Genetic Encoding of 3-Cyano-Tyrosine and Its Use in Controlling the Chromophore Isomeric State of the Fluorescent Protein mKate

Abstract

Switchable β-barrel-type fluorescent proteins are essential genetically encoded probes for super-resolution imaging. The space required for chromophore cis–trans isomerisation can also provide an opportunity to introduce bulkier chemistry at the 3-position of the phenolic ring. Here, we report, to our knowledge, the first successful genetic encoding of 3-cyano-L-tyrosine (3CNY) into a protein. Using genetic code expansion, the cyano-containing tyrosine derivative is incorporated directly into the chromophore of mKate, a pH-dependent switchable red fluorescent protein. While mKate adopts a fluorescent phenolate cis-state chromophore at physiological pH, substituting the native tyrosine with 3CNY yields a functional protein exhibiting hypsochromically shifted spectral properties. Time-dependent density functional theory (TD-DFT) calculations indicate that 3CNY incorporation results in a trans state at pH 8 but, unlike mKate, is fluorescent. The electron-withdrawing cyano group potentially perturbs conjugation across the chromophore, thus lowering the barrier to cis–trans isomerisation. The trans form may also be stabilised by hydrogen bonds from the cyano group to the rest of the protein. Overall, the introduction of a genetically encoded 3-CNY tyrosine analogue into a fluorescent protein chromophore expands our mechanistic understanding and enables the incorporation of a new chemical tag directly into the chromophore.

1. Introduction

Classical β-barrel fluorescent proteins (FPs) represent an important class of proteins due to their role as genetically encodable imaging tags [1,2,3]. The modern repertoire of FPs spans the blue to far red region, performing roles as both passive and active imaging tags [2,4] and even venturing into the realms of nanoscience [5,6,7]. The chromophore is central to FP function and is encoded directly within the amino acid sequence. Three contiguous residues, Xaa-Tyr-Gly (where Xaa is variable), enclosed by the β-barrel, comprise the chromophore. In the presence of O₂, the three residues undergo covalent rearrangement to form the chromophore (Figure 1a), with the phenol group (P ring) of the tyrosine linked to the newly formed imidazolinone (I ring) via a β-methylene bridge [8,9]. An additional oxidation event occurs in red FPs to form an N-acylimine, thus extending the conjugated network into the residue preceding Xaa [10].

The chromophore can exist as either the cis or trans isomers centred around β-methylene C=C (Figure 1a). The majority of FPs sample the cis isomer as their fluorescence permissive state. In a few rare cases such as eqFP611 [12] and its engineered derivatives such as mRuby [13], the chromophore’s fluorescent permissive state is the trans isomer. Other FP chromophores can switch between the two states, with each state having discrete spectral properties. Such switchable FPs have proved to be important probes for high-resolution imaging [14], whereby fluorescent proteins can be “switched on” post-translationally and post-maturation when and where required [15,16]. Switching is most commonly induced photochemically, as is the case for FPs such as Dronpa [17], Kaede [18] and the EosFP family [19]. Photoswitching occurs through the application of certain wavelengths of light. For example, Dronpa switches from its dark state to its fluorescent state upon application of UV–violet light (~400 nm) and converts to the dark state on irradiation with light (488 nm) close to its λ_max (503 nm) [17]. Generally, the cis state is the most red-shifted and brightest form. FPs such as pHlourin [20] and mKate [21] exist predominantly in the cis forms at physiological pH but convert to the trans state under acidic conditions [11]. The mechanism is thought to involve protonation of the phenol hydroxyl group, with the cis form favouring the phenolate and trans the phenolic form. As with the photochemically switchable FPs, the trans state is generally blue-shifted compared to the cis state.

