THRONCAT: Efficient metabolic labeling of newly synthesized proteins using a bioorthogonal threonine analog

Profiling the nascent cellular proteome and capturing early proteomic changes in response to external stimuli provides valuable insight into cellular physiology. Existing metabolic protein labeling approaches based on bioorthogonal methionine-or puromycin analogs allow for the selective visualization and enrichment of the newly synthesized proteins. However, their applications are limited as they require methionine-free conditions, auxotrophic cells and/or are toxic to cells. Here, we introduce THRONCAT, a novel threonine-derived non-canonical amino acid tagging method based on bioorthogonal threonine analog β-ethynylserine (βES) that enables efficient and non-toxic labeling of the nascent proteome in complete growth media within minutes. We used THRONCAT for the visualization and enrichment of nascent proteins in bacteria, mammalian cells and Drosophila melanogaster. We profiled immediate proteome dynamics of Ramos B-cells in response to receptor activation, demonstrating the ease-of-use of the method and its potential to address diverse biological questions. In addition, using a Drosophila model of Charcot-Marie-Tooth peripheral neuropathy, we show that THRONCAT enables visualization and quantification of relative protein synthesis rates in vivo.


INTRODUCTION
Cells rapidly respond to environmental changes by synthesizing new proteins to ensure proper cell functioning. Characterizing this nascent proteome is important to understand cell functioning, but remains challenging as it requires an approach to distinguish between newly synthesized proteins (NSPs) and the pre-existing proteome. Metabolic protein labeling uses exogenous amino acid-or puromycin analogs, which are incorporated into nascent proteins by the endogenous biosynthetic machinery. 1  The most commonly used metabolic labeling reporters are bioorthogonal methionine analogs azidohomoalanine (AHA) and homopropargylglycine (HPG; Fig. 2a), [3][4][5]

and puromycin analog O-
propargyl-puromycin (OPP). 6 OPP labels NSPs within minutes and is easy to use, as labeling can be performed in complete growth media. However, OPP is toxic to cells and yields truncated OPP-polypeptide adducts that are unstable and proteolytically degraded within 1 hour. 6 In contrast, bioorthogonal non-canonical amino acid tagging (BONCAT) with AHA and HPG is less toxic and provides stable labeling in NSPs. However, as AHA and HPG are poorly incorporated in proteins in the presence of methionine, BONCAT requires methionine starvation, precluding its use to study protein synthesis dynamics in their native context. 7 These challenges highlight the need for a metabolic protein labeling method that allows fast and efficient incorporation of amino acid analogs in nascent proteins and can be used under native conditions. Threonine analogs 4-fluorothreonine (4-FT) 8 and β-hydroxynorvaline (β-HNV) 9,10 are excellent substrates for the threonine tRNA synthetase (ThrRS) and are efficiently incorporated into the nascent proteome of bacteria in complete growth media. We envisioned that the bioorthogonal threonine analog β-ethynylserine (βES; Fig. 2b), [11][12][13] may also be efficiently incorporated into nascent proteins, facilitating their labeling in cells grown in complete medium.
Here, we introduce THRONCAT, a novel metabolic labeling method based on threonine-derived non-canonical amino acid tagging. We show that threonine analog βES is efficiently incorporated into NSPs, is non-toxic and allows labeling of the nascent cellular proteome in complete growth medium within minutes. We demonstrate that THRONCAT has high labeling efficiency and compare our method to BONCAT for the visualization of NSPs in bacteria, mammalian cells and

βES is efficiently incorporated into the nascent prototrophic E. coli proteome
To explore the use of threonine analogs in metabolic labeling experiments, we first devised a stereoselective synthesis route towards βES ( Supplementary Fig. 1). Starting from commercially available materials and using an asymmetric aminohydroxylation as key step in our synthetic route, we obtained βES in four steps in a 22% overall yield (>98% de, 86% ee).
Next, we aimed to explore whether βES is incorporated into the nascent proteome of bacteria. As prototrophic bacteria can synthesize their own pool of methionine, tagging NSPs using methionine derivatives is generally challenged by the need of methionine-auxotrophic bacteria and the use of methionine-free growth medium. Indeed, we observed no labeling using methionine derivative HPG in prototrophic E. coli BL21 cells that were grown in complete growth medium (Fig. 2c, d), while we confirmed strong incorporation of HPG in auxotrophic E. coli B834 cells in methioninefree growth medium (Fig. 2e). In contrast to HPG, supplementing prototrophic E. coli BL21 cells growing in complete medium with βES resulted in strong labeling of the nascent proteome (Fig.   2c, d). The labeling ensued throughout the proteome, in a time-dependent manner and without significantly affecting bacterial growth (Fig. 2d, f; Supplementary Fig. 2). Incorporation of βES was abrogated by the addition of protein synthesis inhibitor chloramphenicol (CAP) or a 50-fold excess of threonine, suggesting that βES is incorporated exclusively into NSPs at positions encoding for threonine (Fig. 2g). Interestingly, looking at the relative labeling intensities of individual protein bands in the SDS-PAGE gels, we observed a clear difference between βES-and HPG-labeled E.
coli lysate, possibly because of varying numbers of threonine-and methionine residues in individual proteins (Fig. 2e). in E. coli BL21 after 1 h incubation with 1 mM βES in LB medium. No βES incorporation is detected upon co-incubation with chloramphenicol or an excess of threonine. CAP, chloramphenicol; Thr, L-threonine; kD, Kilodalton.

