Cover Page
The handle
http://hdl.handle.net/1887/90130
holds various files of this Leiden University
dissertation.
Author: Witting, K.F.
Chapter 2
Covalently capturing E2-E3 enzymes during
Ubiquitin transfer with a genetically encoded
2
Preface
Reprogramming the translational machinery to genetically
encode unnatural amino acids
Site-specific modification of proteins is critical for investigation and manipulation of
structure-function relationships[1], protein localization[2], as well as protein-protein
interactions and function[3]. Although chemical methodologies such as expressed protein
ligation (EPL) have been devised to modify proteins, these approaches suffer from several
limitations, most notably, protein size and specificity[4, 5]. Thus, exploiting the endogenous
cellular biosynthetic machinery for co-translational incorporation of an unnatural amino acid (UAA)—amino acids that are genetically not encoded—enables these modifications to be introduced site-specifically in vivo. While early attempts at incorporation of unnatural amino
acids utilized chemically acetylated suppressor tRNAs[6, 7], the breakthrough was achieved by
reassignment of the amber codon (UAG)—the least used stop codon in Escherichia coli [8- 10].
Undoubtedly, the utility of genetic code expansion is exemplified by the incorporation of a broad spectrum of unnatural amino acids ranging from photo-crosslinkers for studying
protein-protein or protein-ligand interactions[11, 12] to fluorophores[13-15], linkers for
antibody-drug conjugates[16-18], and photocaged amino acids for temporal and spatial control of
proteins. Additionally, new unnatural amino acids have been added to the genetic lexicon
to site-specially insert post-translational modifications (PTMs) such as methylation[19-21],
crotonylation[22, 23], phospho-tyrosine,[24-26] or by incorporation of a chemical group allowing
introduction of PTMs such as Ubiquitination[27, 28].
The genetic code predetermines which amino acid is inserted by the aminoacyl-tRNA in response to the codon triplet. Since the genetic code encodes for 20 natural amino acids using 61 sense codons and three nonsense codons (stop codons), utilizing these enables an unnatural amino acid to be genetically inserted into the growing polypeptide chain. Protein synthesis is achieved by decoding the genetic information of messenger RNA (mRNA) into the corresponding amino acids by the concerted action of the ribosome, a multiprotein complex
devoted to synthesizing proteins[29]. Initially, the appropriate amino acid is covalently attached
to the tRNA molecule by the tRNA synthetase under the consumption of ATP (Figure 1A)[29].
During translation, the anticodon region of appropriately charged transfer RNAs (tRNAs) form base pairs with the complementary three nucleotide codon on the mRNA, thereby inserting the correct amino acid into the growing polypeptide chain (Figure 1B). Termination of prokaryotic protein synthesis is achieved by recognition of the stop codon by release factors—RF1 (for
UAA and UAG) and RF2 (for UAA and UGA)[30]. Interestingly, RF1 structurally resembles a tRNA
and thus directly competes with the suppressor tRNAs for the amber codon[31], dramatically
reducing UAA incorporation efficiency and increasing protein truncation[32] (Figure 1B).
Figure 1| Concept of ribosomal protein translation and amber codon suppression. A) An appropriate tRNA is charged with the amino acids (natural or unnatural) by the corresponding tRNA synthetase in an ATP-dependent reaction. B) During ribosomal protein translation, a charged tRNA is inserted in response to the corresponding anticodon. For amber suppression, the tRNA recognizes the amber codon (AUG) and inserts the unnatural amino acid at this position in the mRNA. Protein synthesis is terminated by release factor 1 (RF1), releasing the folded polypeptide.
Genetic code expansion relies on the principle of establishing anticodon recognition between a charged tRNA and the complementary mRNA base pairs, resulting in the incorporation of the charged non-proteinogenic amino acid. Thus, in principle, mutagenesis of the anticodon region of the tRNA as well as evolution of the synthetase should suffice to co-translationally introduce unnatural amino acids into proteins. However, in practice, engineering such an aminoacyl-tRNA/tRNA synthetase pair is not trivial requiring a robust selection methodology
evolving both the tRNA and the tRNA synthetase[33]. But how is the evolution of an orthogonal
tRNA synthetase that does not cross-react with endogenous towards recognition of an unnatural amino acid achieved? While early attempts have utilized chemically acylated
tRNAs and cell-free transcription/translation systems[6,7], directed evolution permitted the
creation of orthogonal tRNA/ tRNA synthetase pairs[34, 35]. An important aspect for evolving
a tRNA/tRNA synthetase pair endowed with the ability to recognize and charge a specific unnatural amino acid, is the generation of a diverse yet manageable mutagenesis library. Typically, such a mutagenesis library is generated by randomizing the appropriate codons to encode each of the 19 natural amino acids using degenerate NNK (N = A, C, G, or T, and
K = G or T) codons[34]. However, one caveat of such an approach is that the libraries easily
reach the limit of bacterial transformation (109 colony forming units (cfu)) thereby greatly
restricting the number of residues that can be randomized[36]. Given these limitations,
2
A) B) Transformation of mutagenesis library Chloramphenicol + UAAOnly clones with aaRS charged with unnatural amio acid
Transformation
Plasmid contains CAT gene with TAG codon
Plasmid contains UPRT gene with TAG codon
No UAA Only clones with
aaRS charged with unnatural amio acid
POSIT IVE SELECTION N EG AT IVE S ELE CT IO N 1 2 3 4 5 + 5-FU Unnatural amino acid PylRS synthetase ATP Multiple rounds
cat-uprt fusion gene
tetR urpt cat p15a ori PylRS PylT pREPCM D181_TAG Q98_TAG
Figure 2| Construct and general selection procedure using dual-selection strategy to evolve a PylRS capable of charging an unnatural amino acid. A) Composition of dual selection vector (adapted from Melçanon et al.[33]). B) Selection procedure involves multiple rounds of both positive (in the presence of the UAA and chloramphenicol) and negative selection (in the presence of 5-FU).
