OS-FRET: A New One-Sample Method for Improved FRET Measurements

Abstract

Fluorescence resonance energy transfer (FRET) is a powerful tool to study macromolecular assemblies in vitro under near physiological conditions. Here we present a new type of One Sample FRET (OS-FRET) method employing a novel, non-fluorescent methanethiosulfonate-linked acceptor that can be reversibly coupled to a target sulfhydryl residue via a disulfide bond. After the quenched donor emission is quantitated the acceptor is removed by reduction, enabling measurement of unquenched donor emission in the same sample. Previous one-sample methods provide distinct advantages in specific FRET applications. OS-FRET is a generalizable spectrochemical approach that can be applied to macromolecular systems lacking essential disulfide bonds and eliminates the potential systematic errors of some earlier one-sample methods. In addition, OS-FRET enables quantitative FRET measurements in virtually any fluorescence spectrometer or detection device. Compared to conventional multi-sample FRET methods, OS-FRET conserves sample, increases data precision, and shortens time per measurement. The utility of the method is illustrated by its application to a protein complex of known structure formed by the P4–P5 fragment of CheA and CheW, both from Thermotoga maritima. The findings confirm the practicality and advantages of OS-FRET. Anticipated applications of OS-FRET include analysis of macromolecular structure, binding and conformational dynamics, as well as high-throughput screening for interactions and inhibitors.

Fluorescence resonance energy transfer (FRET) is widely used to study protein-protein interactions and conformational changes in soluble and membrane-associated complexes. Its resolution is lower than crystallography or NMR, but it can be applied to systems of arbitrary size and often can be carried out using near physiological protein concentrations and buffer conditions.

FRET is a non-radiative energy transfer process in which the excitation energy of an excited donor fluorophor is transferred to a nearby ground state acceptor chromophore, thereby returning the donor to the ground state while exciting the acceptor. To a first approximation, the efficiency of energy transfer E is given by:

$$E=R_0^6/(R_0^6+r^6)$$ (Eq. 1)

and

$$R_0\propto {[{\kappa }^2{\eta }^{-4}{Q}_{D}J(\lambda )]}^{1/6}$$ (Eq. 2)

where r is the distance between the chromophores and R₀ is the Foerster distance at which the transfer efficiency is E = 1/2. For a given donor-acceptor pair, κ² is the orientation factor describing the relative geometries of the donor and acceptor transition dipoles, η is the refractive index of the medium, J(λ) is the integrated overlap of the donor and acceptor emission and absorption spectra, respectively, and Q_D is the fluorescence quantum yield of the donor in the absence of FRET (1–3).

FRET efficiency is often quantitated as the loss of donor fluorescence intensity in the presence of an acceptor fluor, which requires up to three samples (Figure 1a, upper panel): 1) a donor-only sample to measure the initial donor fluorescence, 2) a donor-acceptor sample to measure the donor fluorescence in the presence of FRET, and 3) an acceptor-only sample to quantitate fluorescence at the donor emission wavelength arising from direct acceptor excitation. In recent years some studies have employed a non-fluorescent acceptor, termed a dark acceptor or quencher, which eliminates the need for the acceptor-only sample, but such studies still require both donor-only and donor-quencher samples to measure FRET efficiency (reviewed in (4)). When multiple samples are used to measure FRET, small variations in sample composition or cuvette positioning decrease precision. Moreover, when a donor-only sample is employed, acceptor inner-filter effects may not be adequately quantitated. Measurement of FRET by donor lifetime analysis can overcome some of these weaknesses, but generally requires two samples (reviewed in (2)).

Figure 1. Principle of One-Sample FRET (OS-FRET).

(a) Comparison of the conventional three-sample FRET method employing a fluorescent donor-acceptor pair (upper panel) to the new OS-FRET method utilizing a fluorescent donor and a reversible, non-fluorescent acceptor (or quencher, lower panel). (b) Illustration of the reversible coupling chemistry used in OS-FRET. The label is attached via disulfide formation and can be quantitatively released by addition of a reducing agent like TCEP. (c) Chemical structure of the new Non-Fluorescent Quencher 1 (NFQ1). (d) Spectral overlay of the NFQ1 absorption spectrum (brown) with the fluorescein excitation (black) and emission (gray) spectra.

