Synthesis and Evaluation of Novel Ring‐Strained Noncanonical Amino Acids for Residue‐Specific Bioorthogonal Reactions in Living Cells

Abstract Bioorthogonal reactions are ideally suited to selectively modify proteins in complex environments, even in vivo. Kinetics and product stability of these reactions are crucial parameters to evaluate their usefulness for specific applications. Strain promoted inverse electron demand Diels–Alder cycloadditions (SPIEDAC) between tetrazines and strained alkenes or alkynes are particularly popular, as they allow ultrafast labeling inside cells. In combination with genetic code expansion (GCE)‐a method that allows to incorporate noncanonical amino acids (ncAAs) site‐specifically into proteins in vivo. These reactions enable residue‐specific fluorophore attachment to proteins in living mammalian cells. Several SPIEDAC capable ncAAs have been presented and studied under diverse conditions, revealing different instabilities ranging from educt decomposition to product loss due to β‐elimination. To identify which compounds yield the best labeling inside living mammalian cells has frequently been difficult. In this study we present a) the synthesis of four new SPIEDAC reactive ncAAs that cannot undergo β‐elimination and b) a fluorescence flow cytometry based FRET‐assay to measure reaction kinetics inside living cells. Our results, which at first sight can be seen conflicting with some other studies, capture GCE‐specific experimental conditions, such as long‐term exposure of the ring‐strained ncAA to living cells, that are not taken into account in other assays.


Cloning of constructs
For the expression in mammalian cells, EGFP was cloned with an N-terminal FLAG tag and a C-terminal 6xHis-tag into the commercial pCI plasmid (Promega, E1731), containing an amber stop codon at position Y39, resulting in the plasmid pCI-FLAG-EGFP Y39TAG -6His. As a second reporter plasmid, we are using the pCI-iRFP-EGFP Y39TAG -6His plasmid, as published earlier by us [1] . Plasmid pCMV-NES-PylRS AF -U6tRNArv is carrying the PylRS tRNA-synthetase from Methanosarcina mazei (Mm PylRS AF ), containing the Y306A and the Y384F mutation respectively and an N-terminal NES signal, as well as the tRNA expression cassette in an opposite direction, consisting of an U6 promoter signal and the Methanosarcina mazei tRNA Pyl gene.
For the in vitro studies, we used the construct pALS-Flag-EGFP Y39TAG -6His, which we obtained by cloning FLAG-EGFP Y39TAG -6His into the plasmid pALS-sfGFP N150TAG -MbPyl-tRNA replacing the sfGFP gene. For the expression of the synthetase, we cloned the different PylRS genes into the pBK plasmid [2] , a generous gift from Ryan Mehl. First, we cloned Mm PylRS WT into the pBK plasmid, using the restriction sites NdeI and PstI, after introducing a BglII site into the PylRS gene at bps 870-875, resulting into pBK-PylRS WT plasmid, containing a Kanamycin resistance cassette (Kan). The BglII site was further used together with PstI to clone the specific binding pocket of the Mm PylRS AF mutant (pBK-PylRS AF ) and Mm PylRS AF-A1 mutant into the pBK plasmid (pBK-PylRS AF-A1 ), by replacing this part of the PylRS WT synthetase.
The PylRS synthetase variant AF-A1 was cloned out of the pBK-PylRS AF-A1 plasmid extracted from the synthetase selection procedure and after digesting with BglII and PstI inserted into pCMV-NES-PylRS AF -U6tRNArv, replacing the corresponding nucleotides of PylRS AF .
E. coli expression and protein purification DH10B cells were transfected with both the pALS-Flag-EGFP Y39TAG -6His and the required pBK-PylRS plasmid (PylRS AF or PylRS AF-A1 ). 2 ng of each plasmid was used together with 50 µl of electrocompetent DH10B cells. The electroshock was performed in 1 mm cuvettes using a Bio-Rad Gene Pulser Xcell™ Electroporation Systems. Directly after the electroshock 200 µl of SOC medium were added to the cells. After incubation at 37 °C shaking at 800 rpm for at least 30 minutes, the cells were plated on LB-agar plates containing Tetracycline (Tet, 12.5 µg/ml) and Kanamycin (Kan, 50 µg/ml) as resistance markers. The plates were incubating at 37 °C for at least 16 hours. For the expression, one single colony was incubated overnight shaking at 37 °C in 5 ml LB medium containing the appropriate antibiotics. Next day, 500 µl of this pre-culture were used to inoculate 250-500 ml of LB medium with Tet and Kan and the cultures were incubated again shaking at 37 °C. When the cultures reach an OD600 of 0.2-0-4, the noncanonical amino acids were added in a final concentration of 1 mM. Therefore, each non-canonical amino acid was dissolved in 0.1 mM NaOH to 100 mM stock concentration. At OD600 of 0.4-0.6, the protein expression was induced, using 0.02% Arabinose. The cultures were incubated for another 16 hours, harvested and pellets stored at -20 °C if not directly used for purification.
