Top Banner
THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc. Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 Printed in U.S.A. The Quaternary Structure of Streptavidin in Urea* (Received for publication, February 6, 1991) Gary P. KurzbanSf, Edward A. Bayerll, Meir Wilchekq, and Paul M. HorowitzSII From the $Department of Biochemistry, University of Texas Health Science Center at Sun Antonio, Sun Antonio, Texas 78284- 7760 and the TDepartment of Biophysics, Weizmann Institute of Science, Rehouot, 76100 Israel We report on the interactions of urea and guanidi- nium salts with streptavidin. Gel filtration chromatog- raphy in 0, 4, 6, and 7 M urea indicates that the strep- tavidin tetramer remains intact in urea. Biotin alters the electrophoretic mobility of streptavidin whether or not 6 M urea is present. The intrinsic fluorescence of streptavidin is increased and blue-shifted in 6 M urea. The fluorescence changes indicate the absence of un- folding. A conformational response to urea is possible, but much of the fluorescence change is due to urea binding as a weak biotin analog (ICa = 1.3 M-’). The resistance to structural perturbation by urea reflects the structural stability of streptavidin’s anti- parallel &barrel motif. Unfolding is sluggish in 6 M guanidinium hydrochloride (half-time, =50 days). After guanidinium thiocyanate unfolding, streptavidin can be refolded, but the unfolding and refolding tran- sitions are centered at different concentrations of per- turbant. Slow unfolding, with a 15th power depend- ence on guanidinium thiocyanate concentration, may be partially responsible for thenoncoincidence of the unfolding and refolding processes. Nonequilibrium be- havior is also seen in 6 M urea, as native streptavidin does not unfold and guanidinium thiocyanate unfolded streptavidin does not refold. Refolding does occur at lower concentrations of urea. Guanidinium thiocya- nate only slowly unfolds the biotin-streptavidin com- plex. In the presence of biotin, unfolded streptavidin does not refold in 6 M guanidinium thiocyanate or in 6 M urea. Streptavidin is a protein secreted by the soil bacterium Streptomyces auidinii and functions as an antibiotic by reduc- ing the free biotin concentration (Chaiet and Wolf, 1964). Streptavidin is heavily employed in biotechnology (Wilchek and Bayer, 1988, 1989). Its utility derives from three proper- ties: the tight (effectively irreversible) binding of biotin, four biotin-binding sites/protein molecule, and the availability of the carboxylate group of biotin for the synthesis of modified ligands. Streptavidin’s applied value has sparked interest in * This research was supported by Welch Grant AQ723 and Re- search GrantGM25177 from the National Institutes of Health (to P. M. H.) and by the Fund for Basic Research administered by the Israel Academy of Sciences and Humanities and a grant from the United States-Israel Binational Science Foundation, Jerusalem, Israel (to E. A. B. and M. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Robert A. Welch Fellow. (1 To whom correspondence and requests for reprints should be addressed: Dept. of Biochemistry, University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Dr., San Antonio, T X 78284- 7760. Tel.: 512-567-3737. its protein chemistry (Wilchek and Bayer, 1989;Green, 1990). Streptavidin is a tetramer of identical subunits. Each sub- unit is an eight-stranded antiparallel @-barrel (Weber et al., 1989; Hendrickson et al., 1989). The structure of the tetramer suggests that dissociation into stable dimers might be achiev- able. Although the subunits are all in identical positions within the tetramer, their quaternary relationships are dis- tinct. The tetramer looks like a dimer of dimers, with a narrow waistline between dimers. Interactions across the waistline are minimal and involve only one @-strand from each subunit (Weber et al., 1989; Hendrickson et al., 1989). Thus, dissocia- tion along the plane of the waistline is plausible. In the dimers that would be formed, four strands from each subunit would have extensive contacts with the other subunit (Weber et al., 1989; Hendrickson et al., 1989). This would stabilize the dimers against further dissociation to monomers. And, since in the tetramer each subunit has one extensive subunit- subunit interface, it is unlikely that single subunits readily dissociate from the intact tetramer. Thus, we propose a pos- sible pathway of dissociation that includes an obligatory “wa- istline dimer.” tetramer -+ waistline dimers -+ monomers I X I trimer + monomer Three types of dimer are theoretically possible, and in this model the waistline dimers, which are those dimers that retain the extensive subunit-subunit interface, are the only ones to form. Each subunit contains one biotin-binding site and 6 tryp- tophan residues. In the absence of biotin, each binding site is located within a single subunit. Biotin binding confers sub- stantial conformational changes (Weber et al., 1989).Trp-120 is of particular interest, asbiotin binding shifts it from being in the tetramer waistline region into contact with the bound biotin. The movement of Trp-120 may help to seal the biotin site, which is consistent with the slow dissociation of biotin (Green, 1990). Critically, in the biotin-streptavidin complex, Trp-120 is located on a subunit across the tetramer waistline from the biotin site with which it interacts. Thus, with biotin bound (and only with biotin bound), biotin sites involve two subunits. Since thesesubunitsare from distinctpotential waistline dimers, biotin binding would be expected to retard the dissociation of the streptavidin tetramer, if our model for dissociation is correct (see above). Of the other 5 tryptophans, Trp-21 and -75 are never within the biotin-binding sites and Trp-79, -92, and -108 always line the binding sites. Two recent observations (Sano and Cantor, 1990a) prompted our investigation of the interactions of urea with streptavidin. First, they reported that biotin increased the 14470
8

THE OF BIOLOGICAL Vol. 266, No. 22, Issue of August 5, pp ... · Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 ... Samples were applied in 0.5 ml with an of 0.6-1.1. The

Jul 20, 2020

Download

Documents

dariahiddleston
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: THE OF BIOLOGICAL Vol. 266, No. 22, Issue of August 5, pp ... · Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 ... Samples were applied in 0.5 ml with an of 0.6-1.1. The

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 Printed in U.S.A.

The Quaternary Structure of Streptavidin in Urea*

(Received for publication, February 6, 1991)

Gary P. KurzbanSf, Edward A. Bayerll, Meir Wilchekq, and Paul M. HorowitzSII From the $Department of Biochemistry, University of Texas Health Science Center at Sun Antonio, Sun Antonio, Texas 78284- 7760 and the TDepartment of Biophysics, Weizmann Institute of Science, Rehouot, 76100 Israel

We report on the interactions of urea and guanidi- nium salts with streptavidin. Gel filtration chromatog- raphy in 0, 4, 6, and 7 M urea indicates that the strep- tavidin tetramer remains intact in urea. Biotin alters the electrophoretic mobility of streptavidin whether or not 6 M urea is present. The intrinsic fluorescence of streptavidin is increased and blue-shifted in 6 M urea. The fluorescence changes indicate the absence of un- folding. A conformational response to urea is possible, but much of the fluorescence change is due to urea binding as a weak biotin analog (ICa = 1.3 M-’).

