Crystal structures of green fluorescent protein with the unnatural amino acid 4-nitro-l-phenylalanine

X-ray crystal structures of superfolder green fluorescent protein with the unnatural amino acid 4-nitro-l-phenylalanine (pNO₂F) independently incorporated at two sites in the protein are reported. These structures support the use of pNO₂F as a relatively nonperturbative spectroscopic reporter of local protein environment.

Abstract

The X-ray crystal structures of two superfolder green fluorescent protein (sfGFP) constructs containing a genetically incorporated spectroscopic reporter unnatural amino acid, 4-nitro-l-phenylalanine (pNO₂F), at two unique sites in the protein have been determined. Amber codon-suppression methodology was used to site-specifically incorporate pNO₂F at a solvent-accessible (Asp133) and a partially buried (Asn149) site in sfGFP. The Asp133pNO₂F sfGFP construct crystallized with two molecules per asymmetric unit in space group P3₂21 and the crystal structure was refined to 2.05 Å resolution. Crystals of Asn149pNO₂F sfGFP contained one molecule of sfGFP per asymmetric unit in space group P4₁22 and the structure was refined to 1.60 Å resolution. The alignment of Asp133pNO₂F or Asn149pNO₂F sfGFP with wild-type sfGFP resulted in small root-mean-square deviations, illustrating that these residues do not significantly alter the protein structure and supporting the use of pNO₂F as an effective spectroscopic reporter of local protein structure and dynamics.

1. Introduction  

The ease of incorporation of a wide variety of unnatural amino acids (UAAs) site-specifically into proteins provides scientists with the opportunity to expand the tools with which they study local protein structure and dynamics. Spectroscopic reporter-containing UAAs represent one subclass of UAAs that can be utilized to study these features in proteins. UAAs containing vibrational reporters, fluorescent groups and NMR probes have all been genetically incorporated into protein systems (Schultz et al., 2006 ▸; Jackson et al., 2007 ▸; Cellitti et al., 2008 ▸; Miyake-Stoner et al., 2010 ▸; Ye et al., 2010 ▸; Smith et al., 2011 ▸; Thielges et al., 2011 ▸; Bazewicz et al., 2012 ▸, 2013 ▸; Adhikary et al., 2014 ▸; Ma et al., 2015 ▸; Tookmanian, Fenlon et al., 2015 ▸; Tookmanian, Phillips-Piro et al., 2015 ▸; Dippel et al., 2016 ▸; Hartley et al., 2016 ▸; Slocum & Webb, 2016 ▸). In order to be effective probes of protein structure, the probes must have both a specific spectroscopic signal sensitive to the local environment and be minimally perturbative to the local protein environment under investigation.

The spectroscopic reporter-containing UAA 4-nitro-l-phenylalanine (pNO₂F; Fig. 1 ▸) has been shown to act as both a fluorescence quencher (Taki et al., 2002 ▸; Tsao et al., 2006 ▸) and a vibrational reporter (Smith et al., 2011 ▸). Two earlier studies incorporating pNO₂F into either streptavidin or the leucine zipper illustrated that pNO₂F acts as a quencher of tryptophan fluorescence in a distance-dependent manner (Taki et al., 2002 ▸; Tsao et al., 2006 ▸). Studies using difference Fourier-transform infrared (FTIR) spectroscopy with isotopically labeled p¹⁵NO₂F-containing and naturally abundant p¹⁴NO₂F-containing superfolder green fluorescent protein (sfGFP) allowed the elimination of vibrational absorptions other than the nitro group and allowed the location of the nitro symmetric stretching frequency to be observed (Smith et al., 2011 ▸) for both the Asp133pNO₂F and Asn149pNO₂F sfGFP constructs. Once identified, the specific nitro symmetric stretching frequency was correlated with reference spectra of p¹⁴NO₂F in aqueous solvent or tetrahydrofuran (THF), which mimicked solvated and buried sites on proteins, respectively. These studies were completed on Asp133pNO₂F sfGFP and Asn149pNO₂F sfGFP (or sites 134 and 150 according to the numbering used in Smith et al., 2011 ▸) and correlate with the structures that we report here.