For the expression of sfGFP N150TAG , the plasmid pALS-sfGFP N150TAG -MbPyl-tRNA was co-transformed with pBK-PylRS AF-A1 into DH10B cells, following the same procedure as above. The expression was carried out the same way as for the EGFP expression.
For the purification cells were resuspended in 2.5-5 ml of 4xPBS, 0.2 mM TCEP, 1 mM PMSF and 5 mM Imidazole pH 8 (lysis buffer) and brought over into 15 ml falcon tubes. After sonication for 30 seconds on ice, cells were centrifuged at 25000 rcf at 4 °C for 30 minutes. The supernatant was incubated in 15 ml Eppendorf tubes on 250-500 µl nickel beads, equilibrated in lysis buffer, for 1-2 hours at 4 °C on a rocker, packed in aluminum foil. The solution was poured into PD10 columns and tubes were cleaned out with 5 ml lysis buffer. Nickel beads were washed with 10 mM Imidazole in lysis buffer for 50 ml and protein was eluted in 5 ml of 4xPBS, 0.2 mM TCEP, 1 mM PMSF and 500 mM Imidazole pH 8. To further purify the samples, a size-exclusion column was used (Superdex S200 Increase 10/300, GE Healthcare), equilibrated in 1xPBS pH 8. Fractions were analyzed by SDS-PAGE and pure proteins were concentrated in an Amicon filter device (3 kDa cutoff, Merck Millipore).

In vitro kinetic studies
Concentrations of purified EGFP Y39ncAA were measured by Absorbance scan (Duetta -Fluorescence and Absorbance Spectrometer, HORIBA) and 10 µM stock solutions of each protein sample was prepared. Kinetic measurements were done with 100 nM of protein in a 1 mm quartz cuvette, diluting the 10 µM stock 1:100 in 1xPBS. The zero point was measured using a fluorescent scan, exciting at 450 nm and measuring emission form 470-700 nm (Duetta, Horiba). Then Cy5-H-tetrazine (Cy5-tet) was added into the cuvette, to obtain a final concentration of 1 µM of dye. The resulting FRET signal was measured with the same fluorescent scan as before, measuring every 5 seconds for 600 seconds long, in the case of the EGFP Y39TCO*-A and EGFP Y39TCO-E . For EGFP Y39SCO and EGFP Y39BCN , the measuring time was elongated to 120 minutes, measured every 60 seconds. For the 10-hour measurement, the data were recorded every 5 minutes. For each in vitro kinetics analysis, 120 single measurements were taken and the maximum value at 500-520 nm and 660-680 nm emission was extracted out of the data to calculate the EFRET value: Measurement of the reactivity and stability of the free ncAA To confirm reactivity of ncAAs they were first analyzed via LC-MS to identify when they elute from the column. Subsequently, equimolar amounts of 1 mM ncAA in MeCN and 1 mM 3,6-dimethyl-1,2,4,5-tetrazine (dimethyl-tetrazine) were mixed and analyzed after five minutes via LC-MS.
ncAA stability was either analyzed directly in LB medium by incubating it at a concentration of 1 mM at 37 °C, shaking at 180 rpm followed by LC-MS analysis at appropriate time points or directly in bacterial cultures. To this end, an overnight culture of E. coli Top10 cells was diluted to an OD600 of 0.1. For the experiment 5 ml cultures in 14 ml test tubes were set up with a concentration of the respective ncAA of 1 mmol/l or without any ncAA as a control. 500 µl samples were taken at the appropriate time points and cells were lysed via sonication on ice in a Bandelin Sonopuls for 2 times 20 sec at 40% power. Samples were spun down (25000 rcf) and the supernatant analyzed via LC-MS. For identification of the ncAAs the extracted ion count (EIC) of the respective mass of the used ncAA was extracted from the obtained mass spectra.