Several specialized FRET methods requiring just one sample have been developed, each with unique advantages. However, some of these approaches may introduce systematic errors in FRET quantitation, while others may be difficult to generalize to different macromolecular systems or to macro-scale fluorescence detection devices (including fluorescence spectrometers). For example, one approach uncouples the donor-acceptor pair by proteolytically degrading the complex or by chemically releasing the donor (5, 6), which may trigger significant changes in donor environment and spectral properties. A second approach bleaches or photoswitches the acceptor on-off using ultra-high intensity irradiation (7–13), but this method alters acceptor inner filter effects and may photodamage the sample. A third approach switches the acceptor on or off via a chemical modification, but such a modification again alters acceptor inner-filtering and a second, donor-only sample is typically required (14, 15). A fourth approach utilizes a transition metal ion bound to a pair of engineered His residues as a reversible quencher released by addition of a metal chelator (16), but the R₀ ≤ 10 Å for such quenching is too short for many applications and the need for two closely spaced His residues complicates the design of non-perturbing labeling sites. Finally, some single molecule or nano-scale FRET methods (13, 17–21) utilize only one sample but require highly specialized instrumentation and are not suitable for bulk samples nor high-throughput assays.

Here we present a novel One-Sample-FRET (OS-FRET) method designed to minimize systematic errors while at the same time eliminating the need for multiple samples, thereby improving data precision, reducing measurement time, and conserving materials relative to multi-sample approaches. The OS-FRET strategy quantitates FRET from a standard donor to a new type of acceptor combining two existing technologies into a single molecule: (i) the chromophore is a dark acceptor or quencher, thereby eliminating possible acceptor fluorescence at the donor emission wavelength due to direct acceptor excitation, and (ii) the coupling chemistry is a reversible disulfide, allowing measurement of quenched donor fluorescence, reductive release of quencher into solution, and measurement of unquenched donor fluorescence all in the same sample (Figure 1a, lower panel). The utility of the new method is tested in a well-characterized protein complex formed between the CheA and CheW proteins of bacterial chemotaxis. These proteins normally associate in the ultrastable, membrane-bound chemosensory signaling complex (22), but the isolated P4–P5 fragment of CheA (CheA_F) yields a soluble complex with CheW. The high resolution crystal structure of the CheA_F-CheW complex has been solved for the Thermotoga maritima proteins (PDB-ID 2CH4) (23). This representative application of OS-FRET confirms its simplicity and utility in the analysis of binding interactions, its suitability for standard fluorescence detection systems (including spectrofluorimeters), and its advantages over conventional multi-sample FRET methods.

MATERIALS AND METHODS

Materials

Reagents were obtained from the following sources: mutagenic oligo-nucleotides from Integrated DNA Technologies, sulfhydryl-specific probes fluorescein-5-maleimide (5FM) and Alexa 750 from Invitrogen Molecular Probes, Ni-NTA from Qiagen, BSA and TCEP from Thermo Fisher, DMSO and DMF spectroscopic grade from Sigma. All other chemicals were analytical grade from Sigma unless noted otherwise.

Mutagenesis

The plasmids encoding N-terminal 6-His-tagged Thermotoga maritima CheW in the pET28a vector, and N-terminal 6-His-tagged Thermotoga maritima CheA P4–P5 domain (CheA_F) in pET28a were a generous gift from Dr. Brian Crane at Cornell University (23). Site-directed Cys mutants were generated using the PCR-based QuickChange XLII mutagenesis kit (Agilent). All mutants were confirmed by sequencing the entire coding region.

Protein purification

His-tagged proteins were isolated by Ni-NTA affinity chromatography as described (24). Purified proteins were stored at −80°C in Buffer A (50 mM Tris-HCl pH 7.5, 50 mM NaCl, 0.5 mM EDTA, 10 μM DTT, 10% glycerol). Protein concentration was measured by Bradford assay using BSA as standard. Purity was quantitated by SDS PAGE and Coomassie staining and was generally >98%.

