Selenomethionine, p-Cyanophenylalanine Pairs Provide a Convenient, Sensitive, Non-Perturbing Fluorescent Probe of Local Helical Structure

Abstract

The use of selenomethionine (M_Se)-p-cyanophenylalanine (F_CN) pairs to probe protein structure is demonstrated. M_Se quenches F_CN fluorescence via electron transfer. Both residues can be incorporated recombinantly or by peptide synthesis. Time-resolved and steady-state fluorescence measurements demonstrate that M_Se-F_CN pairs provide specific local probes of helical structure.

Fluorescence measurements are widely employed in studies of protein dynamics, folding, stability, and aggregation.^{1, 2} Trp has the highest quantum yield of the naturally occurring fluorescent residues in proteins, but its quantum yield depends on a variety of factors, and proteins often contain multiple Trp residues. Both factors can make structural interpretation of Trp fluorescence changes ambiguous. Trp-His pairs have been used to probe secondary structure and rely on the quenching of Trp fluorescence by the His sidechain.³ However, only the protonated form of the His sidechain is an effective quencher of Trp fluorescence, limiting the approach to pH values at which the imidazole group is protonated. The covalent attachment of fluorescent dyes is another popular approach, particularly for use as Förster resonance energy transfer (FRET) pairs, however the method requires selective attachment of two dyes and often requires the introduction of Cys mutations. Furthermore, while typical dyes are very bright, they can perturb the properties of the protein of interest as they are usually built around large polyaromatic cores. In many cases FRET pairs have large R₀ values making it difficult to probe smaller local changes in structure. For these applications short range quenchers are desired.⁴ A simple non-perturbing approach which involves a fluorophore that can be selectively excited and which provides easily interpreted structural information would be a useful addition to the arsenal of fluorescent methods.

We demonstrate that selenomethionine (M_Se) and p-cyanophenylalanine (F_CN) (Fig. 1a) can be used as a minimally perturbative fluorescent probe of protein structure. The pair has been used to examine short oligoproline peptides.⁵ F_CN is the cyano analogue of Tyr. The residue can be incorporated into proteins recombinantly using 21^st pair technology or by solid phase peptide synthesis and represents a conservative substitution for Tyr or Phe.^6–11 F_CN fluorescence can be selectively excited in the presence of Tyr and Trp.⁶ The quantum yield of F_CN is controlled by solvation; fluorescence is high when the cyano group forms a hydrogen bond in a polar protic solvent and is low when it is buried in a hydrophobic environment.^6–12 F_CN fluorescence is also quenched by deprotonated His and Lys as well as via FRET to Tyr or Trp.^{6, 7, 10, 11, 13} The fluorescence lifetime decay of free F_CN is single exponential and has been reported to be 7.5 ns for a G-F_CN-G tripeptide.⁵ In contrast, the fluorescence decay of Trp is multi-exponential, even for simple peptides. M_Se is the selenium analogue of Met and has been widely used in multi-wavelength anomalous diffraction (MAD) phasing for X-ray crystallography and has seen some applications as an NMR probe.¹⁴ The residue can be easily incorporated into proteins in very high yield via recombinant expression and is also compatible with solid phase peptide synthesis. M_Se quenches F_CN fluorescence via electron transfer.⁵ The short range of the quenching effect suggests that F_CN-M_Se pairs could be used to design fluorescence-based probes of local secondary structure. Here we illustrate its use to monitor helical structure.

Fig. 1.

(a) Structure of selenomethionine (M_Se) and p-cyanophenylalanine (F_CN). (b) Sequence of the designed helical peptide. The M_Se and F_CN residues are coloured red. (c) Ribbon diagram of an idealized helix showing the interaction of F_CN and M_Se where the residues are located at positions i and i+4. (d) CD spectra of the peptide in buffer (blue) and in 8 M urea (red). (e) Fluorescence emission spectra of the α-helical state (blue) and the 8 M urea unfolded state (red). Experiments were conducted in 10 mM sodium acetate at pH 5.5 and 25 °C. The concentration of peptide in the samples was 25 μM.

