Protein Spin Labeling with a Photocaged Nitroxide Using Diels–Alder Chemistry

Abstract EPR spectroscopy of diamagnetic bio‐macromolecules is based on site‐directed spin labeling (SDSL). Herein, a novel labeling strategy for proteins is presented. A nitroxide‐based spin label has been developed and synthesized that can be ligated to proteins by an inverse‐electron‐demand Diels–Alder (DAinv) cycloaddition to genetically encoded noncanonical amino acids. The nitroxide moiety is shielded by a photoremovable protecting group with an attached tetra(ethylene glycol) unit to achieve water solubility. SDSL is demonstrated on two model proteins with the photoactivatable nitroxide for DAinv reaction (PaNDA) label. The strategy features high reaction rates, combined with high selectivity, and the possibility to deprotect the nitroxide in Escherichia coli lysate.


Experimental Procedures
Synthesis of the PaNDA spin label (1)

General methods
Technical solvents were distilled prior to use. Dry solvents were either purchased from Sigma Aldrich, Acros or were dried and distilled. For degassing of dry solvents, the freeze-pump-thaw method was applied. Deuterated solvents for NMR spectroscopy were purchased from Deutero.
All reactions were monitored by Thin Layer Chromatography (TLC) with silica gel 60 F254 coated on aluminum sheets from Merck. UV active compounds were detected at 254 nm. Additionally, different staining solutions followed by gentle heating were used for the visualization of the reactants (the composition of the solvents is stated as a ratio of volumes (v/v)):  Anisaldehyde solution: EtOH (150 mL), acetic acid (15 mL), conc. H2SO4 (5 mL), p-methoxybenzaldehyde (3.7 mL)  Ninhydrin solution: EtOH (200 mL), acetic acid (3 mL), ninhydrin (0.2 g)  Potassium permanganate solution: 0.1 % KMnO4 in 1 M NaOH For the preparative Flash Chromatography (FC) silica gel 60 (Geduran Si 60, 0.040-0.063 mm particle size) from Merck was used. Solvent mixtures are specified as volume ratio (v/v) and all solvents were distilled prior to usage. Additionally, FC was performed on a MPLC-Reveleris X2 system from Grace.
Nuclear magnetic resonance (NMR) spectra were recorded on spectrometers Avance III 600 MHz and on Avance III 400 MHz from Bruker at room temperature. The resonance signals of different deuterated solvents were used as internal standards: CDCl3 (δH = 7.26 ppm, δC = 77.16 ppm), DMSO-d6 (δH = 2.50 ppm, δC = 39.5 ppm). In addition to first-order analysis, 1 H, 1 H homo-and 1 H, 13 C hetero nuclear two-dimensional correlation spectra like HSQC, COSY, TOCSY and HMBC were recorded for the assignment of signals. 1 JH-1/C-1 couplings were determined from non-decoupled HSQC and HMBC spectra. The multiplicities of the resonances are abbreviated as followed: s (singlet), d (doublet), dd (doublet of doublets), ddd (doublet of doublets of doublets), t (triplet), dt (doublet of triplets), td (triplet of doublets), m (multiplet). The recorded NMR spectra were analyzed by using the software MestReNova v12.0.0-20080 by Mestrelab Research. NMR spectra can be found under the "Results and Discussion" section in Figure S1-S10 of this Supporting Information.
High resolution masses were measured on a micrOTOF II instrument from Bruker in positive mode. Electrospray was used as ionization method (ESI) and the time of flight (TOF) method was used for detection. The recorded mass spectra were analyzed by using the software Xcalibur v3.0 by Thermo Fischer Scientific.