The resistance to structural perturbation by urea reflects the structural stability of streptavidin’s anti- parallel &barrel motif. Unfolding is sluggish in 6 M guanidinium hydrochloride (half-time, =50 days). After guanidinium thiocyanate unfolding, streptavidin can be refolded, but the unfolding and refolding tran- sitions are centered at different concentrations of per- turbant. Slow unfolding, with a 15th power depend- ence on guanidinium thiocyanate concentration, may be partially responsible for the noncoincidence of the unfolding and refolding processes. Nonequilibrium be- havior is also seen in 6 M urea, as native streptavidin does not unfold and guanidinium thiocyanate unfolded streptavidin does not refold. Refolding does occur at lower concentrations of urea. Guanidinium thiocya- nate only slowly unfolds the biotin-streptavidin com- plex. In the presence of biotin, unfolded streptavidin does not refold in 6 M guanidinium thiocyanate or in 6 M urea.

Streptavidin is a protein secreted by the soil bacterium Streptomyces auidinii and functions as an antibiotic by reduc- ing the free biotin concentration (Chaiet and Wolf, 1964). Streptavidin is heavily employed in biotechnology (Wilchek and Bayer, 1988, 1989). Its utility derives from three proper- ties: the tight (effectively irreversible) binding of biotin, four biotin-binding sites/protein molecule, and the availability of the carboxylate group of biotin for the synthesis of modified ligands. Streptavidin’s applied value has sparked interest in

* This research was supported by Welch Grant AQ723 and Re- search Grant GM25177 from the National Institutes of Health (to P. M. H.) and by the Fund for Basic Research administered by the Israel Academy of Sciences and Humanities and a grant from the United States-Israel Binational Science Foundation, Jerusalem, Israel (to E. A. B. and M. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Robert A. Welch Fellow. (1 To whom correspondence and requests for reprints should be

addressed: Dept. of Biochemistry, University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Dr., San Antonio, T X 78284- 7760. Tel.: 512-567-3737.

its protein chemistry (Wilchek and Bayer, 1989; Green, 1990). Streptavidin is a tetramer of identical subunits. Each sub-

unit is an eight-stranded antiparallel @-barrel (Weber et al., 1989; Hendrickson et al., 1989). The structure of the tetramer suggests that dissociation into stable dimers might be achiev- able. Although the subunits are all in identical positions within the tetramer, their quaternary relationships are dis- tinct. The tetramer looks like a dimer of dimers, with a narrow waistline between dimers. Interactions across the waistline are minimal and involve only one @-strand from each subunit (Weber et al., 1989; Hendrickson et al., 1989). Thus, dissocia- tion along the plane of the waistline is plausible. In the dimers that would be formed, four strands from each subunit would have extensive contacts with the other subunit (Weber et al., 1989; Hendrickson et al., 1989). This would stabilize the dimers against further dissociation to monomers. And, since in the tetramer each subunit has one extensive subunit- subunit interface, it is unlikely that single subunits readily dissociate from the intact tetramer. Thus, we propose a pos- sible pathway of dissociation that includes an obligatory “wa- istline dimer.”

tetramer -+ waistline dimers -+ monomers

I X

I trimer + monomer

Three types of dimer are theoretically possible, and in this model the waistline dimers, which are those dimers that retain the extensive subunit-subunit interface, are the only ones to form.

Each subunit contains one biotin-binding site and 6 tryp- tophan residues. In the absence of biotin, each binding site is located within a single subunit. Biotin binding confers sub- stantial conformational changes (Weber et al., 1989). Trp-120 is of particular interest, as biotin binding shifts it from being in the tetramer waistline region into contact with the bound biotin. The movement of Trp-120 may help to seal the biotin site, which is consistent with the slow dissociation of biotin (Green, 1990). Critically, in the biotin-streptavidin complex, Trp-120 is located on a subunit across the tetramer waistline from the biotin site with which it interacts. Thus, with biotin bound (and only with biotin bound), biotin sites involve two subunits. Since these subunits are from distinct potential waistline dimers, biotin binding would be expected to retard the dissociation of the streptavidin tetramer, if our model for dissociation is correct (see above). Of the other 5 tryptophans, Trp-21 and -75 are never within the biotin-binding sites and Trp-79, -92, and -108 always line the binding sites.

Two recent observations (Sano and Cantor, 1990a) prompted our investigation of the interactions of urea with streptavidin. First, they reported that biotin increased the

14470

Page 2: THE OF BIOLOGICAL Vol. 266, No. 22, Issue of August 5, pp ... · Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 ... Samples were applied in 0.5 ml with an of 0.6-1.1. The

Streptavidin Structure in Urea 14471

mobility of streptavidin upon native PAGE’ but only if 6 M urea was present. This behavior indicated both that streptav- idin retained, minimally, a large amount of tertiary structure in urea and that a conformational change had occurred. Second, the conformational change was characterized by gel filtration chromatography in 6 M urea and was interpreted as indicating that streptavidin had dissociated to stable dimers (Sano and Cantor, 1990a).

Here, we investigate the interactions of urea and guanidi- nium salts with streptavidin. We find that urea imposes little conformational change upon streptavidin and cannot unfold the protein, and that the tetramer remains intact in 6 M urea. The nonresponsiveness to urea reflects the relative weakness of urea as a chaotropic reagent, the binding of urea as a weak biotin analog, and the unusual stability of streptavidin.

EXPERIMENTAL PROCEDURES

Chemicals-Tris, urea, glycine, TEMED, ammonium persulfate, and N,N’-methylenebisacrylamide were “electrophoresis grade” from Bio-Rad. Electrophoresis grade acrylamide was from Kodak. Electro- phoresis grade guanidinium hydrochloride was from Fisher, and guan- idinium thiocyanate (GdmSCN) was from Sigma.

Streptauidin-Biotin-free streptavidin was purified from the cul- ture media of S. auidinii and proteolyzed to “core streptavidin” as described (Bayer et al., 1986, 1990; Kurzban et al., 1990). Proteolysis removes all nonbarrel residues from both the amino and carboxyl termini. The streptavidin was homogeneous as judged by PAGE under denaturing conditions (Laemmli, 1970). Concentration was assessed by absorbance a t 282 nm (Esubunit = 56 mM” cm” (Green, 1975)) assuming M , 13,273, calculated for residues 13-139, the most com- monly reported core species (Pahler et al., 1987; Bayer et al., 1989).

Streptavidin often had multiple bands upon native PAGE or iso- electric focusing PAGE (not shown). For native PAGE experiments only, heterogeneity was reduced by anion exchange chromatography on a Mono Q column (Pharmacia Fine Chemicals AB, Uppsala, Sweden).’ Streptavidin was applied a t low ionic strength and eluted with a gradient of NaC1.

Gel Filtration Chromatography-Sephacryl S-300 SF (superfine, Pharmacia) was poured according to the manufacturer’s instructions to make a 1.0 X 35-cm bed. Chromatography was performed at room temperature at several concentrations of urea a t a flow rate of 7.7 ml/h. All column buffers were 25 mM in Tris, 80 mM in glycine, and 200 mM in NaC1, pH 8.6. When urea was present, methylamine HCl was added to 20 mM, the pH was readjusted to 8.6 with a few drops of HC1, and the solution was employed within 1 day. Blue dextran 2000 (Pharmacia) eluted in a broad peak on the S-300 column. Therefore, the first fractions containing blue dextran were collected and subsequently used to measure the void volume.