For the chromophore to sample both isomeric states, space within the chromophore pocket must be available to accommodate both forms and allow for rotation of the phenol ring. For example, both isomeric forms are observed in the crystal structure of mKate at pH 4.2 [11]. Ergo, space is available to incorporate new chemical entities at the 3 (meta) position of the P ring through incorporation of tyrosine derivatives using an expanded genetic code [22,23,24]. Genetic code expansion enables new chemistry to be incorporated into a protein, expanding its functional landscape [25]. Many genetically encodable non-canonical amino acids (ncAAs) are based on tyrosine [25] and thus can theoretically be incorporated directly into the FP chromophores. Incorporation of ncAAs with the tyrosyl OH replaced by another chemical entity is relatively common and has been applied to FPs, resulting in new versions with useful spectral properties such as photo-activation [26,27,28,29,30]. However, the phenolic/phenolate (Ph-OH/Ph-O⁻) group transition is critical to the spectral properties of FPs [1,8]; replacing the hydroxyl group with another group (e.g., amino, azide, nitro, cyano, ethynyl) results in a blue-shifted chromophore with poorer spectral properties. Alternatively, the new chemical group can be placed at the 3-position, but this adds steric bulk within the tight confines of the buried chromophore pocket [22,23,31]. While halo-tyrosine ncAAs have been successfully incorporated into FPs [24,32], larger groups such as NO₂ (3-nitro-tyrosine) are more disruptive, especially in terms of the impact on fluorescence properties [24,33]. Boxer and colleagues addressed this problem by using the photo-switchable FP Dronpa2 and a circular permutated version of GFP. They successfully incorporated a range of ncAA-version tyrosine including 3-OCH₃, 3-CH₃, and 3-NO₂ [24]. Others have also successfully incorporated bulky groups at the 3-position, including -NH₂ [34].

The cyano (-C≡N) group, also commonly known as a nitrile group, represents a potentially useful chemical entity to incorporate into the chromophore of FPs, especially for alternative, vibrational-based imaging approaches such as stimulated Raman spectroscopy (SRS) [35,36,37,38,39,40]. Incorporating a cyano group in place of the chromophore’s phenolic OH group is possible through the incorporation of 4-cyano-L-phenylalanine but results in hypsochromically shifted, spectrally compromised FPs [40]. Here, we describe the synthesis of 3-cyano-L-tyrosine (3CNY) and demonstrate its ability to be genetically encoded into proteins, including within the chromophore of the pH-switching RFP mKate. We show that the incorporation of 3CNY is tolerated within mKate’s chromophore and results in a blue-shifted conformationally restricted form of the chromophore that we suggest dominantly occupies the trans isomeric state. Density functional theory indicates that the electron-withdrawing nature of the cyano group could promote the trans conformation, which is potentially stabilised by new H-bonds between the cyano group and the protein.

2. Results and Discussion

2.1. Synthesis and Incorporation of 3CNY

As 3CNY (4 in Figure 2) is not commercially available, we synthesised the ncAA over three steps from commercially available 3-iodo-L-tyrosine (1 in Figure 2). Amino acid 1 was protected by reaction with 9-borabicyclononane (9-BBN) to form boroxazolidinone 2 with an 81% yield [41], which was used in Pd-catalysed cyanation to access protected 3-cyano-L-tyrosine 3 in 42% isolated yield after chromatographic purification via silica gel flash chromatography [42]. Deprotection by treatment with MeOH:CHCl₃ (1:7.5) at room temperature for 16 h afforded the desired 3CNY 4 at 78% yield after concentration in vacuo without the need for further purification [43]. Synthesis of 3CNY 4 was confirmed by ¹H NMR (Supplementary Figure S1), ¹³C NMR (Supplementary Figure S2) and mass spectrometry (Supplementary Figure S3).

To incorporate 3CNY, we first tested an existing amino-acyl tRNA synthase (aaS)/tRNA pair engineered to incorporate 3-nitro-L-tyrosine [22]. Given the general promiscuity of engineered aaS/tRNA pairs [44,45], we surmised that the binding pocket for the nitro group could also accommodate the nitrile group. We used a superfolder GFP [46] variant with the codon for Q204 replaced with TAG, as this position is known to incorporate a wide range of different ncAAs while still retaining function [6,7,47,48]. Functional full-length protein was only produced in the presence of 3CNY, and mass spectroscopy confirmed incorporation (Supplementary Figure S4). The yield of the 3CNY-containing variant was lower than the native sfGFP, as estimated by gel electrophoresis (Supplementary Figure S4).

2.2. Comparison of Experimental Versus DFT-Predicted Spectral Characteristics of mKate

The chromophore of mKate switches between a cis (or Z) isomeric state at physiological pH to the trans (or E) state in acidic conditions. Density functional theory (DFT) can be used to predict the optical properties of FPs [49,50,51,52,53] and inform us of the most likely structure formed by each variant. To test this, we compared experimentally observed electronic excitation data (absorbance spectra) with DFT analysis for the four different forms of the mKate chromophore: cis or trans state in either the phenolic (neutral Ph-OH) or the phenolate (anionic Ph-O⁻) form (see Supplementary Table S1).

At pH 8, the observed λ_max is 588 nm with a shoulder around 550 nm (Figure 3a). The closest wavelength from time-dependent DFT (TD-DFT) corresponds to the cis phenolate form (see Supplementary Table S1), which has predicted wavelengths of 565 nm (major) and 526 nm (minor) (Figure 4b). This tallies with previous observations, including of crystal structures, showing that the cis state is present at higher pH values [11]. The deprotonated phenolate form is generally red-shifted compared the protonated forms as the phenolate is a stronger electron donor that increases the number of resonance structures and so promotes electron delocalisation across the chromophore. The cis-phenolate form is also fluorescent, emitting at a maximal wavelength λ_EM) of 622 nm, a quantum yield (QY) of 33% and brightness of 15.2 mM⁻¹cm⁻¹ (molar absorbance 46 mM⁻¹cm⁻¹) [11,21].

From DFT analysis, the absorbance profile of the deprotonated Ph-O^- mKate chromophore differs between cis and trans: the cis form is significantly red-shifted compared to the trans. In both cases the strongest absorbance band is dominated by HOMO to LUMO frontier orbitals (Figure 3b). Deprotonation of the phenolic group raises the energy of both HOMO and LUMO, but the localisation of the former is on the P ring and the latter on the I ring/N-acylimine system. This is raised markedly more for the P ring (~1 eV) than the I ring/N-acylimine (~0.3 eV), and hence there is a notably lower energy gap for electronic excitation. The shift in electron density from the P to I ring suggests an intramolecular charge transfer (ICT) process. The phenolate P ring acts an electron donor while the I ring/N-acylimine is the electron acceptor forming a “push–pull” conjugated system. The implications are an enhanced conjugation system resulting in a smaller HOMO-LUMO gap and thus a red shift in fluorescence.

At pH2, mKate still absorbs visible light but has no detectable fluorescence. The main absorption peak is broad, blue-shifted with a maximum at 447 nm and has a reduced molar absorbance (25,000 M⁻¹cm⁻¹) compared to pH 8 (Figure 3a). From TD-DFT, the closest predicted λ_max is trans phenolic (trans-Ph-OH in Supplementary Table S1 and Figure 3b), with a value of 455 nm, slightly closer to experimental values than the cis phenolic form (cis-Ph-OH) at 462 nm. Given the evidence available from the crystal structure (Figure 1b [11]) at low pH, the trans-phenolic form is the one most likely sampled. DFT indicates that cis–trans isomerism of the mKate chromophore does not strongly affect the electronic structure or predicted absorption spectrum of the protonated OH form (Supplementary Table S1). Frontier orbitals are very similar in energy, as is the energy gap between them. HOMO frontier orbitals are spread across the conjugated bond system, while LUMO is located more on the I ring and linked N-acylimine. Thus, in the trans-phenolic system, ITC plays less of a role with π-π* transitions becoming more significant.

