Structural origins of altered spectroscopic properties upon ligand binding in proteins containing a fluorescent non-canonical amino acid

Abstract

Fluorescent non-canonical amino acids (fNCAAs) could serve as starting points for the rational design of protein-based fluorescent sensors of biological activity. However, efforts toward this goal are likely hampered by a lack of atomic-level characterization of fNCAAs within proteins. Here, we describe the spectroscopic and structural characterization of five streptavidin mutants that contain the fNCAA L-(7-hydroxycoumarin-4-yl)ethylglycine (7-HCAA) at sites proximal to the binding site of its substrate, biotin. Many of the mutants exhibited altered fluorescence spectra in response to biotin binding, which included both increases and decreases in fluorescence intensity as well as red or blue shifted emission maxima. Structural data were also obtained for three of the five mutants. The crystal structures shed light on interactions between 7-HCAA and functional groups—contributed either by the protein or substrate—that may be responsible for the observed changes in the 7-HCAA spectra. These data could be used in future studies aimed at the rational design of fluorescent, protein-based sensors of small molecule binding or dissociation.

INTRODUCTION

Genetically encoded non-canonical amino acids (NCAAs) possess chemical functionalities that are not found among the twenty proteogenic amino acids.¹ The side chains of some NCAAs are fluorescent and have emission wavelengths that fall in the visible range.^2–6 Because such fluorescent non-canonical amino acids (fNCAAs) are translationally incorporated, they are not limited to placement at surface exposed residues or the protein termini, which are common limitations to other frequently used methods of fluorescently labeling proteins. Furthermore, most fNCAAs are similar in size to naturally occurring amino acids, which suggests that they can be incorporated in less accessible areas of a protein without a great risk of disruption of folding or function. Finally, many fNCAAs exhibit fluorescence spectra that are responsive to local environments.^2–4 These features suggest that fNCAAs could represent excellent starting points for the development of novel, protein-based fluorescent biosensors.

Among the fNCAAs, L-(7-hydroxycoumarin-4-yl)ethylglycine³ (7-HCAA), has found extensive use in protein-based, fluorescent sensors that report on protein-ligand interactions,^7–9 protein-protein interactions,^10–12 enzyme-substrate binding,^13,14 and changes in tyrosine phosphorylation state.¹⁵ A likely reason for 7-HCAA’s widespread use is that it possesses tunable functional groups that can be used to modulate its fluorescence output. For example, the 7-hydroxycoumarin (7-HC) moiety can exist in a number of tautomeric or ionic forms in the ground state that each give rise to distinct absorption/emission maxima (Scheme S1).¹⁶ Furthermore, 7-HC is a photoacid wherein the pK_a of the phenol drops from ~7.8 in the ground state to ~0.4 in the excited state (Scheme S1).¹⁶ When a proton acceptor (e.g., water) is present, the phenolic proton is rapidly shuttled from the excited hydroxycoumarin to the proton acceptor in a process termed excited state proton transfer (ESPT).¹⁶ If ESPT is blocked, the predominant emitting species is the phenol rather than the phenolate and the emission maximum is shifted from 450 nm to 380 nm (Scheme S1).^17–19

The diverse chemical environments found in proteins could potentially be used to alter 7-HCAA emission, but only if appropriate sites of incorporation can be identified. In previous studies,^{7–11,13–15} two general approaches were used to select sites of 7-HCAA incorporation: The first approach relies on Förster Resonance Energy Transfer (FRET)^7–9,11,12 between 7-HCAA and a second fluorophore on the target protein. Although powerful, FRET-based techniques are best suited for use in proteins in which altered distances between the two fluorophores are expected as a consequence of conformational changes related to protein function. In a second approach, structure-based analyses were used to identify sites of 7-HCAA incorporation that were close to ligand binding sites,^13,14 protein-protein interfaces,¹⁰ or a residue that was known to be post-translationally modified.¹⁵ Both approaches afforded protein-based sensors in which changes in 7-HCAA fluorescence could be used to monitor protein function; however, none of these 7-HCAA-modified proteins were structurally characterized in the initial studies. Recently, the structure of a 7-HCAA-containing sensor of protein-protein interactions was reported.²⁰ However, this study was limited to a single protein and only one site of 7-HCAA incorporation. Without additional structural detail regarding the response of 7-HCAA to distinct chemical environments, it is difficult to envision how 7-HCAA might be systematically applied to address distinct challenges in the future.