Figure 1.

The structures of two spectroscopic reporter UAAs: 4-nitro-l-phenyl­alanine (pNO₂F) and 4-cyano-l-phenylalanine (pCNF).

While crystal structures have been reported for some spectroscopic reporter UAAs genetically incorporated into proteins (Tookmanian et al., 2015 ▸; Dippel et al., 2016 ▸; Kearney et al., 2018 ▸), to date no structures that contain a genetically incorporated pNO₂F residue have been reported in the Protein Data Bank (PDB; Berman et al., 2000 ▸). Earlier work with another vibrational reporter, 4-cyano-l-phenylalanine (pCNF; Fig. 1 ▸), illustrated that the incorporation of pCNF had a minimal impact on the structures of either GFP or heme nitric oxide and/or oxygen-binding protein (H-NOX; Dippel et al., 2016 ▸; Kearney et al., 2018 ▸). The lack of significant UAA-induced structural perturbation is foundational to the effectiveness of a specific UAA as a tool to study protein structure. Here, we report the crystal structures of Asp133pNO₂F sfGFP and Asn149pNO₂F sfGFP, the first structures in the PDB with the UAA pNO₂F genetically incorporated into a protein via the amber codon-suppression methodology.

2. Materials and methods  

2.1. Macromolecule production  

The preparation of pBAD plasmids containing the gene for superfolder GFP (sfGFP) with a TAG site incorporated at position 133 or 149 has been described previously (Smith et al., 2011 ▸; Dippel et al., 2016 ▸) and the pDULE_pNO₂F plasmid containing the engineered, orthogonal tRNA synthetase selected for pNO₂F was used as described previously (generated by Ryan Mehl, Oregon State University; Smith et al., 2011 ▸; Table 1 ▸). Both plasmids were co-transformed into Escherichia coli DH10B cells. Starter cultures of 5 ml of non-inducing medium were inoculated with the co-transformed cells and were grown overnight at 37°C with shaking at 250 rev min⁻¹. After 15–20 h, 250 ml of autoinduction medium was inoculated with 2.5 ml starter culture and pNO₂F (Peptech) was added to a final concentration of 1 mM (Studier, 2005 ▸; Hammill et al., 2007 ▸). Following incubation at 37°C with shaking at 250 rev min⁻¹ for 24–30 h, the cells were collected using centrifugation and stored at −80°C. To obtain the purified protein, the cells were thawed on ice, sonicated and the supernatant was added to TALON resin similar to previous procedures (Miyake-Stoner et al., 2010 ▸; Bazewicz et al., 2012 ▸; Tookmanian, Phillips-Piro et al., 2015 ▸; Dippel et al., 2016 ▸) to obtain the histidine-tagged construct (Table 1 ▸). Following elution from the TALON resin, the protein was desalted over a PD10 column into 20 mM HEPES pH 7.5 and the histidine tag was removed by adding trypsin protease as described previously (Pédelacq et al., 2006 ▸; Dippel et al., 2016 ▸). Following the removal of the histidine tag, the protein was concentrated and stored in 20 mM HEPES pH 7.5 buffer. Both protein constructs had a vibrant green colour resulting from a properly formed chromophore.

Table 1. Macromolecule-production information.

