Spectral enhancement of proteins: Biological incorporation and ...

Proc. Natl. Acad. Sci. USA Vol. 89, pp. 12023-12027, December 1992 Biochemistry

Spectral enhancement of proteins: Biological incorporation and fluorescence characterization of 5-hydroxytryptophan in bacteriophage A ci repressor (protein fluorescence/protein-protein interaction/protein-nucleic acid interaction)

J. B. ALEXANDER ROSS*t, DONALD F. SENEARtf, EVAN WAXMAN*, BAMENGA B. KOMBO*, ELENA RuSINOVA*, YAO TE HUANGt, WILLIAM R. LAWS*, AND C. A. HASSELBACHER* *Department of Biochemistry, Mount Sinai School of Medicine, One Gustave L. Levy Place, New York, NY 10029; and tDepartment of Molecular Biology and Biochemistry, University of California, Irvine, CA 92717

Communicated by Saul Roseman, September 24, 1992

ABSTRACT We have used a tryptophan-requiiing Escherichia coli auxotroph to replace the three tryptophan residues of A cI repressor with 5-hydroxy-L-tryptophan (5-OHTrp). By using a nonleaky promoter, we have achieved >95% replacement of tryptophan in the repressor. We show that the absorbance and fluorescence properties of 5-OHTrp-A cI are clearly distinct from A cI repressor and that the fluorescence of 5-OHTrp-A cI repressor can be observed selectively in the presence of exogenous tryptophan. We also show that the 5-OHTrp-A cI repressor functional properties, as assessed by measurement of binding constants for self-association and for association to operator DNA, and structural properties, as assessed by fluorescence, are indistinguishable from the native repressor. Based on these results, we anticipate that the availability of spectrally enhanced proteins wiil significantly enhance the utility of both fluorescence and phosphorescence spectroscopies to study protein structure and function in complex interacting systems.

Tryptophan fluorescence has been used widely to study conformation, function, dynamics, and intermolecular interactions of proteins (1). However, the presence of at least one and often several tryptophans in most proteins severely limits the utility oftryptophan fluorescence for studying one protein species in the company of others. For this reason, many fluorescence studies involving proteins and polypeptides make use of extrinsic probes, such as dansyl or fluorescein. Use of extrinsic probes, however, is complicated by the difficulty of specific placement and the risk that chemical modification can alter the functional and structural properties of the labeled protein. Clearly, it would be to great advantage to incorporate into a protein a fluorophore that is spectrally distinct from tryptophan while having none of the disadvantages of extrinsic probes. In this way, a spectrally enhanced protein (SEP) would be generated that would still have "intrinsic" fluorescence. To be useful for the generation of SEPs, a candidate tryptophan analogue mlist be readily incorporated by biosynthesis into the protein of interest, must be fluorescent and have spectral properties distinct from those of tryptophan, and must have little or no effect on the functional and structural properties of the protein into which it has been incorporated. An intriguing possibility for creation of SEPs is suggested by NMR studies in which tryptophan auxotrophs of Escherichia coli were used to incorporate fluorotryptophan derivatives into Salmonella typhimurium histidine binding protein J (2) and rat cellular retinol binding protein (3). The fluorotryptophan derivatives, however, are spectrally very similar The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

to tryptophan. Thus, they do not have features that enhance fluorescence studies. By contrast, the absorbance spectra of another group of tryptophan derivatives, the hydroxytryptophans, are sufficiently red-shifted to allow selective excitation in the presence of tryptophan. In addition, several hydroxytryptophan derivatives have high fluorescence quantum yields (4-6). These properties suggest the use of hydroxytryptophan derivatives for the generation of SEPs, the intrinsic fluorescence of which can be excited selectively in the presence of other, tryptophan-containing proteins. The feasibility of incorporating 5-hydroxytryptophan (5-OHTrp) into a protein expressed in E. coli has recently been demonstrated in an abstract by Szabo and coworkers (7). They incorporated 5-OHTrp into a Tyr-57 to Trp mutant (Y57W) of the calcium binding protein oncomodulin. More recently, they have shown that this 5-OHTrp-containing oncomodulin is recognized by oncomodulin antibodies (8). Using a different expression system, we show here that 5-OHTrp can be successfully incorporated into the bacteriophage A cI repressor, replacing the three tryptophan residues in the C-terminal domain. The absorbance and fluorescence properties of the 5-OHTrp-containing A cI repressor are distinctly different from the wild-type protein, thus demonstrating spectral features desirable in a SEP. Furthermore, this SEP is functionally indistinguishable from the wild-type protein, suggesting that 5-OHTrp SEPs have great potential for investigations of protein-protein and protein-nucleic acid interactions as well as structure/function relationships.