While amber codon suppression technology has been extensively exploited to introduce a myriad of structurally and functionally diverse modifications in proteins, the technology is still in its infancy. Notwithstanding the substantial improvements that have been made since the advent of the genetic code expansion, such as knockout of the release factor 1
(RF1)[30] and engineering of RF2[39], evolution of the ribosome and the elongation factor
(EF-TU[40] as well as the use of an optimized quadruplet codon approach[41], the metabolic
challenges encountered using this technique still need to be addressed. Recently, attempts to address this problem, have been made by using CRISPR interference (CRISPRi) to engineer
RF1[42], foreshadowing its utility for engineering bacteria to accommodate for the metabolic
alterations arising from genetic code reassignment[43]. However, this technique still faces
numerous challenges such as low incorporation levels[44], adaptability for certain unnatural
amino acids, inefficient unnatural amino acid uptake, as well as metabolic alterations[45, 46]
which need to be resolved before being able to incorporate a larger variety of structurally diverse amino acids.
Although, genetic code expansion has undeniably been valuable to the generation of
immuno-conjugates[16, 18, 47], the technology still requires further development to enable
protein expression in vivo for studying protein-function studies in their native biological contexts. Especially installation of expedient unnatural amino acids in Ubiquitin conjugating and ligating enzymes would enable the targeted capture of their interactors and substrates greatly advancing insights into Ubiquitin biology. In an attempt to modify a Ubiquitin-conjugating enzyme to enable the formation of a covalent Ubiquitin intermediate, genetic encoding of L-2,3-diaminopropionic acid (Dap) was endeavored (Chapter 2).
Additionally, the use of newer dual selection systems permitted efficient selection, without
the need for numerous retransformations that compromise the library quality[33, 37]. In contrast
to the traditional selection systems that utilize amber codon suppression of the barnase gene for negative selections, this vector exploits the cat-uracil-phosphoribosyltransferase
(uprt) fusion gene[37]. Upon expression, the Uracil-phosphoribosyltransferase (UPRT)
converts 5-fluorouracil (5-FU) to 5-fluoro-dUMP, an inhibitor of the thymidylate synthetase,
resulting in DNA damage and ultimately cell death[37]. Insertion of an amber codon at a
permissive site (D181) in the cat-portion of the gene permits the negative and positive selection in the presence of both chloramphenicol and the unnatural amino acid (positive
selection) and then in the presence of 5-FU (negative selection)[37]. One advantage of this
dual selection vector system is that the selection can be modulated over a dynamic range of Chloramphenicol (Cm) and 5-FU concentrations, depending on the bacterial strain being used (5-25 µg/mL and 75-125 µg/mL) Cm and 5-FU of 100-200 µg/mL (Figure 2B). This negative selection method utilizing on 5-FU is very effective resulting in sequence
convergence after only three selection rounds[37]. To monitor the incorporation fidelity and
efficiency of the evolved clones, the fluorescence intensity of GFP, which is only fluorescent
if the UAA is inserted into the protein, is measured[37].
Although several (aaRS)-tRNA pairs have been engineered by Schultz and others to introduce a variety of unnatural amino acid into proteins, use of the unique pyrrolysl tRNA synthetase-tRNA pair (PylRS/PylCUA) allows co-translational insertion of unnatural amino acids in response to the amber codon. Pyrrolysine (Pyl or O), the 22nd proteinogenic amino acid, is the only amino acid encoded by the amber codon in the methanogenic archaea Methanosarcina barkeri greatly facilitating the engineering for amber codon suppression. In contrast to selenocysteine (Sec or U), the 21st proteinogenic amino acid, which requires additional post-translational modification, pyrrolysine is directly inserted by a dedicated tRNA synthetase and its cognate tRNA. Additional attributes rendering this tRNA synthetase amenable to genetic code expansion include its high substrate side-chain promiscuity,
relaxed α-amine selectivity, and low tRNAPyl selectivity[38]. Additionally, lack of the editing
domain and non-specific hydrophobic interactions with its substrate common to other canonical amino acid tRNA synthetases (aaRS) and tRNAs immensely facilitates genetic
engineering towards recognition of a variety of unnatural amino acids[38].These unique
features have easily permitted the evolution pyrrolysine tRNA synthetase pair to site-specifically incorporate more than 100 UAAs as well as α-hydroxy amino acids using amber
codon suppression[38]. More recently, the pyrrolysine tRNA synthetase has been successfully
reassigned to the other more frequently used stop codons ochre (UAA) and opal (UGA) as
2
1557-60.