Standard proteins (Boehringer Mannheim GmbH, Mannheim, Federal Republic of Germany) were incubated in each urea-containing buffer for 2-4 h a t room temperature prior to chromatography. Streptavidin was dialyzed into the appropriate buffer overnight a t room temperature. For chromatography of the streptavidin-biotin complex, the column buffer was made 1 p~ in biotin, and biotin was added in 4-fold excess over streptavidin prior to dialysis. For all samples, urea was added primarily by dilution into column buffer. The small dilution of the urea by the sample was compensated for by adding solid urea, using an equation of Kawahara and Tanford (1966), such that the final concentration of urea matched that of the column buffer. Samples were applied in 0.5 ml with an of 0.6-1.1.

The column was monitored at 275 nm with an LKB 2138 Uvicord S flow-through detector. Elution positions were expressed as Kav, according to the following equation.

K a v = ( V - Vvord)/(Vbed - Vuod

where V is the elution volume of the sample, Vvoid is the elution volume of blue dextran, and Vkd is the total bed volume of the column. A K., of 0 corresponds to elution in the void volume. For

I The abbreviations used are: PAGE, polyacrylamide gel electro- phoresis; GdmSCN, guanidinium thiocyanate; HABA, 4-(hydroxy- azobenzene)-2’-benzoic acid; TEMED, N,N,N’,N’-tetramethyleth- ylenediamine.

G . P. Kurzban, unpublished data.

well behaved proteins, a larger value of K., corresponds to a smaller Stokes radius.

Binding of Urea to Streptauidin-We described previously an assay for the weak binding of compounds to streptavidin (Kurzban et al., 1990), based upon the ability to displace a visible wavelength chro- mophore, HABA, from the protein (Green, 1970). In this dual beam spectrophotometric assay, one cuvette contains streptavidin, whereas the other does not. HABA is at identical concentrations in both. Urea is added to both cuvettes, and the displacement of HABA is monitored a t 500 nm.

The data were analyzed as described previously (Kurzban et al., 1990). HABA binding was fit by a nonlinear least squares procedure (Minsq; MicroMath Scientific Software, Salt Lake City, UT) to the following equation.

A ~ O O = Amax([HABAI/([HABAl + Kd))

whereA5,,,, is the corrected absorbance at 500 nm, Amax is the maximum possible level of binding, and [HABA] is the calculated concentration of free HABA. At 0 M urea, the data were fit to Amax = 0.346. This value was 7% lower than that expected for the concentration of streptavidin and the reported extinction coefficient of the HABA- streptavidin complex (Green, 1970). In the presence of urea, satura- tion with HABA was not achieved, and A,., was assumed to be 0.346.

Fluorescence Spectroscopy and Stability Studies-Fluorescence was monitored with an SPF-500C spectrofluorometer (SLM Instruments, Inc., Urbana, IL) at 22 “C, essentially as described (Kurzban et al., 1990). Excitation was a t either 295 or 285 nm; the latter was employed with relatively low concentrations of streptavidin to increase the sensitivity and to shift the Raman peak of the solvent farther away from the peak of the intrinsic fluorescence. Band passes were gener- ally 2.5 and 7.5 nm for excitation and emission, respectively, Emission intensities were measured either directly a t 390 nm (20 nm emission bandwidth) or by averaging the emissions from emission spectra between 380 and 400 nm. The wavelength maxima of emission spectra (corrected for buffer blanks and instrument response factors) were determined by smoothing spectra and then determining the wave- length a t which the first derivative was 0.

The fraction of the streptavidin population that had unfolded was estimated by assuming the presence of only native and unfolded protein. Using the spectra shown in Fig. 4, we constructed a simula- tion of the dependence of the fluorescence intensity and the wave- length maximum upon the fraction of the population that was un- folded. Changes in the fluorescence intensity were strictly linear to the amount of unfolding; intensities were employed for shorter time courses at a single set of conditions. Changes in the wavelength maxima were sufficiently close to being linear to the amount of unfolding to be employed as a measure of unfolding when comparing different concentrations of perturbants; the deviation from linearity is minimal, except at very low and very high fractions of folding. Wavelength maxima were also employed for very slow processes, and a small correction was applied, based upon the spectra in Fig. 4.

Buffer A was 25 mM Tris, 80 mM glycine, pH 8.6. For the perturb- ants, final buffer concentrations were those of Buffer A, and the pH was readjusted to 8.6 with HCl. The isoelectric point of streptavidin, in our hands, is approximately 7.5,” in agreement with a published report (Dittmer et al., 1989). Upon saturation with biotin, the PI of streptavidin is reduced to about 6.1.3 At the pH of Buffer A, both streptavidin and the streptavidin-biotin complex have a net negative charge.

Unfolded streptavidin in 6 M urea was prepared by incubating 1 ml of 10 p M streptavidin in 6 M GdmSCN for 1 h and then dialyzing overnight against 6 M urea (containing 20 mM methylamine HC1 and otherwise as per the preceding paragraph) a t room temperature in five wells of a GIBCO/BRL microdialyzer (6-8-kDa cutoff dialysis membrane), while pumping one change of urea thru the system every 40 min. The residual concentration of thiocyanate was assessed by diluting an aliquot to 1.5 ml with water and adding 1.5 ml of acidic ferric nitrate and measuring the absorbance a t 460 nm of the ferric- thiocyanate complex (Sorbo, 1953) against a standard curve of GdmSCN.

Polyacrylamide Gel Electrophoresis-Gels were prepared in a mini- gel format (Hoefer Scientific Instruments, San Francisco, CA; 8 X 6 X 0.15 cm) with 8% acrylamide in the resolving gels, 3.5% acrylamide in the spacer gels, and a 37.5:l ratio of acrylamide to N,N”methyl- enebisacrylamide. Gels were polymerized with TEMED and ammo-

’ G. P. Kurzban, unpublished data.

Page 3: THE OF BIOLOGICAL Vol. 266, No. 22, Issue of August 5, pp ... · Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 ... Samples were applied in 0.5 ml with an of 0.6-1.1. The

14472 Streptavidin St ruc ture in Urea

nium persulfate. One buffer system employed Buffer A as the gel and electrode buffer (Sano and Cantor, 1990a). The other system em- ployed the discontinuous buffers of Davis (1964). For both systems, gels were polymerized in the presence and absence of 6 M urea. Gels were stained with 0.05% Coomassie Blue R-250, 25% isopropanol, 10% acetic acid.

RESULTS

Streptauidin Is a Tetramer in Urea-Gel Filtration Chroma- tography-In the absence of urea, streptavidin (with or with- out biotin bound), blue dextran 2000, and each of the stand- ards eluted from Sephacryl S-300 as a single, narrow peak (Table I ) . Streptavidin co-eluted with ovalbumin ( M I 45,000). All six proteins eluted within 0.02 K,, of the values expected for globular proteins (Pharmacia, 1982). The data are con- sistent with streptavidin being a tetramer ( M r 53,000) in the native buffer employed in this study.

The quaternary structure of streptavidin in urea was inves- tigated by performing gel filtration chromatography a t several concentrations of urea. The chromatograms at the highest urea concentration (7 M ) are shown in Fig. 1. Several features are noteworthy. First, as the concentration of urea was in- creased, all of the proteins eluted earlier (Table I). However, the bed volume remained constant and the void volume marker always eluted at 0.36 bed volumes. Second, the stand- ards were all homogeneous in the absence of urea (single, narrow peak), but became heterogeneous at some concentra- tion of urea. Third, streptavidin remained homogeneous throughout, with a single peak of invariant, narrow width. Fourth, biotin never altered the elution position or peak shape of streptavidin. Fifth, streptavidin always had nearly the same K., as the last ovalbumin species to elute.