Plasmids and transformation of E. coli
The plasmids pEVOL_PylRS_AF and pBAD_TRX_His6_R74TAG or pBAD_TRX_His6_G34TAG_R74TAG, respectively, were used for the expression of E. coli thioredoxin (TRX-R74→ncAA or TRX-G34/R74→ncAA). For expression of TRX wildtype (TRX wt) only the pBAD_TRX_His6 plasmid was transformed into E. coli. The plasmids pEVOL_PylRS_AF and pBAD-Flag-GFP-Y39TAG-6His were used for the expression of the GFP mutant, which contains the ncAA at amino acid residue 39 instead of tyrosine (GFP-Y39→ncAA). For expression of the GFP wildtype (GFP wt) only the pBAD_Flag-GFPwt-6His plasmid was transformed into E. coli. The plasmids were co-transformed into chemically competent BL21-gold (DE3) E. coli as follows. The reaction tube containing E. coli and an appropriate amount of the particular plasmid(s) was mixed by flicking and incubated 30 min on ice. The cells were then heat shocked at 42 °C for 30 sec and incubated for another 2 min on ice, before being added to 1 mL pre-warmed (37 °C) Super Optimal Broth with catabolite repression (SOC-medium). The cells were then incubated for 1 h at 37 °C and 1400 rpm, before grown on an LB-Agar plate (Lennox; ROTH) containing 34 μg/mL chloramphenicol (ROTH) and 50 μg/mL carbenicillin (ROTH) overnight.

Expression and purification of TRX and GFP and incorporation of genetically encoded artificial amino acids
For all steps performed in medium, transformed E. coli strains were grown in LB-medium (Lennox; ROTH), containing 34 μg/mL chloramphenicol and 50 μg/mL carbenicillin (or only carbenicillin for expression of wildtype proteins). Cells were shaken at 37 °C and 180 rpm. TRX-R74→ncAA, TRX-G34R74→ncAA, TRX wildtype, GFP-Y39→ncAA or GFP wildtype were expressed in E. coli cotransformed with the respective plasmid(s) (see above) as follows. For overnight cultures, 10 mL LB-medium were inoculated with one colony of the particular E. coli from the agar plate. The next day, the overnight culture was diluted 1/100 (typically to a final volume of 1 L), and incubated until an OD600 of 0.2 -0.3 was reached. At this point, 1 mM SCO-L-lysine (2a) or TCO-L-lysine (3a, both ncAA were bought from SICHEM) were added from freshly prepared stock solutions (therefore, the ncAA were dissolved in 0.1 M NaOH in 60 mM or 80 mM for 2a or 3a, respectively). Cells were further grown until an OD600 of 0.4 -0.6 was reached. Protein expression was induced for 4 -6 h with 0.2 % L-arabinose (ROTH) from a 20 % w/v stock solution. For wildtype proteins, only L-arabinose but no ncAA was used for the expression. Expression was stopped and cells were harvested by centrifugation (4 °C, 4000 rpm, 10 min). The supernatant was discarded, and the pellets were stored at -20 °C, until proteins were isolated and purified. Proteins were purified using HisPur Ni-NTA resin (Thermo Fisher Scientific) as described elsewhere. [4] Samples were dialyzed in Slide-A-Lyzer MINI Dialysis Devices (3.5 K MWCO, Thermo Fisher Scientific) against PBS buffer, pH 7.4 (MERCK), at 4 °C. Resulting protein concentration was determined photometrically with the use of an Eppendorf BioPhotometer D30 via absorption at 280 nm (with a Factor Fp = 1/A0.1% = 0.651 g/L for TRX or 1.35475 g/L for GFP; A0.1% = εp/MMp is the absorbance of the protein at 0.1 %, εp is the molar extinction coefficient of the protein, and MMp is the relative molar mass of the protein). Integration of the ncAA and purity were confirmed by SDS-PAGE. Briefly, samples for SDS-PAGE analysis and 5 μL of a BIO-RAD Precision Plus ProteinTM Dual Color Standard were applied to a 15 % SDS-gel and run at 90 V in a BIO-RAD Mini-PROTEAN Tetra System. After Coomassie Blue staining (Brillant Blau R 250, ROTH), gels were imaged using a BIO-RAD ChemiDocTM Imaging System (Figure S11-S13). For additional proof for expression of the correct product and for assessing fidelity of the aminoacyl-tRNA-synthetase, fulllength ESI-MS spectra were recorded (Figure S16-S17).

