Efficient genetic encoding of phosphoserine and its non-hydrolyzable analog

Evolving pSer tRNA enhances phosphoserine incorporation

We began by testing the efficiency of previously reported systems for phosphoserine incorporation using a chloramphenicol resistance assay. Read through of the amber stop codon in a chloramphenicol acetyl-transferase gene at position 112 (cat(112TAG)) by EF-Sep/SepRS/ pSer tRNA(Tran)_CUA, leads to growth on 60 μg/mL chloramphenicol, while in the absence of SepRS cells grow weakly on 20 μg/mL chloramphenicol (Figure 1). Mutating SepRS to SepRS9¹¹ and EF-Sep¹⁰ to EF-Sep21¹¹ did not provide an increase in chloramphenicol resistance in our hands (Figure 1 and Supplementary Results, Supplementary Figure 1). We therefore used SepRS and EF-Sep as a starting point for further experiments.

To investigate whether altering the sequence flanking the anticodon of pSer tRNA_CUA enables efficient phosphoserine incorporation we first created pSer tRNA(N10)_CUA, a 5.1×10⁷ member library that samples all possible combinations of the ten nucleotides G29-U33 and G37-C41 that flank the anticodon. The library was transformed into cells bearing SepRS and EF-Sep and cat(112TAG), and pSer tRNA(N10)_CUA library members that enable efficient read-through of the amber stop codon selected on chloramphenicol. This screen identified 96 colonies, and sequencing revealed that these colonies contained 15 distinct tRNA(XX)_CUA sequences, indicating a large degree of library convergence (Figure 1).

To quantify the efficiency and specificity of phosphoserine incorporation with the selected pSer tRNA(XX)_CUA variants we examined the chloramphenicol resistance of cells containing cat(112TAG) with either EF-Sep/SepRS/ pSer tRNA(XX)_CUA variants or EF-Sep/pSer tRNA(XX)_CUA variants (in which SepRS is omitted). Thirteen out of the fifteen selected EF-Sep/SepRS/pSer tRNA(XX)_CUA combinations led to growth on substantially higher concentrations of chloramphenicol (200 μg/mL to 700 μg/mL) than the starting system while combinations using two selected tRNAs (pSer tRNA(A8)_CUA and pSer tRNA(C9)_CUA) were not active. The most active combinations led to approximately ten times the level of chloramphenicol resistance provided by pSer tRNA(Tran)_CUA(Figure 1b). All 13 functional tRNAs have G37 mutated to A, and conserve C32, A38 and U33. G37A in isolation is known to decrease the efficiency of tRNA_GCA aminoacylation by SepRS²¹, but the net result of the mutations transplanted into the anticodon and co-selected with G37A from pSer tRNA(N10)_CUA on aminoacylation is unknown. We note that A37 is selected¹⁶ in many efficient amber suppressor tRNA and may facilitate the efficient decoding of amber stop codons. Interestingly, tRNAs C2 and C4 are predicted to have an expanded anticodon loop due to mis-pairings between positions 31 and 39. However these tRNAs, like evolved quadruplet decoding PyltRNAs that also lack base pairing at this position¹⁵, are functional.

In the absence of SepRS we observe a range of chloramphenicol resistance levels resulting from mis-aminoacylation of the selected pSer tRNA(XX)_CUA variants by natural synthetases that are constitutively present in the cell. In the absence of SepRS, pSer tRNA(Tran)_CUA can be aminoacylated with glutamine by GlnRS¹⁰, and the tRNA mutations selected from pSer tRNA(N10)_CUA may alter the levels of mis-aminoacylation by GlnRS or other endogenous aminoacyl-tRNAsynthetases. All of the selected active tRNAs show SepRS dependent chloramphenicol resistance, and two tRNAs (pSer tRNA(B2)_CUA, pSer tRNA(C2)_CUA) are approximately ten times more active than pSer tRNA(Tran)_CUA in the presence of EF-Sep/SepRS, but have a low activity in the absence of SepRS, that is comparable to that of pSer tRNA(Tran)_CUA in the absence of SepRS. This data suggests that these tRNA mutants enable much more efficient incorporation of phosphoserine with EF-Sep/SepRS, but do not increase mis-incorporation of natural amino acids in the absence of SepRS.

Next we aimed i) to directly demonstrate the improvement in phosphoserine incorporation efficiency provided by the evolved pSer tRNA(XX)_CUA variants in a recombinant protein and ii) to characterise the fidelity of phosphoserine incorporation by mass spectrometry. We expressed the gene for sperm whale myoglobin bearing a C-terminal His6 tag and an amber codon at position 127 (myo(127TAG)-His6) in cells that also contain EF-Sep/SepRS and either pSer tRNA(Tran)_CUA or one of the fifteen selected pSer tRNA(XX)_CUA variants. Expression using each of the thirteen active, selected pSer tRNA(XX)_CUA variants led to substantially more myoglobin expression than expression using pSer tRNA(Tran)_CUA (Figure 1c and Supplementary Figure 2), consistent with the chloramphenicol resistance assay. Purified myoglobin from each expression was subject to mass spectrometry to demonstrate the incorporation of phosphoserine (Supplementary Figure 3). Despite the variable levels of chloramphenicol resistance observed in the absence of SepRS (Figure 1b), in the presence of SepRS a single protein is produced, for all the pSer tRNA(XX)_CUA variants, in which phosphoserine is the only amino acid that is detectably incorporated in response to the amber codon. This is consistent with previous reports that background misacylation of amber suppressor tRNAs may be effectively out-competed in the presence of an efficient cognate synthetase and its substrate^{22, 23}.

