Both TCEP (from a 10 mM stock solution in water) and NaCl (from a 4 M stock solution in water) were added to final concentrations of 30 µM and 100 mM, respectively. A solution of 4.6 mg of I-Py in 3.85 mL of DMSO was added in two portions (2.2 and 1.65 mL) at the initiation of the reaction and after 8 h, respectively. The reaction was allowed to proceed for 24 h under slow stirring at room temperature (24 °C) protected from light. The reaction was stopped by the addition of DTT to a final concentration of 50 mM. The excess Py derivatives were removed by filtration through a 0.2 mm filter (acrodisc 32, Gelman Science, Ann Arbor, MI) and by gel filtration on a G15 Sephadex column in Tris buffer (pH 8.0). The product mixture was resolved by hydrophobic interaction chromatography. NH4SO4 was added to a final concentration of 0.8 M, and after 10 min, the solution was applied to a Phenyl Sepharose 6 Fast Flow (low sub) (Pharmacia Biotech) column (2.5 cm × 16 cm), pre-equilibrated with 0.6 M NH4SO4 in 20 mM HEPES (pH 7.2). The labeled species were eluted in a linear gradient from 0.6 to 0 NH4SO4 in 350 mL at a flow rate of 3 mL/min (Figure 3). The third eluted peak (0.0 M NH4SO4, ∼100 mL) was concentrated to ∼10 mL in an ultrafiltration cell with a YM membrane with a 10K

cutoff (Amicon) and dialyzed in 20 mM HEPES (pH 7.2) and 2 mM EDTA. Acceptor Labeling. AK(V169C-Py/A188C) was reactivated and dialyzed against 20 mM HEPES and 2 mM EDTA. GuHCl (8.2 M stock solution in water) and TCEP were added to final concentrations of 1.8 M and 30 µM, respectively. The addition of GuHCl shifted the pH to 7.8. Solid Cl-Bi (Molecular Probes) was added to a final concentration of 500 µM in excess of the protein concentration. The reaction was allowed to proceed for 24 h at room temperature (gently stirring, protected from light) and then terminated by the addition of DTT to a final concentration of 20 mM. The unbound reagent was removed by gel filtration on a Sephadex G15 column (2.5 cm × 24 cm) equilibrated with 1.8 M GndHCl in 20 mM Tris (pH 8.0). Refolding was performed by dialysis in 20 mM Tris-HCl (pH 8). After concentration in a Centriprep10 ultrafiltration concentrator (Amicon) and being dialyzed in 50 mM imidazole buffer (pH 7.2), all the labeled protein samples were stored in 50 mM imidazole buffer (pH 7.2) containing 50% glycerol at -18 °C. Large-Scale Preparation of Singly Labeled Mutants. Singly labeled AK(V169C-Py) and AK(A188C-Bi) were prepared from the respective single-cysteine AK mutants. Reactivated AK(V169C) was dialyzed in 20 mM HEPES (pH 7.2) and 2 mM EDTA and labeled in unfolding buffer. The final reaction mixture contained equivalent amounts of protein and dye with an additional portion of 50 µM I-Py in excess of the protein concentration. The total amount of the reagent dissolved in DMSO was added in two equal portions (at the initiation of the reaction and 4 h later). The final DMSO concentration was 15% (v/v), and the reaction was allowed to proceed for 24 h under slow stirring at room temperature (24 °C) protected from light. The reaction was terminated by the addition of DTT to a final concentration of 20 mM. The excess Py derivatives were removed by filtration (0.2 µm) followed by gel filtration on a Sephadex G15 column in 50 mM Tris-HCl (pH 8) containing 1.8 M GuHCl. The unfolded protein was refolded by dialysis in 50 mM TrisHCl (pH 8.0). The protein was then concentrated and stored in 50 mM imidazole buffer (pH 7.2) containing 50% glycerol at -18 °C.