A detailed analysis, indicating that streptavidin is a tetra- mer at all concentrations of urea, is presented under “Discus- sion.”

Urea Binds to, but Does Not Unfold Streptauidin-The intrinsic fluorescence of streptavidin (6 tryptophans/subunit) was examined so as to detect urea-streptavidin interactions. Streptavidin was diluted into 6 M urea at room temperature

TABLE I K,, of four standards and streptavidin at several concentrations of

urea-gel filtration on Sephacryl S-300 Concentration of urea

Protein O M 4 M 6 M 7 M 6 M ”

Bovine serum albumin 0.42 0.34 0.26 0.20 0.24 (MI 68,000) shb 0.15 sh 0.10 sh

0.08 sh Voidb Ovalbumin ( M , 45,000) 0.47 0.39 0.41 0.40’ 0.33

0.27 sh 0.20’ Void

Chymotrypsinogen A 0.57 0.51 0.49 0.37 0.63

Cytochrome c 0.66 0.59 0.56 0.52 0.71

Streptavidin (no biotin) 0.48 0.41 0.40 0.39 0.465 Streptavidin (plus bio- 0.47 0.41 0.41 0.40 0.465

(M, 25,000) sh 0.39 sh sh

( M , 12,500) Wider“

tin) ‘Data were extracted from Fig. 4 of Sano and Cantor (1990a).

Buffer was 6 M in urea, 25 mM in Tris, 80 mM in glycine, pH 8.6, and contained neither NaCl nor methylamine HCI. For the standards, data on peak width, shoulders, and number are not available. For streptavidin, only the column profile in the absence of streptavidin was presented.

sh, shoulder of lower absorbance than the major peak. Peak position of some shoulders could not be determined accurately, and in these cases, no peak position is given. The term wider refers to a peak wider than at lower concentrations of urea; void refers to a peak eluting in the void volume.

’ These two peaks were of equal height.

E 50

u) 40 c

b

4 30 (v

20 .- $ - = 10

n

0)

- 10 15 20 25

50 6o : 40 .

30 -

20 - 10 i

I I

~ I ” ” . . , , , ~ \ . .

0 .’. . . ‘ . . . . 10 15 20 25

Elution Volume, ml FIG. 1. Gel filtration on Sephacryl S-300 in 7 M urea. Top

panel, blue dextran 2000 (- - -), ovalbumin (. . . .), biotin-free strep- tavidin (-). Bottom panel, bovine serum albumin (- - -), chymo- trypsinogen A ( . . . .), cytochrome c (-). The absorbance at 275 nm is in arbitrary units for clarity. Additional information is under “Experimental Procedures” and in Table I.

in Buffer A. Immediately after the dilution, the intrinsic fluorescence was recorded (data not shown). The quantum yield (310-400 nm) was 22% higher than in Buffer A, without urea. The wavelength maximum was shifted in 6 M urea by 2 nm toward the blue. When the same experiment was per- formed in the presence of an excess of biotin, urea caused a 12% increase in quantum yield with no change in the wave- length maximum (data not shown).

The changes in fluorescence seen immediately upon dilu- tion into urea do not correspond to complete unfolding. Un- folding would be expected to red-shift the emissions substan- tially (see below). To establish whether 6 M urea can unfold streptavidin (no biotin), the intrinsic fluorescence was moni- tored at longer times. The emission maximum was unchanged through 66 days. Thus, 6 M urea does not unfold streptavidin at room temperature.

Urea reduces the binding to streptavidin of a visible wave- length chromophore, HABA (Fig. 2, top). The apparent dis- sociation constant of HABA at each of several concentrations of urea was estimated, with the assumption that the maximum level of HABA binding was unchanged by the addition of urea (Fig. 2, bottom). The upward curvature in Fig. 2, bottom, suggests that HABA and urea may not bind in a strictly competitive fashion. The apparent Kd of HABA was doubled at 0.78 M urea. Thus, with some reservations, the avidity of urea binding can be estimated as K, = 1.3 ”I.

The question of whether urea dissociates HABA by binding to a shared site or by altering the structure of streptavidin is considered under “Discussion.”

Biotin Increases Mobility of Streptavidin during Electropho- resis in Both the Presence and Absence of Urea-PAGE of streptavidin is shown in Fig. 3. Two buffer systems were

Page 4: THE OF BIOLOGICAL Vol. 266, No. 22, Issue of August 5, pp ... · Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 ... Samples were applied in 0.5 ml with an of 0.6-1.1. The

Streptavidin Structure in Urea 14473

E O" C 0 0 0 0.2 W 0 Z Q

0.1

I

n ' specifically the lack of unfolding by urea, was a general property of streptavidin. Urea is a relatively weak denaturant (von Hippel and Wong, 1964; Tanford, 1968; Greene and Pace, 1974). Studies were therefore carried out in stronger denaturants, guanidinium hydrochloride and guanidinium thiocyanate (GdmSCN).

At high concentrations of GdmSCN, the wavelength of the maximum intrinsic fluorescence of streptavidin was shifted from 335 to 348 nm (Fig. 4). This red shift is consistent with the complete unfolding of streptavidin, accompanied by ex- posure to tryptophan residues to the solvent. It is noted that lesser structural changes might or might not be accompanied by changes in fluorescence. In guanidinium hydrochloride, unfolding was quite slow, with a half-time of about 50 days (data not shown).

Unfolded streptavidin can spontaneously refold. The pro- tein was unfolded in 6 M GdmSCN and then dialyzed over- night into Buffer A. The intrinsic fluorescence returned to that of native (never unfolded) protein, as judged by the wavelength maximum, bandwidth, and quantum yield. Titra- tion of the intrinsic fluorescence of the refolded streptavidin with biotin (Kurzban et al., 1990) indicated that about two- thirds of the biotin-binding sites had been regained. Similarly, the blue shift and narrowing of the spectral bandwidth of the intrinsic fluorescence caused by biotin binding were about two-thirds of that seen with a control sample of native strep- tavidin (data not shown). These results suggest that two- thirds of the streptavidin unfolded and refolded reversibly. For the other one-third of the streptavidin, the biotin sites became incompetent as a result of the unfolding and refolding cycle, although a high degree of tertiary structure (as intrinsic fluorescence) was regained.

Streptavidin resists unfolding by moderate concentrations of GdmSCN. After a 5-day incubation, an unfolding transition was centered a t about 1.3 M denaturant, and after 39 days, it was below 1 M GdmSCN (Fig. 5).

Equilibrium folding/unfolding at intermediate concentra- tions of GdmSCN was not achieved. Streptavidin was dena- tured, diluted to various concentrations of the denaturant, and then incubated for 6 or 40 days. Refolding was centered

FIG. 2. Urea binds to streptavidin. Binding was indirectly as- sessed by the ability of urea to displace HARA from streptavidin. Top, concentrations of urea were 0 M (o), 0.3 M (a), 1 M (01, 2 M (O), and 3 M (W). In the bottom panel, apparent dissociation constants of' HARA were calculated from the data in the top panel, assuming a fixed maximum level of binding of HABA (see "Experimental Pro- cedures").