In this study, we hoped to gain additional insight into how chemical environments affect the fluorescence properties of 7-HCAA. To achieve this, we genetically encoded 7-HCAA at five positions in the biotin-binding protein streptavidin (SAV) that were in close proximity to the substrate binding pocket. The photochemical properties of each mutant were assessed in the presence and absence of biotin, and biotin-dependent changes in fluorescence were observed in four of the five mutant proteins. Structural characterization of three of these mutants provided insight into how the protein environment surrounding 7-HCAA may alter its photochemical properties. The results of this study should prove useful in future efforts to develop novel protein-based fluorescent sensors of analyte binding.

MATERIALS AND METHODS

Protein Expression & Purification:

7-HCAA incorporation into streptavidin was achieved using the amber codon suppression method developed by Schultz et al.^3,21 Briefly, amber codons were introduced into SAV using site-directed mutagenesis. E. coli BL21 Star (DE3) cells co-transformed with both a pEVOL plasmid²² containing a chloramphenicol resistance marker, two copies of the evolved CouRS synthetase, and an evolved tRNA specific to the evolved CouRS and a pET29 plasmid containing a kanamycin resistance marker and genes encoding the mutant streptavidins. Protein expression was carried out in 2xYT media in the presence of 1 mM 7-HCAA. After harvesting, cells were lysed and the SAV mutants were purified from inclusion bodies using a modified version of a previously described protocol.²³ See the Supporting Information for a detailed description of the protein purification protocol. Ultimately, SAV variants were purified using immobilized metal ion affinity chromatography, ion exchange chromatography and size exclusion chromatography.

Spectroscopic Analysis of SAV Mutants:

All spectroscopic experiments were performed in a Tris-Buffered Saline (TBS) solution (25 mM Tris-HCl pH 7.0, 150 mM NaCl) using a 1 cm quartz cuvette (Starna Cells). Each apo mutant was concentrated to 100 μl and then diluted to a final absorbance of 0.05 at 340 nm. The samples were then split and either biotin (100 μM final concentration, dissolved in 0.3% DMSO) or TBS was added to each. A SpectraMax M5 spectrophotometer was used to take absorbance spectra from 200 nm to 750 nm at 1 nm intervals; measurements were made in triplicate. A Horiba Nanolog fluorimeter was used to measure relative fluorescence intensities. Both the excitation and emission slit widths were set to 2 nm and the spectra were taken from 365 nm to 625 nm in triplicate with excitation at 340 nm.

Crystallization:

A CrystalMation Phoenix (Rigaku) was used to screen a 10 mg/mL solution (10 mM HEPES, pH 5.5, 75 mM NaCl) of both apo and holo L110X, S112X, and W120X against three sitting drop vapor diffusion crystal screening libraries (Hampton Research, Crystal HT, Index HT, and PEG/Ion HT). Each screen contains 96 conditions, and each condition was tested twice (v/v ratios of protein to reservoir drop) for a total of 576 total conditions per protein in 200 and 300 nl drop sizes. Conditions that produced crystals after 5 days were then recapitulated using larger volume sitting drop vapor diffusion. Streptavidin L110X apo and holo crystals were grown with a reservoir solution of 0.1 M Bis-Tris, pH 6.5, 25% w/v polyethylene glycol 3350. Streptavidin S112X holo, and W120X apo crystals were grown with a reservoir solution of 0.1M citric acid, pH 3.5, 3.0 M NaCl. Each drop contained 2 μL of protein solution mixed with 2 μL of reservoir solution, crystals were grown until no new growth was visible (approximately 1 week). Both the S112X and W120X crystals were soaked in a new reservoir solution of 0.1M citric acid, pH 5.5, 3.0 M NaCl. Three 2 μL drop exchanges were performed over the course of 12 hours and the crystals were allowed to soak overnight. The cryoprotectant was 25% PEG 3350 for L110X, S112X and the apo crystal for W120X. 25% glycerol was used as cryoprotectant for the W120X glycerol bound crystal.

Data Collection and Structure Determination.