Bioconjugate Chem., Vol. 13, No. 5, 2002 1167 Table 1. Rate Constants for the Reaction of SH Groups in a Series of AK Mutants with DTNB at pH 7.2 in 20 mM HEPES at Two Temperatures, with and without Salt k (mM-1 s-1) without salt

a

with 100 mM NaCl

mutanta

22.5 °C

40 °C

22.5 °C

40 °C

AK(V203C) AK(V169C) AK(A55C) AK(V142C) AK(A73C) AK(A188C)

5.3 ( 0.2 1.81 ( 0.09 1.13 ( 0.08 0.186 ( 0.009 0.027 ( 0.002 0.017 ( 0.001

0.84 ( 0.04 0.48 ( 0.02 0.154 ( 0.013 0.059 ( 0.004 ND ND

10.9 ( 0.3 3.58 ( 0.04 1.11 ( 0.02 0.383 ( 0.008 0.115 ( 0.005 0.077 ( 0.004

1.78 ( 0.05 0.89 ( 0.05 0.16 ( 0.01 0.133 ( 0.005 0.050 ( 0.003 0.039 ( 0.004

E. coli adenylate kinase in which cysteine 77 was replaced with serine and one surface-exposed residue was replaced with cysteine.

Reactivated, dialyzed AK(A188C) was labeled with ClBi using the same procedure described for AK(V169CPy/A188C); it was then concentrated and stored in 50 mM imidazole buffer (pH 7.2) containing 50% glycerol at -18 °C. The degree of homogeneity of the products AK(V169CPy/A188C-Bi) and AK(V169C-Py) was analyzed using analytical FPLC ion exchange chromatography. Spectroscopic Instrumentation. Rate constants for the reaction of AK mutants with DTNB were determined using a homemade stopped-flow device based on a DX17MV mixing unit (Applied Photophysics, Leatherhead, U.K.) (Ratner and Haas, 1998). Reactions were initiated by mixing equal volumes of 12 µM protein and of 0.5-2.0 mM DTNB in 20 mM HEPES and 5 mM EDTA (pH 7.2). The kinetics of the reactions were monitored by recording the absorbance at 440 nm (selected by an interference filter). Rate constants were extracted by curve fitting of the kinetics data using a fitting module prepared by A. Goldin and implemented in MatLab (Ratner and Haas, 1998). Fluorescence spectra were recorded using an Aviv F105 spectrofluorimeter. Absorbance spectra were recorded using an Aviv 17DS spectrophotometer. RESULTS

Preparation of Cysteine Mutants of AKe. The single native cysteine residue of Ake(Cys77) was replaced with a serine residue without any perturbation of the folding and activity of the enzyme (Sinev et al., 1996). Ake(C77S) (AK) was the base mutant for all further modifications in this project. Single-cysteine mutants, prepared by the replacement of residues 55, 73, 142, 169, 188, and 203, were produced, expressed, and purified in ∼200 mg quantities (Sinev et al., 1996). Determination of Reaction Rate Constants with a Reference Reagent. Elman reagent [5,5′-dithiobis(2-nitrobenzoic acid) (DTNB)] was used as the reference reagent. The rate constants for the reaction of DTNB and each of the single-cysteine mutants were determined in 20 mM HEPES (pH 7.2) and in the same buffer with the addition of 100 mM NaCl at two temperatures. The rate of light absorption change at 440 nm was measured in a stopped-flow device. Table 1 shows that the rate constants for the DTNB-AK reaction varied over a wide range. Pairs of sites for which the ratio of rate constants was nearly 100 were selected for labeling. The first example was the pair AK(V203C) and AK(A73C), which were labeled with lucifer yellow and I-Py, respectively (Ratner et al., 2000). The pair AK(A169C) and AK(A188C), which had a ratio of rate constants as high as 106, was used in the experiment presented here. Fluorescent Probes for Labeling Cysteine 169 and 188. The two fluorescent alkylating reagents used