MATERIALS AND METHODS Chemicals and Reagents. L-Tryptophan, 5-OHTrp, and calf thymus DNA were purchased from Sigma. Glucagon, purchased from Elanco (Indianapolis), was purified by ionexchange chromatography as described by Ross (9). Synthetic adrenocorticotropin-(1-24) was a gift from W. Rittel of CIBA-Geigy. The synthetic peptide Gly-5-OHTrp-(Gly)3Glu-(Gly)3-Tyr-Gly and tert-butyloxycarbonyl (t-boc) (a-amino)-5-OHTrp were gifts from G. P. Schwartz of Mount Sinai School of Medicine. Isopropyl f-D-thiogalactopyranoside was purchased from Boehringer Mannheim. [a-32P]dNTPs (3000 Ci/mmol; 1 Ci = 37 GBq) were obtained from Amersham. Unlabeled dNTPs, Klenow polymerase, and bovine serum albumin (acetylated, nucleic acid grade) were obtained from Bethesda Research Laboratories. Bovine pancreas DNase I (code D), from Worthington, was treated as described (10). The E. coli tryptophan auxotroph strain CY15077 (W311OAtrpEA2) and the lacIq-bearing plasmid Abbreviations: SEP, spectrally enhanced protein; 5-OHTrp, 5-hydroxy-L-tryptophan; NATrpA, N-acetyl-L-tryptophanamide; NATyrA, N-acetyl-L-tyrosinamide; t-boc, tert-butyloxycarbonyl; LINCS, linear combination of spectra.

tTo whom reprint requests should be addressed. 12023

12024

Biochemistry: Ross et al.

pMS421 were generous gifts from Charles Yanofsky (Stanford University). All other reagents were reagent or analytical grade. Production of 5-OHTrp-A cI Repressor. E. coli strain CY15077 was transformed with pMS421 and with the tac promoted A expression vector pEA300WT (11) and maintained under ampicillin (60 ;zg/ml) and streptomycin (20 ,ug/ml) selection. An overnight culture, grown in LB-rich medium, was diluted 1:15 into M9 medium (12) supplemented with 0.5% glucose, 1% Casamino acids, 0.1% thiamine, and 0.25 mM L-tryptophan. The culture was grown with shaking at 3rC to an OD550 of 1.0. The cells were then harvested by centrifugation and resuspended in an equal volume of the same medium, except that 5-OHTrp was substituted for L-tryptophan. After the culture was grown for an additional 30 min to an OD55o of 1.3, the expression of repressor was induced by addition of solid isopropyl /3-D-thiogalactopyranoside (final concentration, 1.0 mM). After 6 h of shaking at 370C, the cells were lysed by two passes through a French press at 15,000 psi (1 psi = 6.9 kPa). 5-OHTrp-A cI repressor protein was purified from the crude lysate as described (10). The purified repressor protein accounted for w1% of the wet cell mass, and the preparation was homogeneous as judged by SDS/PAGE. Individual Site n Experhments. DNase I footprint titrations were conducted as described (13, 14). Experiments were conducted with a buffer consisting of 10 mM Bistris (pH 7.00 ± 0.01 at 200C ± 0.10C), 50 mM KCl, 2.5 mM MgCl2, 1.0 mM CaCl2, 0.1 mM Na4EDTA, 100 usg of bovine serum albumin per ml, and 2 1Lg of calf thymus DNA per ml. Conditions for DNase I exposure were 0.2 ng of DNase I added in a 5-pA vol for 8.0 min. DNase I reactions were quenched by addition of Na4EDTA to a concentration of 7 mM before addition of stop solution (14). Two-dimensional optical scanning of footprint titration autoradiograms and analysis of the digitized autoradiogram images were done as described (14). Spectroscopy. Absorption spectra were obtained at room temperature (-o220C) with a Hitachi 3210 dual-beam spectrometer. All fluorescence measurements were made at 200C. Steady-state fluorescence spectra were measured with an SLM Aminco SPF 500C fluorometer modified by us to include Glan-Thompson polarizers. Fluorescence spectra were obtained by using magic angle conditions to avoid intensity artifacts due to molecular rotation during the lifetime of the excited state (15). The steady-state fluorescence anisotropy, r = (Iv - IH)/(Iv + 2IH), where Iv and IH are the vertically and horizontally polarized emission intensities, was measured with an SLM 4800 fluorometer modified by us for singlephoton counting and interfaced to a personal computer as described by Waxman et al. (16). The anisotropy measurements were repeated and averaged until the standard error was 0.002 or smaller. Time-resolved fluorescence intensity and anisotropy measurements were obtained by the timecorrelated single-photon counting method as described (17). Decay curves were collected into either 2000 or 4000 channels, yielding a resolution of 22 or 11 ps per channel, respectively. Analysis of Absorbance Spectra. The absorbance spectra of guanidinium chloride (GdmCl) denatured proteins were analyzed as a linear combination of spectra (LINCS) as described by Hasselbacher et al. (17). Basis spectra were generated using the model compounds N-acetyltryptophanamide (NATrpA), N-acetyltyrosinamide (NATyrA), and t-boc(a-amino)-5-OHTrp.§ The basis spectra and their relative scaling factors (relative extinction coefficients) were

