Experimental procedures for the synthesis of PazK can be found in the supporting information.

4-Azido-L-phenylalanine (AzF, CAS: 33173-53-4, Santa Cruz Biotechnology) and (S)-2-amino-6-[(2-azidoethoxy)carbonylamino]hexanoic acid (AzeoK, CAS: 1994331-17-7, Iris Biotech) were prepared as 100 mM stock solutions in 200 mM NaOH and 15% DMSO (w/v), and used at the final concentrations indicated.

The constructs for expression of Mma PylT/RS wild type (RRID: Addgene_140009) and AF (RRID: Addgene_140023) variants as well as the sfGFP150TAG reporter constructs (RRID: Addgene_154766) were described previously (Meineke et al., 2018, 2020). We generated analogous constructs for AzFRS with four repeats of Bst TyrT^CUA (RRID: Addgene_140018 and Addgene_174891). The plasmids share a common architecture and are here collectively referred to as “pAS” (Amber Suppression) plasmids: the aaRS, reporter or gene of interest are controlled by an EF1 promoter and followed by an IRES that allows expression of a downstream selection marker. A cassette with four tandem repeats of the tRNA gene, controlled by 7SK Pol III promoter, is placed upstream of the EF1 promoter in antisense orientation. All DNA constructs were verified by Sanger sequencing.

HEK293T cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM, GlutaMAX^TM, Thermo) supplemented with 10% (v/v) FBS at 37°C and 5% CO₂ atmosphere. For transient transfection, 1.5–2.0 × 10⁵ cells/ml were seeded 24 h before transfection with TransIT-LT1 (Mirus) according to the instructions of the manufacturer. ncAAs were added at the time of transfection, and cells were harvested after 24 h.

A modified transfection protocol was used for larger-scale GFP expression for bead purification, increasing the amount of total DNA to 6 μg (Mma PylT/RS AF and PylT/sfGFP150TAG at 2 + 8 ratio) per ml culture and transfecting 5.0–8.0 × 10⁵ cells/ml with 2 μg of polyethylenimine (PEI) per μg of DNA. PazK was supplemented to 0.5 mM at transfection and until harvest after 6 days. Cells were lysed in RIPA buffer supplemented with 1× cOmplete protease inhibitor (Roche). The insoluble fraction was removed by centrifugation. Expressed GFP was captured on GFP-Trap_MA magnetic beads (ChromoTEK), washed with RIPA buffer and PBS, and eluted in 1% (v/v) acetic acid.

Purified GFP samples were desalted and rebuffered into 100 mM ammonium acetate, pH 7.5, using ZebaSpin columns with a 7-kDa cutoff (Thermo). Samples were directly infused into an Orbitrap Fusion Tribrid mass spectrometer equipped with an offline nanospray source using borosilicate capillaries (Thermo). The capillary voltage was 1.5 kV, and the pressure in the ion-routing multipole was maintained at 0.11 torr. Spectra were acquired in the Orbitrap mass analyzer operated in high mass mode at a resolution of 60.000 between 1,000 and 4,000 m/z. Data were analyzed using Excalibur (Thermo).

GFP-expressing HEK293T cells were imaged in a ZOE Fluorescent Cell Imager (BioRad).

HEK293T cells were transfected, cultured in the presence of 0.25 mM ncAA for 24 h and lysed in RIPA buffer with 1× cOmplete protease inhibitor (Roche). The insoluble fraction was removed by centrifugation. CuAAC was carried out on equal volume aliquots in 1 mM CuSO₄, 1 mM TCEP, 100 µM THPTA, and 1 µM AF647 dye (AF647-Alkyne or AF647-Picolyl Azide (Jena Bioscience)) for 1 h at 24°C, 450 rpm followed by incubation at 4°C overnight. Samples were separated on 4%–20% Tris-glycine gels (BioRad) and exposed for in-gel fluorescence at 630 nm in a GE AI600 imager and further analyzed by Western blot.

Transfected HEK293T cells were grown in the presence of 0.25 mM PazK or 0.25 mM AzeoK for 24 h. Cells were washed with PBS and labeled with 5 µM AF647-alkyne dye (Jena Bioscience), 10–50 µM CuSO₄, 50–250 µM THPTA in 2.5 mM ascorbic acid (from a freshly prepared 100 mM stock) for 10 min at room temperature (Hong et al., 2010). Cells were collected in cold PBS, spun down, and lysed in PBS 0.1% (v/v) triton X-100 supplemented with 1× cOmplete protease inhibitor (Roche). Aliquots were separated on 4%–20% Tris-glycine gels (BioRad) and exposed for in-gel fluorescence at 460 and 630 nm in a GE AI600 imager and further analyzed by Western blot.