As a first test of the proposed approach we designed a 21-residue α-helical polypeptide (Fig. 1b, c). M_Se is prone to oxidation, however, we did not detect significant oxidation products after incubating the peptide in buffer (Fig. S1). Reducing agents can be used and oxygen excluded by degassing if this is an issue. The polypeptide contains an F_CN residue at position 16 and an M_Se residue at position 12. The i, i+4 separation brings the two residues into proximity in the α-helical state. Circular dichroism (CD) shows that the peptide is helical in buffer at 25 °C (Fig. 1d) and is much less structured in 8 M urea. The mean residue ellipticity at 222 nm is −12,600 deg cm² dmol⁻¹ in buffer which corresponds to an estimated helical content of 38% (Supporting Information). In the urea unfolded state the mean residue ellipticity at 222 nm is −1,400 deg cm² dmol⁻¹. F_CN fluorescence is high in the urea unfolded state, but is reduced twofold in the α-helical state (Fig. 1e). We next conducted time-resolved fluorescence lifetime measurements. The integrated area under the fluorescence decay curve for for the folded state is much smaller than for the unfolded state, consistent with quenching of the F_CN fluorescence in the helical state. The fluorescence decay for the folded state is fit by two components with lifetimes of 4.12 and 0.93 ns with relative amplitudes of 0.73 and 0.27, respectively (Fig. 2). The multi-exponential decay suggests the presence of two populations with different separations between the sidechains of the two residues. Bi-exponential decays have been reported for an M_Se-F_CN dipeptide indicating the effect is likely local in origin.⁵ The fluorescence decay of the urea unfolded state is fit by two components with lifetimes of 5.72 and 1.12 ns with relative amplitudes of 0.86 and 0.14, respectively. Analysis of the data using a maximum entropy approach yields very similar time constants and relative amplitudes (Fig. S2).

Fig. 2.

Time-resolved fluorescence decays for the α-helical state (blue) and the urea unfolded state (red). Decays were fit using two exponentials. The residuals are also plotted.

We next examined the applicability of the pair to follow helix formation in a globular protein. The 36-residue villin headpiece helical subdomain (HP36) was chosen as a model system (Fig. 3a). HP36 is a three-helix protein which contains a single Trp at position 24 in the third helix.^15–17 The helical subdomain is part of the villin protein and the numbering used here denotes the first residue in our construct as residue 1. The domain has been widely used for studies of protein folding, dynamics and stability.^18–27 We replaced Trp-24 with F_CN and residue 28 with M_Se. These sites are located on the surface of the protein on the C-terminal helix. The CD spectrum of the construct is very similar to that reported for wild-type HP36 indicating that these surface substitutions do not perturb the structure.¹⁸ The hydrophobic core of HP36 contains three closely packed Phe residues, leading to characteristic ring current shifted resonances in the ¹H-NMR spectrum which is indicative of the folded state.^{17, 27, 28} These peaks are observed in the ¹H-NMR spectrum of the F_CN-M_Se variant, providing additional evidence that the substitutions do not perturb the fold (Fig. S3). The HP36 F_CN-M_Se variant, like the wild-type, exhibits a sigmoidal thermal unfolding transition (Fig. S4). The i, i+4 spacing of residues leads to efficient quenching in the folded state, but not in the urea unfolded state. The F_CN intensity is six fold less in the folded state relative to the urea unfolded state (Fig. 3c). The larger change in fluorescence observed for HP36 compared to the peptide reflects the fact that HP36 is fully folded in buffer, while the designed helical peptide is only partially structured. Thus a significant fraction of the helical peptide is unfolded and does not experience effective fluorescence quenching. Fluorescence lifetime studies confirm that there are significant differences in F_CN fluorescence between the folded and unfolded states; the integrated area under the time-resolved decay is much less for the folded state relative to the urea unfolded state. The folded state exhibits a bi-exponential decay with time constants of 4.96 and 0.72 ns. The relative amplitudes of the components are 0.68 and 0.32, respectively. The urea unfolded state also exhibits a bi-exponential decay with time constants of 6.72 and 1.21 ns, and relative amplitudes of 0.87 and 0.13, respectively (Fig. 4). Analysis of the data using a maximum entropy approach yields very similar time constants and relative amplitudes (Fig. S5).