A B C D

"_..I .... .-

FIG. 3. Polyacrylamide gel electrophoresis of streptavidin in the presence and absence of 6 M urea. A, Buffer A, no urea; H , Buffer A, 6 M urea; C, discontinuous system of Davis (1964), no urea; 11, Davis system, 6 M urea. For each gel, the left lane contains streptavidin in the absence of biotin and the rixht lane has an identical amount of streptavidin, which was made 5:l in hiotinstreptavidin subunits immediately prior to electrophoresis. Urea was present only within the polyacrylamide matrix. The anode is at the bottom throughout.

employed, and each was run in 0 and 6 M urea. For all four, the streptavidin-biotin complex electrophoresed faster than the biotin-free streptavidin. PAGE similar to that in panels R and C were previously reported (Sano and Cantor, 1990a). They reported faster migration for the biotin-streptavidin complex in 6 M urea (as per panel R ) but no effect by biotin in the absence of urea (in contrast to panel C).

Resistance to Perturbants Is a General Property of Streptau- idin-We wondered whether the lack of perturbation, and

" . . -I._. 310 EMISSION WAVELENGTH, nm 400

FIG. 4. Intrinsic fluorescence of native and unfolded strep- tavidin in guanidinium thiocyanate. The spectrum with the greater peak intensity shnws the intrinsic fluorescence of streptavidin (1.25 FM in subunits) within 1 min after dilution into 3.5 M GdmSCN. Unfolding was sufficiently slow such that only folded streptavidin was present. The spectrum of lower peak intensity was the same sample, taken 6 h later. Based on the unfolding rate (Fig. 6) and emission spectra in 6 M GdmSCN, this is a spectrum of unfolded streptavidin. Excitation was at 296 nm. A buffer hlank has heen subtracted, and the net spectra have been corrected fnr instrument response factors but not smoothed.

Page 5: THE OF BIOLOGICAL Vol. 266, No. 22, Issue of August 5, pp ... · Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 ... Samples were applied in 0.5 ml with an of 0.6-1.1. The

14474 Streptavidin St ruc ture in Urea

[GdmSCN], M

FIG. 5 . Unfolding and refolding of streptavidin in guani- dine thiocyanate. Unfolding: streptavidin (0.1 PM in subunits) was incubated for 5 (0) or 39 (0) days at the indicated concentrations of GdmSCN. Refolding: streptavidin was unfolded in 6 M GdmSCN and then diluted to 0.1 pM in subunits at the indicated concentrations of GdmSCN. Data a t 6 days have been omitted for clarity. A, 40 days of refolding.

-6 0 1 2 3 4 5

[GdmSCN], M FIG. 6. Rate of unfolding of streptavidin in guanidinium

thiocyanate. At the four highest concentrations of GdmSCN, fluo- rescence intensities were measured, and the unfolding rates were extracted from a semilog plot. For 1.0 M (39 days) and 1.25 M ( 5 days) GdmSCN, the wavelength of maximum fluorescence of the native and unfolded states was taken to be that seen on that day in 0 M and 3 M GdmSCN, respectively (see Fig. 5). The apparent fraction of folded molecules was then calculated assuming a linear relationship between the change in wavelength maximum and the amount of unfolding and was then corrected, based upon the spectra shown in Fig. 4. At these two concentrations of denaturant, data were available only a t a single time point. The rate of unfolding was therefore calculated with the assumptions that unfolding was first-order and that substantial refolding had not occurred.

at 100 mM GdmSCN a t 6 days, with a relatively broad transition starting below 40 mM GdmSCN (data not shown). After 40 days, the refolding transition was centered at about 125 mM GdmSCN, started above 80 mM GdmSCN, and was sharper than at 6 days (Fig. 5).

The lack of coincidence of the unfolding and folding curves indicates that at least one of the processes is not portraying the equilibrium response of streptavidin to guanidine thiocy- anate. Extremely slow folding and/or refolding could explain the lack of equilibria.

Unfolding kinetics were determined a t several concentra- tions of guanidine thiocyanate (Fig. 6). The rate of unfolding was found to obey the following equation.

log kunlold = -5.821 + 1.176[GdmSCN]

where kunfold is the rate of unfolding in units of reciprocal minutes. The slope of the plot indicates that the rate of unfolding is 15th power in GdmSCN. By extrapolation, at low concentrations of GdmSCN the rate of unfolding will be exceedingly slow, and in the absence of GdmSCN, the half- time would be 10 months.

A refolding experiment was conducted in urea. Streptavidin was unfolded in 6 M guanidine thiocyanate and then dialyzed overnight into 6 M urea until the thiocyanate concentration was below 0.3 mM. The fluorescence wavelength maximum

was at 348 nm, indicating that streptavidin remained unfolded in 6 M urea. Refolding did not occur in 6 M urea after 2 days, but did occur a t lower concentrations of urea, with a refolding transition centered at about 1.35 M urea (Fig. 7).

The behavior at high concentrations of urea is another case of nonequilibrium behavior. In 6 M urea, streptavidin does not unfold, but neither does unfolded streptavidin refold.

Biotin Prevents the Unfolding but Does Not Induce Refold- ing of Streptauidin-Biotin was added to streptavidin to a 4:l ratio of biotin to biotin sites, and the sample was immediately made 6 M in GdmSCN or 6 M in guanidinium hydrochloride. The intrinsic fluorescence wavelength maximum was then monitored uersus time. In guanidinium hydrochloride, no change was seen through 71 days. In GdmSCN, no significant change was seen through 21 days, but after 125 days, there was a 4-nm red shift (data not shown). Presuming that this corresponds only to the accumulation of completely unfolded streptavidin, then about 30% of the protein molecules had unfolded.

Unfolded streptavidin, in either 6 M GdmSCN or after dialysis into 6 M urea, did not refold after adding an excess of biotin (through 31 or 5 days, respectively).

DISCUSSION

Quaternary Structure of Streptavidin in Urea-In order to understand the structural dynamics of streptavidin, its qua- ternary structure will need to be dissected. The observed resistance of streptavidin to unfolding does not foreclose the possibility of imposing lesser structural changes, including changes at the quaternary level. In particular, the minimal interactions at the waistline of the tetramer and the extensive subunit-subunit interface within the putative dimers, suggest that dissociation to a pair of stable dimers may be feasible.

The gel filtration experiments reported here provide infor- mation that bears directly upon the quaternary structure of streptavidin. The x-ray structure (Hendrickson et al., 1989; Weber et al., 1989), equilibrium sedimentation (Chaiet et al., 1964), and previous gel filtration experiments (Bayer et al., 1989; Sano and Cantor, 1990b) all indicate that streptavidin is a tetramer under native conditions. The gel filtration re- ported here, in the absence of urea, also indicates a tetrameric structure in nonperturbing conditions.

Gel filtration chromatography reports the molecular size of a given protein, rather than the molecular weight. For a stable protein, a methodology that employs less stable proteins as standards, at concentrations of perturbant where the struc- ture of the standards is unknown, calls for careful interpre- tation. Indeed, the standard proteins all showed signs of being perturbed by urea, with some combination of a shoulder on the leading edge of the main peak, multiple peaks, a peak in

E 345 i 343

.E 337

‘ E 341 E 339

.!? 335 333

E

C

0 1 2 3 4 5 6

[UREA], M FIG. 7. Refolding of streptavidin in urea. Unfolded streptav-

idin was prepared in 6 M urea as described under “Experimental Procedures” and then diluted to 0.5 M in subunits at the indicated concentrations of urea. The intrinsic fluorescence was assessed with excitation at 285 nm 2 days after the dilution.