21. Nguyen, D.P., et al., Genetically encoding N(epsilon)-methyl-L-lysine in recombinant histones. J Am Chem Soc, 2009. 131(40): p. 14194-5.
22. Tan, M., et al., Identification of 67 histone marks and histone lysine crotonylation as a new type of histone modification. Cell, 2011. 146(6): p. 1016-28.
23. Kim, C.H., et al., Site-specific incorporation of epsilon-N-crotonyllysine into histones. Angew Chem Int Ed Engl, 2012. 51(29): p. 7246-9.
24. Fan, C., K. Ip, and D. Soll, Expanding the genetic code of Escherichia coli with phosphotyrosine. FEBS Lett, 2016. 590(17): p. 3040-7.
25. Luo, X., et al., Genetically encoding phosphotyrosine and its nonhydrolyzable analog in bacteria. Nat Chem Biol, 2017. 13(8): p. 845-849.
26. Hoppmann, C., et al., Site-specific incorporation of phosphotyrosine using an expanded genetic code. Nat Chem Biol, 2017. 13(8): p. 842-844.
27. Virdee, S., et al., Traceless and site-specific Ubiquitination of recombinant proteins. J Am Chem Soc, 2011. 133(28): p. 10708-11.
28. Huguenin-Dezot, N., et al., Trapping biosynthetic acyl-enzyme intermediates with encoded 2,3-diaminopropi-onic acid. Nature, 2019. 565(7737): p. 112-117.
29. Clark, D.P., N.J. Pazdernik, and M.R. McGehee, Protein Synthesis, in Molecular Biology (Third Edition). 2019, Cell. p. 1006.
30. Johnson, D.B., et al., RF1 knockout allows ribosomal incorporation of unnatural amino acids at multiple sites. Nat Chem Biol, 2011. 7(11): p. 779-86.
31. Xie, J. and P.G. Schultz, A chemical toolkit for proteins--an expanded genetic code. Nat Rev Mol Cell Biol, 2006. 7(10): p. 775-82.
32. Patrick, W.M. and A.E. Firth, Strategies and computational tools for improving randomized protein libraries. Biomol Eng, 2005. 22(4): p. 105-12.
33. Melancon, C.E., 3rd and P.G. Schultz, One plasmid selection system for the rapid evolution of aminoacyl-tRNA synthetases. Bioorg Med Chem Lett, 2009. 19(14): p. 3845-7.
34. Reetz, M.T. and J.D. Carballeira, Iterative saturation mutagenesis (ISM) for rapid directed evolution of func-tional enzymes. Nat Protoc, 2007. 2(4): p. 891-903.
35. Reetz, M.T., L.W. Wang, and M. Bocola, Directed evolution of enantioselective enzymes: iterative cycles of CASTing for probing protein-sequence space. Angew Chem Int Ed Engl, 2006. 45(8): p. 1236-41.
36. Jacobs, T.M., et al., SwiftLib: rapid degenerate-codon-library optimization through dynamic programming. Nucleic Acids Res, 2015. 43(5): p. e34.
37. Santoro, S.W. and P.G. Schultz, Directed evolution of the substrate specificities of a site-specific recombinase and an aminoacyl-tRNA synthetase using fluorescence-activated cell sorting (FACS). Methods Mol Biol, 2003. 230: p. 291-312.
38. Wan, W., J.M. Tharp, and W.R. Liu, Pyrrolysyl-tRNA synthetase: an ordinary enzyme but an outstanding genet-ic code expansion tool. Biochim Biophys Acta, 2014. 1844(6): p. 1059-70.
39. Korkmaz, G. and S. Sanyal, R213I mutation in release factor 2 (RF2) is one step forward for engineering an omnipotent release factor in bacteria Escherichia coli. J Biol Chem, 2017. 292(36): p. 15134-15142.
References
1. Narancic, T., S.A. Almahboub, and K.E. O’Connor, Unnatural amino acids: production and biotechnological potential. World Journal of Microbiology & Bio technology, 2019. 35(4).
2. Charbon, G., et al., Subcellular Protein Localization by Using a Genetically Encoded Fluorescent Amino Acid. Chembiochem, 2011. 12(12): p. 1818-1821.
3. Nguyen, T.A., M. Cigler, and K. Lang, Expanding the Genetic Code to Study Protein-Protein Interactions. Ang-ewandte Chemie-International Edition, 2018. 57(44): p. 14350-14361.
4. Dawson, P.E., et al., Synthesis of proteins by native chemical ligation. Science, 1994. 266(5186): p. 776-9. 5. Kent, S., Chemical protein synthesis: Inventing synthetic methods to decipher how proteins work. Bioorg Med
Chem, 2017. 25(18): p. 4926-4937.
6. Noren, C.J., et al., A general method for site-specific incorporation of unnatural amino acids into proteins. Science, 1989. 244(4901): p. 182-8.