§The a-amino-blocked t-boc(a-amino)-5-OHTrp

was used as the model compound for the basis spectrum of a 5-OHTrp residue in a protein. This was justified by the excellent fit obtained to the absorbance spectrum ofthe synthetic peptide, Gly-5-OHTrp-(GIy)3Glu-(Gly)3-Tyr-Gly, when this basis spectrum was used in combination with the basis spectrum from NATyrA.

Proc. Natl. Acad. Sci. USA 89 (1992)

verified from spectra of glucagon and adrenocorticotropin(1-24), each of which contains one tryptophan and two tyrosine residues, and the synthetic Tyr- and 5-OHTrpcontaining peptide. All absorbance spectra of the model compounds, peptides, and proteins for LINCS analysis were measured in 50 mM Tris.HCI/100 mM NaCl, pH 7.5, or 20 mM Tris HCI/200 mM KCI/5 mM MgCI2/2 mM CaCI2/1 mM NaN3, pH 8, buffers that also contained 6 M Gdm Cl. Analysis of Fluorescence Intensity and Anisotropy Decay Data. Fluorescence intensity and anisotropy decay data were analyzed by nonlinear least-squares regression (18) using iterative reconvolution (19) (see Tables 1 and 2). The reduced x2, weighted residuals, and autocorrelation of the residuals were used as best-fit criteria.

RESULTS AND DISCUSSION corporation of 5-OHIlrp by A cI Repressor. Fig. 1 shows the 5-OHTrp-A cI and wild-type repressor absorption spectra of the proteins denatured in 6 M Gdm-Cl. It also shows the spectra of NATrpA, NATyrA, and t-boc(a-amino)-5-OHTrp for comparison. It should be noted that the absorbance spectrum of 5-OHTrp (spectrum 2 in Fig. 1B) demonstrates a prominent shoulder beginning at 300 nm and extending to =330 nm, which is absent in NATrpA (spectrum 3 in Fig. 1B). This characteristic shoulder is clearly evident in the absorbance spectrum of 5-OHTrp-A cI repressor (spectrum 1 in Fig. 1B). LINCS analysis (17) of the 5-OHTrp-A cI absorbance spectrum was used to assess the degree of incorporation of 5-OHTrp into the repressor. Using the scaled basis spectra (see Materials and Methods), we obtained a ratio of 2.28 (NATyrA/NATrpA) for the wild-type repressor (Fig. 1A), consistent with the 7:3 (2.33) molar ratio expected from the amino acid composition of repressor (20). The 5-OHTrp-A cI repressor spectrum was best fit when all of the three basis spectra described above were used in the LINCS analysis (Fig. 1B). According to this analysis, 95% ofthe incorporated tryptophan residues are 5-OHTrp. The tyrosine/tryptophan

CO

0 0) C) 0)

C) 0

C,)

0.