Given the known limitations of standard functionals and finite basis sets in TD-DFT calculations, as well as the fact that these only used the chromophore, omitting the rest of the protein and water that are known to influence the fine spectral properties of FPs [49,51,54], the agreement with the experiment noted above is likely to have arisen from some error cancellation. Nevertheless, DFT is a good way of assigning different isomeric forms sampled based on comparison with observed spectral properties.

2.3. Incorporation of ncAAs into the Chromophore of mKate

We next assessed the ability of mKate to tolerate the incorporation of 3CNY within the chromophore. As well as the cis-trans and phenolic/phenolate forms, the cyano group can occupy two atropisomeric rotamer forms, termed here the U and D (Figure 1b). Simple modelling was undertaken using the low and high pH crystal structures to ascertain which rotamer forms can be introduced into the chromophore without generating steric clashes. On analysis of the models using MolProbity [55], the cis-D had some potentially significant clashes with L199 (Supplementary Figure S5).

Incorporation of 3CNY in place of the mKate chromophore forming residue Y64 (via amber stop codon reprogramming [31]) generated a functional, fluorescent protein (termed here mKate-CRO-CNY) that is spectrally different compared to mKate (Figure 4a). At pH 8, a major peak is observed at 452 nm (molar absorbance 14,000 M⁻¹cm⁻¹) together with a minor peak at 544 nm (molar absorbance 6000 M⁻¹cm⁻¹). Excitation at 544 nm generated a single fluorescence emission peak at 589 nm (Figure 4a) with a QY of 44%. Based on the TD-DFT analysis, the 544 nm peak is likely due to the trans-phenolate form (trans 3-CN X Ph-O⁻), which has a predicted λ_max of 537 and 538 nm for the D and U forms, respectively. In the trans phenolate form, both HOMO and LUMO are lower in energy compared to mKate, but here the HOMO is reduced slightly more, giving a wider energy gap (Figure 4c), resulting in the blue shift of the main absorbance band. As with the native mKate chromophore, the HOMO electron density is largely centred on the P ring, whereas the LUMO has more I ring/N-acylimine character (Figure 4c), suggesting absorption has a similar ICT nature to mKate. However, the presence of the electron-withdrawing 3-cyano group reduces the density of the P ring in the HOMO (Figure 4c), reducing the donor strength of the phenolate compared to mKate and making it less able to support the delocalised negative charge, resulting in the increased HOMO-LUMO energy gap. The lower electron density on the HOMO P ring may result in a more directional P ring to I ring charge transfer and could make the chromophore more sensitive to solvent effects and protonation.

Excitation at 454 nm results in a dual emission profile with a major emission peak at 480 nm (estimated QY approximately 35%) and a less intense peak at 579 nm (estimated QY approximately 7%). While the origin of the secondary fluorescence emission peak is unknown, we speculate that excited state deprotonation, more commonly known as excited state proton transfer (ESPT), could partially be occurring here [8,56,57]. ESPT is observed in the original GFP from Aequorea victoria, in which the ground state is dominated by the phenolic chromophore but emission is primarily from the phenolate chromophore [1]. This results in a large Stokes shift (~100 nm) between the ground state absorption and excited state emission wavelength. However, it is suggested here that the mKate-CRO-CNY emissive form is chemically similar in part to the ground state (i.e., the phenolic form) due to the major emission peak having a relatively small Stokes shift (33 nm). The 452 nm peak likely results from the protonated, trans-phenolic form (trans 3-CN X Ph-OH), which is based on the TD-DFT predicted wavelengths of 444 nm and 446 nm for the D and U forms, respectively (Supplementary Table S1). Unlike mKate, mKate-CRO-CNY does not show any pH-responsive switching (Supplementary Figure S6). At pH 4, the absorbance spectra is too noisy for any clear peaks to be discerned; fluorescence emission on excitation at either peak absorbance wavelength drops. This suggests that mKate-CRO-CNY is structurally more sensitive to pH than the original mKate; rather than switching chromophore states, lowering pH could lead to significant structural perturbation or even unfolding.