in this study were the donor I-Py and Cl-Bi, the acceptor of excitation energy. The alkylation with the pyrene reagent creates a very hydrophobic product, which is a possible source of perturbation of the protein structure or folding. Yet this reagent was used for its favorable spectroscopic properties (very long fluorescence lifetime and high extinction coefficient). The bimane reagent is a comparatively small, uncharged probe. Preparation of Singly Labeled Mutants. The two single-cysteine mutants, AK(V169C) and AK(A188C) (∼40 µM), were labeled with I-Py (∼100 µM) under reducing-unfolding conditions where every protein sulfhydryl group was accessible to the dye. The solubility of I-Py in folding buffer did not exceed ∼10 µM at 10% DMSO. Therefore, a heterogenic (pseudo-first-order) reaction was used. The insoluble excess of the reagent formed microaggregates, which served as a stock. This stock provided a saturated concentration of the reagent, and was gradually depleted during the reaction. The reaction was complete (100% labeling) within 4 h at room temperature. The quantum yields of pyrene fluorescence were 0.33 and 0.06 for AK(V169C-Py) and AK(A188C-Py), respectively. Cysteine 169 was chosen as the target for Py conjugation for the planned FRET experiments since it was difficult to find an acceptor probe that has a suitable Ro value when paired with the low-quantum yield pyrene attached to residue 188. Singly labeled AK(A188C-Bi) was obtained by the reaction of 40 µM protein in the unfolding buffer with 550 µM Cl-Bi for 24 h at room temperature. Figure 2 presents the fluorescence spectrum of AK(V169C-Py) and the absorbance spectrum of AK(A188C-Bi). The Fo¨rster critical distance, Ro, of the pair N-(1-pyrenylmethyl)acetamide (Py) and bimane (Bi) attached to residues 169 and 188, respectively (calculated assuming an orientation factor κ2 of 2/3), is 26 Å. This length is close to the separation of these two residues in the crystal structure (28.9 Å). The wild-type level of specific enzymatic activity was measured for both singly labeled mutants. This activity is an indication that the above-referenced genetic and chemical modifications did not significantly perturb the native conformation of the protein. Chromatographic Purification. The search for a method of separating the reaction products was conducted using the singly labeled mutants. Nonlabeled protein and the two singly Py-labeled mutants (at residues 169 and 188) were separated on an analytical MonoQ ion exchange column. Larger-scale preparative separations were performed by hydrophobic chromatography using a Phenyl Sepharose 6 Fast Flow column. The two species of singly Py-labeled AK [AK(V169C-Py) and AK(A188C-Py)], which differed only in the attachment sites of the probe, were resolved with this column. Figure 3 presents the results of the separation of the alkylation reaction products.

1168 Bioconjugate Chem., Vol. 13, No. 5, 2002

Preparation of the Doubly Labeled Mutant. The rate constants for the reaction of the two single-cysteine mutants with I-Py were measured under reducingfolding conditions at room temperature. The following values were obtained: 0.002 and 0.08 h-1 for I-Py and cysteine 188 and 169, respectively. These rate constants were used for optimization of the site-directed labeling of the two cysteine residues in AK(V169C/A188C) as outlined above. An ∼40 µM solution of AK(V169C/A188C) was first reacted with I-Py (∼100 µM) in folding buffer at room temperature. Under these conditions, the rate constants obtained for AK(V169C/A188C) matched the rates obtained for single-cysteine mutants AK(V169C) and AK(A188C). The reaction mixture was resolved by ion exchange chromatography on a Mono-Q column. Large-Scale (∼100 mg) Production of the Doubly Labeled Mutant. With a 10-fold increase in the concentration of the protein in the reaction mixture, the amount of I-Py in this mixture was increased accordingly. Under these conditions, the insoluble excess of I-Py formed large flakes. In addition, the ratio of the reaction rates was reduced due to a cooperative effect, which increased the fraction of AK(V169C-Py/A188C-Py) and reduced the yield of the desired product. Adsorption of protein molecules on the surface of the reagent flakes possibly enhanced the reactivity of the slow-reacting side chain and caused the cooperativity. This undesired cooperative effect was reduced to some extent by the stepwise addition of the reagent (two steps). The reaction proceeded for 24 h and was terminated by the addition of excess DTT. The resulting product mixture contained ∼70% AK(V169C-Py/A188C) along with smaller fractions of AK(V169C/A188C), AK(V169C/ A188C-Py), and AK(V169C-Py/A188C-Py). This mixture was separated by hydrophobic interaction chromatography on a Phenyl Sepharose 6 Fast Flow column (Figure 3). Second-Site Labeling. The pure AK(V169C-Py/ A188C) was reacted with excess Cl-Bi in unfolding buffer. Here, the low reactivity of the side chain of Cys188 was enhanced. After 24 h, the reaction was terminated by the addition of DTT (final concentration of 20 mM). The excess reagent and its products were removed by gel filtration [Sephadex G15 column in 50 mM Tris buffer (pH 8.0) and 1.8 M GndHCl], followed by dialysis against 50 mM Tris buffer. The homogeneity of the final doubly labeled product, AK(V169C-Ph/A188C-Bi), was analyzed by analytical chromatography on a Mono-Q FPLC column. In a control experiment, the base mutant AK (which contains the C77S mutation) was subjected to the same labeling procedure. No trace of fluorescence was detected in the final protein fraction. This finding indicates that only the specific covalent attachment of the probes to the cysteine residues was involved. The chromatography and dialysis steps efficiently removed nonspecifically adsorbed reagents. Observation of FRET. The bimane fluorescence that was directly excited in AK(A188C-Bi) in the native state was very close to the value obtained for that probe in AK(V169C-Py/A188C-Bi) under those conditions. Under denaturing conditions (in 1.8 M GuHCl), the fluorescence of bimane decreased by ∼10 and ∼70% for AK(A188CBi) and AK(V169C-Py/A188C-Bi), respectively, relative to the corresponding values obtained in the native state. The decrease in the fluorescence of bimane in the doubly labeled mutant occurred at the same concentration of GuHCl as the change of the pyrene fluorescence. Figure 4 shows the denaturant concentration dependence of the fluorescence emission intensities of the Py probe in the