Estimation of incorporation efficiency of different ncAAs
To analyze the incorporation efficiency for each ncAA used in the in-cell FRET studies, we performed transient transfection with different amounts of ncAAs and followed the EGFP signal after 24 hours post transfection. Therefore, HEK293T cells were seeded in a 24-well plate 16 hours prior to transfection. For the two-plasmid transfection (pCI-FLAG-iRFP-25Helix-EGFP Y39TAG -6His and pCMV-NES-PylRS AF -U6tRNArv), 1 µg of total DNA was used per well. The DNA was mixed with 50 µl of DMEM medium without Phenol Red and 3 µl of PEI were added. After 10 seconds of vortexing and a short centrifugation step, the DNA mixture was incubated 15 minutes in the hood, before it was added drop-wise to the well. A master mix was prepared for all wells whenever suitable. After four hours of incubation, the medium was aspirated and fresh medium with different concentrations of ncAAs was added. The concentration of ncAA was ranging from 2 µM up to 500 µM. After 20 hours of incubation, the EGFP signal was analyzed with a flow cytometer analyzer (LSRFortessa™, BD Biosciences) using the green laser with a 530-30 filter. Stock solutions for all ncAAs were prepared as described previously [4] , ncAAs not synthesized by us were purchased from SiChem (SIRIUS FINE CHEMICALS, SICHEM GMBH, Germany). N  -tert-butyloxycarbonyl-L-lysine (Boc, 11) was purchased from Iris Biotech GmbH, Germany) In vivo kinetic measurements (in cellulo FRET) For each ncAA to be tested, three 60 mm petri dishes were seeded 16 hours prior to transfection with 500.000 cells/ml, 3 ml per dish, as mentioned above. In addition, two rows in a 24-well plate were seeded for control experiments. In the morning, cells were transfected with pCI-FLAG-EGFP Y39TAG -6His and pCMV-NES-PylRS AF -U6tRNArv in a 1:1 ratio, using 17 µg of total DNA per 60 mm petri dish. A master mix for three 60 mm petri dishes was set up by using 53 µg total DNA mixed with 1500 µl DMEM medium without Phenol Red and 153 µl PEI. The master mix was vortexed for 10 seconds, centrifuged down for 4-5 seconds and incubated at room temperature (RT) for 15 minutes. The master mix was distributed equally throughout the three 60 mm petri dishes. For the control plate, the master mix contained 6 µg total DNA, 600 µl of medium and 36 µl of PEI, was vortexed for 10 seconds, spun down and also incubated for 15 minutes at RT before it was pipetted drop-wise into 12-wells. The cells were incubated in the incubator for at least 4 hours, then the medium was changed and the appropriate amount of each ncAA was added to the medium. Cells were further incubated for 18 hours in the incubator. Three hours before the in vivo kinetic measurements, the washing procedures was started. Therefore, the medium was aspirated from the petri dishes and the 24-well plate and replaced by fresh medium without any ncAAs. After 30 minutes in the incubator, the cells were washed with DMEM without Phenol Red and incubated another 30 minutes, this washing step was repeated another time and the cells were stored for another hour in the incubator.
The kinetic measurements were performed on a flow cytometry analyzer (LSRFortessa™, BD Biosciences) using the green laser (488 nm) with a 530/30 filter to monitor EGFP signal and a 710/50 filter to capture the FRET signal. The measurements were started with aspirating the medium of the three 60 mm petri dishes from the same ncAA. The cells from each 60 mm petri dish were resuspended in 1.2 ml of the same DMEM medium without Phenol Red, filtered and transferred into test tubes suitable for the LSRFortessa. 1 ml of these cells was used for the labeling reaction, while the rest was used to measure the zero-time point. Three different kinds of SiR-tet concentrations were used for the measurements, ranging from 100 nM to 1 µM of dye. The appropriate amount of dye was mixed with the cells, vortexed briefly and incubated in a cell culture incubator at 37 °C, 5% CO2. The labeling reactions were set up in a 2-minute time interval, to give enough time to measure all three concentrations at a specific time point. Measurements were done after 5, 10, 30, 60, 180, 240 and 300 minutes. If the reaction between the ncAA and the tetrazine was very low, a 360-minute time point was also included as well. After 1 hour and 6 minutes a second batch of cells with another ncAA was labeled with the dye and the measurement was started.
For the measurements on ice, cells were cooled down on ice before labeling and kept on ice after labelling as well as throughout the whole measurements.
To analyze the EFRET-MAX and the kon for each ncAA-tetrazine reaction, data was evaluated with FlowJo™ (Becton, Dickinson & Company). Cells were gated for live cells (FSC vs SSC), followed by single cells gating (FSC vs SSW). After gating for EGFP positive cells, the mean values for the 530 nm and 710 nm signal were extracted, out of which the EFRET for each time point can be calculated as follows: The resulting EFRET values were corrected with the EFRET value at time point zero. The data was plotted in Igor and fitted with an exponential fit: from which the kObs for each concentration curve could be obtained. Plotting the three kObs of each ncAA versus the concentrations used and fitting these data points with a linear fit: one can obtain the kOn of the reaction between this ncAA and the tetrazine, which is the incline of the linear fit.
We note that the concentration of SiR-tet given is the concentration pipetted in the medium. For the reaction to take place, the dye also has to cross the membrane to get into the cell. All those steps are absorbed into the reaction rate constant we observe, as opposed to fitting with more complex models, as other than the dye concentration all other concentration were chosen to be as similar as possible.