Page 6: THE OF BIOLOGICAL Vol. 266, No. 22, Issue of August 5, pp ... · Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 ... Samples were applied in 0.5 ml with an of 0.6-1.1. The

Streptauidin Structure in Urea 14475

the void volume, or a widened single peak. The Stokes radius of single-subunit, globular proteins increases substantially upon unfolding (Fish et al., 1970; Tanford et al., 1974). Aggre- gation of bovine serum albumin and ovalbumin upon unfold- ing has been reported and may include covalent species in- volving disulfide bonds (Tanford, 1968). The peaks of largest K,, can thus be assigned to the most native species, and the earlier peaks, to unfolded and aggregated species.

As the concentration of urea in the column buffer was increased, the last (most native) peak of all of the proteins eluted earlier. Gel filtration media are known to interact with solutes by mechanisms other than passive gel filtration. Ion exchange, hydrophobic or aromatic interactions, and ionic gating of pores have been reported (Khym, 1976; Belew et al., 1978; Pharmacia, 1982; Washabaugh and Collins, 1986). The exact interactions of urea with Sephacryl S-300, resulting in the observed earlier elution of the proteins, cannot be stated. However, a detailed study of the effects of chaotropes on gel filtration chromatography indicates that the chaotropes ad- sorb to the gel filtration media and also increase the hydro- dynamic radii of small solutes (Washabaugh and Collins, 1986). Similar mechanisms may have been operative during the gel filtration studies reported here. The changes in elution position of the proteins are also reminiscent of the changes in protein mobility, apart from the effects of unfolding, seen in PAGE experiments in which the polyacrylamide gel con- tains a gradient of urea (i.e. tilted base lines) (Creighton, 1979).

The coelution of streptavidin at 0, 4, 6, and 7 M urea with the most native ovalbumin species (Mr 45,000) is consistent with streptavidin retaining the same quaternary and tertiary structure in all concentrations of urea. Further, streptavidin eluted as a single, narrow peak at all urea concentrations, consistent with streptavidin being homogeneous (with regard to Stokes radius) during each experiment. Had there been a change in quaternary structure, a transition region would be anticipated in which the now heterogeneous streptavidin would be expected to elute in multiple or broadened peaks. The observed single, narrow peaks are thus consistent with a lack of quaternary changes in streptavidin. We can rule out the possibility that, in urea, one is observing unfolded mono- mers (whose Stokes radii would be very similar to that of the native tetramer (Fish et al., 1970; Tanford et al., 1974)), since unfolding was not observed in 6 M urea in the fluorescence experiments. Since it is established that streptavidin is a tetramer in the absence of perturbants (see above), the gel filtration data are best interpreted as indicating that strep- tavidin is a tetramer under all conditions employed.

Sano and Cantor (1990a) reached a different conclusion from a gel filtration experiment in 6 M urea. Using the same gel filtration media as here, streptavidin eluted at an inter- mediate position between ovalbumin (Mr 45,000) and chy- motrypsinogen A (Mr 25,000) (data shown in the last column of Table I). This result was interpreted as indicating the dissociation of the streptavidin tetramer (Mr 53,000) to stable dimers (MI 26,500).

We reanalyze their data as follows. Bovine serum albumin eluted at nearly the same position in 6 M urea in both laboratories. Ovalbumin eluted at an intermediate position between the two peaks seen in our experiments. This may reflect the slow unfolding kinetics of ovalbumin (Creighton, 1979), particularly near room temperature (e.g. a 1-h half- time in 7.3 M urea) (Simpson and Kauzmann, 1953).

Chymotrypsinogen A and cytochrome c eluted far later in the experiment of Sano and Cantor (1990a) than reported here. Sephacryls are cation exchangers above pH 8 (Belew et

al., 1978). Chymotrypsinogen A and cytochrome c have iso- electric points of 9.5 and 10.6, respectively (Lehninger, 1975), and will be cations at pH 8.6. At the low ionic strength employed by Sano and Cantor (1990a) (Buffer A, pH 8.6, ~0 .006 M in Tris+/glycinate-), ionic interactions of chymo- trypsinogen A and cytochrome c with the Sephacryl S-300 could be substantial, resulting in these two standards eluting later than warranted solely by their Stokes radii. The higher ionic strength employed in the present study (0.22 M) appears to have been adequate to prevent ionic interactions. Bovine serum albumin (Lehninger, 1975), ovalbumin (Lehninger, 1975), and streptavidin (with or without biotin) (Dittmer et al., 1989; Sano and Cantor, 1990a; this report, Fig. 3) all have net negative charge at pH 8.6, and ionic interactions with the Sephacryl should not be a problem.

In summary, the gel filtration study reported here indicates a tetrameric structure for streptavidin under all conditions of gel filtration. The gel filtration experiment of Sano and Can- tor (1990a) reported that streptavidin eluted in 6 M urea between ovalbumin and chymotrypsinogen A. This result should not be interpreted as evidence for the presence of the streptavidin dimer, since the chromatography of the flanking standards was affected by nongel filtration processes, thereby reducing their value as standards.

Urea Binding within Biotin Sites and the Conformational Consequences of 6 M Urea-Urea did not unfold streptavidin. However, there were changes in the intrinsic fluorescence, both in the absence (22% increase in intensity and a 2-nm blue shift) and presence (12% increase, no blue shift) of biotin. Possible causes for these changes are: (a) changes in the hydration shell around streptavidin, ( b ) specific interactions of urea with the 6 tryptophans of each subunit, or (c) confor- mational changes.

In the absence of biotin, urea likely alters the fluorescence of tryptophan residues that line the biotin-binding sites, doing so by being in physical proximity to these residues. That is, the second mechanism described above holds, especially for binding-site tryptophans. The affects of urea upon these residues would correspond to much of the difference between the changes imparted in the absence and presence of biotin (i.e. ~ 1 0 % intensity increase and the 2-nm blue shift). For urea to have directly altered the fluorescence of binding-site tryptophans, four conditions must be met. First, there must be tryptophans within the binding site. Second, urea must have access to the binding site. Third, urea within the binding site must be able to come sufficiently close to the relevant tryptophans to be able to alter their fluorescence. And, fourth, it must be clear that urea, at the concentrations likely to be seen locally, can alter the fluorescence of tryptophans in the directions observed. We consider these conditions below.

First, are there tryptophans within the biotin-binding sites? The x-ray studies place 3 tryptophans (Trp-79, -92, and -108) within the binding site in the absence of biotin (Weber et al., 1989) and 4 tryptophans (Trp-79, -92, -108, and -120) in the binding site once biotin is bound (Hendrickson et al., 1989; Weber et al., 1989).