Diffraction data from the L110X apo and W120X apo crystals were collected at the Berkeley Center for Structural Biology (BCSB) from the Advanced Light Source (beamline 5.0.2) on a Dectris Pilatus3 6M detector. Data from the L110X holo and W120X glycerol bound crystals were collected at the Argonne National Laboratory Advanced Photon Source (beamline 19-ID) on a Dectris Pilatus 6M detector. Diffraction data from the S112X holo crystals were collected at the Stanford Synchrotron Radiation LightSource (beamline BL9–2) on a Dectris Pilatus 6M detector.

Crystals were flash frozen in liquid nitrogen prior to data collection at 100 K. Datasets for the L110X and W120X crystals were indexed, refined, integrated, and scaled using the HKL-3000 software package²⁴. The S112X dataset was indexed, refined and integrated using MosFLM²⁵ and scaled using Pointless.²⁶ Structures were then solved by molecular replacement using Phaser19²⁷ using an all glycine model of SAV with loops removed as the search model (PDB ID: 1swt). All models were refined using Refmac520²⁸ and model building was carried out with the program Coot.²⁹ The chemical description of 7-HCAA was taken from Henderson et al.²⁰ All structural figures were made with the PyMOL molecular graphics software.³⁰ Structures and all supporting data have been deposited in the Protein Databank.

RESULTS AND DISCUSSION

Selection of sites of 7-HCAA incorporation.

Our primary goal in selecting potential sites of 7-HCAA incorporation was to identify residues that were likely to experience distinct chemical environments in the unbound (apo) and biotin-bound (holo) forms of SAV. We began by analyzing a high-resolution (0.95 Å) structure of SAV bound to biotin (PDB ID: 3ry2)³¹ and identified residues in close proximity to the biotin binding site. Because 7-HCAA contains a flexible ethylene linker between the peptide backbone and the 7-HC side chain (Scheme S1), it is difficult to confidently predict the orientation that the 7-HCAA side chain would adopt within the folded SAV. However, we reasoned that substitution of 7-HCAA at sites suitably close to the substrate binding pocket could result in biotin-dependent alterations of the chemical environments in the vicinity of the fNCAA. Importantly, our approach was agnostic with respect to the nature of the changes in fluorescence that might occur in going from apo to holo SAV; rather, we simply hoped to identify sites where a biotin-dependent change in chemical environment seemed possible. Ultimately, we identified five residues surrounding the biotin binding pocket (L25, L110, S112, W120, and L124; Figure 1) that were not directly involved in hydrogen bonding interactions with biotin. Although the side chains of residues 45–50 are in close proximity to the biotin substrate, they fall on the flexible “loop 3/4”, which adopts “open” and “closed” conformations in the absence and presence of biotin, respectively; thus, these residues were excluded from consideration.³²

Figure 1.

Sites of 7-HCAA incorporation in streptavidin. The structure of one subunit of the streptavidin tetramer (Green) and a portion of an adjacent subunit (Orange) are shown in cartoon form (PDB ID: 3ry2). Biotin (BTN; teal) and the side chains of the five residues that were substituted with 7-HCAA (dark green) are shown as sticks. Residue W120 (dark orange sticks) lies on an adjacent streptavidin subunit but interacts with the biotin molecule in the subunit in green. Loop ¾ is shown in the closed position (dark purple). The structure of 7-HCAA is shown in the inset.

Spectroscopic and structural characterization of 7-HCAA containing SAV mutants.

7-HCAA incorporation was achieved using the Amber codon suppression methodology developed by Schultz and co-workers.^3,21 Full-length protein expression was observed for all five mutants only when 7-HCAA was present in the growth media (Figure S2a). Furthermore, fluorescent bands of the correct size were observed for all mutants, which confirmed 7-HCAA incorporation (Figure S2b). Mutant proteins were refolded from inclusion bodies using a modified version of a previously reported protocol²³ and were further purified using immobilized metal ion, anion exchange, and size exclusion chromatography (see Experimental Procedures in the Supplemental Information for full expression and purification and refolding details).

In order to explore the effect of biotin binding on the spectral properties of 7-HCAA, we collected absorbance and fluorescence spectra of each SAV variant in the absence and presence of saturating amounts (100 μM) of biotin. At neutral pH, 7-HC exhibits two absorbance maxima at ~325 nm and ~360 nm that correspond to the neutral and anionic species, respectively. Because the pK_a of 7-HC is ~7.8,¹⁶ both the phenol and phenolate species should have been present under the assay conditions (20 mM Tris-HCl, pH 7.0, 150 mM NaCl) unless the protein environment had altered the apparent pK_a of the 7-HC phenol. All proteins were excited at 340 nm, which corresponds to the isosbestic point of 7-HCAA.