The highest EFRET value, which was reached throughout the experiment was defined as EFRET-MAX .

Evolution of PylRS synthetase for incorporation of TCO-E
A NNK synthetic library, which was based on the Methanosarcina mazei PylRS AF variant (Y306A, Y384F) was ordered from GenScript Biotech Corp. with five sites (L305, L309, C348, I405 and W417) mutated to contain any of the 20 amino acids at these positions excluding two stop codons, which leads to 32 possible variants at each of the five sides, resulting in a 3.3 x 10 7 library size. The library was cloned into the selection plasmid, pBK-PylRS AF , as mentioned in the previous section (Cloning of constructs), replacing the binding pocket of PylRS with the library gene, resulting in pBK-PylRS lib . For the selection pREP-pylT was transformed into DH10B cells and highly competent electro competent cells were prepared freshly, using LB medium with Tet to grow the cells up to OD600 of 0.5. After harvesting the cells, they were washed with 10% Glycerol. To resuspend the cells after each harvesting step, the cells should be gently mixed by shaking the harvesting bottles, avoiding any pipetting step. The washing step was repeated two times and finally cells were taken up in as little volume as possible (for 1-liter expression in 1 ml 10% Glycerol) and aliquoted. 100 ng of pBK-PylRS lib were transformed into 50 µl DH10B (pREP-PylT) cells by electroporation in a 1 mm cuvette and 800 µl of SOC medium were directly added. This step was repeated ten times and cells were combined into a 50 ml shaking flask to incubate for 1 hour at 37 °C shaking at 200 rpm. To estimate the library coverage after the transformation, LB-agar plates were prepared, containing Tet and Kan, and 10 µl of the cell suspension were used to make serial dilutions (1:10 2 to 1:10 7 ). On each LB-agar plate 100 µl of dilution was plated and incubated overnight in a 37 °C incubator. The rest of the 10 ml transformation mixture was added into 500 ml 2xYT medium, containing Tet and Kan as antibiotics, and incubated over night at 37 °C on a shaker. To estimate the library coverage, 45 colonies on the 1:10 6 plate were counted, which means 4.5 x 10 9 total amount of cells. This yields in a 140-fold coverage of the library size. Next day, the cells were diluted 1:100 in 500 ml fresh 2xYT medium (Tet, Kan) and grown up to OD600 of 1, which normally took 2-4 hours. For the 1 st positive selection, 10 LB-agar plates (150 mm petri dish) were prepared, containing 1 mM of the TCO-E as well as 60 µg/ml Chloramphenicol (Cm), Tet and Kan. As control, one plate was prepared the same way, just without Cm. The plates were prepared under sterile conditions and cooled down until they were dry. To start the selection, 100 µl of the culture were plated on each 15 mm LB-agar plate, spread by using glass beads and dried next to a flame. The plates were incubated overnight at 37 °C, but not longer than 16 hours. The resulting colonies were scratched from plate with 5 ml 2xYT medium per plate and a cell scraper. The cell suspension of all ten plates was pooled into a 50 ml Erlenmeyer flask and shaken for 1 hour at 37 °C. To isolate the library plasmid, the DNA was extracted first with a Miniprep Kit (Invitrogen) followed by gel extraction. Therefore, the DNA was loaded on a 1% agarose gel and the lowest band was cut. The DNA was isolated by a gel extraction kit (Invitrogen). The plasmid for the negative selection, pYOBB2-PylT, which contains a Barnase gen harboring two amber sites (Gln2 and Asp44), was transformed into DH10B cells and electro competent cells were prepared as mentioned above. 10 ng of gel extracted library plasmid were transformed into 50 µl freshly prepared DH19B (pYOBB2-PylT) cells and recovered after electroporation with 800 µl SOC medium for 1 hour at 37 °C, shaking at 200 rpm in a 14 ml test tube. During the incubation time, the plates for the negative selection were prepared. Six 15 mm petri dishes were casted with LB-Agar, three of them containing 0.2% Arabinose to induce the Barnase expression. All of them containing 50 µg/ml Kan (pBK-PylRS lib plasmid) and 33 µg/ml Cm (pYOBB2-PylT plasmid). 100 µl of the pure cells, as well as 100 µl of a 1:10 and 1:100 dilutions, were plated each on two plates (one with and one without Arabinose). After drying the plates next to a flame, they were incubated overnight in a 37 °C incubator. Colonies were scratched from plates and the DNA was extracted as mentioned above. The positive selection was repeated once more transforming 10 ng of gel extracted library plasmid in 50 µl of DH10B (pREP-PylT), recovering in 800 µl of SOC for 1 hour shaking at 37 °C. This time only three 15 mm LB-agar plates containing 33 µg/ml Cm (Kan, Tet) with 1 mM TCO-E and one control plate without TCO-E were used for the selection. 