270

290

330

310

Wavelength, FIG. 1. LINCS analysis of protein absorption spectra. Spectra are arbitrarily scaled and do not reflect relative extinctions. Absorbance spectra were fit to the relationship AT(A) = ZaiAi(A), where AT(A) is the absorbance of the protein at each wavelength A, Ai(A) are the basis spectra, and aj are the relative molar amount of each basis spectrum in the protein spectrum. (A) A cI repressor: dotted line represents the fit to the protein absorbance using the NATrpA and NATyrA basis spectra. (B) 5-OHTrp-A cI repressor: dotted line represents the fit to the protein absorbance using the t-boc-5-OHTrp, NATrpA, and NATyrA basis spectra. nm


Biochemistry: Ross et al. ratio in the SEP repressor {NATyrA/[t-boc(a-amino)-5OHTrp + NATrpA]} was 2.34. The recovery of the appropriate ratios and the excellent fits of the LINCS-derived spectra to the actual spectra for both wild-type and 5-OHTrp-A cI repressor are convincing evidence for efficient incorporation of 5-OHTrp into the repressor. Steady-State and Time-Resolved Fluorescence of 5-OHTrp in A cI Repressor. The excitation and emission spectra of the model compounds NATrpA and 5-OHTrp are compared in Fig. 2A. The fluorescence spectra of the natural and 5-OHTrp-containing repressor proteins are compared in Fig. 2B. The emission maxima of the two proteins are blue-shifted compared to the emission maxima of the corresponding model compounds. This blue shift is typical of tryptophan residues that are not fully exposed to solvent. As expected from the shoulder in the absorption spectra (Fig. 1) extending to wavelengths longer than 300 nm, it is possible to obtain selective excitation of 5-OHTrp residues. This is demonstrated in Fig. 3 in which the emission spectra of 5-OHTrp-A cI repressor alone and in the presence of a 15- and 30-fold molar excess of NATrpA are compared. It is clear from these spectra that the presence of tryptophan would not interfere with observation of 5-OHTrp fluorescence in a SEP. The time-resolved fluorescence parameters of model compounds and the two repressor proteins are compared in Table 1. The intensity decays of both NATrpA and 5-OHTrp are single exponential. By contrast, the intensity decays of the two repressor proteins are complex and can be fit by a sum of three exponentials. Since single tryptophan proteins can exhibit multiexponential fluorescence intensity decays (1, 21), the multiexponential decay of the repressor proteins could reflect either complex fluorescence decay of a single residue, assuming that the other two are nonfluorescent, or decay from more than one residue. Solvent Accessibility of 5-OHTrp in A cI Repressor. Fluorescence quenching by KI was used to compare the exposure of the tryptophan and 5-OHTrp residues in A cI repressor to solvent. The quenching of A cI and 5-OHTrp repressor was measured at a constant ionic strength of 700 mM. Assuming that all emitting residues can be treated as a single class, fluorescence intensity ratios obtained from integrated spectra as a function of KI concentration can be fit to the SternVolmer equation (22, 23). This analysis yields Stern-Volmer Z 11

1 2

A

1

E

1

2

/\0

y(0B

/''''y1/ 260

310

360

410

460

Wavelength, nm FIG. 2. Peak normalized fluorescence excitation (1 and 2) and emission (1' and 2') spectra with respective detection at 350 nm and excitation at 295 nm. Excitation and emission bandpasses were 7.5 nm. All samples were 1-2 LM in pH 8 buffer. (A) NATrpA (-) and 5-OHTrp (-). (B) A cI repressor (-=) and 5-OHTrp-A cI repressor (-

)-

12025

2l)

C

0)

1

0)0

a)C 0

320

420 370 Wavelength, nm

470

FIG. 3. Fluorescence emission spectra of 5-OHTrp-A cI repressor (2 jM monomer) with excitation at 315 nm (-) and after addition of a 15-fold (---) and a 30-fold (-) molar excess of NATrpA. Excitation and emission bandpasses were 4 and 5 nm, respectively. All other experimental conditions are as described in Fig. 2.