Four of our five mutants exhibited significant biotin-dependent changes in their fluorescence emission spectra. Three mutants (L25X, S112X and W120X) showed increases in fluorescence intensity upon the addition of biotin while the emission intensity of the L110X mutant decreased in the presence of biotin. Additionally, two SAV variants (L110X and W120X) exhibited substantial (>10 nm) blue shifts in their emission maxima in the presence of biotin. In an effort to better understand the origin of the observed alterations in fluorescence, we subjected the four mutants with altered fluorescence spectra to crystallization trials. We were ultimately able to solve the structures of the L110X (apo and bound forms), S112X (bound form) and W120X (apo and glycerol bound forms) mutants. These structures provide insight into how protein environments can tune the properties of 7-HCAA as described below. Analyses of SAV mutants for which structures were not obtained can be found in the supporting information.

L110X Characterization.

The L110X mutant showed a 44% decrease in emission intensity and a 14 nm blue shift in emission maximum (from 449 nm to 435 nm) with the addition of biotin (Table 1, Figure 2c). To better understand the origin of these changes, we solved the crystal structures of the L110X mutant in both the absence and presence of biotin to 1.55 Å and 2.10 Å resolution, respectively.

Table 1.

Spectroscopic data for SAV mutants compared with free 7-HCAA.

Mutant	Abs. Max (apo / holo)	Δ Abs. Intensity (holo - apo)	Em. Max (apo / holo)	Δ Em. Intensity (holo - apo)
7-HCAAα	326	N/Aβ	452	N/ Aβ
L25X	325 / 325	−3%	453 / 455 (+2 nm)	+40%
L110X	324 / 324	12%	449 / 435 (−14 nm)	−44%
S112X	331 / 325 (−6 nm)	10%	451 / 455 (+4 nm)	+72%
W120X	329 / 332 (+3 nm)	−18%	453 /433 (−20 nm)	+135%
L124X	N/Aβ	N/Aβ	450 /450	+6%

^αUnincorporated L-(7-hydroxycoumarin-4-yl)ethylglycine in TBS (25 mM Tris-HCl, pH 7.0, 150 mM NaCl).

^βValues were either not applicable or not obtainable.

Figure 2.

Normalized fluorescence and absorbance spectra of 7-HCAA (a), and the L25X (b), L110X (c), S112X (d), L124X (e), and W120X (f) mutants. Fluorescence spectra of the apo and holo forms of each mutant are shown in black and cyan lines, respectively. The fluorescence intensity of the apo form of each protein was set to unity in each case. Absorption spectra for each mutant are shown in the insets; black and cyan lines again represent the apo and holo forms of SAV, respectively. Studies were carried out at ~10 uM protein with 100 μM biotin at an excitation wavelength of 340 nm; all spectra are the average of three independent measurements.

In the apo form (Figure 3a), 7-HCAA extends into the biotin-binding pocket, and the 7-HC side chain is in van der Waals contact with residues W120 and L124; the C_β atoms of S88 and S112 pack against the opposite face of the coumarin ring. In the holo structure, clear electron density is observed for the biotin molecule (Figure 3b) and alignment of this structure with a previously reported, high resolution structure of biotin-bound SAV (PDB ID: 3ry2) indicated that very little structural rearrangement had occurred due to the 7-HCAA substitution; biotin was observed to adopt its native binding orientation in this mutant (Figure S4). Furthermore, loop 3/4 is in the closed configuration in all subunits of L110X in the biotin-bound structure, which suggests that the placement of 7-HCAA within the binding site does not preclude the large structural rearrangements that occur upon biotin binding.^31,32

Figure 3.

Crystal structures of the 7-HCAA substituted streptavidins. Structural data for the L110X mutant in the apo and holo forms are shown in panels a and b, respectively. An overlay of the apo (green) and holo (orange) forms of the L110X mutant are shown in panel c. Structural data for the W120X mutant in the apo form is shown in panel d. Panel e shows the W120X mutant in the presence of glycerol, which had been used as a cryo-protectant. An overlay of the apo (green) and holo (orange) structures of the W120X mutant is shown in panel f. 7-HCAA, biotin (BTN) and glycerol (GOL) and residues in the vicinity of 7-HCAA are shown as sticks. Electron density around each residue is shown as a 2FO-FC map contoured to 1 σ.