100 µl of cell suspension were plated and incubated overnight at 37 °C. The surviving colonies were scratched from plate and the library plasmid was extracted out of a 1% agarose gel. To test the amber suppression efficiency of the remaining PylRS variants, an expression test based on superfolder GFP (sfGFP) was performed. Therefore, the library DNA was transformed together with the plasmid pALS-sfGFP N150TAG -MbPyl-tRNA, encoding sfGFP with an amber site at position N150 and the tRNA Pyl from Methanosarcina barkeri into DH10B cells. To include some controls, the pBK-PylRS plasmid, encoding for PylRS WT as well one encoding or PylRS AF were also co-transformed with the pALS plasmid. The transformations were recovered in SOC medium for 1 hour and different amounts (50, 100 and 200 µl) were plated on auto-inducing minimal media plates (Kan, Tet) with 1 mM TCO-E and without ncAA. Colonies were allowed to grow for 24 hours at 37 °C and if needed, another 24 hours at RT. Green colonies were picked and grown overnight in a 96-well plate, containing 480 µl of non-inducing minimal media in each well. The controls (PylRS WT and PylRS AF ) were also picked and included in this 96-well plate. Next day, two 96-well plates were prepared, each with 480 µl of auto-inducing minimal medium per well, one plate with 1 mM TCO-E and one plate without. 20 µl of each overnight culture were pipetted into corresponding wells in the new 96-well plates and incubated for 24 hours at 37 °C shaking at 250 rpm. The sfGFP expression was analyzed with a fluorescent measurement (BIOTEK Synergy 2 Microplate Reader). Therefore, a 96-well plate, containing 180 µl water per well was prepared, in which 20 µl of the sfGFP expression culture was pipetted in the corresponding well. To correct for different OD600 of each expression, also an optical density scan at 600 nm was performed (150 µl water + 50 µl expression culture).
Out of the 96-well plate with the non-inducing minimal media, each well of interest can be analyzed further, e.g. the DNA of the PylRS variant can be extracted to send for sequencing and further subcloning.

Synthetic procedures
General Information for TCO*N and TCO*C-E:  13 C NMR spectra were run at 100 MHz using a proton-decoupled pulse sequence with a d1 of 0 second unless otherwise noted, and are tabulated by observed peak. High-resolution mass spectra were obtained on a Bruker Apex-Qe mass spectrometer using electrospray ionization (ESI).

Continuous flow synthesis of trans-cyclooctenes
To be able to perform efficient photoisomerization of cis-cyclooctenes we have built a continuous flow photoreactor. Inspired by work of Fox [6] and Mikula [7] et. al, we thought to further optimize the existing designs and additionally, improve on the safety features. Based on the previous observations, cooling of the system improves stereo-selectivity of the isomerization reaction [8] .
Enhanced cooling of the reaction mixture is achieved by distinct design features of this reactor: re-packable glass column with a cooling jacket and fan-ventilated light-impermeable aluminum wall of the UV-unit casing (Figure below: In house built photoreactor). Fan ventilation is necessary to dissipate the heat from the lamps and prevent evaporation of the solvents (hexanes, Et2O) during the long runs. Additionally, cooling bath can be applied to the reaction flask. Electrical unit allows independent switching of the low pressure Hg-lamps of variable intensity (2x55 W and 1x95 W) to control the intensity of the UV-light (Figure below, 5). The lamps are positioned closely around the quartz tube to allow for efficient light penetration. This minimizes the space required for the UV-unit and eliminates the need for expansive quartz flasks.
In house built photoreactor: General procedure for the synthesis of trans-cyclooctenes using in-house built photoreactor.
Glass column with a cooling jacket was packed with an oven-dried unmodified SiO2 (bottom; 2 g per mole of the starting material cis-cyclooctene) and 10%AgNO3/SiO2 (top; 2 g per mole of the starting material cis-cyclooctene). The column was equilibrated at 6-10 °C at 100 ml/min with 5:1 Et2O-hexanes for 30 minutes using HPLC pump. Starting material ciscyclooctene (1.0 equiv.) was placed on the ice bath in the 3-neck flask and dissolved in 5:1 Et2O-hexanes (~30 ml per mole of starting material cis-cyclooctene). The continuous flow was achieved between flask, quartz tube and column and the solution was equilibrated at 6-10 °C at 100 ml/min for 20 minutes. Methyl benzoate (2.0 equiv.) was added into the flask and equilibration continued for another 30 minutes while degassing with nitrogen. The quartz tube was then irradiated with UV light under 100 ml/min continuous flow of the reaction mixture (4-8 h). The reaction progress was monitored by the consumption of the starting material using TLC (EtOAc/heptane 2:1). After completion of the reaction the UV light was switched off and the flow continued for another 10 minutes. The column was briefly flushed with 5:1 Et2O-hexanes and Et2O then dried using nitrogen. The content of the column was then transferred into Erlenmeyer`s flask containing NaCl (4 g per mole of the starting material cis-cyclooctene) and ice-water slurry (10 ml per mole of the starting material cis-cyclooctene) and stirred vigorously. Ammonium hydroxide (2.0 equiv.) was added as 25% aqueous solution and resultant heterogeneous mixture was vigorously stirred for 1 hour. The mixture was then filtered and the filter cake was excessively washed with Et2O. Aqueous layer was extracted with Et2O twice and combined ethereal phase was dried over magnesium sulfate, filtered and evaporated to afford corresponding trans-cyclooctene in purity sufficient for the next step.