The second condition is whether urea has access to the binding site. Urea dissociates HABA from streptavidin (Fig. 2). The data can be interpreted as a somewhat competitive binding of urea and HABA or, alternatively, as being due to a conformational mechanism. There is evidence in favor of competitive binding. HABA is considered to be a biotin- binding site probe. One molecule of HABA is displaced upon binding one molecule of biotin (Green, 1964; Green, 1970), consistent with HABA binding within biotin sites. HABA also bears structural resemblance to biotin. At one end is a hydro-

Page 7: THE OF BIOLOGICAL Vol. 266, No. 22, Issue of August 5, pp ... · Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 ... Samples were applied in 0.5 ml with an of 0.6-1.1. The

14476 Streptavidin Structure in Urea

gen bond donor (a hydroxyl), the middle of the molecule is hydrophobic, and the other end is a carboxylate. These prop- erties match those of biotin. The shared ability of urea and biotin to dissociate HABA support a model in which urea and HABA bind competitively to streptavidin. Moreover, this model places the bound urea within the biotin-binding sites.

Previous studies of the interactions of acrylamide with streptavidin (Kurzban et al., 1990) provide additional evidence that urea binds within the biotin-binding site. Acrylamide was shown to bind to streptavidin, with an avidity ( K , = 5 "I) similar to that of urea. Binding was competitive with HABA. Acrylamide quenched the fluorescence of streptavidin by a static mechanism, as might be expected for a bound quencher. Acrylamide quenching was largely prevented by biotin binding, which is consistent with acrylamide binding within biotin-binding sites.

The structural similarity between acrylamide and urea ar- gue for a shared ability to enter and bind in the biotin site, so long as biotin is absent. Indeed, even more than acrylamide, urea bears a structural resemblance to biotin. Urea resembles the ureido portion of the biotin structure, which is critical to biotin binding. The x-ray structures show that both nitrogen atoms of biotin are hydrogen bond donors and that the ureido oxygen atom is a potent hydrogen bond acceptor (Hendrick- son et al., 1989; Weber et al., 1989).

Experiments with hen egg white avidin, which has struc- tural similarities to streptavidin (Argaraiia et al., 1986; Wil- chek and Bayer, 1989; Green, 1990), provide a consistent picture of urea being a biotin-binding site analog. For avidin, a continuum of biotin analogs was constructed that indicated that urea binds as a weak biotin analog (Green, 1963, 1975). Moreover, urea interactions are of the same avidity whether assessed by UV difference spectroscopy or by the ability to dissociate HABA,* which is consistent with HABA function- ing as a biotin-binding site probe. As with streptavidin, ac- rylamide quenching of avidin is largely prevented upon biotin binding (Kurzban et al., 1989).

The weight of the evidence is that urea binds to streptavidin within the biotin site. Since the association constant was estimated as -1.3 M-', it is clear that in 6 M urea the binding site would have a high local concentration of urea.

The third condition necessary for urea binding to alter the intrinsic fluorescence is for binding site urea to be sufficiently close to tryptophans to alter their fluorescence. Our acryl- amide studies (Kurzban et al., 1990) indicate that this is the case (see above), as does the accessibility of Trp-92 and -108 (within the biotin-binding site) to chemical modification (Git- lin et al., 1988).

Lastly, urea has the ability to alter tryptophan fluorescence in the directions observed with streptavidin. In 6 M urea, the fluorescence of the tryptophan model compound, N-acetyl- tryptophanamide, is increased 47% and is slightly (0.5 nm) blue-shifted (data not shown).

From these arguments, we conclude that ( a ) urea binds to streptavidin, with an association constant of approximately 1.3 M-', and ( 6 ) urea binds within the biotin-binding sites, where ( c ) urea blue shifts and increases the fluorescence intensity of some of the 3 binding site tryptophans.

A previous report indicated that urea enables biotin to confer changes in the structure of streptavidin such that the mobility in native PAGE is increased (Sano and Cantor, 1990a). This would clearly indicate a conformational change being induced by urea. However, our data are in direct con- trast, as biotin increased the electrophoretic mobility of strep- tavidin whether or not 6 M urea was present (Fig. 3). The

G. P. Kurzban, unpublished observations.

reason for the discrepancy in the data obtained between the two laboratories is not clear.

The PAGE experiment reported here shows that the bind- ing of biotin makes streptavidin migrate faster, whether or not urea was present. The simplest explanation for this is that biotin increased the net negative charge of the protein by adding its own negative charge to that of the protein. It still remains possible that the conformational changes im- posed by biotin binding also alter the net charge of streptav- idin.

The inability of urea to change the effect of biotin binding upon the electrophoretic mobility of streptavidin neither sup- ports nor eliminates the possibility of a conformational effect of urea upon streptavidin. The ability of biotin to increase mobility in PAGE in the presence of urea implies that strep- tavidin retains its tertiary structure (as biotin binding) in urea, which is consistent with the gel filtration and fluores- cence data.

In summary, the fluorescence changes imposed by urea are at least largely due to urea binding weakly to streptavidin in the biotin-binding sites, where it perturbs the 3 binding site tryptophans. It is not necessary to invoke a conformational change to explain the fluorescence changes imposed by urea, although a conformational change cannot be excluded. If urea does impose a conformational change, the change does not include dissociation of the tetramer.

Stability of Streptauidin and Stabilization by Biotin-Strep- tavidin is a potentially exciting system for the study of protein folding, assembly, and ligand interactions. With the proteo- lyzed core streptavidin employed here, all nonbarrel residues at both the amino and carboxyl termini have been removed. The resulting all antiparallel &barrel structure allows exper- imental information to be related to a single structural motif. The x-ray structure has been reported in both the absence (Weber et al., 1989) and presence (Hendrickson et al., 1989; Weber et al., 1989) of biotin, and the streptavidin gene has been cloned (Argaraiia et al., 1986; Sano and Cantor, 1990a) and expressed in Escherichia coli (Sano and Cantor, 1990a), offering the potential for studying site-directed mutants.

Streptavidin is remarkably resistant to unfolding. Streptav- idin does not unfold in 6 M urea and unfolds very slowly in 6 M guanidinium chloride. In contrast, many proteins unfold at about 6 M urea (Creighton, 1979). Streptavidin is more resist- ant to denaturation than is hen egg white avidin5, which shares many physical properties and substantial sequence homology with streptavidin (Argaraiia et al., 1986; Wilchek and Bayer, 1989; Green, 1990). Only in the particularly pow- erful denaturant, guanidinium thiocyanate, did substantial, rapid unfolding of streptavidin occur.

Refolding is often highly successful for single subunit pro- teins of the approximate size of the streptavidin subunit. And, the lack of cysteine residues in streptavidin avoids the for- mation of cross-linked or otherwise oxidized species, which could impair refolding. Nonetheless, only two-thirds of the biotin-binding ability was regained after refolding GdmSCN unfolded streptavidin. No attempt was made to increase the recovery of ligand binding. The nature of the impediment(s1 to a complete recovery of biotin binding cannot be stated, but may reflect the need to properly assemble the tetramer.

Equilibrium behavior at intermediate concentrations of GdmSCN was not observed (Fig. 2). Rather, refolding oc- curred at far lower concentrations of guanidine thiocyanate than did unfolding. The unfolding rate data (Fig. 6) provide a simple explanation for this phenomenon. The rate of un- folding (and presumably also the rate of refolding) was highly

' P. M. Horowitz, unpublished observations.