Using βES for fast and non-toxic visualization of mammalian NSPs
Encouraged by the observation that βES is vastly incorporated in NSP of prototrophic E. coli in complete medium, we next explored the efficiency of βES labeling of nascent proteins in mammalian cells. We treated HeLa cells for 1 h with βES in complete growth medium and conjugated incorporated βES to a Cy5-azide fluorophore. Fluorescence analysis of HeLa cells by flow cytometry and in-gel fluorescence revealed efficient and concentration-dependent incorporation of βES (Fig. 3a, b; Supplementary Fig. 3). Co-incubation with protein synthesis inhibitor cycloheximide or excess threonine abolished all fluorescence, confirming incorporation of βES at the position of threonine ( Supplementary Fig. 4). The fluorescence signal was discernible from background using the lowest (4 μM βES) concentration tested and increased dose-dependently, giving a ~200-fold increase in signal over background at the highest concentration tested (4 mM βES; Fig. 3a). In contrast, 1 h incubation of HeLa cells with 4 mM HPG in complete medium yielded only minimal HPG incorporation in NSPs (Fig. 3a, Fig. 3b).
Importantly, we could obtain an additional 1000-fold increase in detection sensitivity using βES when cells were grown in medium that was depleted from threonine (e.g. 200-fold signal over background at 4 μM βES, Supplementary Fig. 5).
Fluorescence microscopy and flow cytometry revealed that labeling with 4 mM βES in complete medium gave a strong fluorescent labeling of the nascent HeLa proteome after Cy5-azide conjugation within minutes (Fig 3c, d). The NSPs were distributed throughout the whole cell and strongest fluorescence was observed in the nucleoli (Fig. 3c), which is consistent with previous results obtained with HPG and OPP labeling. 6 The results above suggest that βES incorporation is less sensitive to competition of threonine than HPG for methionine. Indeed, while competition of threonine decreased βES incorporation in HeLa cells, we observed a much stronger inhibitory effect for methionine on HPG incorporation ( Fig. 3e). Next, we assessed the incorporation of βES over longer time periods. We incubated HeLa cells up to 24 h with βES in complete medium and observed a steady increase of βES incorporation as quantified by flow cytometry (Supplementary Fig. 6). Importantly, we did not observe a reduction in cell viability at 24 h in the presence of 0.4 -4 mM βES as measured by propidium iodide exclusion ( Supplementary Fig. 7). In addition, to reveal any adverse effects on cell proliferation, we observed no effect of protein labeling on HeLa cells exposed to up to 1 mM βES for 24 h. Incubation of cells with higher βES concentration, however, resulted in slightly decreased proliferation rate compared to control cells, which we also observed in HeLa cells treated with the lowest concentration HPG tested (e.g. 0.1 mM) for 24 h in methionine-free medium (Fig. 3f).

THRONCAT enables identification and quantification of the dynamic proteome
Next, we were interested to explore THRONCAT in the enrichment and detection of NSPs of HeLa cells and compared the nature of the identified proteins with that observed when using BONCAT in a mass-spectrometry-based proteomics setup. To this end, we incubated HeLa cells in complete medium for 5 h with 4 mM βES in 3 biological replicates, enriched NSPs and subjected digested peptides to LC-MS/MS analysis (Fig. 4a). By selecting for proteins that were identified in all 3 biological replicates, we confidently identified 3073 unique HeLa NSPs (  9). Using BONCAT -4 mM HPG in methionine-free conditions -we identified a similar number of HeLa NSPs as with 4 mM βES in complete medium (Fig. 4b). Overlap between NSPs identified by THRONCAT and BONCAT was large (81%) and only a fraction of NSPs were identified solely by THRONCAT (9.4%) or BONCAT (9.5%) (Fig. 4c). Interestingly, the THRONCAT-only fraction was enriched in proteins low in methionine content and conversely, the BONCAT-only fraction was enriched with proteins low in threonine content ( Supplementary Fig. 10

THRONCAT allows in vivo analysis of protein synthesis in Drosophila melanogaster
We next explored whether βES is incorporated into NSPs in Drosophila melanogaster, a living and behaving model organism with a rich genetic tool kit. We fed second instar Drosophila larvae that selectively express a membrane-tethered green fluorescent protein in motor neurons (OK371-GAL4>UAS-mCD8::GFP) for 48 h on Drosophila medium containing 4 mM βES (Fig. 5a).
Following labeling, the larval central nervous system (CNS) was dissected and subjected to conjugation chemistry with TAMRA-azide. Although βES is expected to be incorporated in all cells in the body, we decided to focus on the ventral nerve cord (VNC, equivalent to the vertebrate spinal cord) and in particular on motor neurons because of the relevance of protein synthesis (defects) in this cell type for neuromuscular diseases. Confocal microscopy revealed robust THRONCAT labeling in VNC of larvae incubated with 4 mM βES compared to an untreated control ( Fig. 5b,c). We observed that βES incorporation was strongly concentration dependent, decreasing ~ 90% at a 10-fold lower concentration of βES (0.4 mM) (Fig. 5d).
Next, we investigated the in vivo labeling kinetics of βES in comparison to HPG by incubating larvae for different durations with either analog. Signal intensity increased significantly over time for both analogs, but increased faster for βES and reached a higher maximum level (Table S1, Finally, we determined whether in vivo THRONCAT can be used to quantify changes in protein synthesis rates. Using in vivo cell-type-specific MetRS L262G -ANL FUNCAT, 18 we previously showed that expression of human glycyl-tRNA synthetase (GlyRS) carrying mutations that cause Charcot-Marie-Tooth (CMT) peripheral neuropathy reduce global protein synthesis in Drosophila motor and sensory neurons by ~30 to 60%, depending on the specific mutation, cell type, and ANL labeling time. 19,20 This inhibition of protein synthesis is attributable to sequestration of tRNA Gly by CMT-mutant GlyRS, resulting in insufficient supply of glycyl-tRNA Gly to the ribosome and ribosome stalling on glycine codons. 20 Thus, we generated Drosophila lines that co-express G240R mutant GlyRS and mCD8::GFP in motor neurons (OK371-GAL4), and exposed larvae to 4 mM βES for 48 h. In vivo THRONCAT revealed that in motor neurons (identified as mCD8::GFP expressing cells in the VNC) the labeling intensity was reduced by ~40% upon expression of GlyRS_G240R, as compared to motor neurons expressing mCD8::GFP alone (Fig. 5f,g). This result demonstrates that THRONCAT can be used to quantify and detect cell-type-specific changes in protein synthesis rates in Drosophila when combined with fluorescent cell markers.

Discussion
In this work we have used non-canonical amino acid βES to establish a new metabolic protein labeling technique, THRONCAT. βES is incorporated biosynthetically into nascent proteins in the position of threonine, presumably through ThrRS catalyzed aminoacylation of tRNA Thr . We showed that βES competes efficiently with threonine for incorporation into NSPs, suggesting a similar specificity constant for aminoacylation by ThrRS as observed for threonine. Also, we note that threonine residues, compared to methionine, are more abundant in the human proteome (5.2% vs 2.5%) and are often solvent exposed in proteins. 21 Together, these factors likely contribute to the strong NSP labeling observed when using βES.
Methionine starvation, essential to ensure efficient incorporation of methionine analogs in proteins using BONCAT, has been reported to reduce histone synthesis, biomolecule methylation and to impede cell cycle progression. 22 In experiments where exhaustive identification of NSPs is crucial, a combination of THRONCAT and BONCAT may be used to increase proteomic coverage. We believe THRONCAT will find wide application for analysis of proteomic changes such as cellular activation and differentiation, by facilitating simple and fast pulse labeling of NSPs. Although label-free quantification (LFQ) provides simple quantification of NSPs, 25 we suggest that THRONCAT is compatible with stable isotope labeling by amino acids in culture (SILAC) or tandem mass tags (TMT) for more accurate quantification of protein expression.
Using a Drosophila model of CMT peripheral neuropathy, we showed that THRONCAT enables in vivo quantification of protein synthesis rates. Although βES incorporation is not inherently celltype specific, a combination of THRONCAT and GFP-expression in a cell type of interest enabled cell-type specific visualization of protein synthesis. We envision that a similar approach in (co)cultured cells or organisms may enable cell-specific protein synthesis visualization and quantification.
In summary, we introduced THRONCAT, a novel non-canonical amino acid tagging method based on bioorthogonal threonine analog βES, which enables efficient and non-toxic labeling of newly synthesized proteins in complete growth media within minutes. We foresee that the efficient incorporation of βES in NSPs without the need to deplete threonine and combined with the ease of use, creates unprecedented opportunities to examine cell responses and mechanisms in models that are currently challenging to study.