Synthesis of TCO*N (Z)-9-azabicyclo[6.2.0]dec-6-en-10-one (12) [9]
To a cooled (0 °C, ice bath) stirred solution of cycloocta-1,3-diene (3.0 ml, 24.1 mmol) in 6.0 ml of anhydrous DCM under an inert atmosphere was slowly added chlorosulfonyl isocyanate (CSI, 2.1 ml. 24.1 mmol). Reaction mixture was allowed to warm up and stirred at RT overnight (14 h). Reaction was then diluted with 3 ml of DCM, drawn into a syringe and added slowly into a vigorously stirring mixture of DCM-H2O (40 ml, 1:1) and 5.2 g of Na2SO3. Simultaneously, via another syringe 10% w/w KOH solution was added so that pH was maintained between 6-8. Resultant reaction mixture was vigorously stirred for 2 hours at RT in an opened flask and then transferred into a separatory funnel. Organic layer (as milky suspension) was separated and aqueous layer was extracted with DCM (150 ml x 4). Combined DCM was removed in vacuo and resultant solid was purified by flash column chromatography on SiO2 (EtOAc/heptane, 10 to 100% EtOAc) to provide 3.

N6-((E)-10-oxo-9-azabicyclo[6.2.0]dec-6-ene-9-carbonyl)-L-lysine (7)
13 (137 mg, 0.91 mmol, 1.0 equiv.) was dissolved in dry DCM (0.1 M) and the solution was placed in an ice bath and cooled to 0 °C under an inert atmosphere of argon. Pyridine (184 l, 2.28 mmol, 2.50 equiv.) was added and the reaction mixture was stirred for 5 minutes. 4-Nitrophenylchloroformate (204 mg, 1.01 mmol, 1.1 equiv.) in 1.0 ml of dry DCM was added slowly dropwise, ice bath was removed and the reaction mixture allowed to warm up and stirred at RT for 6 hours. Solvent was removed in vacuo and crude orange oil was dissolved in dry acetonitrile (18 ml) under an inert atmosphere of argon. Fmoclysine hydrochloride (387 mg, 0.96 mmol, 1.05 equiv.) was then added in one portion followed by slow, dropwise addition of DIPEA (476 l, 2.73 mmol, 3.0 equiv.) and reaction mixture was stirred at room temperature overnight (19 h). The reaction mixture was concentrated in vacuo crude residue was well mixed with 3 g SiO2 in small amount of DCM, excess of DCM was removed in vacuo and resultant yellow solid was semi-purified by flash column chromatography on SiO2 (DCM/MeOH, 1 to 11% MeOH) to provide 14 as a white foamy oil which was immediately carried to the next step. 14 was dissolved in 3.30 ml of DMF and piperidine (0.36 ml, 3.67 mmol) was added dropwise. Reaction mixture was stirred at RT for 45 minutes. DMF and excess piperidine were then removed in vacuo and the resultant crude material was purified by reverse phase Synthesis of TCO*C-E:

triethyl-cyclooct-2-en-1-ylmethanetricarboxylate (15)
To a solution of commercially available trans-TCO-OH equatorial alcohol (4.10 mmol, 520 mg) in toluene (10 mL) was added PMe3 (1 M in toluene, 7 ml) and triethyl methanetricarboxylate (6.2 mmol, 1.3 ml) at 0 °C followed by slow addition of DEAD (3.19 ml, 7.0 mmol) as 40% v/v solution in toluene, and the resulting solution was then heated at 90 °C for 3 hours. Toluene was then evaporated in vacuo and resultant orange oil was dry loaded on 3 g of SiO2 and purified by flash column chromatography on SiO2 (EtOAc/heptane, 1 to 20% EtOAc) to provide 1.21 g (86%) of 15 as clear oil. 1  In this reaction we obtained only one stereoisomer. This is most likely the equatorial isomer as for Mitsunobu reactions SN2' type mechanisms have been described that lead to stereochemical [10] . This could be further investigated by obtaining a crystal structure, however as the final amino acid did not exhibit high reaction rate constants we did not investigate this compound further. 13 C NMR:   , 19 mm x 150 mm). The purifications were carried out at a flow rate of 30 mL/min using water and acetonitrile as eluents with 0.2% formic acid as modifier.