Site-directed spin labeling of TRX and GFP with the PaNDA spin label
The labeling protocol was based on a reaction mixture of 200 μL of 40 μM protein (in PBS, pH 7.4) containing a 10-fold molar excess of PaNDA spin label (from a 10 mM stock solution in DMSO, which was stored at -20 °C). The mixture was incubated for 30 min at 20 °C and 0 rpm in an Eppendorf ThermoMixer C. The excess reactants were then removed by using HisPur Ni-NTA resin (Thermo Fisher Scientific). 120 µL of bead slurry were washed three times with MQ-H2O in a centrifuge tube, before it was incubated together with the labeling mixture for 30 min at 4 °C under constant rotation. In a centrifuge column (Pierce Centrifuge Columns, 0.8 mL, Thermo Fisher Scientific) the beads with adherent proteins were washed approx. ten times with washing buffer (50 mM NaH2PO4, 300 mM NaCl, pH 8). For elution of the spin-labeled proteins, the beads were incubated for 20 min with 60 µL of washing buffer containing 500 mM imidazole, before proteins were collected by centrifugation. All wash steps and elution were performed in a benchtop centrifuge. Wildtype proteins were also incubated with the PaNDA spin label as described above to exclude unspecific labeling ( Figure  S18-S19). Full-length ESI-MS of spin labeled proteins (after irradiation) proves the conversion of the ncAA-containing proteins into labeled ones (Figure S18-S19).

Deprotection of the PaNDA spin label by irradiation
A typical sample of 30 μL was filled into a glass capillary (HIRSCHMANN® ringcaps®; inner diameter 1.02 mm), and sealed with tube sealing compound (Kimble Cha-Seal) on one end. The capillary was placed on a table, and covered with a 302 nm handheld UV lamp (UVLM-28 EL Series UV Lamp, 8 Watt, 302/365 nm, analytikjena). Samples were irradiated for 2 min (or longer, if indicated; Figure S21). The sample in the capillary was directly used for subsequent EPR spectrometric measurements.

EPR measurements
EPR spectra were recorded at a BRUKER EMXnano X-band continuous wave EPR spectrometer at room temperature (approx. 22 °C). A typical sample volume of 30 μL was filled into a glass capillary (HIRSCHMANN® ringcaps®; inner diameter 1.02 mm). Spectra were recorded at a modulation amplitude of 1 G, microwave attenuation 15 dB, and a sweep width of 150 G. Typically, 20 scans of 60.06 sec scan time each were accumulated to improve the signal-to-noise ratio. Quantitative spin concentrations of samples were obtained with the use of the built-in EMXnano reference-free spin counting module (Xenon software, Bruker). Spectra were plotted with MATLAB R2018a (The MatWorks, Inc. 3 Apple Hill Drive, Natick, MA 01760-2098, USA).

Full-length mass spectrometry of proteins
Before being subjected to mass spectrometry, the buffer of the protein samples was replaced by MQ-water in 3K spin filters (Amicon Ultra-0.5 mL Centrifugal Filters, MERCK). Protein masses were recorded by an amaZon speed ETD mass spectrometer (Bruker) with a flow rate of 4 µl/min at the Proteomics Facility of the University of Konstanz. Mass spectrometric data were evaluated using the Data Analysis Version 4.4 (Bruker) software (Figure S16-S19).

Measurement of circular dichroism (CD) spectra
Before CD measurements, the buffer of the protein samples was replaced by MQ-water using 3K spin filters. CD spectra were recorded in a JASCO J-715 Spectropolarimeter. Spectra were recorded at room temperature, using a 0.5 mm cuvette. Ten scans each were accumulated to improve signal-to-noise levels. 1000 data points were received between 280 and 180 nm. The data were baseline-corrected, and subsequently background-corrected with a sample containing only MQ-water. Noisy data at low and high wavelengths were cut off for the final presentation of the data. For the calculation of the molar residue ellipticity out of the given spectrometer unit (CD-signal in mdeg), the formula MRE = [MRW*CDsignal/1000]/(10*d*c) with M in g/mol, Naa (number of amino acid residues in the protein), MRW = M/Naa, c in g/mL, and d = 0.05 cm was used ( Figure S20).