7. Young, T.S. and P.G. Schultz, Beyond the canonical 20 amino acids: expanding the genetic lexicon. J Biol Chem, 2010. 285(15): p. 11039-44.
8. Wang, L. and P.G. Schultz, A general approach for the generation of orthogonal tRNAs. Chem Biol, 2001. 8(9): p. 883-90.
9. Sun, J., et al., Relationships among stop codon usage bias, its context, isochores, and gene expression level in various eukaryotes. J Mol Evol, 2005. 61(4): p. 437-44.
10. Korkmaz, G., et al., Comprehensive analysis of stop codon usage in bacteria and its correlation with release factor abundance. J Biol Chem, 2014. 289(44): p. 30334-42.
11. Tian, Y. and Q. Lin, Genetic encoding of 2-aryl-5-carboxytetrazole-based protein photo-cross-linkers. Chem Commun (Camb), 2018. 54(35): p. 4449-4452.
12. Yang, Y., H. Song, and P.R. Chen, Genetically encoded photocrosslinkers for identifying and mapping pro-tein-protein interactions in living cells. IUBMB Life, 2016. 68(11): p. 879-886.
13. Summerer, D., et al., A genetically encoded fluorescent amino acid. Proc Natl Acad Sci U S A, 2006. 103(26): p. 9785-9.
14. Lang, K., et al., Genetically encoded norbornene directs site-specific cellular protein labelling via a rapid bio-orthogonal reaction. Nat Chem, 2012. 4(4): p. 298-304.
15. Lemke, E.A., Site-specific labeling of proteins for single-molecule FRET measurements using genetically en-coded ketone functionalities. Methods Mol Biol, 2011. 751: p. 3-15.
16. Tian, F., et al., A general approach to site-specific antibody drug conjugates. Proc Natl Acad Sci U S A, 2014. 111(5): p. 1766-71.
17. Kim, C.H., et al., Synthesis of bispecific antibodies using genetically encoded unnatural amino acids. J Am Chem Soc, 2012. 134(24): p. 9918-21.
18. Axup, J.Y., et al., Synthesis of site-specific antibody-drug conjugates using unnatural amino acids. Proc Natl Acad Sci U S A, 2012. 109(40): p. 16101-6.
19. Groff, D., et al., A genetically encoded epsilon-N-methyl lysine in mammalian cells. Chembiochem, 2010. 11(8): p. 1066-8.
2
Abstract
Capturing the conformational changes occurring during E3 ligase binding to cognate Ub-charged E2 enzymes necessitates the generation of stable thioester mimics to enable structural studies. Covalent bond formation between the E2 enzyme and Ubiquitin can be created by genetically introducing an unnatural amino acid (UAA) amenable to reacting with Ubiquitin to yield a non-hydrolyzable E2-Ub complex retaining structural features of the thioester intermediate. Incorporation of L-azidoalanine, an unnatural amino acid which can be reduced to yield 2,3-diaminopropionic acid (Dap), into an E2 enzyme was undertaken by evolving PylRS/PylCUA. Since this approach was unsuccessful, chemical modification of the active-site cysteine into a dehydroalananine (Dha) and subsequent amination was alternatively attempted. The difficulty to alter the active site of an enzyme using either genetic encoding or chemical modification underscores the challenge of this specific protein manipulation and warrants the exploration of alternative methodologies.
40. Gan, R., et al., Translation system engineering in Escherichia coli enhances non-canonical amino acid incorpo-ration into proteins. Biotechnol Bioeng, 2017. 114(5): p. 1074-1086.
41. Chatterjee, A., et al., A bacterial strain with a unique quadruplet codon specifying non-native amino acids. Chembiochem, 2014. 15(12): p. 1782-6.
42. Zhang, B., et al., CRISPRi-Manipulation of Genetic Code Expansion via RF1 for Reassignment of Amber Codon in Bacteria. Sci Rep, 2016. 6: p. 20000.
43. Wang, Q., et al., Response and adaptation of Escherichia coli to suppression of the amber stop codon. Chem-biochem, 2014. 15(12): p. 1744-9.
44. Wang, J. and A.C. Forster, Ribosomal incorporation of unnatural amino acids: lessons and improvements from fast kinetics studies. Curr Opin Chem Biol, 2018. 46: p. 180-187.
45. Wandrey, G., et al., Probing unnatural amino acid integration into enhanced green fluorescent protein by genetic code expansion with a high-throughput screening platform. J Biol Eng, 2016. 10: p. 11.
2
genetic incorporation mild reduction E1, ATP, MgCl2 Ub L-Azidoalanine (1) N3 H2N COOH A) AMP + PPi PylRS ATPunnatural amino acid (UAA)
E. coli 3 N E2 2 NH E2 NH Ub E2 B) NH E2 E3 Ub E3 HOOC NH2 HN O O Boc-Lysine (2) Coomassie GFP BocL ys AzAla No UAA 26 kDa AmpR UbcH5c_TAG85 E2-TAG85
Figure 1| “Covalent substrate capture” of E3-ligases. A) Schematic representation of envisioned strategy depicting the covalent trapping of an E2-Ub- thioester intermediate by genetically incorporating L-azidoalanine (1) in the active site, followed by mild reduction to yield the amine nucleophile. After charging the modified E2-enzyme with Ubiquitin in an E1- catalyzed reaction, this conjugate can be used to bind cognate RING-E3 ligases. B) Coomassie-stained SDS-PAGE gel visualizing purified GFP that had been expressed in the presence of either L-azidoalanine (1) or Boc-Lysine (2) or in the absence of unnatural amino acids.