Page 8: THE OF BIOLOGICAL Vol. 266, No. 22, Issue of August 5, pp ... · Vol. 266, No. 22, Issue of August 5, pp. 14470-14477,1991 ... Samples were applied in 0.5 ml with an of 0.6-1.1. The

Streptavidin Structure in Urea 14477

dependent upon the concentration of the denaturant. Thus, at the concentrations under which the two curves are not coincident, extremely slow kinetics prevented equilibria from being established during the time of observation. The molec- ular details behind the slow kinetics are unknown.

Nonequilibrium behavior was seen in urea as well. While 6 M urea could not unfold streptavidin, neither could the un- folded streptavidin be refolded in 6 M urea, even in the presence of biotin. We do not know, in this case, which process may be rate-limiting, and thus, the equilibrium structure of streptavidin in 6 M urea is unknown. Nonequilibrium folding/ unfolding was also evident for the streptavidin-biotin complex in GdmSCN, as unfolding was quite slow and refolding was absent. Stabilization by ligands is a common phenomenon, and is to be expected as a consequence of the tight binding of biotin.

Previous studies support the present results of the excep- tional stability of streptavidin. Streptavidin is generally iso- lated from culture supernatants after a week or more of growth (Suter et al., 1988; Bayer et al., 1990). After proteolysis of amino- and carboxyl-terminal peptides, core streptavidin re- sists further proteolysis (Pahler et al., 1987; Bayer et al., 1989). Streptavidin is stable against sodium dodecyl sulfate below 60 “C (Bayer et al., 1986, 1990), and biotin binding is retained in sodium dodecyl sulfate (Bayer et al., 1990) and in 6 M urea (Sano and Cantor, 1990a; this report). We have noted diffi- culty in fixing and staining streptavidin on isoelectric focusing gels, specifically due to the presence of ampholytes, and especially for the streptavidin-biotin ~ o m p l e x . ~

The experiments and analysis reported here place the quest for the streptavidin dimer in its infancy. However, the behav- ior of streptavidin during unfolding and refolding offers clues as to where to begin. As discussed, the streptavidin structure suggests the feasibility of producing stable waistline dimers. The stability of streptavidin is consistent, however, with the native structure being “locked in” such that perturbants are resisted. Thus, conditions likely to dissociate the tetramer to dimers within a reasonable amount of time are likely to be excessively harsh, such that the dimers are not stable. In contrast, milder conditions can be employed during refolding. Since biotin binding pulls Trp-120 across the waistline, cre- ating quaternary interactions between potential waistline di- mers, we anticipate that these dimers may form poorly, if at all, in the presence of bound biotin. Thus, a reasoned starting point in looking for dimers is to examine the refolding path- way, in the absence of biotin, for thermodynamic and/or kinetic intermediates.

REFERENCES

Argaraiia, C. E., Kuntz, 1. D., Birken, S., Axel. R., and Cantor, C. R.

Bayer, E. A., Ben-Hur, H., Gitlin, G., and Wilchek, M. (1986) J. (1986) Nucleic Acids Res. 14, 1871-1882

Biochem. Biophys. Methods 13, 103-112

Bayer, E. A., Ben-Hur, H., Hiller, Y., and Wilchek, M. (1989)

Bayer, E. A., Ben-Hur, H., and Wilchek, M. (1990) Methods Enzymol.

Belew, M., Porath, J., Fohlman, J., and Janson, J.-C. (1978) J.

Chaiet, L., and Wolf, F. J. (1964) Arch. Biochem. Biophys. 106, 1-5 Chaiet, L., Miller, T. W., Tausig, F., and Wolf, F. J. (1964) in

Proceedings of the Third Znterscience Conference on Antimicrobial Agents and Chemotherapy, Washington, D. C., October 28-30, 1963 (Sylvester, J . C., ed) pp. 28-32, American Society for Microbiology, Ann Arbor, MI

Biochem. J . 259,369-376

1 8 4 , 8 0 4 9

Chromatogr. 147, 205-212

Creighton, T. E. (1979) J. Mol. Biol. 129, 235-264 Davis, B. J . (1964) Ann. N. Y. Acad. Sci. 121, 404-427 Dittmer, J., Dittmer, A., Bruna, R. D., and Kasche, V. (1989) Electro-

Fish, W. W., Reynolds, J. A., and Tanford, C. (1970) J. Biol. Chem.

Gitlin, G., Bayer, E. A,, and Wilchek, M. (1988) Biochem. J. 256,

Green, N. M. (1963) Biochem J . 89, 599-609 Green, N. M. (1964) Biochem J. 94, 23C-24C Green, N. M. (1970) Methods Enzymol. 18A, 418-424 Green, N. M. (1975) Adv. Protein Chem. 29, 85-133 Green, N. M. (1990) Methods Enzymol. 184, 51-67 Greene, R. F., Jr., and Pace, C. N. (1974) J. B i d . Chem. 249, 5388-

5393 Hendrickson, W. A., Pahler, A., Smith, J. L., Satow, Y., Merritt, E.

A., and Phizackerley, R. P. (1989) Proc. Natl. Acad. Sci. U. S. A. 86,2190-2194

Kawahara, K., and Tanford, C. (1966) J. Bid. Chem. 241,3228-3232 Khym, J . X. (1976) Anal. Biochem. 71,231-242 Kurzban, G. P., Gitlin, G., Bayer, E. A., Wilchek, M., and Horowitz,

P. M. (1989) Biochemistry 28,8537-8542 Kurzban, G. P., Gitlin, G., Bayer, E. A., Wilchek, M., and Horowitz,

P. M. (1990) J. Protein Chem. 9, 673-682 Laemmli, U. K. (1970) Nature 227, 680-685 Lehninger, A. L. (1975) Biochemistry, 2nd Ed., p. 162, Worth Pub-

lishers, Inc., New York Pahler, A., Hendrickson, W. A., Gawinowicz Kolks, M. A., Argaraiia,

C. E., and Cantor, C. R. (1987) J. Bid. Chem. 262, 13933-13937 Pharmacia (1982) Gel Filtration Theory and Practice, Pharmacia Fine

Chemicals AB, Uppsala, Sweden Sano, T., and Cantor, C. R. (1990a) J. Biol. Chem. 265, 3369-3373 Sano, T., and Cantor, C. R. (1990b) Proc. Natl. Acad. Sci. U. S. A.

Simpson, R. B., and Kauzmann, W. (1953) J. Am. Chem. SOC. 75,

Sorbo, B. H. (1953) Acta Chem. Scand. 7, 1129-1136 Suter, M., Cazin, J., Jr., Butler, J . E., and Mock, D. M. (1988) J.

Tanford, C. (1968) Adu. Protein Chem. 23, 121-282 Tanford, C., Nozaki, Y., Reynolds, J. A,, and Makino, S. (1974)

von Hippel, P. H., and Wong, K.-Y. (1964) Science 145,577-580 Washabaugh, M. W., and Collins, K. D. (1986) J. Bid. Chem. 261,

Weber, P. C., Ohlendorf, D. H., Wendoloski, J. J., and Salemme, F.

Wilchek, M., and Bayer, E. A. (1988) Anal. Biochem. 171, 1-32 Wilchek, M., and Bayer, E. A. (1989) Trends Biochem. Sci. 14, 408-

phoresis 10, 762-765

245,5166-5168

279-282

87,142-146

5139-5152

Immunol. Methods 113, 83-91

Biochemistry 13,2369-2376

12477-12485

R. (1989) Science 243, 85-88

412