All interactions between 7-HCAA and residues in SAV that are observed in the apo protein are maintained in the biotin-bound form (Figure 3b). This suggests that the close packing interaction between the negatively charged biotin carboxylate and 7-HCAA was likely a primary determinant of the observed alterations in the fluorescence spectrum. Interestingly, in previous work on the “fruit” series of fluorescent proteins, red- and blue-shifted emission maxima were observed as a consequence of altered electrostatic environments in the vicinity of the chromophore.³³ In a related study, red shifted emission maxima were observed for the cyan fluorescent protein amFP486 (whose chromophore is a photoacid like 7-HC) when either of two negatively charged glutamates in close proximity to the chromophore were mutated to neutral glutamine residues.³⁴ Thus, our observations are consistent with previous studies in which negatively charged residues induce blue shifts in emission maxima when in proximity to photoacidic chromophores.

Although these literature precedents provide a potential explanation for the observed blue shift in 7-HCAA fluorescence, the origin of the substantial decrease in emission intensity in the biotin-bound protein is less readily explained. Because the environments experienced by 7-HCAA in the apo and holo forms are so similar, it seems likely that the close packing interactions with biotin are responsible for the observed quenching of fluorescence. The mechanism through which this occurs (e.g., collisional quenching or photoinduced electron transfer) is less clear beyond the scope of this work.

S112X characterization.

The S112X mutant exhibited a 71% increase in fluorescence intensity upon biotin binding (Figure 2d). Furthermore, a notable shoulder at ~380 nm is observed in the emission spectrum of the holo form that could arise from interactions that block ESPT in a population of 7-HCAA. Despite repeated attempts, we were unable to obtain crystals of the S112X mutant in the apo form. However, a structure of the S112X mutant in the holo form was solved to 1.8 Å resolution; both the biotin and 7-HCAA are fully resolved in this structure (Figure S5).

The position biotin adopts in this structure is identical to that observed in the wild type protein (Figure S6), which suggests that the addition of 7-HCAA did not substantially disrupt the native biotin binding interactions. While density for the 7-HCAA is clearly visible in both monomers of the asymmetric unit, it is weaker than the density of the surrounding residues. Furthermore, the 7-HCAA adopts two distinct conformations: In one of the monomers (chain A), the 7-HCAA is involved in a crystal contact via a hydrogen bond between the 7-HC phenol and the backbone carbonyl of Y83 (Figure S7a). In the second subunit (chain B), 7-HCAA adopts a conformation that places its phenol in a reasonably hydrophobic environment generated by the side chains of L124 in the same subunit and L124 on an adjacent subunit (Figure S7b).

Because 7-HCAA in chain A is involved in a crystal contact, it is unlikely that this orientation in present in solution. We therefore cannot draw conclusions regarding the effects of this chemical environment on 7-HCAA fluorescence. In chain B, no solvent molecules are observed within the hydrophobic environment surrounding the 7-HC side chain, which suggests that a neutral form of the phenol would likely be favored. This is consistent with the increase in 325 nm absorbance that occurs as a consequence of biotin binding as well as the slight increase in 380 nm emission that is observed upon biotin-binding. The predominant 450 nm emission that is observed for both the apo and holo forms of the S112X mutant is less readily explained by our crystallographic data. It is therefore likely that the orientation of 7-HCAA in chain A that is adopted in solution is responsible for the observed fluorescence spectra.

W120X characterization.

The emission intensity of W120X increases 135% upon substrate binding and represents the largest biotin-dependent change in fluorescence observed in this study. Similar to the L110X mutant, a blue shift in emission maximum from 453 nm (apo form) to 433 nm (holo form) is evident (Table 1, Figure 2e).

We initially solved a structure of the apo W120X mutant to 1.17 Å resolution at pH 3.5. No significant electron density was observed for the 7-HCAA; however, the presence of positive electron density near residue D128 suggested the possibility that a population of 7-HCAA was interacting with D128 through a hydrogen bond. We rationalized that the low pH of the crystallization conditions may have produced a significant population of protonated D128 that would preclude the formation of this hydrogen bond. We therefore soaked crystals obtained at pH 3.5 in buffers with increasing pHs up to a maximum of 6.0. We ultimately collected diffraction data to 1.55 Å resolution from a crystal that had been soaked to pH 5.5; electron density for 7-HCAA is clearly visible in this structure (Figure 3d). Notably, an anomalously short (< 2.4 Å) hydrogen bond between the phenolic oxygen of 7-HCAA and the carboxylate of residue D128 was observed in both chains of the asymmetric unit (Figure S8). Additional hydrogen bonding interactions were identified between the phenol of 7-HCAA, an ordered water molecule, and the phenol of tyrosine 43 (Figure S8).