While DFT cannot differentiate between the D and U forms, we propose that the trans 3-CN D form is the most likely to form. Analysis by LigPlot+ [58] suggests that for both rotameric forms, the cyano group will make contact with the rest of the protein (Figure 4b). The U rotamer will potentially make what LigPlot+ classes as covalent bonds with the carboxyl group E145 and the hydroxyl group of Y178. As it is unlikely that such covalent bonds are formed, this suggests that these groups are very close, if not sterically prohibitive, possibly hindering this conformation. In the D form, the cyano group is modelled to make H-bonds with S143 and S158. In mKate, S143 is thought to play a role in stabilising the cis state by forming an H-bond with the phenolate oxygen [11]; if S143 is trapped in a H-bond with the 3-cyano group, this may promote the trans form and hinder ground state deprotonation. Previous structural studies on GFP-type proteins suggest that the cis isomer is formed with the equivalent of the U rotamer observed for the majority of the 3-tyrosine derived variants, including nitro and O-methyl [24,33]; the caveat is that in the cis isomer, the cyano group would occupy a similar spatial position in the U form as it would in the D form for the trans isomer (Supplementary Figure S7).

In the phenolic form, HOMO exhibits minimal electron density across the conjugated backbone, with a drop in energy (Figure 4d) reflecting the loss of phenolate-driven π-donation, while the LUMO remains localised on the I ring/N-acylimine, indicating a transition to an electronically depleted, acceptor-dominated excited state with reduced ICT. Furthermore, in the phenolic cyano-substituted chromophore, loss of phenolate donation and strong electron withdrawal means HOMO is S-centred, such that overlap with LUMO, centred on the I ring/N-acylimine, is weak. Strong absorption therefore stems from excitation from HOMO-1 (−6.50 eV) and HOMO-2 (−6.78 eV) (see Supplementary Figure S8 for frontier molecular orbital map). Under these conditions, the trans isomer is favoured because it maximises residual π-overlap and stabilises the electronic structure, whereas the cis form cannot compensate for geometric distortion between the P and I rings through sufficient π-conjugation. This is borne out experimentally as the P ring shifts out of plane with the I ring less in the trans form crystal structure compared to the cis form [11].

Incorporating a chloro-group at the 3-tyrosine position will still introduce steric bulk compared to the native mKate chromophore as chlorine will project further from the phenol ring (~3.5 Å) than hydrogen (~2.3 Å) but less than the cyano group (~4.1 Å), therefore acting as a good intermediate between H and CN; Cl will have a larger effective radius (~1.75 Å) than both hydrogen (~1.2 Å) and cyano (~1.5 Å), though. The incorporation of 3ClY in place of Y64 (mKate-CRO-3ClY) results in a complex absorbance spectrum (Figure 5a). Two spectral peaks are observed: a major form with a λ_max at 586 nm (e = 24,000 M⁻¹cm⁻¹) and a minor form at 493 nm (e = 10,300 M⁻¹cm⁻¹). The major peak is likely to represent the cis-phenolate form, based on TD-DFT predictions (Figure 5b and Supplementary Table S1). The λ_max lies either side of the DFT predicted value for the U (578 nm) and the D (598 nm), so it is difficult to ascertain whether one or the other is favoured. The slight hypsochromic shift compared to mKate is indicative of the electron-withdrawing nature of chlorine, albeit weaker than the cyano group, which has a relatively weak inductive and resonance action. As with the cis phenolate form of mKate, a similar picture emerges from 3-chloro-substituted cis phenolate (Figure 5c), which also lowers the energy of HOMO and LUMO by the same amount but is predicted to have slight red-shift compared to mKate. Chlorine can donate electron lone pairs to the conjugated system so the P ring retains electron density and acts an effective donor to the I ring/N-acylimine acceptor system in ICT, as is the case for mKate (Figure 3c). Bond orders for mKate as well as CN and Cl variants are reported in Supplementary Table S1, and show that both electron-withdrawing groups reduce conjugation across the β-methylene bridge, and so presumably promote cis–trans isomerisation. We therefore posit that the CN apparently exists solely in trans form due to a combination of this reduced barrier to isomerisation and local interactions of the cyano group, assuming the cis form is the most likely initial state formed on chromophore maturation. S143 also does not interact with the 3-Cl group as it potentially does in the presence of 3-CN, so will not stabilise the trans state and promote the phenolate cis state through H-bonding with the phenol hydroxyl group [11].