NFQ1 labeling

NFQ1 stock solutions were prepared by dissolving 20 mM NFQ1 in DMSO. Single Cys mutant proteins were first incubated in Buffer A with 1 mM DTT for 60 min at room temperature to reduce possible disulfides. DTT was then removed by size exclusion chromatography over a PD10 desalting column (GE Healthcare). Protein was collected directly in sample vials holding buffer B (50 mM Tris-HCl pH 7.5, 150 mM NaCl) and free label. Final concentrations during labeling reactions were 5 μM protein and 25 μM label. The reaction was allowed to proceed for two hours at room temperature in the dark while gently stirring. Subsequently, protein was bound to a Ni-NTA column and the remaining free label removed by washing the column extensively with 10 column volumes of buffer B. The labeled protein was eluted with Buffer B containing 500 mM imidazole and dialyzed twice against buffer A without DTT. Finally, the samples were incubated with 5 mM N-ethylmaleimide for 30 min to block unlabelled Cys residues. Samples were aliquoted, snap frozen in liquid N₂ and stored at −80°C. Each aliquot was used only once and not refrozen. Labeling efficiency was quantitated as described (25), and no unlabeled protein was detected by electrospray mass spectrometry on an Applied Biosystems QStar Pulsar instrument (Supplemental Fig. S1).

5FM and Alexa 750 labeling

Labeling of single-Cys mutant proteins with 5FM or Alexa 750 was carried out as described for NFQ1 labeling with the following modifications. Stock solutions were 20 mM 5FM in DMF and 40 mM Alexa 750 in DMSO. Labeling reactions were carried out for 30 min at room temperature in the dark. The masses of labeled proteins were verified by mass spectrometry on a PerSeptive Voyager DE-STR MALDI-TOF instrument.

FRET measurements

a. Experimental conditions

Steady-state fluorescence experiments utilized a Photon Technology International QM-2000-6SE fluorescence spectrometer and were carried out at 25°C in buffer C (50 mM Tris-HCl pH 7.5, 50 mM NaCl, 0.5 mM EDTA, 0.2 μM BSA, 10% glycerol). Excitation wavelength was 492 nm (slit width 4 nm), and the emission wavelength was 515 nm (slit width 8 nm). All fluorescence measurements were made in ratiometric mode, in which the sample fluorescence is ratioed to a reference channel monitoring lamp fluctuations, thereby eliminating the contributions of lamp noise and intensity drift.

Due to the hydrophobicity of chromophores, particularly NFQ1, BSA (0.2 μM) was included in all samples to minimize nonspecific binding of labeled proteins to glass surfaces and to minimize nonspecific binding of free probe to protein, particularly after release of NFQ1 by reduction. All pipette tips used for the FRET experiment were pre-treated by rinsing with a 2 mg/ml BSA solution, pipetting up and down several times then drying, to prevent nonspecific binding of labeled protein or free probe.

b. Complex formation

Complexes for the different experiments were formed by incubating 0.1 μM donor-labeled CheW mutant (or the unlabeled, NEM-treated protein), with 0.5 μM acceptor-labeled CheA_F S660C (or the unlabeled, NEM-treated protein), for 15 min at room temperature.

c. OS-FRET

First, the donor fluorescence intensity of the donor-quencher complex between a 5FM-labeled CheW mutant and NFQ1-labeled CheA_F S660C was measured. Subsequently, TCEP was added to a final concentration of 1 mM in the cuvette, still in the sample holder, and the sample was gently stirred in place for 15 min. This reduction step releases NFQ1 from the complex into solution out of FRET range, enabling measurement of the unquenched donor fluorescence intensity. Electrospray mass spectrometric analysis of the products confirmed that NFQ1 release was quantitative within the limit of detection (Supplemental Fig. S1). The apparent FRET efficiency was calculated as

$$E_{\mathit{app}}=1-\frac{F_{DQ}}{F_{DQ+\mathit{TCEP}}}$$ (Eq. 3)

where F_DQ is the donor fluorescence intensity of the donor-quencher complex and F_DQ+TCEP is the unquenched donor fluorescence intensity after release of quencher by TCEP.