Several attempts were made to either co-crystallize W120X with biotin or to soak the substrate into the apo W120X crystals, but these efforts were unsuccessful. However, in a previously solved structure of the W120X mutant (1.50 Å resolution, pH 5.5) where glycerol was used as a cryoprotectant, two glycerol molecules were observed in the biotin binding pocket (Figure 3e). When superimposed with the 3ry2 structure, the bound glycerol molecules directly overlap with a number of atoms in the biotin molecule (Figure S9), which served to displace the 7-HCAA from the biotin binding pocket. This displacement is primarily a consequence of a substantial (116°) rotation about χ₂ of 7-HCAA (Figure 3f). A spectrum of the W120X mutant in the presence of 50% (v/v) glycerol showed an 11% increase in fluorescence at 450 nm, which supports the notion that displacement of 7-HCAA from the binding pocket is primarily responsible for the observed change in fluorescence. Interestingly, the blue shift in emission maximum that was observed after the addition of biotin was not as pronounced in the presence of glycerol (Figure S10). One possible explanation for this is that a functional group on the biotin molecule is responsible for the shift in emission maximum that is observed in holo W120X. Additionally, we note that loop 3/4, which closes over the binding site when biotin binds, is in the open position in the glycerol bound structure. Any interactions between this loop and the 7-HCAA that are present in the holo form of SAV would not be apparent in this structure.

Given these data, it seems likely that the particularly short hydrogen bond observed between the 7-HCAA phenol and D128 in the adjacent SAV monomer (Figure S8) could be responsible for the quenched fluorescence in the apo state. Similarly short hydrogen bonds have been observed between acidic residues and conjugated phenols (including 7-HCAA) in other proteins.^20,35,36 It has previously been suggested²⁰ that a short hydrogen bond between an acidic residue and the 7-HCAA could be responsible for quenching via backward transfer of a proton.^37–41 Because this hydrogen bonding interaction is disrupted in the glycerol-bound structure of W120X, we believe that this interaction is likely responsible for the quenching observed in the apo state.

Affinity of the streptavidin mutants for biotin.

Structural characterization of three of the five SAV mutants indicated that the introduction of 7-HCAA had not caused substantial rearrangement of residues within the biotin binding pocket. However, it seemed likely that the presence of 7-HCAA within the substrate binding site could have altered the affinity of our mutant proteins for biotin. To test this, we measured the apparent dissociation constants for each of the stable SAV mutants (L110X, S112X, and W120X) using changes in the fluorescence intensity of 7-HCAA that occurred upon biotin binding. Because NCAAs are sometimes incorporated with less than 100% efficiency, we subjected each of the variants to mass spectrometric analysis to assess whether or not other amino acids (e.g., tyrosine) had been aberrantly incorporated in any of our variants. As shown in Figure S11, all three mutants gave the expected masses; furthermore, no incorporation of amino acids other than 7-HCAA was observed.

In order to obtain a K_d for the S112X mutant, the increase in emission at 450 nm was plotted against biotin concentration and an apparent K_d was established by non-linear regression (Table S1, Figure 4). For the L110X and W120X mutants, biotin binding resulted not only in a decrease and an increase in fluorescence emission intensity, respectively, but also showed a blue shifted emission maximum. We therefore plotted the ratio of 450 nm to 438 nm emission (the emission maxima of the apo and holo forms, respectively) against biotin concentration and derived apparent K_ds using non-linear regression (Table S1, Figure 4). Apparent K_ds of 568.6 nM for L110X, 1.6 μM for S112X and 6.2 nM for W120X were ultimately measured. Because the reported K_d of SAV for biotin is in the picomolar to high femtomolar range,⁴² it appears that the introduction of 7-HCAA resulted in a substantial loss in affinity relative to the wild type protein in all mutants for which apparent K_ds were determined; however, reasonable affinities for biotin were still maintained. Nevertheless, the fact that a range of affinities was observed among the mutants suggests a direct path to the development of fluorescent sensors of small molecule metabolites that could function over a broad range of concentrations.