The minor peak does not correspond to any obvious predicted peak from TD-DFT analysis (Figure 5b and Supplementary Table S1). However, the excitation at this wavelength does results in fluorescence emission, with two emission peaks observed at 509 nm and 619 nm (Figure 5a). Dual peak emission was also observed for mKate-CRO-3CNY, which was ascribed to trans-Ph-OH form (Figure 3a). The trans-Ph-OH of mKate-CRO-3ClY is the closest predicted species from DFT analysis, but the two differ by 32–33 nm. Excitation of the major peak results in an emission peak at 619 nm, the same as that of the secondary peak observed on excitation at the lower wavelength (Figure 5a).

3. Methods and Materials

3.1. 3CNY Synthesis

To a flame-dried round-bottomed flask fitted with a condenser was added 3-Iodo-L-tyrosine (3.00 g, 9.8 mmol, 1 equiv.) and dry tetrahydrofuran (THF) (45 mL) under N₂. (Scheme 1) The suspension was vigorously stirred for 5 min followed by the addition of 9-Borabicyclo[3.3.1]nonane (9-BBN) (0.5 M in THF, 27.4 mL, 13.7 mmol, 1.4 equiv.) slowly via syringe. The resulting solution was stirred under N₂ at reflux for 3 h. The mixture was cooled and concentrated in vacuo, yielding a yellow foam. The foam was scrapped, washed with hexanes (45 mL), filtrated and dried under vacuum to give the intended compound as an off-white solid (3.4 g, 81%), which was used directly in the next step.

Scheme 1. 3-Iodo-L-tyrosinato-bicyclononylboron (Scheme 1).

Scheme 1. 3-Iodo-L-tyrosinato-bicyclononylboron (Scheme 1).

To 5 separate oven-dried 20 mL microwave vials fitted with rubber septa were added 3-iodo-L-tyrosinato-bicyclononylboron (600 mg, 1.40 mmol), 2-Di-tert-butylphosphino-2′,4′,6′-triisopropylbiphenyl (tBuXPhos) Pd(II) G3 (70.2 mg, 0.08 mmol, 6 mol%) and ZnCN₂ (131 mg, 1.1 mmol, 0.8 equiv.) under N₂. (Scheme 2) The solids were stirred under high vacuum for 15 min. Degassed THF (4 mL) and degassed water (8 mL) were added, and the resulting yellow solution was stirred for at 50 °C for 15 h. Reaction mixtures of the 5 vials were combined, cooled to rt and quenched with a saturated solution of NaHCO₃ and diluted with ethyl acetate (EtOAc) (200 mL). The organic layer was separated, and the aqueous layer was extracted with EtOAc (2 × 200 mL). The organics were dried over MgSO₄, filtered and concentrated in vacuo. Purification by column chromatography (50–70% EtOAc in hexanes, silica gel) gave 3-cyano-L-tyrosinato-bicyclononylboron as a beige solid (962 mg, 42%).