d. Two sample-FRET using NFQ1 as acceptor

Two different complexes with identical protein concentrations were formed in two separate samples. One (donor-quencher) contained 5FM-labeled CheW mutant and NFQ1-labeled CheA_F S660C, while the other (donor-only) contained 5FM labeled CheW mutant and unlabeled CheA_F S660C. The donor fluorescence intensity of both samples was measured and the apparent FRET efficiency was calculated as

$$E_{\mathit{app}}=1-\frac{F_{DQ}}{F_D}$$ (Eq. 4)

where F_DQ is the donor fluorescence intensity of the donor-quencher complex and F_D is the donor fluorescence intensity of the donor-only complex.

e. Three sample-FRET using Alexa 750 as acceptor

Three different complexes with identical protein concentrations were formed in three separate samples. One (donor-acceptor) contained 5FM-labeled CheW mutant and Alexa 750-labeled CheA_F S660C, one (donor-only) contained 5FM-labeled CheW mutant and unlabeled CheA_F S660C, and the other (acceptor-only) contained unlabeled CheW mutant and Alexa 750-labeled CheA_F S660C. The donor fluorescence intensity of all three samples was measured and the apparent FRET efficiency was calculated as

$$E_{\mathit{app}}=1-\frac{F_{DA}-F_{A(D)}}{F_D}$$ (Eq. 5)

where F_DA is the donor fluorescence intensity of the donor-acceptor complex, F_A(D) is the acceptor fluorescence at the donor emission wavelength due to both direct acceptor excitation and to FRET, and F_D is the donor fluorescence intensity of the donor-only complex. In practice, it is not possible to measure F_A(D) directly, however the contribution due to direct acceptor fluorescence can be determined as the fluorescence intensity of the acceptor-only complex (F_A). In the present system, due to the Stokes shifts of fluorescein and Alexa 750, acceptor fluorescence at the donor emission wavelength arising from FRET must be vanishingly small and can be neglected, thus the approximation F_A(D) ~ F_A is used in Eq. 5. Note this approximation will not be valid for all FRET pairs.

Determination of K_D values for complex formation and correction of FRET efficiencies

For each donor-acceptor complex, the equilibrium dissociation constant (K_D) was determined by titrating donor-labeled CheW mutant with quencher- or fluorescent acceptor-labeled CheA_F S660C. The increasing FRET signal was used to quantitate complex formation. NFQ1-labeled or Alexa 750-labeled CheA_F S660C was titrated into a cuvette containing 0.1 μM 5FM-labeled CheW. Following each addition, the sample was incubated at 25ºC for 10 min to reach equilibrium then the donor fluorescence intensity was measured. FRET efficiency at each titration point was calculated as

$$E=1-\frac{F_T}{F_D}$$ (Eq. 6)

where F_T is the donor fluorescence intensity at the titration point and F_D is the donor fluorescence intensity of the donor-only sample prior to titration. Normalized FRET efficiency was plotted against the total CheW concentration and K_D was determined by using KaleidaGraph 4.0 to best-fit the quadratic binding equation, which accounts for the distribution of total CheW between bound and free populations:

$$\begin{array}{l}E=E_{max}\times P_B\\ =E_{max}\times \frac{\frac{{[{\mathit{CheA}}_F]}_{\mathit{total}}+{[\mathit{CheW}]}_{\mathit{total}}+K_D}{2}-\sqrt{\frac{{({[{\mathit{CheA}}_F]}_{\mathit{total}}+{[\mathit{CheW}]}_{\mathit{total}}+K_D)}^2}{4}-{[{\mathit{CheA}}_F]}_{\mathit{total}}+{[\mathit{CheW}]}_{\mathit{total}}}}{{[\mathit{CheW}]}_{\mathit{total}}}\end{array}$$ (Eq. 7)

where E is the measured FRET efficiency, E_max is FRET efficiency at saturation, P_B is the fraction of CheW bound in complex, [CheA_F]_total is the total concentration of labeled CheA_F, [CheW]_total is the total concentration of labeled CheW and K_D is the equilibrium dissociation constant.