Results
Unnatural amino acid L-azidoalanine (AzAla) does not adversely affect bacterial growth
Prerequisite for incorporation of an unnatural amino acid is its cellular uptake and metabolic compatibility with the expression host. Uptake of proteinogenic amino acids into the bacteria occurs through dedicated protein-based transport mechanisms with varying degrees of specificity. However, due to their relaxed specificity, these same transporters also shuttle non-canonical amino acids across the bacterial cell wall allowing them to be metabolically processed and utilized in protein synthesis. Prior evaluation of an unnatural amino acid’s toxicity using standard growth assays is critical as some non-proteinogenic amino acids can have unexpected metabolic effects. To evaluate the toxicity of L-azidoalanine, an overnight culture of E. coli BL21(DE3) or the release factor 1 (RF1) deficient JX33 E. coli strain harboring either the wild type PylRS or the evolved PylRS KW1 and the pBAD-GFP35-TAG
Introduction
Ubiquitination is orchestrated by the sequential action of three enzymes—E1, E2, and E3 enzymes to covalently modify the lysine residues of its substrate proteins. Through the initial adenylation of the C-terminal glycine of Ubiquitin and subsequent E1-mediated thioester formation, transfer to the active site cysteine of the E2 enzyme is achieved. Upon formation of a E2-Ub thioester intermediate, this complex is poised to undergo either nucleophilic attack by a HECT/RBR E3-ligase or interaction with a RING E3 ligase to be subsequently form an isopeptide bond with the substrate lysine. While these intermediate Ub-E2 thioester complexes are energetically activated to facilitate enzymatic transfer, this integral feature also renders them labile, often with half-lives of a few hours thus complicating structural studies[48, 49].
Transfer of activated Ub from E2 enzyme via the E3 ligase to the substrate is a highly dynamic process associated with a myriad of conformational changes in both enzymes[50]. However, the exact mechanistic details on how Ubiquitin transfer from the E2 to the E3 enzyme occurs
as well as the accompanying conformational changes remain enigmatic[51]. Thus, generating
stable E2-Ub thioester complexes that can be used to dissect the binding interfaces and
the dynamics during Ubiquitin transfer is of importance[50, 52]. Previously, stabilization of
E2-Ub thioesters to enable structural characterization has been performed by either using
oxyesters[50], disulfides[51, 53], or more recently lysines[54]. Yet, one of the greatest caveats in
recapitulating the thioester bond between Ubiquitin (Ub) and the E2-enzyme as closely in length and chemical properties as possible is finding a subtly altered yet consummate substitute for the cysteine active site. Nonetheless, the cysteine surrogates such as the oxyester, disulfide and lysine used to stabilize the thioester bond do not reflect the chemical nature of Ubiquitin transfer, hampering the dissection of the underlying molecular
mechanisms[50, 55, 56]. While Plechanovova et al. have demonstrated that substitution of
the active site cysteine of an E2 enzyme (UBCH5a) with a lysine residue permitted the necessary structural rearrangement of the E2 to facilitate formation of an isopeptide bond
with Ubiquitin[54], the newly formed bond does not accurately reflect the length of an E2-Ub
thioester conjugate[54].
Hence, alteration of the active-site cysteine of an E2 enzyme with a suitable isostere such as L-2,3-diaminopropionic acid (Figure 1) would be attractive. However, genetically encoding this unnatural amino acid in an unprotected form is currently unfeasible due to the mutagenicity
arising from its metabolites [57, 58]. To circumvent this, we opted to mask the free amino group,
for example in the form of an azide (L-azidoalanine) (Figure 1), thus permitting the facile
reduction to an amine under reductive mild conditions[59, 60]. This derivative might prove
to be amenable to amber codon suppression technology since numerous azido-derivatives
have been incorporated into proteins employing this methodology[61-64]. Such modification
2
Due to the short length of the amino side chain of L-azidoalanine, the six residues critical for substrate binding were randomized in order to drastically reduce the PylRS binding cavity size. Based on previous studies demonstrating that the residues Asp346, Cys348, and Val401 in the binding cavity of the synthetase are predominantly responsible for recognition and
binding of substrates with shorter side chains[65], these positions were chosen for
random-ization (Figure 3A and Figure S3B). To generate a mutagenesis library, the reported PylRS
fixed mutant Y384F, reported to improve the catalytic activity of the tRNA synthetase[66]
was used, with each of these three residues (N346, C348, and V401) randomized by using degenerate NNK codons.