Synthesis of AmTCO amino acids:

H-Lys(AmTCO)-OH (24)
Fmoc-Lys(AmTCO)-OMe diastereomers (21) (21 mg, 0.039 mmol, 1 eq) were dissolved in methanol (2 mL). 0.2 M aqueous sodium hydroxide (600 µl, 0.120 mmol, 3 eq) was added. A precipitate formed upon addition but disappeared quickly with stirring. The reaction mixture was stirred at RT overnight. Methanol was removed under reduced pressure and the residue was purified by reverse phase flash chromatography (SNAP Ultra C18 12 g column, gradient from 0 to 30% acetonitrile in water in 15 column volumes). Pure fractions were freeze-dried to yield H-Lys(AmTCO)-OH (24) as a white powder (11 mg, 87% yield, >95% purity by NMR). 1    Each plot contains the data from three different dye concentrations used for the experiment, the highest concentration in cyan, the middle concentration in light blue and the lowest concentration in dark blue. The data points were fitted with an exponential fit using Igor Pro (WaveMetrics), to obtain kObs for each dye concentration. In the case of SCO, the three exponential fits do not reach the same end points, because the reaction between SCO and the dye (SiR-tet) is very slow and therefore, it is impossible to reach full labeling in the measured time frame. Hence the calculated kObs represents an upper limit. For AmTCO-E it was not possible to fit the highest concentration data points, because the reaction is very fast and also shows a fast elimination or quenching mechanism, which does not allow robust fitting. Thus this compound was measured again with cells pre incubated on ice ( Figure S4). g) The kObs, resulting from the kinetic plots, were then plotted over the corresponding concentrations for each compound (AmTCO-E in light green, TCO-E in dark green, BCN in cyan, TCO*-A in blue and SCO in black). For all compounds, except for AmTCO-E, the highest and the middle concentrations were plotted. For AmTCO-E only the lowest and the middle concentration were plotted, because the highest concentration could not be fitted. The data points were fitted with a linear fit to obtain the reaction rate constant (kOn) for each compound. Each plot contains the data from the three different dye concentrations used for the experiment, the highest concentration in cyan, the middle concentration in light blue and the lowest concentration in dark blue. The data points were fitted with an exponential fit using Igor Pro (WaveMetrics), to obtain kObs for each dye concentration. g) The kObs, resulting from the kinetic plots, were then plotted over the corresponding concentrations for each compound (AmTCO-A in green, TCO*N in purple, CpK in light blue, TCO*linker-A in blue and TCO*C-E in magenta). For all compounds, except for AmTCO-A, the highest and the middle concentrations were plotted. For AmTCO-A only the lowest and the middle concentration were plotted, because the highest concentration was too fast to be fitted. The data points were fitted with a linear fit to obtain the reaction rate constant (kOn) for each compound. h) Bar graph showing the measured in vivo reaction rates for the ncAAs. Bar graph showing the measured relative EFRET-MAX values for the ncAAs as well as the EFRET-Final (relative EFRET after 4 hours). These measurements were performed once. Each plot contains the data from the three different dye concentrations used for the experiment, the highest concentration in cyan, the middle concentration in light blue and the lowest concentration in dark blue. The data points were fitted with an exponential fit using Igor Pro (WaveMetrics), to obtain the kObs for each dye concentration. e) Overlay of the FFC data for each compound for three different time points (0 (in black), 30 (in magenta) and 240 (in cyan) minutes). f) The kObs, resulting from the kinetic plots, were then plotted over the corresponding concentrations for each compound. The data points were fitted with a linear fit to obtain the reaction rate constant for each compound, kOn, which is shown in the bar plot (left panel) for each compound. The relative EFRET-MAX for each compound is shown in the bar plot on the right panel. Figure S5. Results of the evolution of PylRS synthetase. a) After sfGFP test expression of the PylRS library in the 96-well plates, the best expressing clones were chosen for sequencing. The sequencing results are shown in the table in comparison to PylRS AF . All PylRS variants show a Leucine residue at position 305 and a Tryptophan at position 417, which are the same for the PylRS AF synthetase. At the three other mutations sites (309, 348 and 405) are mutated to different amino acids. b) The best hits from the screening of the PylRS library after two rounds of selection were cloned into pcDNA3.1 plasmid harboring an U6tRNA expression gene, as well as a CMV promoter for the PylRS expression. The resulting plasmids were