To 9 separate 20 mL microwave vials were added 3-cyano-L-tyrosinato-bicyclononylboron (106 mg, 0.32 mmol) and methanol (MeOH) (1 mL). The suspension was stirred for 30 s and then diluted with CHCl₃ (7.5 mL). The clear solution was stirred at room temperature for 16 h under air. (Scheme 3) The precipitated products of the 9 vials were combined, filtered, washed with CHCl₃ and MeOH and dried under vacuum to yield the pure product as a white solid which darkens on stand (466 mg, 78% yield). ν_max/cm⁻¹ (thin film) 3348, 2492, 1637, 1438; ¹H NMR (500 MHz, DMSO-d₆) δ_H (Supplementary Figure S1): 7.38 (d, J = 2.2 Hz, 1H), 7.28 (dd, J = 8.7, 2.3 Hz, ¹H), 6.96 (d, J = 8.6 Hz, ¹H), 3.48 (dd, J = 6.6, 5.3 Hz, ¹H), 3.36 (d, J = 24.0 Hz, 1H), 2.93 (qd, J = 14.2, 6.0 Hz, 2H); ¹³C NMR (126 MHz, DMSO-d₆) δ_C (Supplementary Figure S2): 169.6, 159.9, 136.0, 133.5, 127.3, 117.6, 116.8, 98.4, 55.1, 35.3; HRMS (ES⁻) [C₁₀H₉N₂O₃] requires [M⁺-H⁺] 205.0613, found 205.0613 (0.0 ppm) (Supplementary Figure S3).

3.2. TD-DFT and DFT Analysis

Chromophore structures were taken from the available crystal structures of the low pH trans state (PDB 3bx9) and high pH cis state (PDB 3bxb) [11]. The 3-cyano and 3-chloro groups were added using the Builder function in PyMOL version 3.1.8 [59]. All DFT analysis employed the Orca 6.1.0 package [60]. Geometry optimisation used the PBE-D3BJ [61,62]/def2-SVP [63] level without any constraint, and minima were confirmed using harmonic frequency calculation. Absorption spectra and frontier molecular orbitals were then calculated at optimal geometry at PBE0 [64]/def2-TZVP-level in the CPCM [65] model of aqueous solution. PBE0 has been used previously to assess the chromophore properties of fluorescent proteins [53,66]. Orbital plots were obtained using Avogadro v 1.2.0.

3.3. Incorporation of ncAA, Recombinant Protein Production and Purification

The gene encoding mKate was present in the pBAD plasmid and was supplied by Addgene Inc (plasmid #54826) [21]. The amber stop codon was introduced in place of the Y64 codon using a Q5 mutagenesis kit by New England Biolab using the forward primer 5′-CATGTAGGGCAGCAA-3′ and reverse primer 5′-AAGCTGGTAGCCAGG-3′ using the manufacturer’s guidelines. 3CNY was incorporated using the 3-nitro-L-tyrosine aminoacyl-tRNA synthase-tRNA cognate pair (pDuleA7; AddGene #174078) [67]. 3ClY was incorporated using the 3-chloro-L-tyrosine aminoacyl tRNA synthase-tRNA pair pEVOL plasmid from Professor Jiangyun Wang at Chinese Academy of Sciences [23]. The sfGFP Q204ncAA mutant resident in the pBAD plasmid has been described previously [47]. E. coli TOP10 cells (Invitrogen/Thermo Fisher Scientific) were transformed with the mKate pBAD plasmid (ampicillin; 100 mg/mL) alone for native mKate production or co-transformed together with either pDuleA7 (tetracycline; 100 mg/mL) or pEVOL (chloramphenicol; 50 mg/mL) onto LB agar plates containing the appropriate antibiotics. Recombinant mKate expression was performed as described previously for other RFPs [53]. The production of sfGFP Q204CNY followed the same procedure as for mKate-CRO-3CNY. Briefly, a single colony was used to inoculate a 5 mL overnight culture, which was then used to inoculate 2xTY media supplemented with 50 μg/mL ampicillin. The cultures were left to grow at 37 °C until they reached an OD₆₀₀ of 0.6, when 0.2% (w/v) arabinose was added to induce expression. The cultures were left to incubate overnight at 37° in a shaking incubator. For ncAA incorporation, an autoinduction medium was used as described previously [67,68,69]; the media was supplemented with 1 mM of 3ClY or 0.1 mM of 3CNY. 3ClY was supplied by Fluorochem.