Subsequently, this library was transformed into competent E. Coli BL21(DE3) to yield a
library of 107 colony forming units (cfu) and subjected to several rounds of positive and
negative selection. Firstly, the positive selection was performed in the presence of 100 µg Chloramphenicol (Cm) and 1 mM L-azidoalanine followed by a negative selection of the
survivors using 200 µg 5-Fluorouracil (5-FU)[33]. As a control, Boc-Lysine which has been
demonstrated to be incorporated by the WT-PylRS was used. After three rounds of both positive and negative selection in the presence of the UAA and Cm or 5-FU, respectively, the library converged to yield one clone KW1 (Y384F, A302T, C346W, N348V, V401L). Interestingly, this mutant PylRS has been reported in literature to accept UAAs structurally
not related to lysine[65]. Incorporation efficiency and fidelity were assayed by co-expressing
the mutant PylRS (KW1) with GFP-containing an amber codon at a permissive site (GFP-TAG35) in the presence of 1mM L-azidoalanine (1), 1mM Boc-Lysine (2) or the absence of unnatural amino acid as controls, respectively, and measuring the GFP-fluorescence
intensity[67]. Both fluorescence and gel-based assays reveal that L-azidoalanine seems to be
incorporated in response to the amber codon in GFP, resulting in the full-length protein as assessed by fluorescence intensity, SDS-PAGE gel and immunoblotting (Figure 3C and D). Furthermore, to assess what effect the individual mutations would have on the L-azidoalanine recognition and subsequent incorporation, each of these individual mutants obtained from this library were assayed for incorporation of the unnatural amino acid (Figure 3E).
Contribution of the individual PylRS residues to L-Azidoalanine recognition
Given the complexity of the mutagenesis library upon randomization of six residues, the individual contributions of the individual positions in recognizing the UAA have previously not been examined. To investigate this, the cooperative effect of the individual mutations was examined by screening for L-azidoalanine incorporation into GFP-TAG35 using a fluorescent-based readout. Given the relevance of the A302T mutation in adenylate recognition of
the aminoacylated UAA[65], this mutant together with the fixed Y384F mutant were used
to generate the mutagenesis libraries. Firstly, the effect of the “gatekeeper’’ position N346 on the incorporation of AzAla was examined in the presence of either A302T—a plasmid was diluted (1:100) in LB media and supplemented with 100 µg ampicillin, 34 µg /
µL chloramphenicol, and 1-10 mM of L-azidoalanine or 1 mM Boc-Lysine (or no unnatural amino acid as a control), the bacterial growth was measured at 600 nm every 20 minutes at 37°C for 8 hours. Addition of up to 10 mM L-azidoalanine had negligible effects in E. coli BL21(DE3) while the RF-1 deficient E. coli strain JX33 displayed slightly decreased growth in the presence of 5-10 mM unnatural amino acid. In contrast, however, high concentrations of L-azidoalanine (10 mM) promoted bacterial growth of E. coli BL21(DE3) co-expressing the mutant PylRS (KW1) and the pBAD-GFP35_TAG, while concentrations of 1 mM accelerated bacterial growth of the JX33 E.coli strain (Figure S1). Since both E. coli strains were tolerant towards the unnatural amino acid L-azidoalanine, the evolution of the pyrrolysyl tRNA synthetase was undertaken.
Given the tolerance of both E. coli strains towards the unnatural amino acid L-azidoalanine, the evolution of the pyrrolysine tRNA synthetase (PylRS) towards recognizing L-azidoalanine (AzAla), was undertaken by designing a saturation mutagenesis library containing degen-erate NNK (N = A, C, G or T and K = G or T) codons encoding all 19 natural amino acids at
permissive sites[32]. However, one caveat of such an approach is that the libraries easily reach
the limit of bacterial transformation (109 colony forming units (cfu)) thereby allowing 5-6
positions to randomized at once[36].