The resulting K_D values were used to correct the apparent FRET efficiencies of one, two-, and three-sample FRET measurements for incomplete complex formation at the protein concentrations employed. When complex formation was incomplete, a small percentage of donor-labeled CheW molecules were not bound to quencher- or acceptor-labeled CheA_F, thereby artificially decreasing the apparent FRET efficiency (E_app). The fraction of the donor-labeled CheW population successfully bound to CheA_F (P_B) was determined from the measured K_D of the complex and the total concentration of labeled CheA_F ([CheA_F]_total) using equation 7. The corrected FRET efficiency (E_cor) was calculated as

$$E_{\mathit{cor}}=\frac{E_{\mathit{app}}}{P_B}$$ (Eq. 8)

which fully accounts for the effects of incomplete complex formation.

Determination of quantum yield of 5FM labeled CheW

Quantum yield was determined as described at the website http://www.jobinyvon.com/usadivisions/Fluorescence/applications/quantumyieldstrad.pdf. Fluorescein in 0.1 M NaOH was used as standard with a quantum yield of 0.93 (2). Samples of individual 5FM-labeled CheW mutants were measured in Buffer A without DTT. To minimize errors, the absorbance of each solution was measured in a 5 cm pathlength cuvette.

RESULTS

A new reversible dark acceptor

To implement the OS-FRET strategy we designed and synthesized a novel, reversibly coupled dark acceptor (or quencher) by combining the non-fluorescent chromophore of QSY35 (Invitrogen) (26), an NBD derivative, with the methanethiosulfonate functionality. Currently, methanethiosulfonate derivatives are widely employed to couple spectroscopic probes to specific protein Cys residues via disulfide exchange as Figure 1b illustrates (26–28). Figure 1c shows the chemical structure of the new probe, non-fluorescent quencher 1 (NFQ1). The structure of NFQ1 was confirmed by NMR (data not shown). NFQ1 has an extinction coefficient of ε_485nm = 10030 M⁻¹cm⁻¹ in aqueous buffer at pH 7.5. Since NFQ1 has limited solubility in aqueous buffer, stock solutions were made in DMSO. Fig. 1d shows the donor fluorescein-5-maleimide excitation and emission spectra, and the NFQ1 absorption spectrum, where both probes are coupled to free Cys in buffer at pH 7.5. We estimate the Foerster distance for the fluorescein-NFQ1 pair to be R₀ ~ 39 Å based on (i) a quantum yield of Q_D = 0.69 measured for fluorescein-coupled CheW (see Methods) and (ii) an average orientation factor of κ² ~ 2/3 assuming rapid, isotropic tumbling of one or both transition dipoles. Notably, the unusually broad absorption spectrum of NFQ1 ensures its utility as a dark acceptor for many donors.

Preparing the protein complex for FRET studies

Figure 2a illustrates the known structure of the soluble complex formed by association of the CheA_F fragment (residues 289–671 of CheA) with CheW (23). Both proteins from Thermotoga maritima lack Cys residues and the K_D for their association is submicromolar, making the complex well-suited for FRET measurements. Engineered Cys residues were introduced as labeling sites (Fig. 2a): five CheW sites (yellow spheres) were selected for donor fluorescein, and one CheA_F site (S660C, brown sphere) for dark quencher NFQ1. In the crystal structure, the selected donor-quencher pairings exhibit Cβ-Cβ distances from 8 to 48 Å, designed to yield a large range of FRET efficiencies.

Figure 2. Application of OS-FRET to a complex of known structure.