constants (KSV) of 1.2 and 0.4 M-1 for A cI and 5-OHTrp repressor, respectively. From the mean fluorescence lifetime, (r) (Table 1), we calculate global bimolecular collisional quenching constants, kq = K.,/I(), of 2.6 and 1.4 x 108 M-1-s-1 for wild-type and 5-OHTrp-A cI repressor, respectively. The corresponding kq values for NATrpA and 5-OHTrp are 29 and 23 x 108 M-1-s-1, respectively, indicating that the small difference in the fluorescence quenching of the two proteins is not significant. The kq values for the proteins are thus an order of magnitude smaller than values typical for exposed tryptophan residues (22, 24). This further suggests that the emitting tryptophan and 5-OHTrp residues are relatively inaccessible to solvent. Fluorescence Resonance Energy Transfer from Tyrosine to 5-OHTrp. Resonance energy transfer from tyrosine to tryptophan can be used to observe changes in three-dimensional structure of proteins. The energy transfer can be monitored directly by a change in the contribution of the donor (tyrosine) absorption to the excitation spectrum of the acceptor (tryptophan) fluorescence. We find that 5-OHTrp is also a good energy transfer acceptor for tyrosine since, upon denaturation of 5-OHTrp-A cI repressor, we observe a substantial reduction in the fluorescence excitation spectrum in the region where tyrosine absorbs. Calculation of the overlap integral, J = fFd.,(A)e,,pt(A)A4dA, for the fluorescence emission spectrum of NATyrA [Fdo.,o(A)] and the absorbance spectrum of 5-OHTrp [ErptO(A)] in water is 2.2 x 10-12 cm6mol-1. By comparison, the overlap integral with NATrpA as an acceptor is 6.6 x 10-13 cM6imol-l (9). Thus, for the same distance of separation and relative orientation of donor and acceptor transition dipoles, the fluorescence energy transfer rate from tyrosine to 5-OHTrp will be -3 times that to tryptophan. Accordingly, energy transfer that takes advantage of5-OHTrp spectral characteristics in a SEP has excellent potential for Table 1. Fluorescence decay parameters Sample al Ti, ns a2 T2, ns a3 T3, ns (T), ns A cI 4.63 0.10 0.75 0.54 3.12 0.36 5.% NATrpA 1.00 3.01 3.01 2.83 5-OHTrp-A cI 0.21 0.25 0.22 1.33 0.57 3.15 3.67 1.00 3.67 5-OHTrp Fluorescence lifetimes were measured as described with detection through a 10-nm bandpass filter centered at 350 nm. Excitation of A cI and NATrpA was at 295 nm. Excitation of 5-OHTrp-A cI and 5-OHTrp was at 315 nm. The solvent was pH 8.0 Tris buffer. Protein samples were 2-3 ,uM (monomer concentration). Fluorescence intensity decay, IM(t), was fit to a sum of exponentials, IM(t) = YcaIexp(-t/T,), where Tj is the lifetime and ai is the amplitude of the ith component. Amplitudes are normalized to unity, and mean lifetime is defined by (7) = XarrFlari.

12026


monitoring conformational change and intermolecular interactions. Time-Resolved Anisotropy of 5-OHTrp and 5-OfHrp-A cI Repressor. To evaluate the utility of 5-OHTrp as an anisotropy probe for studying protein dynamics, we measured the anisotropy decay of the amino acid as a function of viscosity in water, ethylene glycol, and 99%o glycerol (1%o water) with excitation at 315 nm. These results are given in Table 2. When water is the solvent, a time zero anisotropy value, ro, lower than the theoretical maximum of 0.4 was obtained in an unrestricted analysis. A nearly equivalent fit could be obtained, however, when ro was fixed at 0.4. Fits to the anisotropy decay of 5-OHTrp in ethylene glycol and glycerol were significantly improved by including an r. term, and the anisotropy decays were consistent with ro values of 0.4. It is possible that the r. term represents restriction of the out-ofplane rotation in these viscous solvents (26). An exact value for the r. term cannot be resolved from these data since the fluorescence lifetime of 5-OHTrp is more than an order of magnitude shorter than the expected correlation time. The short correlation times from the analyses increased as the viscosity of the solvents increased as would be expected from the Stokes-Einstein relationship (27). In a protein, depolarization of fluorescence may result from a combination of rotational motions of the protein, including segmental fluctuations of the portion of the protein containing the fluorophore and rotations of the entire protein about its molecular axes. For example, we might expect that a spherical, hydrated dimer of 50 kDa (the approximate mass of the dimeric repressor) would have a global correlation time of =18 ns. Segmental motions will be associated with shorter correlation times. In the case where there is more than one fluorophore in a protein, the anisotropy decay can be even more complex since each fluorophore may have a different intensity decay and different segmental freedom. In addition, when there is strong overlap between the emission and the absorption spectrum, as is the case for 5-OHTrp, there is the possibility of depolarization due to resonance energy transfer when more than one such fluorophore is present in a protein. Consequently, it is not surprising that 5-OHTrp--A cI repressor shows a complex anisotropy decay (Table 2). It is important to point out, however, that the correlation times are shorter than predicted for a 50-kDa sphere under conditions where the protein is >95% dimer. Short correlation times are also observed for the wild-type repressor. A third, longer correlation time is required to fit the anisotropy decay Table 2. Fluorescence anisotropy parameters Sample P1i 4, ns 82 42, ns 83 43, ns 0.26 0.07 5-OHTrp in water [0.40] 0.05 5-OHTrp w in ethylene glycol 0.323 1.62 0.108 5-OHTrp 0 in glycerol 0.152 22.3 0.272 0.08o 0.5o 0.102 3.95 5-OHTrp-A cI A cI (wild type) 0.097 0.24 0.031 1.8 0.12 [18] Fluorescence anisotropy decays were measured using the conditions described in Table 1. Decay of emission anisotropy was analyzed as a sum of exponentials, r(t) = 18,jexp(-t/1j), where £,Bj is the limiting time zero anisotropy, ro, and the 4j values are rotational correlation times. These parameters were obtained by simultaneously fitting vertically and horizontally polarized emission decay curves [Iv(t) and IH(t)], using an approach similar to that described by Cross and Fleming (25). To constrain the intensity decay parameters [IM(t); see Table 1] and provide a means for scaling Iv(t) and IH(t) to each other, we included IM(t) in the analysis and fit the data according to the relationships Iv(t) = (1/3)[Im(t){1 + 2I83jexp(-t/4j)}] and IH(t) = (A3)[IM(t){1 - IBjexp(-t/1j)}]. Values in brackets were held constant during analysis (see text).