Figure 4.

Biotin binding analysis for the L110X (a), S112X (b), and W120X (c) SAV mutants. For the L110X and W120X mutants, the ratio of 450 nm to 438 nm emission is plotted against biotin concentration. For the S112X mutant, the emission intensity at 450 nm is plotted directly against biotin concentration. Some error bars are too small to be visible.

Implications of this study on the design of new fluorescent sensors.

The primary goal of this study was to gain a deeper understanding of how protein environments can alter the fluorescence properties of 7-HCAA. Here, we summarize some of our findings and relate them to future rational design efforts aimed at generating new fluorescent sensors of metabolite binding.

One result of note is the ease with which fluorescent sensors of small molecule binding can apparently be developed using 7-HCAA. We employed a simplistic approach in which residues surrounding a small molecule binding site were mutated to 7-HCAA without further consideration of the local environments surrounding these sites. The fact that substrate-dependent changes to the fluorescence properties of 7-HCAA were observed in four of the five mutants suggests the versatility of this approach. These data suggest that significant flexibility exists with respect to potential sites of 7-HCAA incorporation, and that changes in its fluorescence properties are likely if they are in proximity to the binding pocket. However, the fact that biotin-SAV interactions are among the strongest non-covalent interactions known was likely beneficial to this study; it is unknown whether other proteins with lower affinities would be as amenable to modification as SAV.

Our structural data provide atomic-level portraits of 7-HCAA in a variety of chemically distinct environments that suggest how the fluorescence properties of this fNCAA can be tuned by the surrounding environment. For example, a significant blue-shift in emission maximum in one of our mutants appears to be a consequence of the electronic nature of the environment surrounding the 7-HC moiety. Furthermore, it is notable that interactions between 7-HCAA and the ligand itself can modulate 7-HC fluorescence. This suggests that that interactions between the functional groups contained within the ligand and 7-HC should be considered when identifying potential sites of 7-HCAA incorporation.

We also observed two interactions that appear to favor quenched states of 7-HCAA: In the W120X mutant, we attribute quenching to the presence of a short hydrogen bond. This suggests that it could be sufficient to bias sites of 7-HCAA incorporation to regions of target proteins that are both proximal to the substrate binding site and also contain acidic amino acids. Another apparent mode for 7-HCAA quenching was observed in the L110X mutant, in which van der Waals contacts between the biotin substrate and 7-HCAA side chain appear to have quenched 7-HC fluorescence. While we believe that the presence of a negatively charged carboxylate group on biotin may have caused a blue shift in fluorescence in this mutant, it is possible that neutral portions of other substrates could quench, but not shift 7-HCAA fluorescence if bound in close proximity to the fluorophore.

Finally, in one instance, we observed a conformation of 7-HCAA in which the phenol resides within a hydrophobic environment that at least partially favors a neutral form of 7-HC. Because emission from the protonated form of 7-HC occurs in the UV range (380 nm), this principle could potentially be used to develop fluorescent sensors in which substrate binding is detectable by eye and therefore would not require the use of sophisticated instrumentation.

CONCLUSION

Despite a wealth of fluorescent tools that have proven useful in the study of biological systems, the ability to develop fluorescent sensors of protein-small molecule interactions remains a challenge. Fluorescent non-canonical amino acids possess a number of properties (e.g., small sizes and versatility with respect to sites of incorporation) that may offer advantages for use in these studies. However, a lack of characterization of fNCAAs in proteins has likely limited their use. In this study, we incorporate the fNCAA L-(7-hydroxycoumarin-4-yl)ethylglycine into the biotin-binding protein streptavidin and demonstrate its ability to report on substrate binding events. Major advantages of this method relative to others (e.g., those in which FRET plays a central role) are that only a single fluorophore is required to generate a fluorescent sensor and that no substantial conformational changes are required to sense small molecule binding. Structural characterization of three SAV mutants provided an atomic-level understanding of the manner in which protein environments respond to 7-HCAA incorporation and also reported on the manner in which changes in local environments affect its fluorescence properties. In addition to demonstrating that 7-HCAA is a highly versatile, genetically encodable fluorophore, the spectroscopic and structural characterizations reported herein could help guide future efforts to rationally design new fluorescent sensors of protein-ligand interactions.