(a) Complex between CheW (blue) and the P4–P5 domain fragment of CheA (CheA_F, green), both from Thermotoga maritima. Positions chosen for single Cys mutations are highlighted as spheres: the NFQ1 labeling site is brown, while the fluorescein labeling sites are yellow. [Figure created in PyMol (DeLano Scientific LLC) using PDB-ID 2CH4]. (b) Emission spectra of the double-labeled complex ((−)TCEP, solid black line) and of the single-labeled complex after reductive release of the quencher ((+)TCEP, broken black line). Also shown is the emission spectrum of the donor-only complex lacking quencher altogether (broken green line), and of the donor-only complex after addition of reducing agent ((+)TCEP, open green circles) or free quencher and reducing agent ((+)NFQ1 (+)TCEP, solid green line under broken black line). The left panel presents data for a short donor-acceptor separation (CheW L55C-5FM / CheA_F S660C-NFQ1, Cβ-Cβ = 16 Å) while the right panel presents data for a longer separation (CheW V114C-5FM / CheA_F S660C-NFQ1, Cβ-Cβ = 47 Å). (c) FRET titration to determine the equilibrium dissociation constant for a representative complex (CheW V114C-5FM / CheA_F S660C-NFQ1). Titration points (open circles) were fitted to the quadratic binding equation (Methods Eq. 7, fitted in KaleidaGraph 4) yielding the best-fit curve (black line) and K_D.

Standard methods were employed to couple donor fluorescein to the five CheW Cys mutants (25). The Methods describe the labeling reaction used to couple the new NFQ1 probe to CheA_F S660C, yielding an NFQ1 labeling efficiency of 93 ± 1 % or higher as determined biochemically (see Methods) and by mass spec analysis (Supplemental Fig. S1). Thus, nearly all available Cys residues were labeled with quencher. The high labeling efficiency is especially important for the quencher, since on average each complex containing a donor-labeled CheW must also contain quencher-labeled CheA_F for accurate FRET efficiency measurements.

FRET efficiencies measured by OS-FRET

Figures 2b,c present spectra for two representative donor-quencher pairs and Table 1 summarizes the OS-FRET results for all five pairs. As detailed in Methods, complexes were formed by incubating donor-labeled CheW and quencher-labeled CheA_F for 15 min, then the quenched donor emission spectrum was recorded (solid line Figure 2b). Subsequently, to release the quencher, reducing agent (TCEP) was added and the sample incubated 15 min without removing it from the spectrometer. Finally, the unquenched emission spectrum of donor-only complex was recorded (dashed black line, Figure 2b). Release of the quencher out of FRET range yielded a significant donor emission increase, and the increase was greater for the closer representative pair (Fig. 2b, compare left, right panels). This procedure was used to measure the uncorrected FRET efficiency for each donor-quencher pair at sufficiently high concentrations of quencher-labeled CheA_F to nearly saturate donor-labeled CheW. In a separate titration experiment to measure the K_D of each donor-quencher pair, FRET efficiencies were measured as quencher-labeled CheA_F was titrated into the sample (Fig. 2c, see Methods). The final, corrected FRET efficiencies presented in Table 1 include corrections for incomplete saturation of donor-labeled protein with quench-labeled protein.

Table 1.

Comparison of OS-FRET with two- and three-sample FRET methods¹⁾

Donor	Acceptor			FRET efficiency2)
CheW Probe	CheAF S660C	KD [nM]	Cβ - Cβ[Å]	Predicted	OS-FRET	Multi-sample FRET
D38C-5FM	NFQ1	80 ± 10	9	99.9 %	83.7 ± 0.5 %	86.6 ± 0.8 %3)
L55C-5FM	”	110 ± 30	16	99.6 %	93.5 ± 0.5 %	95.6 ± 0.8 %3)
K120C-5FM	”	80 ± 20	40	48 %	34.0 ± 0.6 %	38.7 ± 1.2 %3)
V114C-5FM	”	100 ± 20	47	26 %	31.8 ± 0.7 %	38.9 ± 1.2 %3)
E106C-5FM	”	100 ± 10	48	24 %	28.3 ± 1.0 %	32.5 ± 1.5 %3)
D38C-5FM	Alexa 750	150 ± 20	9	99.9 %	-	74.1 ± 2.7 %4)
L55C-5FM	”	20 ± 10	16	99.7 %	-	65.8 ± 3.2 %4)
K120C-5FM	”	90 ± 30	40	56 %	-	35.4 ± 2.0 %4)
V114C-5FM	”	130 ± 30	47	33 %	-	34.7 ± 2.8 %4)
E106C-5FM	”	110 ± 20	48	30 %	-	32.2 ± 3.0 %4)

¹⁾All errors are standard errors of the mean, calculated for an average of three means, where each mean is an average of triplicate samples.