data of the wild-type protein. The value of this correlation time is in the range expected for the dimer, but its accurate determination is precluded by the short mean lifetime for the intensity decay. Since the mean lifetime for the intensity decay of 5-OHTrp-A cI repressor is shorter (Table 1) and there is the increased possibility of homo-resonance energy transfer, it is not surprising that a correlation time corresponding to the global rotation of the 5-OHTrp-containing dimer is not observed. The presence of shorter correlation times for the anisotropy decay of both proteins suggests that their dynamics are dominated by segmental flexibility and local motions, as has been frequently reported for the anisotropy decay of tryptophan residues (1). Comparison of Self-Assodatlion and Opertor Binding Actities of S-OHTrp-A cI and A cI Repressor. Structural and functional integrity of the 5-OHTrp-A cI repressor was assessed by comparison of the binding constants for the selfassociation reactions of the 5-OHTrp-containing and wildtype repressor monomers and the cooperative binding interactions with A right operator sites OR2 and OR3. The binding constants for the monomer-dimer equilibrium of both 5-OHTrp-containing and wild-type A cI repressor were evaluated by monitoring the change in steady-state fluorescence anisotropy as a function of protein concentration. Wild-type and 5-OHTrp-A cI repressor were excited at 295 and 315 nm, respectively. Fluorescence detection was at 350 nm (15-nm bandpass) for the wild-type and at 380 nm (10-nm bandpass) for the 5-OHTrp-containing repressor. The presence of a small amount of background fluorescence from the buffer precluded measurement of protein anisotropies at monomer concentrations below 10 nM. As a result, it was impossible to obtain rin, the anisotropy for a solution containing only monomers. However, rj. must be >0 and less than or equal to the value obtained at the lowest protein concentration examined. This establishes upper and lower limits for the monomer-dimer dissociation constant. The range of Kd values obtained was identical for the two proteins, strongly suggesting that the free energies of the monomer-dimer interactions are indistinguishable. Furthermore, the upper limit on Kd was 30 nM, a value consistent with the 40 nM Kd measured by Koblan and Ackers (28) using analytical gel chromatography of A cI repressor metabolically labeled with Tran35S-label (29). The cooperative site-specific binding of 5-OHTrp-A cI to the right operator of bacteriophage A (OR) was evaluated by the DNase I footprint titration method (10). The wild-type operator contains three closely spaced operator sites. Binding of repressor dimers to these sites is characterized by pairwise cooperative interactions between repressor dimers bound to adjacent operator sites (30). A statisticalmechanical model for these interactions has been thoroughly evaluated (31). For this characterization of the 5-OHTrp-.A cI repressor-OR interactions, we used reduced valency mutant operators in which site-specific binding to one or more sites was eliminated by the introduction of a single base-pair substitution into each of the mutated sites. The binding competencies of the operators used were OR2- OR1_, and OR1-3-. By simultaneous numerical analysis of the individual site isotherms obtained from these experiments, the free energy changes can be obtained for the intrinsic binding to sites OR2 and OR3 (AG2 and AG3) and for the pairwise cooperative interaction between repressor dimers bound simultaneously to OR2 and OR3 (AG23). Fig. 4 presents the data for the binding of the 5-OHTrp-A cI repressor to the reduced valency operators. Repressor dimer concentrations were calculated assuming a value of 30 nM for the dimer dissociation constant. The values obtained for the free energy changes are AG2 = -14.9 ± 0.2, AG3 = -12.5 ± 0.4, and AG23 = -3.0 ± 0.5 kcal/mol (1 cal = 4.184 J). These free energy changes are indistinguishable from the