Transfected HEK293T were grown on poly-L-lysine-coated 18-well imaging slides (Ibidi) in the presence of 0.25 mM PazK or AzeoK for 24 h. Cells were washed with PBS and labeled with 5 µM alkyne dye (AFdye 647 alkyne, Jena Bioscience) in 50 µM CuSO₄, 250 µM THPTA, and 2.5 mM ascorbic acid for 10 min at room temperature. Subsequently, the cells were washed with PBS, counterstained with 2 µM Hoechst33342 (Life Technologies) in PBS for 30 min, washed again, and fixed in 4% formaldehyde for 10 min. The cells were washed and imaged in PBS on a Nikon Eclipse Ti2 inverted widefield microscope, using a ×20 (0.75 NA) objective and filter sets for DAPI and Cy5 fluorescence.

Aliquots of cell lysates were separated on 4–20% Tris-glycine gels (BioRad) and transferred to nitrocellulose membranes. Expression of GFP reporter and FLAG-aaRS was confirmed by immunoblotting with antibodies against GFP (Santa Cruz, RRID:AB_627695), HA-HRP (Roche, RRID:AB_390917), FLAG-HRP (Sigma, RRID:AB_439702), GAPDH (Millipore, RRID:AB_10615768), and corresponding secondary HRP-conjugated antibodies when needed (BioRad, RRID:AB_11125936 and Invitrogen, RRID:AB_2534727). Quantitative analysis of gel lanes was performed using ImageJ software.

To synthesize picolyl azide-lysine (PazK), two building blocks were required, lysine derivative 3 and azide 7 (Figure 1). The synthesis of 3 commenced with orthogonally protected Nα-Boc-Nε-Cbz-Lysine 1. After methylation under standard conditions, hydrogenation afforded 3 (Schnell et al., 2020), which contains the free side chain amine, in 83% yield over two steps. To access azide 7, dipicolinic acid dimethyl ester 4 was selectively reduced with NaBH₄ to alcohol 5 in 65% yield. Installation of the requisite azide functionality was effected using a one-pot process where the hydroxyl group of 5 was converted into the corresponding alkyl bromide (PPh₃ and CBr₄), followed by displacement with sodium azide without isolation of the bromide intermediate. Subsequent ester hydrolysis under basic conditions afforded carboxylic acid 7 (Hanna et al., 2017). With building blocks 3 and 7 in hand, amide formation was performed using standard conditions (EDCI, HOBt, DIPEA in DMF). Lithium hydroxide-mediated ester hydrolysis, followed by Boc deprotection under acidic conditions afforded the desired PazK, as the hydrochloride salt.

Next, we needed to establish that PazK can be accepted as a substrate for tRNA aminoacylation by a tRNA^CUA/aaRS pair orthogonal in mammalian cells. We used a GFP reporter with an amber codon at position 150, allowing the use of fluorescence as a readout for incorporation efficiency. If the ncAA added to the medium is accepted by the aaRS to aminoacylate the cognate tRNA^CUA, the amber stop codon is suppressed, and full-length fluorescent GFP bearing PazK at position 150 (GFP^150PazK) is produced (Figure 2A).

We tested the incorporation of PazK by Mma PylRS and its variant with mutations Y306A and Y384F: Mma PylRS AF (Yanagisawa et al., 2008). Wild-type Mma PylRS can accommodate a variety of ncAA substrates in its active site, but the Pyl binding pocket cannot accommodate large or bulky lysine adducts. The Mma PylRS AF mutant has been rationally designed to enlarge the ncAA binding pocket (Yanagisawa et al., 2008; Yanagisawa et al., 2019) and has enabled incorporation of lysine derivatives with aromatic and larger hydrocarbon rings (Borrmann et al., 2012; Nikić et al., 2014; Ge et al., 2016).

The Mma PylT/RS pairs were cotransfected with a PylT/sfGFP150TAG amber suppression reporter in HEK293T cells. We assayed AzeoK (Figure 2B) and PazK against a control with no ncAA, which showed no GFP fluorescence; AzeoK is an excellent substrate for wild-type Mma PylRS (Meineke et al., 2020) and, as expected, produced strong GFP fluorescence. PazK only yielded low GFP fluorescence with the same wild-type PylT/RS-transfected cells (Figure 2C). By adding AzeoK and PazK to Mma PylT/RS AF-expressing cells, we observed similar GFP fluorescence levels for both ncAAs (Figure 2C, right). For comparison, we also tested the incorporation of AzF (Figure 2B) and 6-azido-lysine (6AzK) (Supplementary Figure S1). AzF was efficiently incorporated in Bst TyrT^CUA/AzFRS-expressing cells as judged by GFP fluorescence (Figure 2C), while 6AzK was not a substrate for Mma PylT/RS (Supplementary Figure S1). We further tested Methanogenic archaeon ISO4-G1 (G1) PylT/RS and G1 PylT/RS^Y125A pairs (Meineke et al., 2020) and found that wild-type G1 PylRS accepted PazK with low efficiency, but G1 PylRS^Y125A showed high incorporation efficiency for PazK (Supplementary Figure S2). Hence, we conclude that azide-bearing ncAAs can be incorporated well in mammalian cells with existing tRNA/aaRS pairs.