²⁾Predicted FRET efficiencies calculated using Ro values of 39.7 Å for 5FM-NFQ1 and 41.7 Å for 5FM-Alexa 750, estimated using approximations indicated in text. Experimental FRET efficiencies are corrected for incomplete saturation of donor-labeled protein with acceptor-labeled protein as described in Methods.

³⁾Two samples: donor-only complex and donor-acceptor complex

⁴⁾Three samples: donor-only complex, acceptor-only complex, and donor-acceptor complex

Separate control studies were carried out to ensure the OS-FRET method was properly optimized for the CheA_F-CheW complex. Addition of TCEP to donor-only complex had no measurable effect on donor fluorescence (Fig. 2b), indicating that the effect of TCEP on the donor-quencher complex arose simply from quencher release as intended. Addition of NFQ1 and TCEP to donor-only complex yielded the same level of donor fluorescence as TCEP-triggered release of NFQ1 in donor-quencher samples (Fig. 2b), providing optical evidence that TCEP quantitatively releases the quencher and that quencher inner-filtering (see Discussion) is unaltered by quencher release. Finally, mass spec analysis of products provided chemical evidence that the TCEP treatment yielded quantitative quencher release (Supplemental Fig. S1).

FRET efficiencies measured by multi-sample FRET

For comparison, we also measured FRET efficiencies for the same five donor-quencher pairs using conventional two- and three-sample methods, again employing fluorescein as donor. The two-sample method used NFQ1 as non-fluorescent acceptor and calculated FRET efficiencies from the emissions of donor-quencher and donor-only samples. The three-sample method employed Alexa-750 as a fluorescent acceptor and determined FRET efficiencies from donor-acceptor, acceptor-only and donor-only samples. The estimated Foerster distance for the fluorescein-Alexa750 donor-acceptor pair is R₀ ~ 42 Å, similar to that of the fluorescein-NFQ1 donor-quencher pair. Table 1 compares the one-, two-, and three-sample FRET efficiencies, all corrected for incomplete saturation of CheW with CheA_F.

DISCUSSION

Examination of the results reveals that the OS-FRET method is superior to the conventional two- and three-sample methods in several ways. OS-FRET uses 50% and 67% less sample than the two- and three-sample methods, respectively, and reduces time per measurement approximately 30–70%. (Time estimates are based on optimized protocols for triplicate measurements using a 4 sample cuvette holder and include the time required for reductive quencher release, but do not include additional time savings provided by simpler data analysis and fewer protein purifications needed for OS-FRET). Table 1 further shows that the precision of OS-FRET is on average a factor of 1.7- and 4.2-fold better than the two- and three-sample methods, respectively, owing to the error introduced by multiple samples. Moreover, in contrast to OS-FRET, the accuracy of two- and three-sample methods can be compromised by acceptor inner-filter effects. As illustrated in Fig. 2b for the fluorescein-NFQ1 pairs, reduced donor-quencher samples exhibit slightly lower donor emission intensities than those of donor-only samples. This is due to NFQ1 absorbance, or inner filtering, at the donor excitation and emission wavelengths as directly confirmed by adding free NFQ1 to the donor-only labeled sample (Fig. 2b). In OS-FRET the quencher and its inner filtering is present in both the FRET and non-FRET states, thus inner filtering is automatically divided out in OS-FRET efficiency calculations (Eq. 3). As a result, OS-FRET yields more accurate FRET efficiencies that are slightly lower than measured by methods employing donor-only samples. This explains the small but reproducible differences between OS-FRET and two-sample FRET efficiencies for the 5FM-NFQ1 pair in Table 1, where the OS-FRET efficiencies are 1–7% lower.