Experiments with E. coli lysate
This section describes the experiment shown in Figure 4 (main text); EPR spectral raw data of every time point are given in Figure S22. For the preparation of E. coli lysate 5 mL of E. coli overnight culture (carrying pBAD_TRX_His6 plasmids, but without expression) were prepared. After 17 hours cells were harvested by centrifugation for 10 min at 4 °C and 4000 rpm, before the supernatant was discarded. Cells were lysed with 200 µL of B-PER reagent (Thermo Fisher Scientific) containing 1 mM PMSF, and incubated for 10 min on ice with regular vortexing. The cell lysate was cleared by centrifugation for 2 min at 4 °C and 14000 rcf, and the resulting supernatant was stored on ice and further used as reducing environment for experiments. In the first part of the experiment a TRX-R74→2b sample was irradiated and a spin concentration of 89 µM was determined. Then 15 µL of this protein were mixed with 15 µL of the E. coli lysate, and EPR spectra were measured every ten min. For these experiments three scans at each time point were accumulated. The resulting plot of the spin concentration against the time showed the expected degradation kinetics of nitroxides in reducing environments. In the second part of this experiment, 15 µL of the protein, which was not irradiated in advance, were mixed with 15 µL of E. coli lysate, and EPR spectra were measured every twenty min. After 80 min the sample was irradiated and EPR spectra were measured every ten min to see first an increase followed by degradation of the signal (Figure 4).

S16
Full-length ESI-MS spectra (Figures S16-S19) Figure S16. Full-length ESI-MS spectra of GFP samples after expression in E. coli and purification. A depicts the wildtype GFP, while B and C show the spectrum of GFP-Y39→3a or GFP-Y39→2a, respectively. The respective mass assigned to the desired main product is underlined. A B C S18 Figure S18. Full-length ESI-MS spectra of GFP samples incubated with the PaNDA spin label, then purified and irradiated. Spectrum A shows the peaks yielded after incubation of GFP wildtype with the PaNDA spin label, while B and C show the spectra after spin-labeling of the ncAA-containing GFP mutants. The respective mass assigned to the desired main product is underlined.
A 28709 Da: GFP wildtype (compare Figure S16 The asterisk (*) indicates the elimination of the oxygen from the nitroxide, leading to the corresponding amine, due to mass spectrometry-induced fragmentation. This effect was also seen for the unbound spin label in solution (see Figure 2 B).
In the spectrum in the middle (GFP-Y39→3a spin labeled with the PanDA spin label) no peak can be assigned to successfully spin labeled protein (GFP-Y39→3b). However, in the EPR spectrum ( Figure 3) successful labeling was evidenced.
A B C S19 Figure S19. Full-length ESI-MS spectra of TRX samples incubated with the PaNDA spin label, then purified and irradiated. Spectrum A shows the peaks yielded after incubation of TRX wildtype with the PaNDA spin label, while B and C show the spectra after spin-labeling of the ncAA-containing TRX mutants. The respective mass assigned to the desired main product is underlined. The asterisk (*) indicates the elimination of the oxygen from the nitroxide, leading to the corresponding amine, due to mass spectrometry-induced fragmentation. This effect was also seen for the unbound spin label in solution (see . CD analysis of GFP (left) and TRX (right). Wildtype proteins (blue line) were measured after expression (without labeling), while the red and black lines refer to labeled proteins. All samples were desalted before measurement, and spectra were baseline-and background-corrected. The qualitative spectral shape is found to be unchanged. Quantitative deviations upon labeling are due to limited accuracy of the protein concentration determination.  Figure S22. EPR spectra of the experiment shown in Figure 4 (main text). All spectra are drawn on the same y-axis range. The spectrum "10 min after irradiation" shown in the 2 nd column on the right was measured with 25 dB microwave attenuation instead of 15 dB, leading to a higher signalto-noise ratio (SNR).