A) AmpR ori pBAD-GFP_TAG35 araC araBAD promoter GFP-TAG35 PylRS p15a ori PylT pEVOL CmR araC E. coli
Culture in 96-well plate (OD measurement over 6h)
1-10 mM AzAla/ 1 mM BocLysine time (min) 0 200 400 600 0.0 0.3 0.6 0.9 1.2 B) C) OD 600 (nm) No UAA 1mM AzAla 2mM AzAla BocLysine 5mM AzAla 10mM AzAla JX33 0 200 400 600 0.0 0.5 1.0 1.5 time (min) OD 600 (nm) 0 200 400 600 0.0 0.3 0.6 0.9 1.2 time (min) BL21(DE3) OD 600 (nm) No UAA 1mM AzAla 2mM AzAla 5mM AzAla 10mM AzAla BocLysine
2
A) A302 N346 C348 V401 Boc-Lysine B) Y384F positive/negative selection PylRS tetR cat-uprt fusion gene pREPCM TAG Y384F A302T N346NNKC348NNK V401NNK catalytic domain cat-uprt fusion gene pREPCM TAG tetR N346V C348W V401L A302T KW1 Mutagenesis Library C) D) IB: α-GFP GFP 26 Coomassie GFP BocL ys AzAla No UAA 26 kDa 0.0 0.5 1.0 1.5 Boc-L ysine-UAA WT KW 1KW 2 KW 3 KW 4KW 5 KW 1: N346V/C348W/A302T/V401L KW 2: N346V KW 3: N346V/A302T KW 4: N346V/C348W/A302T KW 5: N346V/A302T/V401L E) 0.0 0.5 1.0 1.5 PylRS-KW1 PylRS-WT OD 600 (nm) AzAla BocL ysine No UAA Relative Fluorescence (%)Figure 3 | Mutant Pyrrolsyine tRNA synthetase incorporating AzAla into GFP. A) Structure of M. mazei PylRS (gray) in complex with the Boc-Lysine (magenta) in complex with the adenylate (green) coordinated by water molecules (light blue) are also shown. Zoom-in of PylRS binding cavity with respective amino acids (A302, N346, C348, and V401) critical for amino acid recognition highlighted in blue (based on PDB: 2ZIN). B) Schematic overview of mutagenesis strategy and C) relative GFP-fluorescence after incorporation of either Boc-Lysine (using the wild type PylRS) or L-Azidoalanine (using PylRS-KW5). D) Coomassie stained SDS-PAGE gel of expressed and purified GFP containing Boc-Lysine or L-Azidoalanine and corresponding immunoblot. E) Relative fluorescence of GFP after incorporation of L-azidoalanine using the different PylRS mutants KW 1-5, with both variants KW1 and KW4 revealing the highest unnatural amino acid incorporation efficiency.
fixed mutation reported to significantly increases the recognition of the unnatural amino
acid[66]—and Y384F[65]. While mutation of asparagine 346 by itself doesn’t lead to significant
L-azidoalanine incorporation, the addition of the fixed A302T and Y384F mutations leads to a significant improvement, suggesting a cooperative effect. Additionally, the mutation
V401L, which has been demonstrated to decrease the PylRS binding cavity size[65], slightly
improves the unnatural amino acid recognition in combination with the N346V, A302T, and Y384F mutations (Figure 3F).
These results highlight the significance of both the N346 and the V401 residue—both critical for the size of the PylRS binding cavity—with N346 establishing hydrogen bonds with the Nε-carbonyl group of pyrrolysine. In contrast, the leucine at position 401, plays a significant role in further decreasing the binding cavity volume by establishing contacts with the shorter
amino acid side chain (Figure 3B and 3F) [65, 68]. Since both Asp346 and Cys348 function
cooperatively to accommodate the large side chain of pyrrolysine, they were examined as well, but did not increase the incorporation efficiency any further (Figure 3F). To further improve incorporation efficiency, co-expression of the evolved PylRS (KW1) and GFP-TAG35 was performed in release factor 1 (RF1) deficient E. coli JX33 which has been reported to
improve incorporation efficiency and fidelity[30].
Unexpectedly, the evolved PylRS (KW1) expressed in this optimized bacteria strain revealed enhanced growth in the presence of 1mM L-azidoalanine (Figure S1D) and was therefore used in subsequent experiments. Since mutant KW1 showed the most optimal incorporation efficiency of L-azidoalanine, as assessed by fluorescence measurements and protein expression levels (Figure 3D and F), it was used for incorporation of the UAA in an E2 enzyme—UbcH5c.
2
UV NH 2 OMe OMe NH O 2N NH2 MeO MeO NO2UCH-L3 Sulfonium intermediate UCH-L3-Dha
NaP, pH 8.0 37°C, 30 min 37°C, 16h Calc. mass: Observed mass: 2 2N Br H O NH O Br 26181 26183 26149 26147 26369 26367 5000 10000 15000 20000 25000 30000 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 0 100 % 0 100 % 26147 26367 UCH-L3-Dha 5000 10000 15000 20000 25000 30000 0 % 100 26181 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 100 % UCH-L3 A) S H2N O NH2 O Br B) Calc. mass:
Observed mass: Observed mass:Calc. mass:
Figure 5| Conversion of the active-site cysteine of UCH-L3 to dehydroalanine (Dha). A) Reaction scheme depicting the transformation of the active-site cysteine with the dibromide and subsequent elimination of the sulfonium intermediate yielding the dehydroalanine moiety, which then is amenable to amination. B) LC-MS -ESI trace and deconvoluted MS spectrum of UCH-L3-Dha (Observed mass: 26147; calculated mass: 26142). Analytical data (LC-MS traces) of UCH-L3 and the sulfonium intermediate can be found in the Supplemental Information (Figure S4).
Site-specific modification of the active-site of UCH-L3
In order to assess whether the active-site cysteine of an enzyme could be modified using this approach, we chose to use the deubiquitinating enzyme UCH-L3 as a model protein because of its high thermal stability and its suitability for LC-MS/MS analysis.
MS/MS
A) B)
UbcH5c WTUbcH5c (AzAla)
GST-UbcH5c
kDa
42
Figure 4 | Incorporation of L-azidoalanine into UbcH5c using the evolved PylRS KW5 shown by A) Coomassie-stained gel and B) subsequent validation by tandem MS/MS analysis. From the MS/MS spectrum, it becomes clear that a cysteine was incorporated at position 85 instead of the desired L-Azidoalanine.