Not surprisingly, the FRET efficiencies measured by all approaches deviate somewhat from the values predicted by standard Foerster theory for the crystal Cβ-Cβ distances. The measured OS-FRET efficiencies, for example, roughly follow the predicted r^-6 dependence on donor-quencher separation (Table 1), but deviations are observed. Assuming the crystal structure is an accurate representation of the solution structure, the observed deviations could arise from multiple factors. For a given donor-acceptor pair, the true donor-acceptor distance could differ substantially from the crystal Cβ-Cβ separation due to the large sizes of the probes and the flexibilities of their linkers. In addition, the assumption of isotropic, rapid probe motion (κ² = 2/3) could be unjustified if the local environments of the probes produce strong steric hindrances or binding interactions. Finally, for small donor-acceptor separations approaching the dimensions of the probes, the point-dipole approximation employed by Foerster theory could break down (29, 30). Thus, as for other FRET methods, the structural information provided by OS-FRET distance measurements is low resolution (Table 1). It follows that OS-FRET is most useful for monitoring macromolecular binding titrations (Fig. 2C and Table 1), for low resolution structural analysis, and for detection of function-associated conformational changes (Erbse, Berlinberg and Falke, unpublished).

Overall, OS-FRET provides a unique combination of six useful properties. OS-FRET (i) measures FRET efficiency in just one sample; (ii) typically preserves the same donor local environment when the quencher is released (assuming that the quencher is neither contacting nor allosterically coupled to the donor); (iii) ensures that quencher inner filtering does not introduce systematic error; (iv) provides R_o values on a useful macromolecular distance scale; (v) utilizes a sulfhydryl-specific, reversible coupling technology and engineered Cys residues known to be useful in protein studies; and (vi) can be implemented using a conventional fluorimeter or other macro-scale fluorescence detector without potentially damaging, ultra-high intensity irradiation. The present studies directly demonstrate the practicality of OS-FRET and its advantages relative to multi-sample FRET in studies of a representative macromolecular complex.

The limitations of OS-FRET are relatively minor. (i) NFQ1 is quite hydrophobic, which can cause it to bind to a protein surface rather than freely tumbling while coupled, or to remain non-covalently bound to protein after reductive release. The latter concern is eliminated by the present experimental design that ensures complete quencher dissociation following reductive release (see Methods). The development of new generation, more hydrophilic probes may provide an alternative solution. (ii) The use of a dark quencher prevents direct confirmation that FRET is the quenching mechanism; if desired, one can test whether quenching exhibits the r⁻⁶ distance-dependence characteristic of FRET within practical constraints (Table 1, see above Discussion). (iii) The use of a dark quencher prevents full fluorescence anisotropy analysis of donor and acceptor tumbling useful in minimizing κ² uncertainty. This limitation is deemed minor since many FRET applications, such as qualitative characterization of conformational changes, do not require full knowledge of κ² (most published FRET studies, even those employing fluorescent acceptors, do not investigate probe tumbling).

In summary, the OS-FRET method employing the novel dark quencher NFQ1 enables precise, accurate measurement of FRET in a single sample. The method optimizes sample efficiency and reduces measurement time while minimizing errors relative to conventional multi-sample FRET methods. We expect that OS-FRET will be useful in a wide range of experiments probing macromolecules and their complexes, and will be successfully measured in an array of fluorescence detection devices. OS-FRET is especially useful in systems where sample availability is limited and small changes in FRET must be quantitated with great accuracy. In our laboratory, for example, OS-FRET is proving useful in studies of on-off switching in membrane associated signaling complexes (Erbse, Berlinberg & Falke, unpublished). An important future application of OS-FRET will be high throughput studies of protein-protein binding or inhibitors that block binding, since the OS-FRET method can be easily extended to multi-well plate formats.

ABBREVIATIONS

OS-FRET

one-sample FRET

NFQ1

non-fluorescent quencher 1

MTS

Methanethiosulfonate

DTT

dithiothreitol

Ni-NTA

Ni(II)-nitrilotriacetic acid

BSA

bovine serum albumin

TCEP

tris(2-carboxyethyl)phosphine

DMSO

dimethyl sulfoxide

DMF

dimethyl formamide