*z

0.2 00 s

A'5A

f

0

0.8

0.6

12027

fluorescence of one protein in the presence of others. In addition, we have shown that 5-OHTrp is easily and efficiently incorporated into A cI repressor. 5-OHTrp thus meets all the requirements necessary to be useful in generating SEPs. It is clear that 5-OHTrp-A cI repressor will be valuable for the investigation of this protein's structure and function. It is also clear that SEPs will open up many avenues for the investigation of other proteins where the omnipresence of tryptophan limits the application of intrinsic fluorescence spectroscopy. In the future, it should be possible to use a 5-OHTrp-charged tRNA that has a unique anticodon to replace a single specific residue in vivo. This would make it possible to use intrinsic fluorescence to spectrally probe a specific site without altering all the tryptophan residues in a multitryptophan protein.

0.4

0.2A

0.0

*J

-12

-11

-10

-9

-8

.7

Log [Repressor] FIG. 4. Individual site isotherms of 5-OHTrp-A cI repressor binding to operator sites OR2 (squares) and OR3 (triangles) in reduced valency, mutant bacteriophage A right operators. Operatorcontaining DNA fragments are described by Senear and Ackers (31). Open and solid symbols represent replicate experiments. (A) Noncooperative binding, in which two sites are titrated separately, with no ligation at the other site. OR2 and OR3 data are derived from separate DNase I footprint titration experiments on OR-3- and on OR2-, respectively. (B) Cooperative binding to OR2 and OR3 in which the two sites are titrated together. Data are derived from titrations of an OR1' operator. Solid curves through the points are results of simultaneous analysis of all the data presented. Numerical analysis was conducted as described by Senear and Batey (14).

values previously reported for the wild-type A ci repressor-OR interactions under identical experimental conditions (14), which yielded AG2 = -14.9 + 0.2, AG3 = -12.1 ± 0.3, and AG23 = -2.8 0.5 kcal/mol. These results indicate that replacement of tryptophan by 5-OHTrp has no observable effect on site-specific binding to OR2 and OR3, nor does it affect pairwise cooperativity. The free energy change represents nearly 7 orders of magnitude higher affinity of the repressor for OR2 as compared to random sequence, nonoperator DNA (14). Consequently, even relatively small changes in site specificity should be readily apparent. Experiments to quantitate the interaction of 5-OHTrp-,A cI repressor with OR1 remain to be done. Preliminary results suggest a reduction in the intrinsic affinity for OR1. Nevertheless, this reduction would not affect the ability of the repressor to discriminate among the different operators. The observation that pairwise cooperativity is the same for both wild-type and 5-OHTrp SEP repressors is noteworthy since cooperativity is widely believed to be mediated by the C-terminal domain, which contains the three tryptophan residues in the repressor. Moreover, the C-terminal domain is critically involved in the monomer-monomer association. ±

CONCLUSION The DNA binding and protein self-association data show that 5-OHTrp-A cI repressor is functionally indistinguishable from wild-type repressor. The structural characteristics of 5-OHTrp-A cI repressor, as assessed by fluorescence, are also indistinguishable from those of the wild-type repressor. Furthermore, the spectral results, in particular the unchanged emission spectrum of 5-OHTrp-A cI repressor in the presence of tryptophan (Fig. 3), indicate that 5-OHTrp is a suitable tryptophan derivative for measuring the intrinsic