co-transfected with the reporter plasmid, pCI-FLAG-iRFP-25Helix-EGFP Y39TAG -6His in HEK293T cells and the expression upon addition of TCO-E was analyzed after 24 hours using FFC. The signals were divided into transfected and untransfected cells, depending on their iRFP signal. The geometric mean (GM) of GFP and iRFP was extracted for the transfected cells signal and the ratio of GM GFP to GM iRFP was calculated. The bar graph shows the different expression ratios and indicates the highest ratio for the PylRS variant AF-A1. Figure S6. TCO-A in cellulo FRET kinetics and in vitro reactivity (sfGFP purified). a) Shown are the raw data of the in cellulo FRET FFC analysis for TCO-A as described in the supplementary methods. In brief, HEK293T cells expressing EGFP with TCO-A were labeled with the indicated concentration of SiR-tet and analyzed via FFC after indicated time points. The FFC plots show FRET signal [A.U.] over the GFP fluorescence signal [A.U.] for different time points (0-300 minutes). b) Kinetic plot containing the data from the three different dye concentrations used for the experiment, the highest concentration in cyan, the middle concentration in light blue and the lowest concentration in dark blue. The data points were fitted with an exponential fit using Igor Pro (WaveMetrics), to obtain kObs for each dye concentration. c) and d) show labeling experiments of sfGFP expressed in E. coli with the TCO-E (c) and TCO-A (d) respectively (upper panels) and the resulting EFRET signal plotted over time for each compound (lower panels), explained more in details in the experimental details. Briefly, 100 nM of the corresponding sfGFP in a cuvette was mixed with 1 µM Cy5-tet, the mixture was excited at 450 nm and the emission measured from 470-700 nm every 5 seconds for 10 minutes. Figure S7. In vitro FRET studies of recombinant EGFP Y39ncAA . Shown are the in vitro FRET analysis of different EGFP Y39ncAA expressed in E. coli with the corresponding ncAA. After purification, EGFP Y39ncAA was labeled with the 1 µM of Cy5-tet and monitored over 10 minutes and 2 or 10 hours (as indicated) in a fluorescence spectrometer (excitation 450 nm, emission detection 470-750 nm). Fluorescence emission spectra are shown in the left column (color code indicates different time points, a spectrum was taken every 5 seconds in the case of the 10 min measurement, every 60 seconds for the 120 minutes measurements and every 5 minutes for the 10 hours measurement). The right column shows the EFRET over time as calculated from the fluorescence spectra. EGFP Y39TCO-E only reaches an EFRET-Max of around 0.3, but therefore shows a fast kinetic. This is due to the trans-to cis-isomerization of TCO-E in LB medium (as Figure S9 illustrates). EGFP Y39TCO*-A reaches a high EFRET-Max of around 0.8, but shows elimination long term (2 hours). EGFP Y39BCN is not reacting with the dye and shows nearly the same EFRET-MAX as EGFP Y39BOC , which is due to its degradation in LB medium (as Figure S9 illustrates). EGFP Y39SCO can reach a high EFRET-MAX of 0.7, but only if monitored over a long time (10 hours). Figure S8. LC-MS analysis of free ncAAs. The ncAAs TCO-A and TCO-E were first analyzed via LC-MS to identy where the elute from the column (top panels). Subesquently, 1 mM TCO-A or TCO-E in MeCN and were mixed with equal amounts of 1 mM dimethyl-tetrazine and analyzed after five minutes via LC-MS (bottom panels). The ncAAs elute as three peaks after reaction with the dimethyl-tetrazine, which all show the expected mass (lower 3 MS plots). The analysis shows fast and complete product formation for both ncAAs as no starting material can be detected after five minutes. Figure S9. IEC-MS analysis of ncAAs in LB medium and E. coli cultures. The plots represent IEC-analysis at different time points for different ncAAs. For the analysis ncAAs were incubated either in LB medium or E. coli cultures and processed for analysis as described in the experimental details. ncAAs in LB medium were analyzed without further processing, while ncAAs in E. coli cultures were subjected to IEC-MS analysis after cell lysis. In the case of TCO-A, TCO-E and TCO*-A, the IEC-MS analysis show two peaks, one for the trans and one for the cis isomer of the corresponding ncAA. Each plot contains also a table, representing the percentage of trans-and cis-isomer at the specific time points. The control experiment for the molecular weight of 299 g/mol (corresponding to the molecular weight of TCO and TCO*) is shown directly after the chromatograms for TCO-A. For BCN it can be shown, that the overall quantity of the compound decreases over time. The control experiment for the molecular weight of 322 g/mol (corresponding to the molecular weight of BCN) is directly shown after the BCN data.
IEC analysis of different ncAAs in E. coli cultures: