Continuous hypermutation and evolution of noncanonical amino acid synthases

33 34 Genetic code expansion (GCE) enables the site -specific incorporation of noncanonical amino acids 35 (ncAAs) into proteins but is constrained by reliance on exogenously supplied chiral ncAAs. Achieving 36 intracellular ncAA biosynthesis would enable more scalable and cost-effective GCE. Here, we report 37 the continuous hypermutation and evolution of amino acid synthases that produce high levels of ncAAs 38 inside yeast, thus supporting GCE from simple ncAA precursors. We encoded an engineered ‘tyrosine 39 synthase’ (TmTyrS) on an error -prone orthogonal DNA replication system (OrthoRep) and selected 40 variants based on ncAA biosynthesis from readily available phenol analogs and intracellular L-serine. 41 Our selection employed orthogonal ncAA-specific aminoacyl-tRNA synthetases (aaRSs) as biosensors 42 whereby target ncAA production leads to aminoacylation of an amber suppressor tRNA and the 43 translation of a selectable reporter containing an amber stop codon. Our evolution successfully yielded 44 TmTyrS variants that efficiently produced 3-iodo-, 3 -bromo-, 3 -chloro-, and 3 -methyl-L-tyrosine, 45 enabling amber codon-specified ncAA-dependent translation, in some cases at levels comparable to 46 sense codon -specified natural amino acid translation. This work reduces barriers for expressing 47 proteins containing substituted tyrosines . Moreover, because aaRSs can themselves be evolved 48 (including with OrthoRep) for a flexible range of ncAA specificities, these results establish an end-to-49 end framework for evolving ncAA biosynthetic enzymes in vivo. 50 51 52 Graphical abstract 53 54 55 56 57 58 59 60 61 62 63 We describe an OrthoRep -driven platform for evolving noncanonical amino acid (ncAA) synthases. 64 Hypermutation of ncAA synthase genes enables evolution of ncAA biosynthesis from simple precursors, 65 while intracellular ncAA production is linked to fluorescence via an orthogonal aaRS/tRNA system, 66 allowing FACS enrichment of improved variants through iterative cycles. 67 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 3

68 69 The natural translation machinery uses only 20 canonical amino acids for protein synthesis. Over the 70 past decades, genetic code expansion (GCE) has emerged as a powerful strategy to overcome this 71 rule by enabling the site -specific incorporation of noncanonical amino acids (ncAAs) into proteins in 72 living cells[1]. Much of GCE relies on engineered orthogonal aminoacyl-tRNA synthetases (aaRSs) that 73 recognize ncAAs of interest and charge them onto cognate tRNAs that decode blank codons, most 74 commonly amber suppressor tRNAs [2,3]. Such engineered aaRS/tRNA pairs have enabled the site-75 specific translational incorporation of over 200 types of ncAAs[4–6], advancing protein biology[7–9], cell 76 biology[10,11], enzyme engineering[12–14], chemical biology[15–17], imaging[18–20], and therapeutics[21–23]. 77 78 While engineered/evolved aaRSs have substantially expanded the translational repertoire, a practical 79 bottleneck in many GCE applications is the reliance on externally synthesized ncAAs [24,25]. Because 80 ncAAs are usually absent from endogenous cellular metabolism, they must be chemically synthesized 81 and exogenously supplied at high concentrations, imposing constraints on cost, cellular uptake, and 82 scalability. Biosynthesis of ncAAs integrated into cellular metabolism represents a promising strategy 83 to alleviate these bottlenecks by enabling cost-effective and scalable ncAA production that directly feeds 84 GCE systems in vivo. To date, a variety of ncAAs have been biosynthesized using native or engineered 85 enzymes, including phenylalanine [26–28], tyrosine [29,30], tryptophan [31–33], and cysteine [34,35] analogs. 86 However, due to the lack of known biosynthetic pathways for most ncAAs and generally low production 87 efficiencies, the diversity of ncAAs accessible via biosynthesis remains significantly limited compared 88 to the scope of engineered/evolved aaRSs[25]. Accordingly, establishing a broadly applicable platform 89 to develop ncAA biosynthetic pathways that both expand the accessible ncAA repertoire and improve 90 biosynthetic efficiency remains a major challenge. 91 92 We envisioned an end -to-end experimental evolution framework for the creation of organisms with 93 GCEs that biosynthesize their own ncAAs (Fig. 1). In the first part of this framework, aaRSs are evolved 94 to incorporate ncAAs, achieving GCE from exogenously supplied ncAAs. This has been done previously 95 via a number of strategies [3,36] including using orthogonal DNA replication (OrthoRep) to drive 96 continuous aaRS hypermutation, which yielded highly efficient aaRS/tRNA systems for GCE in a 97 streamlined manner [37]. In the second part of this framework, evolved aaRSs are repurposed as 98 biosensors that couple in vivo ncAA production to the translation of selectable reporters, while amino 99 acid synthases—rather than aaRSs—are encoded on OrthoRep to drive evolution of ncAA biosynthesis 100 from simple precursors. Here, we describe the second part of this framework to complete our envisioned 101 platform by evolving a synthase for the in vivo production of noncanonical L-tyrosines (ncTyrs) from 102 readily available phenol analogs. Since ncTyrs constitute a major class of ncAAs used in GCE, this 103 represents an ideal test and demonstration of our framework. We present the evolution of tyrosine 104 synthase[38] (TyrS) variants that efficiently convert simple phenol analogs, including 2-iodo-, 2-chloro-, 105 2-bromo-, and 2-methylphenol, into their corresponding ncTyrs in vivo, which are then directly used in 106 translation of ncAA-containing proteins. 107 108

and Discussions 109 110 Evolution of TmTyrS with OrthoRep 111 112 Tyrosine synthase (TmTyrS), an enzyme engineered from the Thermotoga maritima tryptophan 113 synthase β-subunit (TmTrpB), catalyzes the formation of diverse ncTyrs from the corresponding phenol 114 analog and L-serine in vitro[38]. Evolved under high concentrations of substrates, the most active variant 115 identified in a previous study ( TmTyrS6) has only moderate catalytic efficiency [38], leaving room for 116 evolutionary improvement in our platform under physiologically relevant substrate concentrations. Our 117 starting point for evolution was TmTyrSc, an in vivo-adapted TmTyrS variant evolved from TmTyrS6 118 through an extensive but incomplete directed evolution campaign for ncTyr biosynthesis in yeast 119 (Supplementary Results and Supplementary Fig. 1). To couple the production of ncTyrs to cellular 120 fitness, we employed NitroY-F5, a Methanomethylophilus alvus pyrrolysyl-tRNA synthetase variant that 121 accepts a range of ncTyrs[39], or its OrthoRep -evolved variant NitroY-F5/3FY-D[37] as the biosensor. 122 TmTyrS6 itself exhibited insufficient ncTyr biosynthesis activity from the corresponding substituted 123 phenols to support survival through suppression of an amber stop codon -containing selection marker 124 by tRNAs aminoacylated with ncTyrs by NitroY-F5 (Supplementary Results and Supplementary Fig. 1). 125 126 For selection, we adopted our previously reported fluorescence -activated cell sorting (FACS) -based 127 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 4 aaRS evolution system based on a ratiometric ‘RXG’ reporter in which RFP and GFP are connected by 128 a linker containing an amber stop codon [37,40,41]. In this system, p roduction of ncTyrs increases the 129 amount of aminoacylated tRNA molecules which in turn increases the probability that the amber codon 130 between RFP and GFP is suppressed. This reporter configuration enabled enrichment of cells that 131 sampled highly active TmTyrS variants through mutation by gating on cells exhibiting higher GFP 132 expression relative to RFP in the presence of chosen phenol analogs during sorting. To normalize RXG 133 reporter fluorescence, we measured GFP and RFP fluorescence of cells expressing an ‘RYG’ reporter, 134 in which the amber codon was replaced with a tyrosine -encoding sense codon. Relative readthrough 135 efficiency (RRE) is defined as the ratio of !"# $"# for RXG to that for RYG [37,40,41], such that an RRE of 1 136 indicates amber suppression is as efficient as native translation at that position. Accordingly, we aimed 137 to evolve TmTyrS variants that exhibited high RREs in the presence of phenol analogs while maintaining 138 low RREs in their absence . Because this selection system directly selects cells based on the 139 readthrough ratio, which must depend on the level of aminoacylated orthogonal tRNA rather than on 140 reporter protein expression levels, it is well suited to select cells with enhanced ncAA production. 141 142 To establish the strain for OrthoRep-mediated evolution of TmTyrS, we started with Saccharomyces 143 cerevisiae GR-Y567, which carries deletions in LEU2 and HIS3, a split-LEU2 landing pad p1[42], and a 144 wild-type orthogonal DNA polymerase (DNAP) at the CAN1 locus[42]. We transformed this base strain 145 with the reporter plasmid, whic h encodes the RXG reporter, an engineered/evolved aaRS , and its 146 orthogonal amber-suppressor tRNA. We then integrated the TmTyrS gene onto the split-LEU2 landing 147 pad p1 and subsequently replaced the wild -type orthogonal DNAP cassette with error -prone DNAPs 148 (epDNAPs; Fig. 2a) to begin evolution. In the resulting strains, the epDNAP continuously replicates the 149 TmTyrS sequence at a high mutation rate of 1.6 × 10⁻⁵ and 3.9 × 10⁻⁵ substitutions per base for BB-Tv 150 and Trixy epDNAPs[42], respectively , while insulating the nuclear genome from the elevated 151 mutagenesis. 152 153 Next, we performed iterative FACS-based evolution cycles. After installation of the epDNAP, cells were 154 first grown for approximately 35 generations at 30 °C to allow diversification before starting selection. 155 Each evolution cycle proceeded as follows: (1) induction of the RXG reporter with galactose in the 156 presence or absence of phenol analogs for 48 h at 30 °C, corresponding to ~3 generations due to slow 157 growth under induction; (2a) a positive selection sort in which ~10,000,000 cells were screened and the 158 top 0.05% exhibiting the highest !"# $"# ratio in the presence of phenol analogs were collected, or (2b) a 159 negative selection sort in which ~1,000,000 cells were screened and the 5% displaying the lowest !"# $"# 160 ratio in the absence of phenol analogs were collected; and (3) regrowth of the sorted populations to 161 saturation at 30 °C, corresponding to ~13–15 generations depending on the number of recovered cells 162 (Fig. 2a). Negative selection was applied every two to four rounds to eliminate cells capable of 163 expressing GFP in the absence of phenol analogs, such as those harboring mutations that converted 164 the amber codon in the RXG reporter to a sense codon. We used both NitroY-F5 and its OrthoRep -165 evolved variant NitroY-F5/3FY-D as biosensors for ncTyrs. NitroY-F5/3FY-D was evolved with 3-fluoro-166 L-tyrosine and exhibited higher activity toward many ncTyrs compared to NitroY-F5[37]. 167 168 As TmTyrS genes on the p1 plasmid autonomously diversified during cultivation, repeated cycles 169 progressively enriched TmTyrS variants with improved activity toward ncTyr biosynthesis. After 170 completing the final cycle, TmTyrS genes were amplified from p1 by PCR, cloned under the TDH3 171 promoter on a CEN/ARS plasmid in a library format, and transformed into a fresh yeast strain carrying 172 the same reporter plasmid. The resulting library underwent a final round of positive selection without 173 hypermutation to observe clonal TmTyrS mutant fitnesses . Approximately 40 clones were then 174 randomly picked and evaluated, and the 4–6 clones exhibiting the highest RRE values with their target 175 phenol analogs (and minimal RRE in their absence) were selected for further characterization and 176 engineering. 177 178 Starting from TmTyrSc, we sequentially performed three evolution campaigns and obtained a series of 179 evolved variants ( TmTyrS7–9). We chose 2-iodophenol and 2 -chlorophenol as evolution substrates, 180 because TmTyrS exhibits catalytic efficiency toward these substrates [38,43] and both Nitro Y-F5 and 181 NitroY-F5/3FY-D efficiently aminoacylate the orthogonal tRNA with the corresponding reaction products, 182 3-iodo-L-tyrosine and 3 -chloro-L-tyrosine[37,39]. The detailed conditions for each evolution experiment 183 and the acquired mutations are provided in Supplementary Data 1 and 2, respectively. TmTyrS8 and 184 TmTyrS9 were generated by combining beneficial mutations identified in parallel evolution experiments 185 with the best-performing variants (Supplementary Fig. 2 and 3). The activities of these TmTyrS variants 186 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 5 toward 2-iodophenol and 2 -chlorophenol are shown in Figures 2b and 2c. Compared to TmTyrS6, 187 TmTyrSc exhibited enhanced production of 3-iodo-L-tyrosine, which is consistent to the results of the 188 URA3 reporter-based growth assay (Supplementary Information). The production of 3-iodo-L-tyrosine 189 increased progressively as evolution proceeded from TmTyrSc to TmTyrS9 (Fig. 2b). A similar 190 evolutionary trend was observed for 3-chloro-L-tyrosine production, with pronounced improvements in 191 TmTyrS8 and TmTyrS9, which were obtained from evolution experiments using 2-chlorophenol as the 192 substrate (Fig. 2c). Overall, these results demonstrate a substantial enhancement of ncTyr biosynthesis 193 through the OrthoRep-mediated evolution campaigns. 194 195 Evolutionary outcomes 196 197 Figures 2d and 2e show the amino acid substitutions accumulated during the FACS -based evolution 198 campaigns. Using TmTyrSc as a reference sequence, mutations acquired sequentially in TmTyrS7 (teal, 199 1 substitution ), TmTyrS8 (salmon pink , 10 substitutions ), and TmTyrS9 (gold , 8 substitutions ) are 200 highlighted, reflecting their stepwise evolutionary accumulation. In total, the final variant TmTyrS9 201 harbors 19 cumulative amino acid substitutions through the OrthoRep-mediated FACS-based evolution 202 campaign described here. 203 204 Among the 19 new substitutions, D300N is particularly notable. This residue is located within the 205 catalytic pocket of TmTyrS and highly conserved among TrpB enzymes, found as either aspartate 206 (D300), in biosynthetic TrpB enzymes, or arginine (R300), in a class of stand-alone, indole-scavenging 207 TrpB2 enzymes [44]. The asparagine (N300) substitution obtained through this OrthoRep -mediated 208 evolution campaign is practically absent from the TrpB evolutionary record, present in only 0.3 3% 209 (60/18,051) of TrpB-like sequences in a previously compiled dataset[38] (Supplementary Table 1 and 210 Supplementary Data 3). This position is known to be involved in both catalytic and alloster ic 211 mechanisms. Its sidechain interacts directly with the hydroxyl group of the L-serine substrate (or, in 212 TrpB2s, the phosphate group of the phospho-L-serine substrate) when it is bound to the pyridoxal 5’-213 phosphate (PLP) cofactor [45]. In TrpBs, its binding to T292 is associated with an allosteric transition 214 induced by its binding partner TrpA, and transitions between binding T292 and L-serine substrate are 215 associated with a closed conformation of the enzyme that generates its reactive aminoacrylate 216 intermediate[45,46]. The substitution T292S , present in all engineered TmTyrS variants, is known to 217 recapitulate this allosteric effect in the absence of TrpA, enhancing TrpB’s stand-alone activity[47]. While 218 the consequences of converting aspartate’s carboxylate sidechain to a carboxamide—and how this 219 may or may not be specifically activating for tyrosine synthesis—remain unclear, D300N was enriched 220 in two of four independent TmTyrS7 evolution experiments (Supplementary Data 2 ), suggesting an 221 important role in enhanced TyrS activity. 222 223 Also notable is the F200S substitution, as this position is dominated by F (~70%, in TrpBs) and H (~20%, 224 in TrpB2s). S200 is present in only 0.27% (48/18,051) of the TrpB-like sequences (Supplementary Table 225 2 and Supplementary Data 4). F200 interacts with the helix that bears the catalytic lysine, a residue that 226 is used to bind PLP and hypothesized to catalyze a concerted proton transfer during the aminoacrylate-227 mediated alkylation of phenol by TyrS enzymes. 228 229 Since we obtained TmTyrSc by shuffling yeast-adapted TmTrpB sequences with TmTyrS6 230 (Supplementary Information), it is possible that TmTyrSc lost some beneficial mutations for TyrS activity. 231 Interestingly, of the 19 substitutions acquired during the evolution from TmTyrSc to TmTyrS9, four 232 mutations (V20A, E30G, A245V, and S302P) represent reversions to pre-TmTyrSc residues (Fig. 2d 233 and Supplementary Data 5). V20A and A245V represent reversions to residues present in both wild-234 type TmTrpB and TmTyrS6. In contrast, E30G and S302P correspond to substitutions identified during 235 previous engineering efforts: E30G was identified during engineering of TmTrpB for reduced-236 temperature tryptophan synthesis[48], while S302P was acquired during the directed evolution campaign 237 for tyrosine synthase activity[38]. These reversion mutations are likely to play important roles in TyrS 238 activity in the evolved variants , whether by retaining important features for aminoacrylate generation 239 used by all TrpBs or for phenol alkylation used by the new TyrS enzymes. 240 241 Characteristics of the evolved TmTyrS 242 243 TmTyrS6, the immediate precursor of TmTyrSc, has previously been reported to produce ncTyrs not 244 only from 2 -iodophenol and 2 -chlorophenol but also from a variety of phenol analogs, including 2 -245 bromophenol and 2-methylphenol[38,43]. Because these phenol derivatives are highly similar in structure, 246 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 6 we hypothesized that TmTyrS9, which was evolved using 2-iodophenol and 2-chlorophenol, might also 247 exhibit high activity toward other phenol analogs. To test this, we compared the activities of TmTyrS9 248 and TmTyrS6 toward various phenol analogs. 249 250 To reconfirm each ncTyr could be detected by the reporter system, commercially available ncTyrs 251 corresponding to the reaction products from each phenol analog were supplied to reporter cells. In all 252 cases, ncTyr-dependent reporter responses were observed (Supplementary Fig. 4). Next, yeast strains 253 harboring TmTyrS9 and the reporter system were evaluated in the presence of phenol analogs at 254 concentrations up to 500 μM, and RRE values were determined across all concentrations. The same 255 experiments were performed for the parental TmTyrS6 for comparison. TmTyrS9 surpassed TmTyrS6 256 toward all tested substrates including non-evolution substrates 2-bromophenol and 2-methylphenol (Fig. 257 3a–d). Notably, the addition of 50 μM 2-iodophenol resulted in an RRE of ~1, which is comparable to 258 that obtained by direct supplementation of 50 μM 3-iodo-L-tyrosine (Fig. 3a). TmTyrS9 also showed 259 high activity toward 2-bromophenol, achieving an RRE of ~ 1 with 158 μM bromophenol, which is 260 comparable to that obtained with 50 μM 3-bromo-L-tyrosine (Fig. 3b). The activities of TmTyrS6 toward 261 2-chlorophenol and 2-methylphenol were nearly undetectable, while TmTyrS9 showed clear responses, 262 with RRE values of 0.36 and 0.23, at 500 μM 2-chlorophenol and 2-methylphenol, respectively (Fig. 3c 263 and d). Given that 3-methyl-L-tyrosine is chemically difficult to synthesize[43,49,50] and thus highly costly 264 (~1,600 USD g ⁻¹), while its precursor 2 -methylphenol is inexpensive (<0.1 USD g ⁻¹), TmTyrS9-265 mediated biosynthesis of 3-methyl-L-tyrosine offers substantial cost reductions. Moreover, TmTyrS 266 catalyzes ncTyr biosynthesis in an effectively irreversible manner by avoiding the thermodynamically 267 favorable degradation of ncTyrs to phenols, pyruvate, and ammonia [38], thereby facilitating efficient 268 coupling of ncTyr production with GCE. 269 270 To evaluate the fidelity of ncTyr biosynthesis mediated by TmTyrS9, we expressed sfGFP containing 271 an amber codon at position 150 together with TmTyrS9 and NitroY-F5/3FY-D in the presence of either 272 2-iodophenol or 3-iodo-L-tyrosine (Fig. 3e–h). The resulting sfGFP proteins were purified and analyzed 273 by whole protein mass spectrometry. The sfGFP produced in the presence of 3-iodo-L-tyrosine exhibited 274 the expected molecular mass corresponding to incorporation of 3-iodo-L-tyrosine at position N150 (Fig. 275 3f). Importantly, the sfGFP purified from cultures supplied with 2 -iodophenol displayed an identical 276 molecular mass, demonstrating that TmTyrS9 indeed produced 3 -iodo-L-tyrosine in vivo with high 277 fidelity (Fig. 3g). 278 279

280 281 In this work, we extended an OrthoRep-driven aaRS evolution platform to enable continuous 282 hypermutation and evolution of ncAA synthases by repurposing aaRSs as biosensors for ncAA 283 production. Using this platform, we demonstrated the evolution of TmTyrS for the biosynthesis of ncTyrs, 284 resulting in several outcomes including TmTyrS9, which supports the efficient production of ncTyrs from 285 2-iodophenol, 2-chlorophenol, 2-bromophenol, and 2-methylphenol. Recently, other work that exploits 286 aaRS-based biosensing to evolve ncAA synthases have been reported. Rubini et al. demonstrated the 287 directed evolution of a carbamoylase catalyzing the conversion of N-carbamoyl-L-3-nitrotyrosine to 3-288 nitro-L-tyrosine[51]. Similarly, Pulschen and Booth et al. reported the evolution of a tryptophan 289 halogenase using phage-assisted continuous evolution (PACE)[52] based on aaRSs that recognize 290 tryptophan analogs. While these approaches are related to ours , our OrthoRep -driven platform is 291 distinct in its coupling of in vivo continuous hypermutation with biosynthetic selection in yeast cells and 292 its completion of an end-to-end framework where both the aaRS and the ncAA synthase are evolved 293 using the same OrthoRep -driven strategy, offering a unified platform for obtaining expanded genetic 294 code organisms that biosynthesize their own ncAAs for GCE. 295 296 The effectiveness of OrthoRep -driven TmTyrS evolution is underscored by two key results ; 1) ncTyr 297 biosynthesis mediated by TmTyrS9 in some cases yielded translation efficiencies, as measured by RRE, 298 comparable to those with sense codons; and 2) TmTyrS9 enabled biosynthesis of 3-methyl-L-tyrosine, 299 a compound whose market price exceeds that of its precursor by more than three orders of magnitude. 300 These results demonstrate the practical utility of the ncAA synthases obtained here in addition to proving 301 a general framework for ncAA synthase and synthetase evolution for the GCE field. 302 303

304 305 DNA plasmid construction 306 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 7 307 Plasmids used in this study are listed in Supplementary Data 6 together with their DNA sequences. All 308 DNA templates for PCR were obtained from previous studies or synthesized as gBlocks (IDT). All 309 primers were synthesized by IDT. Amplicons for plasmid construction were generated using KOD One 310 PCR Master Mix -Blue- (Toyobo). Plasmids were assembled using either Gibson Assembly or Golden 311 Gate Assembly and transformed into chemically competent or electrocompetent E. coli TOP10 cells 312 (ThermoFisher). All plasmids were sequence verified by Sanger sequencing (Azenta) or whole plasmid 313 sequencing (Plasmidsaurus). 314 315 Reagents 316 317 All ncAA stock solutions were prepared at a final concentration of 10 mM of the L-isomer. Deionized 318 water was added to solid ncAA to approximately 90% of the final volume, and the pH was gradually 319 adjusted with NaOH as needed to dissolve the ncAA. Solutions were sterile filtered through 0.2 μm 320 filters and stored at −80 °C. After thawing, stocks were stored at 4 °C for up to 8 weeks prior to use. 321 Commercially available key reagents are listed in Supplementary Data 7. 322 323 Yeast strains and media 324 325 All yeast strains used in this study are listed in Supplementary Data 8. Yeast was incubated at 30 °C 326 and typically cultured in synthetic complete (SC) medium (20 g/L dextrose, 6.7 g/L yeast nitrogen base 327 w/o amino acids (US Biological), and the appropriate dropout mix (US Biological)) or in MSG (L-Glutamic 328 acid monosodium salt) based SC medium ( 20 g/L dextrose, 1.72 g/L yeast nitrogen base w/o 329 ammonium sulfate w/o amino acids (US Biological), appropriate nutrient drop-out mix (US Biological), 330 1 g/L L-Glutamic acid monosodium salt hydrate (ThermoFisher)). Media lacking specific nutrients are 331 denoted as −X, where X indicates the single letter amino acid code of the omitted amino acid or uracil 332 (U). For GAL1 promoter induction, SCGR medium containing 2% galactose and 2% raffinose in place 333 of glucose was used. For selection of the MET15 marker, cells were propagated in media lacking both 334 methionine and cysteine. Liquid cultures (500 μL) in 96 -well deep-well plates were incubated at 750 335 rpm, while all other liquid cultures were incubated at 200 rpm. Agar plates were prepared by mixing 336 equal volumes of 2× molten agar and 2× medium. Prior to experiments, cells were grown to saturation 337 in selective media to maintain plasmids. 338 339 Yeast transformations 340 341 All yeast transformations, including p1 integrations and polymerase replacement integrations, were 342 performed using frozen competent cells as previously described [53]. For p1 and polymerase 343 replacement integrations, 1–5 μg of plasmid DNA was linearized using ScaI -HF or EcoRI-HF (NEB), 344 respectively. For CEN/ARS plasmid transformations, 100 –500 ng of plasmid DNA was used. 345 Transformants were selected on the appropriate selective agar plates. MSG SC −HL agar plates 346 supplemented with 100 mg/L nourseothricin and 200 mg/L L-canavanine were used for polymerase 347 replacement integration. Plates were incubated at 30 °C for 2 days for nuclear plasmid transformations 348 and genomic integrations, and for 4 days for p1 integrations. 349 All linearized plasmids for p1 integration were integrated into a split LEU2 landing pad to generate the 350 desired p1 constructs[42]. Genomic DNA and p1/p2 plasmids were extracted as previously described[33]. 351 Briefly, 1.5 mL of yeast culture was pelleted, washed with 0.9% NaCl, and resuspended in 250  μL 352 Zymolyase solution (0.9 M D-sorbitol, 0.1 M EDTA, 10 U/mL Zymolyase (US Biological) ). After 353 incubation at 37 °C for 1 h, cells were lysed with proteinase K solution and treated at 65 °C for 30 min. 354 Following potassium acetate precipitation and ethanol precipitation, nucleic acids were resuspended in 355 TE buffer, treated with RNase A, and reprecipitated with isopropanol. The final pellet was resuspended 356 in 30 μL water. Proper integration was confirmed by agarose gel electrophoresis of recombinant p1 357 DNA. The presence of recombinant p1 was also confirmed after polymerase replacement and evolution 358 campaigns. 359 360 FACS-based TmTyrS evolution and selection with OrthoRep 361 362 Prior to each round of FACS selection, yeast strains harboring TmTyrS on p1, a reporter plasmid, and 363 an error-prone orthogonal DNAP (BB-Tv or Trixy)[42] integrated at the CAN1 locus were grown in SC−HL 364 at 30 °C to saturation. Cultures were diluted to OD600 = 0.6 in 2 mL medium and grown to OD600 = 1.5–365 3 (4–7 h). The cells were then induced in SCGR−HL at OD600 = 0.6 supplemented with 20 mM L-serine 366 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 8 and with or without phenol analogs. Cultures were incubated at 30 °C for 2 days. After culture saturation, 367 the cells were washed and resuspended in HBSM buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 5 mM 368 maltose) and sorted using a Sony SH800S cell sorter with a 100 μm sorting chip. For positive selection, 369 cells were grown with phenol analogs, whereas for negative selection, no phenol analog was supplied. 370 Approximately 10 million (positive) or 1 million (negative) events were measured per sort. GFP and RFP 371 were detected using 488 nm excitation with 525/50 and 617/30 filters, respectively. The top 0.05% of 372 cells positive for both GFP and RFP were recovered for positive sorts while the top 5% of cells positive 373 for RFP and most negative for GFP were recovered for negative sorts. Cells were recovered in 2 mL of 374 SC−HL at 30 °C until saturation and were used for the next round. 375 After multiple selection rounds, TmTyrS genes were amplified from p1 and subcloned into CEN/ARS 376 plasmids to prevent further hypermutation. Libraries were reintroduced into yeast strains with a reporter 377 plasmid and subjected to final positive FACS selections. Approximately 1 million (positive) events were 378 measured per sort. The top 0.1% of cells positive for both GFP and RFP were recovered for positive 379 sorts. After positive sorts, cells were grown on SC−HL media agar plates to isolate individual clones for 380 sequencing and further characterization. 381 382 Flow cytometry 383 Cells were grown in SC−HL at 30 °C to saturation. Cultures were diluted to OD600 = 0.6 in 500 μL and 384 grown to OD600 = 1.5–3 (4–7 h). The cells were then induced in 200 μL of SCGR−HL at OD600 = 0.6 385 supplemented with 20 mM L-serine and with or without phenol analogs. The induced c ultures were 386 incubated at 30 °C for 2 days. After culture saturation, cells were diluted into 0.9% NaCl and analyzed 387 on an Attune NxT flow cytometer (Life Technologies). The fluorescence of RFP and GFP from 20,000 388 single cells was recorded, and the mean fluorescence for each population was determined. Data were 389 analyzed using FlowJo v10.10.0. Autofluorescence of cells was subtracted using uninduced cells grown 390 in SC media. Fold changes and RREs were calculated as previously described[37]. Cells transformed 391 with the plasmid encoding the RYG reporter for RRE calculation were induced in the absence of phenol 392 analog. 393 394 Protein purification and mass spectrometry 395 396 TmTyrS9-expressing yeast harboring an sfGFP -150TAG reporter was grown in SC −HL media to 397 saturation. The culture was diluted to OD600 = 0.6 in 20 mL and grown to OD600 = 1.5–3 (4–7 h). The 398 cells were then induced in 40 mL of SCGR−HL media at OD600 = 0.6 supplemented with 2-iodophenol 399 or 3-iodo-L-tyrosine. The induced cultures were incubated for 2 days. After culture saturation, cells were 400 washed with 0.9% NaCl . P roteins were extracted from yeast cells using Y -PER (ThermoFisher) 401 containing cOmplete, EDTA-free protease inhibitor cocktail (MilliporeSigma), purified using HisPur Ni-402 NTA resin (ThermoFisher) and eluted with elution buffer (20 mM sodium phosphate, 300 mM NaCl, 250 403 mM imidazole, pH 8.0). Intact proteins were analyzed by LC/MS (ACQUITY UPLC H-class system and 404 Xevo G2-XS QTof, Waters). Proteins were separated using an ACQUITY UPLC BEH Phenyl VanGuard 405 Pre-column (130Å, 1.7 μm, 2.1 mm × 5 mm, Waters) at 45 °C. The 5-minute method used 0.2 mL/min 406 flow rate of a gradient of Buffer A consisting of 0.1% formic acid in water and Buffer B, acetonitrile. The 407 Xevo Z-spray source was operated in positive MS resolution mode, 400 –4,000 Da with a capillary 408 voltage of 3000 V and a cone voltage of 40 V (NaCsI calibration, Leu-enkephalin lock-mass). Nitrogen 409 was used as the desolvation gas at 350 °C and a total flow of 800 L/h. Total average mass spectra were 410 reconstructed from the charge state ion series using the MaxEnt1 algorithm from MassLynx software 411 (Waters) according to the manufacturer’s instructions. To obtain the ion series described, the major 412 peak of the chromatogram was selected for integration before further analysis. The theoretical 413 molecular weight of a protein with 3-iodo-L-tyrosine was calculated by first computing the theoretical 414 molecular weight of wild-type sfGFP and then manually correcting for the theoretical molecular weight 415 of 3-iodo-L-tyrosine. 416 417 Statistics and reproducibility 418 419 Microsoft Excel was used for all statistical analyses. Replicate numbers are provided in the figure 420 legends. No statistical methods were used to predetermine sample size, and no data were excluded. 421 422 Data Availability 423 424 All data related to evolution campaigns, mutations, sequence alignments, reagents, plasmids, yeast 425 strains, and mass spectrometry are provided in the Supplementary Data. 426 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 9

427 428 [1] M. A. Shandell, Z. Tan, V. W. Cornish, Biochemistry 2021, 60, 3455–3469. 429 [2] T. Mukai, M. J. Lajoie, M. Englert, D. Söll, Annu. Rev. Microbiol. 2017, 71, 557–577. 430 [3] J. W. Chin, Annu. Rev. Biochem. 2014, 83, 379–408. 431 [4] L.-S. Icking, A. M. Riedlberger, F. Krause, J. Widder, A. S. Frederiksen, F. Stockert, M. Spädt, N. 432 Edel, D. Armbruster, G. Forlani, S. Franchini, P. Kaas, B. M. Kırpat Konak, F. Krier, M. Lefebvre, 433 D. Mazraeh, J. Ranniger, J. Gerstenecker, P. Gescher, K. Voigt, P. Salavei, N. Gensch, B. Di 434 Ventura, M. A. Öztürk, Nucleic Acids Res. 2023, DOI 10.1093/nar/gkad1090. 435 [5] C. S. Diercks, D. A. Dik, P. G. Schultz, Chem 2021, 7, 2883–2895. 436 [6] N. G. Koch, N. Budisa, Chem. Rev. 2024, DOI 10.1021/acs.chemrev.4c00031. 437 [7] P. M. England, Biochemistry 2004, 43, 11623–11629. 438 [8] W. Wang, J. K. Takimoto, G. V. Louie, T. J. Baiga, J. P. Noel, K.-F. Lee, P. A. Slesinger, L. Wang, 439 Nat. Neurosci. 2007, 10, 1063–1072. 440 [9] X. Steinberg, M. A. Kasimova, D. Cabezas-Bratesco, J. D. Galpin, E. Ladron-de-Guevara, F. 441 Villa, V. Carnevale, L. Islas, C. A. Ahern, S. E. Brauchi, Elife 2017, 6, DOI 10.7554/eLife.28626. 442 [10] L. Cao, L. Wang, Chem. Rev. 2024, 124, 8516–8549. 443 [11] B. J. Wilkins, N. A. Rall, Y. Ostwal, T. Kruitwagen, K. Hiragami-Hamada, M. Winkler, Y. Barral, W. 444 Fischle, H. Neumann, Science 2014, 343, 77–80. 445 [12] E. L. Bell, A. E. Hutton, A. J. Burke, A. O’Connell, A. Barry, E. O’Reilly, A. P. Green, Chem. Soc. 446 Rev. 2024, 53, 2851–2862. 447 [13] I. Drienovská, C. Mayer, C. Dulson, G. Roelfes, Nat. Chem. 2018, 10, 946–952. 448 [14] A. J. Burke, S. L. Lovelock, A. Frese, R. Crawshaw, M. Ortmayer, M. Dunstan, C. Levy, A. P. 449 Green, Nature 2019, 1. 450 [15] Q. Gan, C. Fan, Chem. Rev. 2024, 124, 2805–2838. 451 [16] C. C. Liu, P. G. Schultz, Nat. Biotechnol. 2006, 24, 1436–1440. 452 [17] H.-S. Park, M. J. Hohn, T. Umehara, L.-T. Guo, E. M. Osborne, J. Benner, C. J. Noren, J. 453 Rinehart, D. Söll, Science 2011, 333, 1151–1154. 454 [18] H. B. Yi, S. Lee, K. Seo, H. Kim, M. Kim, H. S. Lee, Chem. Rev. 2024, DOI 455 10.1021/acs.chemrev.4c00112. 456 [19] A. Chatterjee, J. Guo, H. S. Lee, P. G. Schultz, J. Am. Chem. Soc. 2013, 135, 12540–12543. 457 [20] C. M. Jones, D. M. Robkis, R. J. Blizzard, M. Munari, Y. Venkatesh, T. S. Mihaila, A. J. Eddins, R. 458 A. Mehl, W. N. Zagotta, S. E. Gordon, E. J. Petersson, Chem. Sci. 2021, 12, 11955–11964. 459 [21] S. J. Walsh, J. D. Bargh, F. M. Dannheim, A. R. Hanby, H. Seki, A. J. Counsell, X. Ou, E. Fowler, 460 N. Ashman, Y. Takada, A. Isidro-Llobet, J. S. Parker, J. S. Carroll, D. R. Spring, Chem. Soc. Rev. 461 2021, 50, 1305–1353. 462 [22] S. A. Kularatne, V. Deshmukh, J. Ma, V. Tardif, R. K. V. Lim, H. M. Pugh, Y. Sun, A. Manibusan, A. 463 J. Sellers, R. S. Barnett, S. Srinagesh, J. S. Forsyth, W. Hassenpflug, F. Tian, T. Javahishvili, B. 464 Felding-Habermann, B. R. Lawson, S. A. Kazane, P. G. Schultz, Angew. Chem. Int. Ed Engl. 465 2014, 53, 11863–11867. 466 [23] J. Y. Axup, K. M. Bajjuri, M. Ritland, B. M. Hutchins, C. H. Kim, S. A. Kazane, R. Halder, J. S. 467 Forsyth, A. F. Santidrian, K. Stafin, Y. Lu, H. Tran, A. J. Seller, S. L. Biroc, A. Szydlik, J. K. 468 Pinkstaff, F. Tian, S. C. Sinha, B. Felding-Habermann, V. V. Smider, P. G. Schultz, Proc. Natl. 469 Acad. Sci. U. S. A. 2012, 109, 16101–16106. 470 [24] H. D. Biava, Chembiochem 2020, 21, 1265–1273. 471 [25] Z. Hou, J. Tuo, X. Ma, Y.-X. Huo, Results Eng. 2025, 25, 103641. 472 [26] R. A. Mehl, J. C. Anderson, S. W. Santoro, L. Wang, A. B. Martin, D. S. King, D. M. Horn, P. G. 473 Schultz, J. Am. Chem. Soc. 2003, 125, 935–939. 474 [27] J.-E. Jung, S. Y. Lee, H. Park, H. Cha, W. Ko, K. Sachin, D. W. Kim, D. Y. Chi, H. S. Lee, Chem. 475 Sci. 2014, 5, 1881–1885. 476 [28] J. Zhang, K. Yu, Y. Xu, W. Zhao, Y. Li, Y. Wang, F. P. Seebeck, X.-H. Chen, C. Liao, Nat. 477 Commun. 2025, 16, 1–13. 478 [29] S. Kim, B. H. Sung, S. C. Kim, H. S. Lee, Chem. Commun. (Camb.) 2018, 54, 3002–3005. 479 [30] Y. Chen, S. Jin, M. Zhang, Y. Hu, K.-L. Wu, A. Chung, S. Wang, Z. Tian, Y. Wang, P. G. Wolynes, 480 H. Xiao, Nat. Commun. 2022, 13, 5434. 481 [31] R. S. Phillips, Tetrahedron Asymmetry 2004, 15, 2787–2792. 482 [32] R. J. M. Goss, P. L. A. Newill, Chem. Commun. (Camb.) 2006, 4924–4925. 483 [33] G. Rix, E. J. Watkins-Dulaney, P. J. Almhjell, C. E. Boville, F. H. Arnold, C. C. Liu, Nat. Commun. 484 2020, 11, 5644. 485 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 10 [34] M. P. Exner, T. Kuenzl, T. M. T. To, Z. Ouyang, S. Schwagerus, M. G. Hoesl, C. P. R. 486 Hackenberger, M. C. Lensen, S. Panke, N. Budisa, Chembiochem 2017, 18, 85–90. 487 [35] S. Nojoumi, Y. Ma, S. Schwagerus, C. P. R. Hackenberger, N. Budisa, Int. J. Mol. Sci. 2019, 20, 488 2299. 489 [36] A. O. Osgood, Z. Huang, K. H. Szalay, A. Chatterjee, Chem. Rev. 2025, 125, 2474–2501. 490 [37] Y. Furuhata, G. Rix, J. A. Van Deventer, C. C. Liu, Nat. Commun. 2025, 16, 1–13. 491 [38] P. J. Almhjell, K. E. Johnston, N. J. Porter, J. L. Kennemur, V. C. Bhethanabotla, J. Ducharme, F. 492 H. Arnold, Nat. Chem. Biol. 2024, 20, 1086–1093. 493 [39] S. Avila-Crump, M. L. Hemshorn, C. M. Jones, L. Mbengi, K. Meyer, J. A. Griffis, S. Jana, G. E. 494 Petrina, V. V. Pagar, P. A. Karplus, E. J. Petersson, J. J. Perona, R. A. Mehl, R. B. Cooley, ACS 495 Chem. Biol. 2022, 17, 3458–3469. 496 [40] J. T. Stieglitz, H. P. Kehoe, M. Lei, J. A. Van Deventer, ACS Synth. Biol. 2018, 7, 2256–2269. 497 [41] J. W. Monk, S. P. Leonard, C. W. Brown, M. J. Hammerling, C. Mortensen, A. E. Gutierrez, N. Y. 498 Shin, E. Watkins, D. M. Mishler, J. E. Barrick, ACS Synth. Biol. 2017, 6, 45–54. 499 [42] G. Rix, R. L. Williams, V. J. Hu, A. Spinner, A. O. Pisera, D. S. Marks, C. C. Liu, Science 2024, 500 386, eadm9073. 501 [43] Almhjell P. J., Noncanonical Amino Acid Synthesis by Evolved Tryptophan Synthases, 502 Dissertation (Ph.D.), California Institute of Technology, 2022. 503 [44] F. Busch, C. Rajendran, O. Mayans, P. Löffler, R. Merkl, R. Sterner, Biochemistry 2014, 53, 504 6078–6083. 505 [45] A. R. Buller, S. Brinkmann-Chen, D. K. Romney, M. Herger, J. Murciano-Calles, F. H. Arnold, 506 Proc. Natl. Acad. Sci. U. S. A. 2015, 112, 14599–14604. 507 [46] A. R. Buller, P. van Roye, J. K. B. Cahn, R. A. Scheele, M. Herger, F. H. Arnold, J. Am. Chem. 508 Soc. 2018, 140, 7256–7266. 509 [47] J. Murciano-Calles, D. K. Romney, S. Brinkmann-Chen, A. R. Buller, F. H. Arnold, Angew. Chem. 510 Int. Ed Engl. 2016, 55, 11577–11581. 511 [48] C. E. Boville, D. K. Romney, P. J. Almhjell, M. Sieben, F. H. Arnold, J. Org. Chem. 2018, 83, 512 7447–7452. 513 [49] E. W. Schmidt, J. T. Nelson, J. P. Fillmore, Tetrahedron Lett. 2004, 45, 3921–3924. 514 [50] T. Nagasawa, T. Utagawa, J. Goto, C. J. Kim, Y. Tani, H. Kumagai, H. Yamada, Eur. J. Biochem. 515 1981, 117, 33–40. 516 [51] R. Rubini, S. C. Jansen, H. Beekhuis, H. J. Rozeboom, C. Mayer, Angew. Chem. Int. Ed Engl. 517 2023, 62, e202213942. 518 [52] A. A. Pulschen, J. Booth, A. Satanowski, C. Soudy, J. Caro-Astorga, O. Ather, N. Patel, A. 519 Alidoust, S. Aoudjane, L. Nematollahi, E. DeBenedictis, bioRxiv 2025, 2025.10.08.681035. 520 [53] R. D. Gietz, R. H. Schiestl, Nat. Protoc. 2007, 2, 1–4. 521 522 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 11

523 524 This work was funded by NIH R35GM136297 (C.C.L.), a JSPS Overseas Research Fellowship 525 202260318 (Y.F.), Grant in Aid for Scientific Research (B) 25K00105 (Y.F.), and NIH 1F32GM156066 526 (P.J.A.) 527 528 Author Contributions 529 530 Y.F., G.R., and C.C.L. designed the experiments. Y.F. and G.R. conducted the experiments. Y.F., G.R., 531 P.J.A., and C.C.L. analyzed the data and wrote the manuscript. 532 533 Competing interests 534 535 C.C.L. is a co-founder of Eira Bio, which uses OrthoRep for protein engineering. P.J.A. is an inventor 536 on a patent that covers enzymatic synthesis of tyrosine analogs from analogs of phenol and serine 537 (US12421534). The remaining authors declare no competing interests. 538 539 540 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 12 Figure Legends 541 542 Figure 1. End-to-end experimental evolution framework for the creation of organisms with GCEs that 543 biosynthesize their own ncAAs. 544 Center, conceptual illustration of an ideal GCE -enabled organism. In this organism, ncAA precursors 545 are converted into target ncAAs in vivo by an intracellular ncAA synthase . The synthesized ncAA is 546 recognized by an orthogonal aaRS, charged onto its cognate tRNA, and incorporated at a reassigned 547 codon during translation. To establish this architecture , we develop an end-to-end experimental 548 evolution framework that enables evolution of both an orthogonal aaRS and an ncAA synthase within 549 the same platform and workflow. 550 Left, OrthoRep-driven aaRS evolution platform described previously[37]. Hypermutation is applied to an 551 orthogonal aaRS on an OrthoRep plasmid to evolve its specificity toward a target ncAA . Intracellular 552 aminoacyl-tRNA levels are coupled to a fluorescent readout, enabling FACS enrichment of improved 553 aaRS variants through iterative cycles. 554 Right, OrthoRep-driven ncAA synthase evolution platform developed in this study. Hypermutation is 555 applied to an ncAA synthase on an OrthoRep plasmid to evolve biosynthetic activity of target ncAAs. 556 Intracellular ncAA production is coupled to a fluorescent readout via orthogonal aaRS/tRNA system, 557 enabling FACS enrichment of improved ncAA synthase variants through iterative cycles. 558 559 Figure 2. Directed evolution of TmTyrS with OrthoRep. (a) Schematic for a directed evolution approach 560 to evolve the activity of TmTyrS for the synthesis of tyrosine analogs from phenol analogs. (b–c) RRE 561 measurements of the evolved TmTyrS lineage with 2-iodophenol (b) and 2 -chlorophenol (c). Each 562 condition was measured in biological quadruplicates, and the mean ± one standard deviation (error 563 bars) is shown. GR-Y261 and pYF227 were used as the base strain and reporter plasmid, respectively. 564 (d) Mutations accumulated in TmTyrS during evolution campaigns. Detailed sequence alignments are 565 provided in Supplementary Data 5. (e) Crystal structure of TmTyrS1 (PDB 8EH1) with mutated residues 566 relative to the parental sequence in the evolution experiments that yielded TmTyrS7 (teal), S8 (salmon 567 pink), and S9 (gold) are highlighted. The colors correspond to those shown in panel (d). TmTyrS1 is the 568 ancestral sequence for all TmTyrS variants described in this study. 569 570 Figure 3. Characterization of TmTyrS9. (a–d) Sensitivity to phenol concentration of TmTyrS6 (gray) 571 and TmTyrS9 (colored). 2-iodophenol (a), 2-bromophenol (b), 2-chlorophenol (c), and 2-methylphenol 572 (d) were titrated. A dotted line in each graph represents the RRE value obtained when the indicated 573 concentration of the corresponding tyrosine analog was added instead of the phenol analog . Each 574 condition was measured in biological quadruplicates, and the mean ± one standard deviation (error 575 bars) is shown. GR-Y261 and pYF227 were used as the base strain and reporter plasmid, respectively. 576 (e) Schematic of the plasmids used for sfGFP 150TAG protein expression and purification for protein 577 mass spectrometry. (f–h) Whole-protein mass spectrometry of sfGFP150TAG (f, g) and WT sfGFP (h). 578 sfGFP150TAG was expressed with TmTyrS9 with 500 μM 3-iodo-L-tyrosine (f) or 500 μM 2-iodophenol 579 (g). WT sfGFP was expressed with the empty vector. GR-Y261 and pYF488 were used as the base 580 strain and reporter plasmid, respectively. Source data are provided in Supplementary Data 9. 581 582 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 13 Figure 1 583 584 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 14 Figure 2 585 586 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint 15 Figure 3 587 588 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.11.711232doi: bioRxiv preprint