SepRS evolution enhances phosphoserine incorporation

We next created a SepRS anticodon binding domain library (SepRS(AC6)) and selected SepRS variants that function more efficiently with the thirteen selected, active pSer tRNA(XX)_CUA variants. From the X-ray crystal structure of the Archaeoglobus fulgidus SepRS/tRNA^Cys_GCA complex⁹, we identified six amino acid residues that interact with G34 and C35 of the cognate tRNA’s anticodon (Glu412, Glu414, Thr417, Pro495, Ile496 and Phe526) to target for combinatorial mutagenesis to all twenty amino acids (Figure 2a). The resulting library contains two critical residues (Phe526, Thr417) that pack around and between G34 and C35 of the anticodon (Figure 2a), in addition to the anticodon binding domain residues mutated in previous work¹¹. To generate the SepRS(AC6) library we first created plasmids recoding Phe526 to every amino acid and then used these plasmids as a template for two rounds of Enzymatic Inverse-PCR mediated randomisation with a “small-intelligent” primer set²⁴ to encode all 20 amino acids at each of the other five targeted positions. The library was generated with a diversity of 3.1×10⁹ c.f.u., surpassing the required diversity of 2.9×10⁸.

The SepRS(AC6) library was co-transformed into cells with each of the thirteen selected, active pSer tRNA(XX)_CUA variants, EF-Sep, and cat(112TAG) in thirteen parallel transformations, and subjected to a positive selection on chloramphenicol. Fourteen new SepRS(XX)/pSer tRNA(XX)_CUA pairs, in which the SepRS(XX) variants discovered were convergent at several positions (Supplementary Table 1), were identified from this selection, and their activity determined by chloramphenicol resistance assays (Figure 2b).

The eight most active, selected SepRS(XX)/ pSer tRNA(XX)_CUA pairs survived on 1100 μg/mL of chloramphenicol, indicating that these pairs are greater than eighteen times more efficient than previously reported systems for phosphoserine incorporation^{10, 11} (Figure 1 & Supplementary Figure 1). Seven of the eight most active pairs utilize pSer tRNA(XX)_CUA variants that exhibit the greatest activity with SepRS (Figure 2b).

Protein expression experiments with the selected SepRS (XX)/ pSer tRNA(XX)_CUA pairs, EF-Sep, and myo(127TAG)-His6 confirmed the increase in efficiency provided by the selected pairs, and demonstrate that the pairs we describe are much more efficient than the SepRS/ pSer tRNA(Tran)_CUA pair¹⁰, or mutants of SepRS previously reported to function with pSer tRNA(Tran)_CUA ¹¹(Figure 2c and Supplementary Figure 2). Moreover, mass spectrometry confirms the quantitative incorporation of phosphoserine in recombinant myoglobin produced from cells containing these systems (Supplementary Figure 4).

Ef-Tu mutation is not required to encode phosphoserine

Next we investigated whether EF-Sep is required for the site-specific incorporation of phosphoserine using the evolved pSer tRNA(XX)_CUA variants. Control experiments confirm that the SepRS/pSer tRNA(Tran)_CUA pair does not support chloramphenicol resistance in cells bearing cat(112TAG) when EF-Sep is omitted. In contrast, cells bearing cat(112TAG), and the SepRS, pSer tRNA(XX)_CUA pairs, but lacking EF-Sep, survive on concentrations of chloramphenicol up to 100μg/mL (eg: pSer tRNA(A4)_CUA ) (Figure 3a), demonstrating that the selected SepRS/pSer tRNA(XX)_CUA pairs are sufficient to direct the incorporation of phosphoserine in the absence of EF-Sep. The level of chloramphenicol resistance provided by selected SepRS/pSer tRNA(XX)_CUA pairs exceeds that of the EF-Sep/SepRS/ pSer tRNA(Tran)_CUA system (60μg/mL), providing evidence that the selected tRNA(XX)_CUAs provide a contribution to the efficiency of phosphoserine incorporation that exceeds the contribution provided by EF-Tu engineering to create EF-Sep. Protein expression experiments with the SepRS/ pSer tRNA(XX)_CUA pairs and myo(127TAG)His6 confirmed the increase in efficiency provided by pSer tRNA(XX)_CUA in the absence of EF-Sep (Figure 3b and Supplementary Figure 2). Moreover, mass spectrometry confirms the quantitative incorporation of phosphoserine in recombinant myoglobin produced from cells containing the SepRS/ pSer tRNA(XX)_CUA pairs, but lacking EF-Sep (Supplementary Figure 5).

Next, we asked whether evolving the anticodon binding region of SepRS enables efficient incorporation of phosphoserine in the absence of EF-Sep. To discover SepRS (XX)/ pSer tRNA(XX)_CUA pairs that function in the absence of EF-Sep we co-transformed the SepRS(AC6) library with each of the thirteen active pSer tRNA(XX)_CUAs (Figure 1) and cat(112TAG) and selected cells that survive on increasing concentrations of chloramphenicol. We discovered three SepRS (XX)/ pSer tRNA(XX)_CUA pairs that direct the incorporation of phosphoserine more efficiently that the progenitor SepRS / pSer tRNA(B2)_CUA, and SepRS / pSer tRNA(B4)_CUA pairs (Figure 3c, Supplementary Table 1). The most efficient of these pairs, SepRS (17)/ pSer tRNA(B2)_CUA, confers chloramphenicol resistance to approximately 500μg/mL, approximately ten times the resistance conferred by the SepRS/ pSer tRNA(Tran)_CUA/EF-Sep system or the SepRS9/ pSer tRNA(Tran)_CUA/EF-Sep21 system¹¹ (Figure 3c, Figure 1, and Supplementary Figure 1).

Protein expression experiments with the selected SepRS (XX)/ pSer tRNA(XX)_CUA pairs, and myo(127TAG)His6 confirmed the substantial increase in efficiency provided by the selected SepRS (XX)/ pSer tRNA(XX)_CUA pairs, and further demonstrated that the selected pairs are much more efficient than the SepRS/ pSer tRNA(Tran)_CUA/EF-Sep, system previously reported¹⁰ (Figure 3d and Supplementary Figure 2). Moreover, mass spectrometry confirmed the quantitative incorporation of phosphoserine in recombinant myoglobin produced from cells containing these systems (Supplementary Figure 5). Taken together the chloramphenicol resistance data, protein expression data and mass spectrometry data demonstrate that pSer tRNA_CUA evolution and SepRS evolution lead to much more efficient phosphoserine incorporation than EF-Tu evolution, and demonstrate that, in contrast to previous assumptions¹⁰, alteration of EF-Tu is not required for the incorporation of phosphoserine. Nonetheless, the best EF-Sep/SepRS(XX)/pSer tRNA(XX)_CUA combinations we have discovered outperform the best EF-Sep independent SepRS(XX)/pSer tRNA(XX)_CUA pairs (Supplementary Figure 1), consistent with the sequential roles of aminoacylation, delivery of the aminoacylated tRNA to the ribosome, and ribosomal decoding in the process of translation.

Optimized site-specific phosphoserine incorporation

Having established the optimal SepRS and pSer tRNA sequences we cloned the SepRS(2)/pSer tRNA(B4)_CUA/EF-Sep combination into a single high copy plasmid for convenient use in protein expression experiments when co-transformed with a plasmid encoding the gene of interest. We used a Glutathione-S-transferase-TAG-calmodulin (gst(TAG)cam) reporter with an amber stop codon at position 1 of calmodulin to characterize the efficiency of phosphoserine incorporation (Supplementary Figure 6) via the ratio of GST (synthesized as a result of termination in response to the amber stop codon) to GST-calmodulin (synthesized as a result of unnatural amino acid incorporation in response to the amber stop codon)¹⁵. The optimized system led to the synthesis of full-length GST-calmodulin (Supplementary Figure 6), and only a trace amount of GST. To benchmark the efficiency of the phosphoserine incorporation system against the efficiency of a widely used genetic code expansion platform we expressed (gst(TAG)cam) using the PylRS/tRNA_CUA pair in the presence of (Nε-[(tert-butoxy)carbonyl]-L-lysine), an established and efficient substrate for PylRS²⁵. This led to a greater fraction of protein synthesis resulting in truncation, demonstrating that the new system we have developed here has an efficiency in this assay that is comparable to or greater than that of a system that has been widely adopted for genetic code expansion.

While the efficiency of the system for phosphoserine incorporation was already very good, we tested whether the efficiency could be improved with an orthogonal translation system that does not recognize RF1, or using an RF1 deletion^{15, 23, 26-28} (Supplementary Figure 6). In both cases we see little or no truncated GST indicating near quantitative read through of the amber stop codon. While the use of these systems is unlikely to be necessary for expressing many proteins, these experiments demonstrate their compatibility with the evolved system for phosphoserine incorporation.

Expression and characterization of Ub(pSer65)

To further demonstrate the generality of our approach we produced ubiquitin bearing phosphoserine at position 65, a phosphorylation that has been identified in cells ^17-19 and is installed by PINK1, a kinase that also phosphorylates the E3 ubiquitin ligase Parkin²⁹. Mutations in PINK1 or Parkin are the most common cause of autosomal recessive Parkinson’s disease³⁰. Upon stress, the phosphorylation of both Parkin and ubiquitin at Ser 65 by PINK1 appear to be required for the activation of Parkin and the ubiquitylation and clearance of damaged mitochondria from cells ^17-19. Pure ubiquitin phosphorylated at serine 65 is challenging to generate enzymatically in vitro, and is typically obtained in sub-milligram quantities, though very recently a more efficient route to enzymatic phosphorylation of serine 65 has been reported and used for detailed biochemical characterization of this phosphorylation³¹.

We expressed ubiquitin specifically phosphorylated at position 65 (Ub (pSer65)) in cells containing Ub (65TAG)-His6 and the optimized phosphoserine incorporation system. Ub (pSer65) was purified with a yield of 3 mg per L of culture (Figure 4a,b,c). We investigated whether the Ub(pSer65) we have generated can directly activate Parkin E3 ligase activity to accelerate discharge of ubiquitin from a loaded E2 and ubiquitylation of substrates, as previously demonstrated for enzymatically derived ubiquitin that is phosphorylated at serine 65 ^17-19, which we refer to as Ub(pSer65*) for clarity. Addition of Ub(pSer65) led to near maximal ubiquitin discharge from a ubiquitin loaded E2 (UbcH7) by Parkin (Figure 4d and Supplementary Figure 7) and had a similar effect to Ub(pSer65*) (Fig 4d). Consistent with previous results we did not observe any discharge upon addition of un-phosphorylated ubiquitin (Fig 4d). Parkin activation by Ub(pSer65) led to the formation of polyubiquitin chains, increased auto-ubiquitylation of Parkin and ubiquitylation of a Parkin substrate, Miro 1 (Figure 4e and Supplementary Figure 2). The activity conferred by Ub(pSer65) was comparable to that observed with Ub(pSer65*) (Figure 4e). In control experiments un-phosphorylated ubiquitin did not activate full-length Parkin (Figure 4e), as expected.

Expression of synthetically-activated Nek7 kinase

To demonstrate that our system enables the efficient incorporation of phosphoserine at biologically relevant sites in a protein kinase we expressed Nek7, an essential kinase that is activated by phosphorylation at Serine 195 in mammalian cells^{32, 33}. Nek7 plays an important role in mitotic progression, and is required for robust mitotic spindle formation and cytokinesis^{32, 33}.

We expressed and purified Nek7 (Ser195TAG)-His6, leading to the production of Nek7 (pSer195)-His6 with a purified yield of 2 mg per L of culture (Figure 5a). ESI-MS and ESI MS/MS demonstrates the synthesis of Nek7 phosphorylated at position 195 (Figure 5b,c and Supplementary Figure 8). A small fraction of the purified protein has a mass consistent with the autophosphorylation of the synthetically activated kinase at an additional site (or sites) (Figure 5b). Indeed, we observe autophosphorylation of Nek7 (Supplementary Figure 9), as previously observed for the closely related kinase Nek6³². To further confirm that the minor species results from autophosphorylation we purified a catalytically inactive variant, Nek7 (D179N) (Supplementary Figure 10). Expression of Nek7 (D179N, Ser195TAG)-His6 led to production of Nek7 (D179N, pSer195)-His6 with a purified yield of 2 mg per L of culture. ESI-MS and MS/MS confirmed the site-specific incorporation of phosphoserine with no other phosphorylations.

We compared the activity of wild-type (wt) Nek7, Nek7(pSer195) and Nek7(D179N, pSer195) in phosphorylation of a model substrate (myelin basic protein, MBP)³² (Figure 5d and Supplementary Figure 11). While Nek7 (pSer195) robustly phosphorylates the substrate, neither wtNek7 nor Nek7(D179N, pSer195) efficiently phosphorylates the substrate. The addition of the C-terminal domain of Nek9 (Nek9-CTD) activates Nek7, as expected³⁴, but does not further activate Nek7 (pSer195), indicating that Nek7 (pSer195) bypasses the requirement for activation by Nek9 in vitro.

These experiments provide a first route to expressing and purifying Nek7 with a quantitative and site-specific phosphorylation in the activation loop, and demonstrate that milligram quantities of phosphorylated kinases can be obtained to enable structural and functional studies. Moreover, they demonstrate that synthetic phosphorylation of Nek7 is sufficient to activate the kinase and bypass trans-activation by the Nek9-CTD in vitro.

Encoding a non-hydrolyzable analog of phosphoserine

The genetically directed incorporation of the non-hydrolyzable analog of phosphoserine, 2-amino-4-phosphonobutyric acid (2)^{35, 36}, into recombinant proteins may facilitate the isolation of phospho-protein binders in complex mixtures that contain phosphatases, the creation of constitutively active kinases, as well as the characterization of complexes formed between phosphatase and their substrates.

No unnatural substrates have been incorporated with the SepRS/tRNA system, and a key challenge associated with the incorporation of unnatural substrates with this system results from the abundance of phosphoserine in the cell that will outcompete unnatural substrates at the active site of SepRS (Figure 6a). We asked whether we could manipulate the cell’s phosphoserine biosynthesis and destruction pathways²⁰ to facilitate the incorporation of 2 using the evolved SepRS(XX)/tRNA(XX)_CUA pairs (Figure 6b). Since phosphoserine is synthesized from 3-phosphohydroxypyruvate by phosphoserine aminotransferase (encoded by serC), and converted to serine by phosphoserine phosphatase (encoded by serB) ²⁰we reasoned that the intracellular levels of phosphoserine may be decreased by either overexpressing serB, deleting serC from the genome or by overexpressing serB in a serC deletion strain.

We found that the addition of 2mM 2 to wild-type E. coli provided with the SepRS(2)/tRNA(B4)_CUA /EF-Sep system did not increase the levels of myoglobin synthesized from myo (127TAG)-His6 (Figure 6c), and mass spectrometry confirmed the incorporation of phosphoserine rather than 2 (data not shown). Overexpression of serB led to an increase in myoglobin level, but the level of expression was not dependent upon the addition of 2. These observations are consistent with serB overexpression being insufficient to deplete phosphoserine to enable the specific incorporation of 2. Deletion of serC led to a decrease in myoglobin levels, consistent with a depletion of intra-cellular phosphoserine, but myoglobin levels were not rescued by addition of 2, suggesting that a serC deletion is not sufficient to enable the incorporation of 2. Overexpression of serB in the serC deletion led to an increase in myoglobin levels, as observed for wild-type E.coli. However, the addition of 2 to the serC deletion overexpressing serB led to a clear increase in myoglobin levels, consistent with the specific incorporation of 2 in this strain.

We note that it is currently unclear why serB overexpression leads to an increase in myoglobin levels, or why the levels of myoglobin increase upon addition of 2 to ΔserC +serB, but not upon addition of 2 to ΔserC. Nonetheless, these experiments prompted us to investigate whether 2 can be site-specifically incorporated into proteins produced in E.coli deleted for serC and overexpressing serB. In ΔserC + serB cells containing myo(127TAG)-His6/EF-Sep and the SepRS(2)/tRNA(B4)_CUA pair we observe expression of full-length myoglobin that is dependent on the addition of the phosphoserine analog 2 (Figure 6d). As expected, there is still some full-length protein produced in the absence of the added unnatural amino acid 2. The yield of full-length myoglobin incorporating 2 from ΔserC + serB cells was comparable to the yield for myoglobin incorporating phosphoserine in a serB deletion strain (Figure 6c), suggesting that when phosphoserine levels are reduced, the non-hydrolyzable amino acid (2) is an efficient substrate for the SepRS(2)/tRNA(B4)_CUA pair.

Mass spectrometry and MS/MS of myoglobin expressed in the presence of 2 demonstrates the site-specific incorporation of the analog (Figure 6e,f). This demonstrates that the analog effectively competes with any amino acids that may be incorporated in response to the amber codon in its absence, a common observation in many unnatural amino acid incorporation experiments²³.

Plasmids generation

The Methanocaldococcus jannaschii (Mj) tRNA_CUA, containing a C20U mutation shown to improve acylation by SepRS³⁸, was cloned into (p15A vector) between NotI and PstI to give pSertRNA_CUA. A DNA fragment encoding the Methanococcus maripaludis SepRS was cloned into NdeI/StuI digested pBK-Amp by Gibson cloning to give pBK-SepRS. For testing synthetase dependent incorporation, SepRS was removed from the pBK-SepRS plasmid using NEB Q5 SDM kit (New England Biolabs) and primers DTR634 and DTR635 to give pBK-NoRS. The E.coli EF-Tu gene (tufA) was amplified from DH10B genomic DNA and cloned into pCDFDuet-1 (Novagen) under control of the tac promoter. The phosphoserine permissive EF-Tu mutations (H67R, E216N, D217G, F219Y, T229S and N274W) were introduced by Overlap Extension PCR to give pCDF-EF-Sep. The gene for Chloramphenicol Acetyl-Transferase containing an amber stop codon at position 112 was amplified cloned into pCDF-EF-Sep to give pCDF-EF-Sep-CAT. pCDF-CAT was generated by deletion of EF-Sep gene using the NEB Q5 SDM kit (New England Biolabs) and primers DTR623 and DTR624. The gene for myoglobin (D127TAG)_His6 under the control of the arabinose promoter was amplified by PCR using DTR674 and DTR675. This fragment was inserted by Gibson cloning into pCDF-EF-Sep which had been amplified with DTR676 and DTR678 to give pCDF-EF-Sep-Myo(D127TAG)_His6. pCDF-Myo(D127TAG)_His6 was generated by deletion of EF-Sep from pCDF-EF-Sep-Myo(D127TAG)_His6 using the NEB Q5 SDM kit and primers DTR623 and DTR 624. SepRS mutant 2 and 17, along with the WT, were amplified from pBK-SepRS(2), pBK-SepRS(17) and pBK-SepRS using primers DTR733/734. pKW (Cm^R) was digested using PstI and NdeI, and the amplified SepRS fragment were inserted by Gibson cloning to yield pKWRS2, pKWRS17 and pKWRSWT. Gene blocks encoding tRNA(B4), tRNA(B2) and tRNA(WT) were ordered and Gibson cloned into BglII/SepI digested pKWRS2, pKWRS17 and pKWRSWT respectively, to yield pKW2, pKW17 and pKWWT. EF-Sep was amplified from pCDF-EF-Sep including its promoter and terminator using DTR823/824 and this fragment was Gibson cloned with the backbone of pKW2, which was amplified with DTR837 andDTR838 to yield pKW2-EF-Sep. The pRSF_ribo-Q1_O-gst-cam_1TAG and pRSF_ribo-wt_gst-cam_1TAG plasmids contain the following components, (i) a gene encoding ribosomal RNA (rRNA), former containing ribo-Q1 and latter with wildtype rRNA, (ii) a gene encoding glutathione-S-transferase and calmodulin (gst-cam) which is downstream of either an orthogonal ribosomal binding site (O-gst-cam) or a wildtype ribosomal binding site, and (iii) both the plasmids contain kanamycin resistance marker ¹⁵. pCDF-EF-Sep-serB was generated by PCR amplifying the serB gene from the genome of MegaX DH10B (Invitrogen) using DTR793 and DTR794, and Gibson cloning the fragment into pCDF-EF-Sep-CAT in place of CAT gene by amplification of the backbone with DTR795 and DTR796. All primer sequences can be found in Supplementary Table 2.

pSertRNA(10)_CUA library generation

To create the pSertRNA(10)_CUA library, the 5 bases either side of the anticodon of pSer tRNA_CUA in pSertRNA_CUA were randomised by Enzymatic Inverse PCR (EI-PCR)³⁹ using KOD polymerase (Novagen) and primers DTR195/DTR198. The 3.6kb PCR product was gel extracted, digested with SapI, re-circularised using T4 ligase and electroporated into MegaX DH10B (Invitrogen) cells to afford a library with a diversity of ≈5.1×10⁷ colony forming units (c.f.u), surpassing the required theoretical diversity of 4.8×10⁶ c.f.u. The pool of DNA clones was subsequently extracted by DNA maxiprep and frozen.

SepRS(AC6) library generation

Guided by the structure of the Archaeoglobus fulgidus SepRS–tRNA^Cys–O-phosphoserine ternary complex, residues E412, E414, T417, P495, I496 and F526 were chosen for mutation due to their proximity to G34 and C35 in the tRNA anticodon. Initially pBK-SepRS was mutated using the NEB Q5 site directed mutagenesis kit (New England Biolabs), wherein F526 was mutated to every amino acid using the primers in Supplementary Table 3. Each plasmid, including the WT F526 plasmid, was used as a template for individual EI-PCR mediated mutagenesis reactions using Q5 polymerase. To maximise sequence coverage and minimize degeneracy bias, “small-intelligent” oligos were used ²⁴. The first round of mutation randomised P495, I496 (primers in Supplementary Table 4); the resulting PCR reactions were normalised for concentration, pooled, digested with BsaI, re-circularised by ligation and transformed. A miniprep of this transformation was then used to randomise E412, E414 and T417 (primers in Supplementary Table 4) which was digested with BsaI, ligated and transformed into MegaX DH10B (Invitrogen) to give the SepRS(AC6) library with a diversity of ≈3.1×10⁹ c.f.u, with a theoretical requirement of 2.9×10⁸ c.f.u.

pSertRNA(XX)_CUA selection

The pSertRNA(10)_CUA anticodon library was co-transformed into DH10B ΔserB with pBK-SepRS and pCDF-EF-Sep-CAT to a diversity of 8×10⁹ c.f.u. The pool of transformed cells was grown in LB + 2mM phosphoserine to a density of OD₆₀₀=1.0. 10mL of this culture (approximately 8×10⁹ cells) was transferred to 500mL fresh LB + 2mM phosphoserine + 1mM IPTG and to grown to OD₆₀₀=0.6. 2mL of this culture, equivalent to approximately 1×10⁹ cells, was plated on LB + 2mM phosphoserine + 1mM IPTG and either 100, 250 or 500μg/ml chloramphenicol. From these plates 960 colonies were picked and phenotyped directly on increasing concentrations of chloramphenicol. 96 clones showing improved chloramphenicol resistance over pSertRNA(Tran)_CUA were selected and the tRNA plasmids were isolated by digestion of the pBK-SepRS and pCDF-EF-Sep-CAT. The tRNA plasmids were retransformed individually into chemically competent DH10B and plated to single colony density, then cultured and miniprepped to yield isolated tRNA plasmids. Sequencing of these clones revealed 15 unique tRNA sequences. These 15 tRNA plasmids and the pSertRNA(Tran)_CUA plasmid were transformed into DH10B ΔserB with pCDF-EF_Sep-CAT along with pBk-SepRS or pBK-NoRS, for observing synthetase dependent and independent incorporation respectively, and re-phenotyped on increasing concentrations of chloramphenicol. To test incorporation in the absence of EF-Sep, the 15 mutant tRNAs and pSertRNA(Tran)_CUA were co transformed and re-phenotyped as in the presence of EF-Sep but with the pCDF-CAT plasmid in place of pCDF-EF-Sep-CAT.

SepRS(XX) selection

In separate transformations, electrocompetent DH10B ΔserB cells containing each of the 13 functional mutant tRNAs (excluding the non-functional mutants A8 and C9) and either pCDF-CAT or pCDF-EF-Sep-CAT were transformed with the SepRS(AC6) anticodon library to a diversity of at least 1×10⁹c.f.u, giving full library coverage for every tRNA, and grown overnight. 1mL overnight culture of each pCDF-CAT transformant was used to inoculate 500ml LB + 2mM phosphoserine + 1mM IPTG. 1mL of each overnight culture was used to inoculate 500ml of the LB + 2mM phosphoserine + 1mM IPTG and grown to OD₆₀₀=0.6. Of these cultures 8×10⁹ cells, covering the mutant diversity requirement of 3.7× ×10⁹ c.f.u, was plated separately on LB + 2mM phosphoserine + 1mM IPTG and either 50, 100, 200, 300 and 400 μg/ml chloramphenicol and 600, 700, 800, 900 and 1000 μg/ml chloramphenicol for pCDF-CAT and pCDF-EF-Sep-CAT cultures respectively. 288 colonies were picked for each set and phenotyped directly on increasing chloramphenicol concentration plates. From this phenotyping, 14 EF-Sep dependent and 3 EF-Sep independent tRNA-synthetase pairs were found. These clones were picked, the plasmid DNA isolated, and the tRNA/synthetase for each was sequenced. The tRNA and CAT reporter plasmids were digested away and the synthetase plasmids were isolated by retransformation of each to single colony density, then miniprepping DNA from these colonies and plasmid isolation. The isolated synthetase plasmids were then co-transformed with their respective reporters, in the presence and absence of the corresponding tRNAs, for re-phenotyping on increasing chloramphenicol concentration plates.

Myoglobin expressions

The 15 selected tRNA(XXX)_CUA variants were transformed with pBK-SepRS and either pCDF-EF-Sep-MyoD127TAG_His6 or pCDF-MyoD127TAG_His6 for EF-Sep dependent and independent incorporation. Similarly, the 17 SepRS(XX) variants identified from the SepRS(AC6) library were transformed with their corresponding tRNA and either pCDF-EF-Sep-MyoD127TAG_His6 or pCDF-EF-Sep-MyoD127TAG_His6. 250ul of each overnight culture was inoculated into 50ml LB, grown to OD₆₀₀=0.6 and induced with 1mM IPTG, 0.2% arabinose and 2mM phosphoserine. The culture were grown for 4 hours and pelleted. Samples were boiled in 4× LDS, subject to SDS-PAGE, transferred to nitrocellulose and full-length protein detected using an anti-his HRP conjugate antibody (Novagen, 1:1000 dilution). To purify the protein for mass spectrometry, the pellet was lysed in 1× lysis buffer (1× BugBuster (Novagen), 50mM Tris, 300mM NaCl, 20mM Imidazole, 0.5mg/mL lysozyme, 50μg/mL DNase, 0.2mM PMSF and 1 protease inhibitor cocktail tablet (Roche) per 50ml lysis buffer) and the clarified lysate incubated overnight at 4°C with magnetic Ni-NTA beads (Qiagen). Bound protein was eluted in 50mM Tris, 300mM NaCl and 300mM Imidazole and analysed by Electrospray ionization mass spectrometry.

Expression and purification of recombinant GST-CaM incorporating phosphoserine

For overexpression of GST-CaM incorporating phosphoserine, DH10B ΔserB or C321(DE3) ΔserB E. coli cells were transformed by electroporation with pRSF and pKW vectors, and recovered in 1 ml SOB medium for one hour at 37°C prior to aliquoting to 100 ml LB-KC (LB media with 25 μg/ml kanamycin, and 17.5 μg/ml chloramphenicol) and incubated overnight (37°C, 250 r.p.m., 16 h). The overnight culture was diluted to OD₆₀₀=0.1 in LB-KC. At OD₆₀₀=0.5, IPTG (1 mM final concentration) and phosphoserine (2 mM) were added to the culture. The culture was incubated (37°C, 250 r.p.m.) for 5 h, pelleted (3,000g, 10 min, 4°C) and washed twice with 1 ml PBS. The overexpressed proteins were purified using the glutathione affinity chromatography. Cell pellets were resuspended in 1 ml of BugBuster Protein Extraction Reagent (Novagen) (supplemented with 1× protease inhibitor cocktail tablet (Roche), 1 mg ml⁻¹ lysozyme (Sigma), 1 mg ml⁻¹ DNase I (Sigma)) and lysed (25°C, 250 r.p.m., 1 h). The lysate was clarified by centrifugation (25,000g, 30 min, 4°C). GST containing proteins from the lysate were bound in batch (1 h, 4°C) to 70 μl of glutathione sepharose beads (GE Healthcare). Beads were washed 4 times with 1 ml PBS prior to elution by heating in 1× NuPAGE LDS sample buffer (Invitrogen, 95°C, 5 min) supplemented with 100 mM DTT. All samples were analysed on 4-12% Bis-Tris gels (Invitrogen) with BIO-RAD Low Range Molecular Weight Standard as marker. The gels were subsequently stained with Coomassie Blue (InstantBlue, Expedeon).

Expression and purification of Ubiquitin phosphorylated at serine 65

BL21 ΔserB(DE3) cells containing pKW2-EF-Sep (a chloramphenicol resistant plasmid containing SepRS 2, pSer tRNAB4_CUA and EF-Sep) and pNH-Ub(S65TAG)-His₆ (a tetracycline resistant plasmid containing LacI and Ub(S65TAG)-His₆) were grown overnight (37°C, 220 r.p.m., in TB+CT: TB media containing 25μg/mL chloramphenicol and 25μg/mL tetracycline). The culture was diluted 1:50 into 4L of fresh TB+CT and incubated at 37°C, 220 r.p.m. At OD₆₀₀=0.6 phosphoserine and IPTG were added to a final concentration of 2mM and 1mM respectively. After incubation (37°C, 220 r.p.m., 4hr), cells were harvested by centrifugation and re-suspended in ice-cold lysis buffer (PBS pH 7.4, 20mM imidazole, 0.5mg/mL lysozyme, 50μg/mL DNase, 0.2mM PMSF), then lysed by sonication. The lysate was clarified by centrifugation (39000 × g, 30 min) and purified using a 5mL HisTrap HP (AKTA explorer), with elution in PBS pH 7.4 + 400mM imidazole. Fractions were determined by SDS-PAGE and were pooled and concentrated to <9mL with a Amicon Ultra-15 3kDa MWCO centrifugal filter device (Millipore). The sample was then dialyzed against 10mM Tris pH 7.6 overnight at 4°C. 1mM DTT was added to the Ub(pSer65)-His₆ sample, followed by UCH-L3 to a final concentration of 15μg/mL. The sample was incubated at (37°C, 1 hr) to remove the C-terminal His₆ tag. Ub (pSer65) was then dialyzed against 50mM Ammonium acetate pH 4.5 overnight at 4°C and purified by cation exchange chromatography using a HiTrap SP HP column (ATKA explorer), with elution in 50mM Ammonium acetate pH 4.5 + 400mM NaCl. Pure fractions were confirmed by SDS-PAGE and were pooled and concentrated as before. The sample was then dialyzed against 10mM Tris 7.6, 150 mM NaCl at 4°C. Protein yield was quantified by SDS-PAGE gel relative to a commercial ubiquitin standard and then aliquoted and flash frozen at −80°C. Correct masses and site specific phosphoserine incorporation was confirmed by whole protein ESI-MS and LC-MS/MS.

Expression and purification of Nek7 phosphorylated at serine 195

BL21 Δser(DE3) cells containing pKW2-EF-Sep and pNH-Nek7(D179N/S195TAG)-His₆ (a tetracycline resistant plasmid containing LacI and Nek7(D179N/S195TAG)-His₆) were grown overnight (37°C, 220 r.p.m., in TB+CT: TB media containing 25μg/mL chloramphenicol and 25μg/mL tetracycline). The culture was diluted 1:50 into 4L of fresh TB+CT and incubated at 37°C, 220 r.p.m. At OD₆₀₀=0.6 phosphoserine and IPTG were added to a final concentration of 2mM and 1mM respectively. After incubation (18°C, 220 r.p.m., 16hr), cells were harvested by centrifugation and re-suspended in ice-cold lysis buffer (50mM HEPES pH 7.5, 300mM NaCl, 5% Glycerol, 20mM Imidazole, 1mM MgCl₂, 0.2mM MnCl₂ and 50μg/mL DNase), then lysed by sonication. The lysate was clarified by centrifugation (39000 × g, 30 min) and purified using a 5mL HisTrap HP (AKTA explorer), with elution in 50mM HEPES pH 7.5, 300mM NaCl, 5% Glycerol, 250mM Imidazole. D179N-pSer195containing fractions were determined by SDS-PAGE and were pooled and concentrated to <5mL with a Vivaspin 20 (GE Healthcare) with a molecular weight cut off (MWCO) of 10kDa. The protein was then further purified by size exclusion chromatography using a HiLoad Superdex S200 16/60 (GE Healthcare) gel filtration column and eluted in 50mM HEPES pH 7.5, 300mM NaCl, 5% Glycerol, 5mM DTT, the protein was then aliquoted and flash frozen at −80°C until required. Correct masses and site specific phosphoserine incorporation was confirmed by whole protein ESI-MS and LC-MSMS. Nek7(S195TAG)-His₆ was expressed and characterized as described above.

Phosphoserine analog incorporation

Phosphoserine analog incorporation was performed and analysed as for myoglobin (D127TAG)_His6 expression with phosphoserine, with the exception of using Keio WT strain (BW25113), Keio ΔserB or Keio ΔserC with plasmids pKW2, pCDF-EF-Sep-serB and p15A-Myo(D127TAG)_His6 for the reporter, as described in the text. DL-AP4 sodium salt (CH₂ substituted phosphoserine analogue, 2) was obtained from Abcam and used in place of phosphoserine where indicated at a final concentration of 2mM.

Ub(pSer65) in vitro E2-discharge assay

Wild-type Parkin (His6–SUMO cleaved) and enzymatic derived Ub(pSer65), hereby referred to as Ub(pSer65*), was expressed and purified as previously described ¹⁹. The E2-charging reaction was performed in a 5μl volume containing 0.5 μg Ube1, 2 μg UbcH7, 50 mM Hepes pH 7.5 and 10 μM FLAG-ubiquitin, in the presence of 2 mM magnesium acetate and 0.2 mM ATP. After an initial incubation of 60 min at 30°C, full-length wild-type Parkin was added in the presence of Ub(pSer65), Ub(pSer65*) and non-phospho-ubiquitin. The reaction was allowed to continue for a further 10 min at 30°C. Reactions were terminated by the addition of 5μl of LDS sample loading buffer and were subjected to SDS/PAGE (4–12 % gel) analysis in the absence of any reducing agent. Gels were stained using InstantBlue.

Ub(pSer65) in vitro ubiquitylation assay

Wild-type Parkin (2μg) was added to ubiquitylation assay component mastermix (50mM Tris pH 7.5, 5mM MgCl2, 0.12μM ubiquitin E1, 1μM UbcH7, 2μg of His6–SUMO-Miro1 and 2mM ATP) to a final volume of 50μl. Each reaction contained 0.05mM ubiquitin comprising 25μg of FLAG-ubiquitin (Boston Biochem) mixed with 5μg of Ub(pSer65), Ub(pSer65*) or non-phospho-ubiquitin. Ubiquitylation reactions were carried out at 30°C for 1hr and terminated by the addition of LDS sample buffer. For all assays, reaction mixtures were resolved by SDS/PAGE (4–12% gel). Ubiquitylation reactions were subjected to analysis by immunoblotting as follows: ubiquitin (anti-FLAG antibody; Sigma, 1:10000), Parkin (anti-Parkin antibody; Santa Cruz, 1:5000) and Miro-1 (anti-SUMO1 antibody; 1:2000).

Nek7-His₆ Kinase assay

Nek7-His₆ or the indicated variant and GST-Nek9-CTD were incubated in 50mM HEPES pH7.4, 5mM NaF, 5mM MnCl₂, 5mM β-glycerolphosphate, 1mM DTT and Myelin Basic Protein (MBP) (Room temperature, 1h). After incubation, 1mCi radiolabelled ³²P-ATP and 4μM unlabelled ATP were added and incubated (30°C, 30min, 200r.p.m). LDS sample loading buffer was added to the reactions and they were subjected to SDS/PAGE (4–12% gel). The gel was stained with Coomassie and then analysed by autoradiography. Efficiency of ³²P incorporation into both Nek7(D179N-pSer195)-His₆ by autophosphorylation and MBP substrate was determined by scintillation counting.

Electrospray ionization mass spectrometry

Mass spectra for protein samples were acquired on an Agilent 1200 LC-MS system that employs a 6130 Quadrupole spectrometer. The solvent system used for liquid chromatography (LC) was 0.2 % formic acid in H₂O as buffer A, and 0.2 % formic acid in acetonitrile (MeCN) as buffer B. Samples were injected into Phenomenex Jupiter C4 column (150 × 2 mm, 5 μm) and subsequently into the mass spectrometer using a fully automated system. Spectra were acquired in the positive mode and analyzed using the MS Chemstation software (Agilent Technologies). The deconvolution program provided in the software was used to obtain the mass spectra. Theoretical average molecular weight of proteins with unnatural amino acids was calculated by first computing the theoretical molecular weight of wild-type protein using an online tool (http://www.peptidesynthetics.co.uk/tools/), and then manually correcting for the theoretical molecular weight of unnatural amino acids.

Electrospray ionization tandem mass spectrometry

Polyacrylamide gel slices (1-2 mm) containing the purified proteins or proteins in solution were prepared for mass spectrometric analysis by manual in situ enzymatic digestion. Briefly, the excised protein gel pieces were placed in a well of a 96-well microtitre plate and destained with 50% v/v acetonitrile and 50 mM ammonium bicarbonate, reduced with 10 mM DTT, and alkylated with 55 mM iodoacetamide. After alkylation, proteins were digested with 6 ng/μL Trypsin (Promega, UK) overnight at 37 °C. The resulting peptides were extracted in 2% v/v formic acid, 2% v/v acetonitrile. The digest was analysed by nano-scale capillary LC-MS/MS using a Ultimate U3000 HPLC (ThermoScientific Dionex, San Jose, USA) to deliver a flow of approximately 300 nL/min. A C18 Acclaim PepMap100 5 μm, 100 μm × 20 mm nanoViper (ThermoScientific Dionex, San Jose, USA), trapped the peptides prior to separation on a C18 Acclaim PepMap100 3 μm, 75 μm × 150 mm nanoViper (ThermoScientific Dionex, San Jose, USA). Peptides were eluted with a gradient of acetonitrile. The analytical column outlet was directly interfaced via a modified nano-flow electrospray ionisation source, with a hybrid dual pressure linear ion trap mass spectrometer (Orbitrap Velos, ThermoScientific, San Jose, USA). Data dependent analysis was carried out, using a resolution of 30,000 for the full MS spectrum, followed by ten MS/MS spectra in the linear ion trap. MS spectra were collected over a m/z range of 300–2000. MS/MS scans were collected using an threshold energy of 35 for collision induced dissociation. LC-MS/MS data were then searched against an in house protein sequence database, containing Swiss-Prot and the protein constructs specific to the experiment, using the Mascot search engine programme (Matrix Science, UK). Database search parameters were set with a precursor tolerance of 5 ppm and a fragment ion mass tolerance of 0.8 Da. Two missed enzyme cleavages were allowed and variable modifications for oxidized methionine, carbamidomethyl cysteine, pyroglutamic acid and phosphorylated serine, threonine, tyrosine, were included. MS/MS data were validated using the Scaffold programme (Proteome Software Inc., USA). All data were additionally interrogated manually.

Statistical methods

Bar graphs show the mean of the indicated number of trials. All error bars represent the s.d.