	Asp133pNO2F sfGFP	Asn149pNO2F sfGFP
Source organism	Aequorea victoria	Aequorea victoria
Expression vectors†	pBAD_sfGFP_D133TAG and pDULE_pNO2F	pBAD_sfGFP_N149TAG and pDULE_pNO2F
Expression host	E. coli DH10B	E. coli DH10B
Complete amino-acid sequence of the construct produced‡	MVSKGEELFTGVVPILVELDGDVNGHKFSVRGEGEGDATNGKLTLKFICTTGKLPVPWPTLVTTL(CRO)VQCFSRYPDHMKRHDFFKSAMPEGYVQERTISFKDDGTYKTRAEVKFEGDTLVNRIELKGIDFKE(PPN)GNILGHKLEYNFNSHNVYITADKQKNGIKANFKIRHNVEDGSVQLADHYQQNTPIGDGPVLLPDNHYLSTQSVLSKDPNEKRDHMVLLEFVTAAGITHGMDELYKGSHHHHHH	MVSKGEELFTGVVPILVELDGDVNGHKFSVRGEGEGDATNGKLTLKFICTTGKLPVPWPTLVTTL(CRO)VQCFSRYPDHMKRHDFFKSAMPEGYVQERTISFKDDGTYKTRAEVKFEGDTLVNRIELKGIDFKEDGNILGHKLEYNFNSH(PPN)VYITADKQKNGIKANFKIRHNVEDGSVQLADHYQQNTPIGDGPVLLPDNHYLSTQSVLSKDPNEKRDHMVLLEFVTAAGITHGMDELYKGSHHHHHH

^†Expression vectors have been described in previous publications (Smith et al., 2011 ▸; Dippel et al., 2016 ▸).

^‡These are the full construct sequences that were expressed and purified. Before crystallization the protein was treated with trypsin, removing the last eight amino acids (GSHHHHHH). The second amino acid in the sequence, valine, is an added amino acid resulting from the cloning vector used. We have numbered the residues to be consistent with previous crystal structures, resulting in the first methionine being residue 0 and the valine residue 1. CRO refers to the post-translationally formed chromophore consisting of residues Thr65-Tyr66-Gly67. PPN refers to the translationally incorporated 4-nitro-L-phenylalanine (pNO₂F) residue.

2.2. Crystallization  

The purified pNO₂F mutants of sfGFP were concentrated to between 35 and 45 mg ml⁻¹ and screened against a number of crystal screens in 96-well sitting-drop crystal trays. These initial crystallization screens contained 50 µl of a precipitation condition per reservoir and the sitting drops consisted of 1 µl protein solution and 1 µl precipitation solution. Initial crystallization hits were optimized in Cryschem sitting-drop plates and the final crystallization conditions are summarized in Table 2 ▸.

Table 2. Crystallization.

	Asp133pNO2F sfGFP	Asn149pNO2F sfGFP
Method	Sitting-drop vapor diffusion	Sitting-drop vapor diffusion
Plate type	Cryschem sitting-drop plate	Cryschem sitting-drop plate
Temperature	Room temperature (∼296 K)	Room temperature (∼296 K)
Protein concentration (mg ml−1)	36	42
Buffer composition of protein solution	20 mM HEPES pH 7.5	20 mM HEPES pH 7.5
Composition of reservoir solution	0.2 M ammonium citrate tribasic pH 7.0, 18% PEG 3350	1.2 M malic acid
Volume and ratio of drop	1 µl protein solution and 1 µl reservoir solution	1 µl protein solution and 1 µl reservoir solution
Volume of reservoir (µl)	500	500

2.3. Data collection and processing  

X-ray diffraction data were collected on the Northeastern Collaborative Access Team (NE-CAT) beamlines (either 24-ID-C or 24-ID-E) at the Advanced Photon Source (APS), Argonne National Laboratory (Table 3 ▸). Data for Asp133pNO₂F sfGFP were processed in space group P3₂21 to 2.05 Å resolution using HKL-2000 (Otwinowski & Minor, 1997 ▸). The data for Asn149pNO₂F sfGFP were processed in space group P4₁22 to 1.60 Å resolution using HKL-2000 (Otwinowski & Minor, 1997 ▸).

Table 3. Data-collection and processing statistics for Asp133pNO₂F sfGFP and Asn149pNO₂F sfGFP.