Reagents and chemical synthesis
All commercial chemicals and solvents were purchased from Sigma Aldrich, Fluorochem, TCI or Fisher Scientific and used without further purification. The identity and purity of each product was characterized by 1 H NMR, 13

Metabolic labeling of HeLa cells for in-gel analysis
HeLa cells were seeded at a density of 3  Control cells were left untreated. Amino acid labeling was stopped by quickly aspirating the medium and replacing it by complete medium (DMEM) containing 600 μM cycloheximide. Then, cells were washed twice with cold PBS and fixed with 4% paraformaldehyde for 10 min. washed twice with PBS and stored at 4 °C until further use.

Competition assay of analogs versus natural amino acids
HeLa cells were seeded at a density of 60,000 cells per well in a 48 well plate and grown for 1 d in complete medium (DMEM). Cells were starved of intracellular threonine or methionine by incubating at 37 °C in threonine-free or methionine-free medium for 1 h. Then, cells were incubated 37 °C for 1 h with 50 μM βES and various ratios of threonine or 50 μM HPG and various ratios of methionine in threonine-free or methionine-free medium, respectively. After incubation, cells were washed twice with cold PBS, fixed with 4% paraformaldehyde for 10 minutes, washed twice with PBS and stored at 4 °C until further use. Incorporated analogs were conjugated to Cy5azide and quantified by flow cytometry.

Cy5 conjugation to HeLa proteome via azide-alkyne click chemistry
Metabolic labeling for flow cytometry and microscopy was performed as previously described. Then

Widefield microscopy
HeLa cells were metabolically labeled and subjected to CuAAC with Cy5-azide as previously described. Cells were imaged on a DMi8 widefield microscope (Leica). Images were processed and analyzed in ImageJ/Fiji (National Institutes of Health). Image contrast was enhanced to improve signal visibility by changing the maximum displayed values. Maximum intensity projection was used in all images.

THRONCAT and BONCAT in HeLa cells for proteomics
HeLa cells were seeded at a density of 4

LC-MS/MS measurements and data analysis
Peptide samples were eluted from StageTips with elution buffer (80% acetonitrile, 0.1% formic acid in H2O), reduced to 10% of the original volume by vacuum concentration and diluted in 0.1% formic acid. Peptides were separated using an Easy-nLC 1000 liquid chromatography system (ThermoScientific) with a 44 minute acetonitrile gradient (7-30%), followed by washes at 60% and 95% acetonitrile for a total of 60 minutes data collection. Mass spectrometry was performed on analysis, all proteins that were detected in replicates of the untreated negative control were considered a-specific binders and were removed for downstream analysis for all conditions.
Average threonine-or methionine content of protein fractions was determined by a Python script (Supplementary Data 2). Uniprot reference proteome (UP000005640) was used as complete human proteome to analyze average proteomic threonine and methionine content.
Differentially enriched protein analysis was performed using the DEP package. 31 All proteins that were detected in all replicates of at least one condition were considered for downstream analysis.
Imputation of missing values was performed using the MinProb method with the default settings.
Imputation was performed 1000x and adjusted p-values and fold changes in LFQ intensities were calculated for each round. All proteins that showed an adjusted p-value < 0.05 and a fold change > 1.5 in more than 80% of the iterations were considered to be significantly differentially expressed.
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository 32 with the dataset identifier PXD032368.

Drosophila genetics
Flies were housed in a temperature-controlled incubator with 12:12 h on/off light cycle at 25°C.

THRONCAT in Drosophila melanogaster
For in vivo THRONCAT, previously described FUNCAT procedures [18][19][20]33 were adapted. 8 h egg collections were performed and animals were raised on Jazz-Mix Drosophila medium (Fisher Scientific) at 25 °C. 72 h after egg laying (AEL), larvae were transferred to βES or HPG-containing medium. The standard βES or HPG concentration used was 4 mM, except for the βES dosagetitration experiment (Fig. 5e). The standard exposure time to non-canonical amino acid was 48 h, except in the experiment shown in Fig. 5f, in which larvae were exposed to βES

Statistics
For flow cytometry experiments, at least three replicates were measured for each condition (n = 3). A minimum number of n = 10.000 events were measured (before gating) for each replicate.
For experiments with Drosophila melanogaster, no statistical methods were used to predetermine sample sizes, but sample sizes are similar to those reported in previous publications. 19,20 Whenever possible, data collection and analysis were performed by investigators

CONFLICTS OF INTEREST
There are no conflicts to declare.