We wish to thank Dr. Charles Yanofsky for his generous gift of the E. coli tryptophan auxotroph and the lacIq plasmid. We also thank Dr. Gerald P. Schwartz for synthesizing the 5-OHTrp-containing peptide. We gratefully acknowledge Dr. Arthur G. Szabo and Christopher W. V. Hogue for sharing their results prior to publication (8). B.B.K. is supported by a predoctoral fellowship sponsored by the United States Information Agency Fulbright Program. This work was supported in part by National Institutes of Health Grants GM-39750, HL-29019, and GM41465, and in part by a University of California Faculty Research Fellowship. 1. Beechem, J. M. & Brand, L. (1985) Annu. Rev. Biochem. 54,43-71. 2. Robertson, D. E., Kroon, P. A. & Ho, C. (1977) Biochemistry 16, 1443-1451. 3. Li, E., Qian, S., Nader, L., Yang, N. C., d'Avignon, A., Sacchettini, J. C. & Gordon, J. I. (1989) J. Biol. Chem. 264, 17041-17048. 4. Bridges, J. W. & Williams, R. T. (1968) Biochem. J. 107, 225-237. 5. Chen, R. F. (1968) Proc. Natl. Acad. Sci. USA 60, 598-605. 6. Hott, J. L. & Borkman, R. F. (1989) Biochem. J. 264, 297-299. 7. Rasquinha, I., Hogue, C. W. V., Szabo, A. G. & MacManus, J. P. (1992) Biophys. J. 61, 2610. 8. Hogue, C. W. V., Rasquinha, I., Szabo, A. G. & MacManus, J. P. (1992) FEBS Lett. 310, 269-272. 9. Ross, J. B. A. (1976) Dissertation (Univ. of Washington, Seattle). 10. Brenowitz, M. D., Senear, D. F., Shea, M. A. & Ackers, G. K. (1986) Methods Enzymol. 130, 132-181. 11. Amann, E., Brosius, J. & Ptashne, M. (1983) Gene 25, 167-178. 12. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual (Cold Spring Harbor Lab., Cold Spring Harbor, NY). 13. Brenowitz, M. D. & Senear, D. F. (1989) in Current Protocols in Molecular Biology, eds. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A. & Struhl, K. (Greene/Wiley-Interscience, New York), Vol. 2, Suppl. 7, Unit 12.4. 14. Senear, D. F. & Batey, R. (1991) Biochemistry 30, 6677-6688. 15. Badea, M. G. & Brand, L. (1979) Methods Enzymol. 61, 378-425. 16. Waxman, E., Ross, J. B. A., Laue, T. M., Guha, A., Thiruvikraman, S. V., Lin, T. C., Konigsberg, W. H. & Nemerson, Y. (1992) Biochemistry 31, 3998-4003. 17. Hasselbacher, C. A., Waxman, E., Galati, L. T., Contino, P. B., Ross, J. B. A. & Laws, W. R. (1991) J. Phys. Chem. 95, 2995-3005. 18. Bevington, P. R. (1969) Data Reduction and Error Analysis for the Physical Sciences (McGraw-Hill, New York). 19. Grinvald, A. & Steinberg, I. Z. (1974) Anal. Biochem. 59, 583-598. 20. Sauer, R. T., Reihaar-Olson, J. F. & Sauer, R. T. (1990) Proteins Struct. Funct. Genet. 7, 306-316. 21. Grinvald, A. & Steinberg, I. Z. (1976) Biochim. Biophys. Acta 427, 663-678. 22. Eftink, M. R. & Ghiron, C. A. (1981) Anal. Biochem. 114,199-227. 23. Laws, W. R. & Contino, P. B. (1992) Methods Enzymol. 210, 448-463. 24. Eftink, M. R. (1991) Methods Biochem. Anal. 35, 127-205. 25. Cross, A. J. & Fleming, G. R. (1984) Biophys. J. 46, 45-56. 26. Barkley, M. D., Kowalczyk, A. A. & Brand, L. (1981) J. Chem. Phys. 75, 3581-3593. 27. Einstein, A. (1906) Ann. Phys. (Leipzig) 19, 371-381. 28. Koblan, K. S. & Ackers, G. K. (1991) Biochemistry 30, 7817-7821. 29. Beckett, D., Koblan, K. S. & Ackers, G. K. (1991) Anal. Biochem. 196, 69-75. 30. Ackers, G. K., Johnson, A. D. & Shea, M. A. (1982) Proc. NatI. Acad. Sci. USA 79, 1129-1133. 31. Senear, D. F. & Ackers, G. K. (1990) Biochemistry 29, 6568-6577.