Values in parentheses are for the outer shell.

	Asp133pNO2F sfGFP	Asn149pNO2F sfGFP
Diffraction source	24-ID-C, APS	24-ID-E, APS
Wavelength (Å)	0.979	0.979
Temperature (K)	100	100
Detector	Pilatus 6MF	Quantum 315 CCD
Crystal-to-detector distance (mm)	300	200
Rotation range per image (°)	0.2	1.0
Total rotation range (°)	120	180
Exposure time per image (s)	0.2	1.0
Space group	P3221	P4122
a, b, c (Å)	97.72, 97.72, 154.87	55.30, 55.30, 166.53
α, β, γ (°)	90, 90, 120	90, 90, 90
Mosaicity (°)	0.30	0.15
Resolution range (Å)	50–2.05 (2.09–2.05)	50–1.60 (1.63–1.60)
No. of unique reflections	53525 (2616)	35157 (1706)
Completeness (%)	98.5 (97.7)	99.4 (98.8)
Multiplicity	6.6 (6.2)	13.9 (13.6)
〈I/σ(I)〉	21.36 (1.56)†	36.83 (2.33)
CC1/2	99.5 (74.5)	99.3 (82.0)
R r.i.m. ‡	3.2 (41.1)	2.8 (30.8)
Overall B factor from Wilson plot (Å2)	37.45	21.13

^†The decision to cut the data at 2.05 Å resolution was made because of the high CC_1/2, multiplicity and completeness even though 〈I/σ(I)〉 < 2 at this resolution. The data were collected on a PILATUS 6MF detector and each frame resulted from a 0.2 s exposure, resulting in lower intensities but also less environmental noise. 〈I/σ(I)〉 > 2 at resolutions below 2.12 Å.

^‡ R _r.i.m. is the redundancy-independent merging R factor.

2.4. Structure solution and refinement  

Initial molecular replacement was completed using the wild-type sfGFP model (PDB entry 2b3p; Pédelacq et al., 2006 ▸) with solvent molecules removed, B factors set to 20 Ų and the residue at the location of pNO₂F incorporation mutated to an alanine (D133A or N149A). Molecular replace­ment was completed using Phaser (Storoni et al., 2004 ▸; McCoy et al., 2007 ▸) in PHENIX (Adams et al., 2010 ▸) for both structures. Molecular replacement with the Asp133pNO₂F sfGFP data identified two molecules in the asymmetric unit with a translation-function Z-score (TFZ score) of 7777.4 and a log-likelihood gain (LLG) of 88.4. By contrast, the Asn149pNO₂F sfGFP structure contained only one molecule in the asymmetric unit, and the TFZ score and LLG following molecular replacement in Phaser were 5381.9 and 71.7, respectively.

Following molecular replacement, the Asp133pNO₂F sfGFP structure was automatically refined in PHENIX (Adams et al., 2010 ▸) alternating with manual refinement in Coot (Emsley & Cowtan, 2004 ▸). The initial 2F _o − F _c electron density and F _o − F _c electron density illustrated the presence of the pNO₂F residue (Fig. 2 ▸ b). After the replacement of D133A with the pNO₂F residue and further refinement, water molecules, a sodium ion and ethylene glycol molecules were added, resulting in a final R and R _free of 18.50 and 21.89%, respectively (Table 4 ▸, Figs. 2 ▸ a and 2 ▸ c).

Figure 2.

Crystal structure of Asp133pNO₂F sfGFP (PDB entry 6dq0). The overall structure is shown in cartoon representation in (a) with the Asp133pNO₂F residue and chromophore shown in stick representation. The initial 2F _o − F _c map at 1.0σ and F _o − F _c map at +3.0σ and −3.0σ around the D133A site are shown in (b) in blue, green and red mesh, respectively. (c) The final 2F _o − F _c map at 1.0σ is shown in blue mesh around the pNO₂F modeled at the Asp133 site. Figures were generated in PyMOL (Schrödinger).