Fig. 3.

(a) Ribbon diagram of HP36 showing the location of the F_CN and M_Se residues. The N-terminus is labelled. (b) CD spectra of the protein in buffer (blue) and in 10 M urea (red). (c) Fluorescence emission spectra of the protein in buffer (blue) and in 10 M urea (red). Experiments were performed in 20 mM sodium acetate at pH 5.0 and 25 °C. Protein concentration was 25 μM.

Fig. 4.

Time-resolved fluorescence decays for the protein in buffer (blue) and in 10 M urea (red). Decays were fit using two exponentials. Residuals are plotted below.

In principle F_CN fluorescence quenching can be used to monitor thermally induced protein unfolding. However, like Trp, the quantum yield of F_CN is temperature dependent and decreases with increasing temperature (Fig. S6). This will lead to potential problems with pre- and post-transition baselines, and the ability to unambiguously detect the protein unfolding transition will depend on the magnitude of the fluorescence change due to unfolding.^{1, 3, 29}

His-Trp pairs have been used as probes of α-helical structure and rely on the ability of a protonated His sidechain to quench Trp fluorescence, but this approach is limited to pH values where a significant fraction of the imidazole sidechain is protonated.^{3, 30–32} The quenching of F_CN fluorescence by His is also pH dependent; in the case of F_CN, a deprotonated His sidechain is an effective quencher, but a protonated His sidechain is not.¹³ This is not an issue with the M_Se approach, and effective quenching is observed at both high and low pH (Fig. 5).

Fig. 5.

M_Se quenching of F_CN is pH independent. Fluorescence emission spectra in buffer (blue) and in 10 M urea (red) at (a) pH 5.0 (b) pH 8.5. Experiments were conducted in 20 mM sodium phosphate at 20 °C. Protein concentration was 17 μM.

In summary, we have demonstrated that F_CN-M_Se pairs provide a simple probe of local helical structure in polypeptides and globular proteins. Here we have used equilibrium studies to demonstrate the utility of the method, but the pair is also suitable for kinetic folding and unfolding studies, particularly those that involve dilution out of denaturant, such as stopped-flow experiments. The F_CN-M_Se pair offers several advantages including the ability to selectively excite F_CN fluorescence in the presence of Tyr and Trp, the pH independent response, and the conservative nature of the substitution. In the present case, we illustrated the approach by developing local probes of α-helical structure, but the methodology could also probe β-sheet formation. For example, residues located across from each other on two adjacent β-strands will be close in the native state of a protein, but much more distant in the unfolded state and should experience a significant change in fluorescence. The approach is best suited to solvent exposed sites for the F_CN residue since burial of the F_CN sidechain can quench the fluorescence independent of any M_Se effect. However, the choice of a surface site also ensures that the substitution will be minimally perturbing. As noted earlier, F_CN fluorescence can be quenched by deprotonated His and Lys and by FRET to Tyr or Trp. Interaction with a deprotonated Lys is unlikely given its pK_a, but the presence of a His, Tyr or Trp residue that is in close proximity to the F_CN site should be avoided. We anticipate that the approach will be useful for studies of protein folding and dynamics, protein-protein interactions, and protein aggregation. The approach described here is complementary to the use of thioamides as fluorescence quenchers.^{33, 34} Thioamides offer a quencher localized to the backbone while M_Se provides a sidechain based quencher. Thioamides are commonly incorporated by native chemical ligation while M_Se is incorporated with standard auxotrophic strains.