The production cultures were centrifuged and resuspended in 50 mM Tris HCl pH 8 and the cells lysed using a French pressure cell. Purification was performed with an AKTA Purifier using a 5 mL His Trap™ HP column (Cytiva.Amersham, UK) equilibrated in 50 mM TrisHCl pH8, 10 mM imidazole for Nickel-affinity chromatography. Bound protein was eluted by washing the column with 500 mM imidazole (pH 8.0). Pooled protein samples were then subjected to size exclusion chromatography (SEC), using a Superdex 75/300 increase 10/300 GL (Cytiva, Amersham, UK) that was equilibrated with 50 mM Tris-HCl buffer pH8.0. The purity of the proteins was then checked by SDS-PAGE. Incorporation of 3CNY into the sfGFP Q204 mutant was confirmed by liquid chromatography–mass spectrometry (LCMS) using a Waters Acquity UPLC/Synapt G2-Si QTOF mass spectrometer (Water Limited, Wilmslow, UK) by the Mass Spectrometry facility in the School of Chemistry, Cardiff University.

3.4. Absorbance and Fluorescence Spectroscopy

Absorbance spectra were recorded using an Agilent Cary 600 spectrophotometer (Aglient Techologies, Santa Clara, CA, USA) at 1–5 mM protein in a 1 cm path–length quartz cuvette. Protein concentrations were calculated using the known molar absorbance coefficient for mKate (45,000 M⁻¹cm⁻¹), which was independently verified using a Bradford protein assay. The equivalent value at 280 nm was then determined to be 46,700 M⁻¹cm⁻¹ and applied to the ncAA variants to determine their protein concentration via their A₂₈₀, as has been used previously for other RFPs [53]. Emission and excitation spectra were recorded using a Varian Cary Eclipse Fluorimeter (Aglient Techologies, Santa Clara, CA, USA) at 1–2 mM protein concentration, with the sample loaded into a 400 µL quartz cuvette. Fluorescence emission spectra were recorded using a 5 × 5 mm quartz cuvette, and data were collected with a 5 nm slit width at a rate of 600 nm/min. Each protein was excited at wavelengths stated in the main text. When measured, quantum yields were determined using a comparative method described previously [69]. mCoral [53] was used as the standard for mKate-CRO-3CNY and mCherry [69] for mKate. Buffers used to record spectra at each pH were pH 8, 50 mM TrisHCl; pH 4, 100 mM acetate buffer; pH 2, 100 mM glycine buffer.

4. Conclusions

Switchable fluorescent proteins have classically provided a means to control fluorescence emission post-maturation for super resolution imaging through chromophore cis-trans isomerisation. The space required to accommodate isomerisation can be exploited to incorporate new chemical functionalities within the chromophore [24,34] to tune spectral properties. Ideally, the FP chromophore should retain the phenol group as replacement of the hydroxyl group significantly blue shifts the chromophore’s absorbance [27,28]. Here we report for the first time, to our knowledge, the genetic incorporation of 3-cyano-L-tyrosine into the chromophore of the switchable RFP mKate. Incorporation of 3CNY complements existing 3-tyrosine derivatives such as halides and nitrosyl [24], and adds a potentially useful imaging tag for SRS [38]. It also provides a means to investigate the mechanistic basis of fluorescence and switching.

We show that the 3-CN group potentially traps the chromophore in its trans state through a combination of local interactions with the cyano group and lowering the barrier to cis–trans isomerisation. Sampling the trans state as its default form is very rare for FPs, even those with 3-substituted phenol rings [24,33]. We will need to confirm sampling of the trans state structurally and perform additional mutagenesis to confirm the importance of the new interactions involving the cyano group. However, we speculate that preference for the trans state is partly due to the relatively strong electron-withdrawing nature of the cyano group, the potentially new interactions formed by the cyano group and the prior observation that the trans state has more co-planer P and I rings than the cis state [11]. The stabilisation of the trans state in mKate-CRO-3CNY could also be the reason for it retaining fluorescence emission, which is absent in the original phenolic trans mKate.