Table 4. Structure solution and refinement of the Asp133pNO₂F sfGFP and Asn149pNO₂F sfGFP crystal structures.

	Asp133pNO2F sfGFP	Asn149pNO2F sfGFP
Resolution range (Å)	50–2.05	50–1.60
Final R cryst (%)	18.80	16.05
Final R free (5% of data) (%)	21.89	19.05
R.m.s. deviations
Bonds (Å)	0.010	0.009
Angles (°)	1.026	1.005
No. of atoms
Protein	3685	1931
4-Nitro-L-phenylalanine (PPN)	28	14
Sodium ion(s)	1	1
Ethylene glycol	16
Water	313	272
Overall	4015	2204
Average B factors for non-H atoms (Å2)
Protein	43.37	23.88
4-Nitro-L-phenylalanine (PPN)	60.96	19.41
Sodium ion(s)	45.10	31.63
Ethylene glycol	54.16
Water	49.32	34.89
Overall	43.87	25.24
Ramachandran plot
Most favored (%)	97.05	99.11
Allowed (%)	2.95	0.89

The first round of refinement in PHENIX (Adams et al., 2010 ▸) following molecular replacement with the N149A sfGFP model resulted in clear electron density for the pNO₂F residue (Fig. 3 ▸ b). The pNO₂F residue was added to this site, replacing N149A, and the refinement continued, alternating automatic refinement in PHENIX and manual refinement in Coot (Emsley & Cowtan, 2004 ▸). The parameter files for the chromophore (ligand ID CRO) and pNO₂F (ligand ID PPN) were refined from the starting parameters available in the Protein Data Bank in Grade (Smart et al., 2011 ▸). Following initial adjustments to the backbone and side chains, water molecules, a sodium ion and eventually riding H atoms were added to the 1.60 Å resolution structure. The final R and R _free for the Asn149pNO₂F sfGFP structure were 16.05 and 19.05%, respectively (Table 4 ▸, Figs. 3 ▸ a and 3 ▸ c).

Figure 3.

Crystal structure of Asn149pNO₂F sfGFP (PDB entry 6dq1). The overall structure is shown in cartoon representation in (a) with the Asn149pNO₂F residue and chromophore shown in stick representation. The initial 2F _o − F _c map at 1.0σ and F _o − F _c map at +3.0σ and −3.0σ around the N149A site are shown in (b) in blue, green and red mesh, respectively. (c) The final 2F _o − F _c map at 1.0σ is shown in blue mesh around the pNO₂F modeled at the Asn149 site. Figures were generated in PyMOL (Schrödinger).

3. Results and discussion  

The initial electron densities in the areas of the D133A or N149A sites in the sfGFP crystal structures reported here illustrate that the 4-nitro-l-phenylalanine was successfully site-specifically incorporated into sfGFP (Figs. 2 ▸ b and 3 ▸ b). In addition, the electron density for the refined pNO₂F residues incorporated at either the Asp133 or Asn149 sites illustrate good 2F _o − F _c electron density for the full residue including the NO₂ group, suggesting that the pNO₂F did not undergo photodegredation during protein expression, purification, crystallization and data collection (Figs. 2 ▸ c and 3 ▸ c).