Ratner et al.

Figure 4. Dependence of pyrene fluorescence in singly (asterisks) and doubly (triangles) labeled AK on GuHCl concentration. The lines show the fits to a two-state model with an m of 8.5 kcal/M and midtransitions at 0.77 and 0.64 M for the singly and doubly labeled mutants, respectively. The FRET parameters for the attached Py-Bi pair were as follows: Ro ) 22 Å and Rmean (apparent mean distance) ) 20 Å for the unfolded state and Ro ) 25.5 Å and Rmean ) 32 Å for the folded state of the protein.

singly and doubly labeled mutants AK(V169C-Py) and AK(V169C-Py/A188C-Bi). A sharp drop in the emission intensity at ∼0.7-0.8 M GuHCl appears for both mutants. The drop in Py fluorescence intensity in the singly labeled mutant reflects the change in the local environment of the probe. This change affects the Fo¨rster critical distance, Ro, which is recalculated for each state of the protein (Table 1). The relative amplitude of the decrease of the donor emission intensity observed for the doubly labeled mutant is greater than that of the singly labeled mutant, due to the enhanced efficiency of the FRET effect upon denaturation. The ratio of the emission intensity of the donor in the doubly and singly labeled mutants (absence and presence of FRET) is used as a measure of the extent of FRET efficiency. The increased ratio found at high GuHCl concentrations indicates enhanced transfer efficiency. Upon unfolding, the extended R-helical conformation (residues 161-188) (Schultz et al., 1991) collapsed to an apparent unordered state, where the apparent mean inter-residue distance between residues 169 and 188 was considerably shorter than in the native state. The mean FRET efficiencies were calculated for the two equilibrium states (native and denatured). An apparent mean distance between the two probes was calculated for each state. This distance is a rough approximation, because it is obtained by ignoring the effect of the width of the intramolecular distance distributions and the possible contributions to the FRET efficiencies by the fast segmental fluctuations. The distances were calculated under the assumption that distributions of distances were narrow. This is a reasonable approximation for the native state. However, for the denatured states, this approximation is probably somewhat biased toward lower values. Thus, the apparent change in the mean distance from the denatured to the native states may indeed be partially due to a reduction of the distributions. DISCUSSION

Site-directed mutagenesis provides a straightforward means of creating specific targets for chemical modifica-