We further sought to confirm the selective incorporation and chemical stability of PazK in a target protein. Hence, we purified sfGFP^150PazK from HEK293T cells transfected with PylT/sfGFP150TAG and Mma PylT/RS AF and performed intact mass spectrometry. The calculated mass of 27,089 Da and determined mass of 27,090.2 Da were in agreement, confirming PazK incorporation and the stability of the picolyl-azide moiety in the cellular environment (Figure 2D).

To compare CuAAC labeling of the three azide-containing ncAAs, AzeoK, PazK, and AzF, we reacted GFP^150ncAA with fluorescent AF647-alkyne in HEK293T cell lysates after transient transfection of amber-suppressor tRNA/aaRS, using Mma PylT/RS AF for AzeoK and PazK and Bst TyrT/AzFRS for AzF (Figure 3A). In agreement with fluorescent imaging, anti-GFP Western blot confirmed the efficient incorporation of all three ncAAs, in the order AzF > AzeoK > PazK under the conditions used. For assessing the specificity of CuAAC reaction for the three ncAAs, we reacted a fluorescent dye, AF647-alkyne, via CuAAC in whole-cell lysate. Here, we chose traditional in vitro conditions with excess alkyne dye, high concentration of copper salt (1 mM), 100 µM THPTA, and long reaction time (1 h at RT followed by overnight incubation at 4°C) to reach a reaction end point. CuAAC AF647-alkyne yielded a single band corresponding to the size of GFP visible with in-gel fluorescence imaging at 630 nm (Figure 3A). No other bands are observed, confirming that all ncAA are orthogonal to (i.e., not incorporated by) the endogenous complement of aaRS enzymes. In principle, stoichiometric labeling should be observed under the given reaction conditions for the three azide-modified GFP proteins. However, despite the lower amount of total GFP produced, the signal for AF647-labeled GFP was strongest for PazK and weakest for AzeoK (corresponding to a roughly 7.5-fold higher AF647/GFP ratio for PazK compared to AzeoK) (Figure 3A). These results confirm that AzeoK and AzF are more efficiently incorporated, but suggest that the incorporated PazK has a higher CuAAC reactivity. There are several potential explanations for this observation: terminal azides can undergo reduction to amines, and aromatic azides are known to be photolabile; hence, some of the AzeoK and AzF azide moieties may have been eliminated in cellulo or upon lysis (Milles et al., 2012). On the other hand, natural Cu(I) chelating molecules in the crude lysate and reoxidation of Cu(I) to Cu(II) with atmospheric oxygen may deplete Cu(I) available for CuAAC under elongated reaction conditions. As an additional control, we performed an SPAAC reaction with dibenzocyclooctyne (DBCO)-TAMRA fluorescent dye in lysates of all the three azide-bearing GFP species and again observed an improved reactivity of PazK over AzeoK and AzF (Supplementary Figure S3). This further hinted at the decomposition of the AzeoK and AzF azide moieties in cellulo or upon cell lysis. In summary, these results, together with the intact mass (Figure 2D) suggest that PazK is favorably stable and reactive compared with other available azide ncAAs. Of note, SPAAC labeling with DBCO showed less specific labeling of the azide-bearing GFP and a number of background bands, in line with prior reports that SPAAC reactions are not strictly bioorthogonal due to side reactions with thiols (van Geel et al., 2012).

We moved on to investigate CuAAC labeling on live cells, where the concentration of added copper and labeling conditions must be optimized to find a compromise between reaction efficiency and adverse side effects to proteins and cells. CuSO₄ concentrations of 50 µM in the presence of excess copper chelators have been successfully used for live cell CuAAC labeling, while higher concentrations have been shown to impact cell viability (Hong et al., 2010; Uttamapinant et al., 2012; Meineke et al., 2020). We incorporated PazK and AzeoK with Mma PylT/RS AF into an amber mutant of the class B GPCR corticotropin-releasing factor type 1 receptor (CRFR1 95TAG) (Coin et al., 2013; Serfling et al., 2018, Serfling et al., 2019). CuAAC with AF647-alkyne was performed with 5–50 µM CuSO₄ and a fivefold excess of THPTA on the surface of live cells expressing CRFR1^95AzeoK or CRFR1^95PazK (Figures 3B, C). CRFR1^95AzeoK could be labeled with AF647-alkyne on the surface of live cells with 50 µM CuSO₄, while AF647 fluorescence was barely detectable at 20 µM CuSO₄ and undetectable at 5 µM CuSO₄. CRFR1^95PazK yielded much stronger specific AF647 fluorescence at 50 and 20 µM CuSO₄ despite the lower expression level. Incorporation efficiency of PazK and AzeoK into CRFR1 can be compared via detection of a C-terminal HA tag: at the same ncAA concentration, AzeoK addition allows much more efficient amber suppression (Figure 3C). Thus, PazK demonstrated greatly improved CuAAC reactivity over AzeoK at copper concentrations as low as 20 μM, while further reducing the copper concentration did not support CuAAC with either ncAA. Thus, we conclude that PazK, in combination with THPTA, allows efficient CuAAC reactions on live cells with minimal expected toxicity (Hong et al., 2010).