To accurately report on the local environment and/or dynamics in a protein, a reporter UAA must not cause any significant structural perturbation. The Asp133pNO₂F sfGFP structure was aligned with both wild-type sfGFP [root-mean-square deviation (r.m.s.d.) of 0.229 Å for 195 C^α atoms] and Asp133pCNF sfGFP (PDB entry 5dpg; r.m.s.d. of 0.189 Å for 192 C^α atoms; Dippel et al., 2016 ▸), illustrating minimal structural alteration upon pNO₂F incorporation at the Asp133 site (Fig. 4 ▸ a). The phenyl ring in Asp133pNO₂F sfGFP is in the same orientation as the phenyl ring in the Asp133pCNF sfGFP crystal structure, rotated approximately 180° from the side chain of Asp133 in wild-type sfGFP (Fig. 4 ▸ a). This side-chain rotation is not concerning as the backbone aligns well with the wild-type sfGFP Asp133 residue and its location on a solvent-accessible loop suggests that in solution any residue at site 133 (including Asp133 in wild-type sfGFP) is likely to be experiencing a number of solvent-exposed conformations. Finally, the location of Asp133pNO₂F is consistent with previously published IR results suggesting that this site is solvated (Smith et al., 2011 ▸).

Figure 4.

Structure alignments of pNO₂F-containing crystal structures with wild-type sfGFP and the corresponding pCNF-containing crystal structure. In (a) the Asp133pNO₂F sfGFP structure (PDB entry 6dq0) is shown in green with the Asp133pCNF sfGFP structure (PDB entry 5dpg) in orange and wild-type sfGFP (PDB entry 2b3p) in gray. The residues at the Asp133 site for each of these structures are shown in stick representation and are colored by atom type. In (b) the Asn149pNO₂F sfGFP structure (PDB entry 6dq1) is shown in pink with the Asn149pCNF sfGFP structure (PDB entry 5dph) in teal and wild-type sfGFP (PDB entry 2b3p) in gray. The residues at the Asn149 site in each of these structures are shown in stick representation and are coloured by atom type. Figures were generated in PyMOL (Schrödinger).

The Asn149pNO₂F sfGFP structure was aligned with wild-type sfGFP (r.m.s.d. of 0.180 Å for 190 C^α atoms) and Asn149pCNF sfGFP (PDB entry 5dph; r.m.s.d. of 0.218 Å for 198 C^α atoms; Fig. 4 ▸ b). These alignments illustrated that the backbone, C^β, C^γ and δ position atoms overlaid well for the Asn149pNO₂F, Ala149pCNF and Asn149 residues (Fig. 4 ▸ b). The Asn149pNO₂F residue is partially buried from bulk solvent by neighbouring side chains along the side of the β-barrel, which is consistent with the observed IR frequency for the nitro stretching frequency reported for Asn149pNO₂F sfGFP (Smith et al., 2011 ▸).

The two structures reported here represent the first reported X-ray crystal structures with pNO₂F incorporated into a protein structure using the amber codon-suppression technology. Previous structures containing the pNO₂F residue contained pNO₂F on a synthesized short peptide fragment (4–8 residues; Rose et al., 1996 ▸; Lizak et al., 2011 ▸; Aleem et al., 2016 ▸; Napiórkowska et al., 2017 ▸) or a semi-synthetic protein in which the pNO₂F was synthesized in a peptide that was later ligated to an expressed protein fragment (Baril et al., 2017 ▸). The Asp133pNO₂F sfGFP and Asn149pNO₂F sfGFP crystal structures illustrate clear density for the pNO₂F residue at the specified position, indicating that pNO₂F can be site-specifically incorporated to produce a stable pNO₂F group. Further, these structures illustrate that the incorporation of pNO₂F at the Asp133 or Asn149 sites in sfGFP does not cause a significant structural change, supporting the use of pNO₂F as a spectroscopic probe for various protein systems.

Supplementary Material

PDB reference: sfGFP, with Asp133 mutated to 4-nitro-l-phenylalanine, 6dq0

PDB reference: with Asn149 mutated to 4-nitro-l-phenylalanine, 6dq1

Funding Statement

This work was funded by Camille and Henry Dreyfus Foundation grant TH-15-009 to Scott H. Brewer. National Institute of General Medical Sciences grant P41 GM103403. National Institutes of Health, National Center of Research Resources grant S10RR029205. U.S. Department of Energy, Office of Science grant DE-AC02-06CH11357.