Bioconjugate Chem., Vol. 13, No. 5, 2002 1169

tions of proteins. This capability enhanced the applications of spectroscopic methods adapted for addressing specific structural questions. The procedure described herein is a general outline of procedures, which can meet the double challenge of both site specificity and largescale preparation of doubly labeled proteins. The selectivity of the labeling reaction can be further enhanced by a local increase or reduction in the reactivity of selected side chains by means of conformational effects. Solvent composition (organic solvents, salt components, and pH) and temperature are the most common variables that can differentially affect the reaction rate constants of selected side chains. Drastic effects can be obtained by ligand binding (substrates, inhibitors, and proteinprotein interactions). In this work, the ratio of the reaction rate constants was examined only under a range of temperatures and salt concentrations which yielded satisfactory results for two pairs of sites (residues 73 and 203 and residues 169 and 188). The level of homogeneity required for spectroscopic measurements makes the chromatographic separation steps indispensable. Separation of the reaction products can be enhanced when one of the reagents is either charged or very hydrophobic. Yet either hydrophobicity or extra electrostatic charges can affect the folding pathway of proteins and can be a disadvantage when the modifications are made for structural studies. Affinity chromatography was also used to separate multiply labeled protein derivatives (Amir et al., 1986; Amir and Haas, 1987). It is also possible to use a mixed disulfide resin for separation of fractions with free cysteine residues in the process of labeling two-cysteine mutants. The excellent separation that was achieved between unlabeled AK and AK labeled with I-Py at either residue 169 or 188 was based on the hydrophobicity of the pyrene moiety. The procedure described here is based on the preparation of three mutants: two single-cysteine mutants and one two-cysteine mutant. In principle, it is possible to accomplish the FRET experiments without the preparation of the two single-cysteine mutants. The need to prepare two single-cysteine mutants depends on the required quantities of labeled proteins and the availability of analytical procedures for the identification of the labeling sites. Large-scale preparations can be done either by increasing the volume of the reaction mixture (thus maintaining low protein concentrations) or by increasing the protein concentration. The second procedure was applied in the study presented here since it has two practical advantages: (1) reduction of the waste of precious fluorescent reagents and (2) the prevention of the losses of the labeled protein in the extra concentration step required after a reaction in a large volume. The precipitation of the hydrophobic fluorescent reagent, the ensuing changes of the reaction conditions, and the reduced maximal yield of the labeled protein are the disadvantages of this approach. The optimization of the reaction volume and of the reagent’s concentrations in a large-scale labeling reaction depends on the solubility of the protein and the reagents. FRET Measurements. The reduced quantum yield of the pyrene florescence in the denatured state of the singly labeled mutant AK(V169C-Py) was not due to solvent quenching by the denaturant. This conclusion is evident from the observation that the same probe attached to cysteine 73 exhibited only a minimal reduction of the fluorescence quantum yield under the same conditions (Ratner et al., 2000). A possible source for a specific

quenching of pyrene fluorescence in AK(V169C-Py) might be the side chain of one of the neighboring residues. Collision quenching by an adjacent neighbor side chain was restricted by its orientation in the native state helical structure. When this well-defined structure was collapsed by denaturation, the stereochemical restriction was relaxed and collision quenching was made more efficient. Charged side chains, either positive or negative, are possible quenchers found in the vicinity of both residue 73 and residue 169. The phenol side chain of tyrosine is found only near residue 169. Therefore, it is reasonable to speculate that the likely near-neighbor quencher side chain is Tyr171. Incidentally, the donor’s specific nonFRET quenching helped to improve the resolution of the distance determination in the experiment described here, since it caused a decrease in the Fo¨rster critical distance, Ro, from 26 Å (in the native state) to 21.8 Å. This Ro value is closer to the apparent mean distance between the two residues in the denatured state (Table 2). The mean length of the 20-residue labeled helical segment in the native state (31.7 Å) is close to the theoretically expected value (CR-CR separation of 29.3 Å) and the Cγ-Cβ separation in the crystal structure (30.5 Å). CONCLUSIONS

The preparation of large-scale amounts of site-specific, doubly chemically modified mutant proteins for spectroscopic studies of protein folding and dynamics can be improved. Such improvement can be achieved by following a systematic procedure, which is based on structural and chemical kinetics data. The native three-dimensional structure should be used for the preliminary selection of a set of residues, which can be mutated with minimal structural perturbation, and with a large variation of reactivity of inserted cysteine residues. The determination of individual reaction rate constants of the engineered SH groups and the optimization of reaction conditions allowed high yields (70-90%) of the preparation of the pure, site-specific doubly labeled AK mutant. These high yields make some otherwise difficult experiments, such as the double kinetics time-resolved FRET measurements (Ratner and Haas, 1998; Ratner et al., 2000), feasible. Preliminary FRET measurements show the effectiveness of this experimental approach. ACKNOWLEDGMENT

This work was supported by research and equipment grants from the Israel Science Foundation, by NIH Grant GM39372, and by the Demedian Center for Magnetic Resonance Research at Bar-Ilan University. LITERATURE CITED (1) Amir, D., and Haas, E. (1987) Estimation of Intramolecular Distance Distributions in Bovine Pancreatic Trypsin Inhibitor by Site-Specific Labeling and Nonradiative Excitation Energy Transfer Measurements. Biochemistry 26, 2162-2175. (2) Amir, D., Levy, D. P., Levin, J., and Haas, E. (1986) Selective Fluorescent Labeling of Amino Groups of Bovine Pancreatic Trypsin Inhibitor by Reductive Alkylation. Biopolymers 25, 1645-1658. (3) DeSantis, G., Berglund, P., Stabile, M. R., Gold, M., and Jones, J. B. (1998) Site-directed mutagenesis combined with chemical modification as a strategy for altering the specificity of the S1 and S1′ pockets of subtilisin Bacillus lentus. Biochemistry 37 (17), 5968-5973. (4) Fersht, A. R. (1998) Nucleation mechanisms in protein folding. Curr. Opin. Struct. Biol. 7 (1), 3-9. (5) Forster, Th. (1948) Zwischen Molekulare Energie Wanderung und Fluoreszenz. Ann. Phys. (Berlin) 2, 55-75.

1170 Bioconjugate Chem., Vol. 13, No. 5, 2002 (6) Foucaud, B., Perret, P., Grutter, T., and Goeldner, M. (2001) Cysteine mutants as chemical sensors for ligand-receptor interactions. Trends Pharmacol. Sci. 22 (4), 170-173. (7) Guiso, N., Michelson, S., and Barzu, O. (1984) Inactivation and proteolysis of heat-sensitive adenylate kinase of Escherichia coli CR341 T28. J. Biol. Chem. 259, 8713-8717. (8) Haas, E. (1996) The problem of protein folding and dynamics: time-resolved dynamic non-radiative excitation energy transfer measurements. IEEE-JSTC 2, 1088. (9) Haas, E., Katchalski-Katzir, E., and Steinberg, I. Z. (1978) Brownian motion of the ends of oligopeptide chains in solution as estimated by energy transfer between the chain ends. Biopolymers 17, 11-31. (10) Kuwajima, K. (1989) The molten globule state as a clue for understanding the folding and cooperativity of globularprotein structure. Proteins: Struct., Funct., Genet. 6, 87-103. (11) Miranker, A. D., and Dobson, C. M. (1996) Collapse and cooperativity in protein folding. Curr. Opin. Struct. Biol. 6, 31-42. (12) Mu¨ller, C. W., and Schulz, G. E. (1992) Structure of the complex between adenylate kinase from Escherichia coli and the inhibitor AP5A refined at 1.9 Å resolution. A model for a catalytic transition state. J. Mol. Biol. 224, 159-177. (13) Navon, A., Ittah, V., Scheraga, H. A., and Haas, E. (2001) Distributions of intramolecular distances in the reduced and denatured states of bovine pancreatic ribonuclease A. Folding initiation structures in the C-terminal portions of the reduced protein. Biochemistry 40, 105-118. (14) Noda, L. (1973) Adenylate kinase. In The Enzymes (Boyer, P. D., Ed.) Vol. 8, pp 279-305, Academic Press, New York. (15) Ortiz, J. O., and Bubis, J. (2001) Effects of differential sulfhydryl group-specific labeling on the rhodopsin and guanine nucleotide binding activities of transducin. Arch. Biochem. Biophys. 387 (2), 233-234.

Ratner et al. (16) Pennington, M. W. (1994) Site-specific chemical modification procedures. Methods Mol. Biol. 35, 171-185. (17) Ptitsyn, O. B. (1995) Molten globule and protein folding. Adv. Protein Chem. 47, 83-229. (18) Ratner, V., and Haas, E. (1998) An instrument for timeresolved monitoring of fast chemical transitions: application to the kinetics of refolding of a globular protein. Rev. Sci. Instrum. 69, 2147-2154. (19) Ratner, V., Sinev, M., and Haas, E. (2000) Determination of intramolecular distance distributions during protein folding on the millisecond time scale. J. Mol. Biol. 299, 1363-1371. (20) Schultz, G. E. (1991) Domain motions in proteins. Curr. Opin. Struct. Biol. 1, 883-888. (21) Schultz, G. E., Muller, C. W., and Diederichs, K. (1990) Induced-fit movements in adenylate kinases. J. Mol. Biol. 213, 627-630. (22) Sinev, M. A., Sineva, E. V., Ittah, V., and Haas, E. (1996) Domain closure in adenylate kinase. Biochemistry 35, 64256437. (23) Sinev, M., Landsmann, P., Sineva, E., Ittah, V., and Haas, E. (2000) Bioconjugate Chem. 11, 352-362. (24) Stryer, L. (1978) Fluorescence energy transfer as a spectroscopic ruler. Annu. Rev. Biochem. 47, 819-846. (25) Van der Meer, B. W., Raymer, M. A., Wagoner, S. L., Hackney, R. L., Beechem, J. M., and Gratton, E. (1993) Designing matrix models for fluorescence energy transfer between moving donors and acceptors. Biophys. J. 64, 12431263. (26) Van Der Meer, B. W., Coker, G., III, and Chen, S.-Y. (1994) Resonance energy transfer: Theory and data, VCH Publishers, New York.

BC025537B