Chemical Mutagenesis: An alternative approach to protein modification
Although a variety of methods to site-specifically alter proteins such as amber codon suppression and semi-synthetic approaches have been developed, the scope of
these methodologies remains limited[74].Thus, introducing precise modifications
post-translationally using biocompatible chemical strategies presents an alluring alternative[74-76].
Perhaps one of the most useful chemical “tags” is dehydroalanine—a naturally occurring non-canonical amino acid formed by phosphoserine elimination of either pSer or by serine
dehydration in naturally occurring peptides[75]. Due to the electrophilic nature of this
chemical moiety, it is reactive towards a variety of nucleophiles, enabling a broad range of chemical transformations including sulfonation, O- and N-linked acetylglucosamine (GlcNAc)
modification, lysine, arginine, and glutamine methylation[77- 79].
Another attractive feature promoting the use of dehydroalanine as a “handle” for protein modification is its facile site-specific installation, which is compatible with protein
chemistry[75]. Chalker et al., reported the transformation of cysteine to dehydroalanine using
2,5-dibromohexandiamide in phosphate buffer at pH 8.0, which proceeded under elimination
of the sulfonium intermediate to yield the dehydroalanine[76] (Figure 5A). This mild chemical
modification strategy permits further diversification using a broad variety of nucleophiles
to generate phosphorylated, methylated, glycosylated, or Ubiquitinated proteins[75]. While
thia-Michael additions are the most common modification of dehydroalanine due to the nucleophilic nature of the sulfur atom, bio-compatible Aza-Michael additions, which are typically performed in organic solvents and the presence of metal catalysts, are beginning
2
reaction times (Figure S4B). Despite the inability to react the dehydroalanine in the active-site of UCH-L3 with the photocleavable amine (5), we made a striking observation. Firstly, even after prolonged reaction times of UCH-L3-Dha with 2,5-dibromohexandiamide, selective modification of one cysteine (active site) could be identified (Figure S4B). The inability of modifying the additional cysteines in UCH-L3 might be explained by the difference in chemical environment (i.e. pKa value), which has been demonstrated to play a role in the
modification of cysteines to dehydroalanine[83] .
However, despite the ease of transforming the active site cysteine of UCH-L3 to dehydroalanine, amination under mild conditions proved to be challenging, as sufficiently nucleophilic amines that are reactive in aqueous buffers and mild temperatures still await development.
Discussion
Amber codon reassignment —a double-edged sword
In conclusion, the “misincorporation” of L-cysteine instead of L-azidoalanine by the evolved PylRS suggests that its binding cavity size has been significantly decreased to accommodate such a drastic substrate deviation. However, genetic encoding of L-azidoalanine, reported to be a mutagenic metabolite in Salmonella typhimurium as well as in Bacillus subtilis, activates the L-cysteine biosynthesis pathway in bacteria possibly leading to cysteine incorporation
instead[69, 70]. Microbiological and biochemical studies in S. typhimurium (TA1530) as well as
B. subtilis indicate that the free L-azidoalanine as well as its metabolites contribute to its
mutagenicity by promoting DNA damage [70-73, 84-87]. Given that the enzymes responsible for
cysteine metabolism are conserved between S. typhimurium, B. subtilis and E. coli it is to be expected that L-azidoalanine is metabolized in a similar manner. In order to circumvent azide mutagenicity arising from its bioactivation, bacteria utilize the L-cysteine pathway to inhibit the O-acetyl serine sulfhydrylase—an enzyme catalyzing the L-azidoalanine biosynthesis from the azide and thereby inhibiting the production of the mutagenic
metabolite[88]. Activation of this feedback loop during protein expression in the presence of
L-azidoalanine might have led to an increased cysteine concentration, that competed with the unnatural amino acid to counterbalance its reported mutagenic effects as outlined in
Figure 6[70]. Based on our observations that the bacterial growth was not comprised in the
presence of L-azidoalanine (Figure 2B and C), we conclude that the metabolic processing of this unnatural amino acid led to the increase of L-cysteine to compensate toxic effects. Thus the possible increase of the intracellular L-cysteine concentration in addition to the drastic size decrease of the evolved PylRS may have led to the incorporation of L-cysteine in response to the amber codon.
Using 2,5-dibromohexandiamide, the active site cysteine residue of UCH-L3 was converted into dehydroalanine under elimination of the sulfonium intermediate (Figure 5A). Notably, only modification of the active-site cysteine of UCH-L3 was observed even after prolonged reaction times, as indicated by mass spec analysis and reaction towards Ub-PA probe (Figures S3A, S3B, and S3C). Further modification of the resulting UCH-L3-Dha to yield an amine was attempted; however, addition of an amine (protected or unprotected) was challenging due
to a number of reasons: Firstly, a primary amine (i.e. NH3) by itself is not nucleophilic enough
to react with the electrophilic dehydroalanine. Secondly, the protection groups available for amines are incompatible with enzymes as they commonly require drastic reaction or deprotection conditions (i.e. organic solvents, metal catalysts, high temperatures etc.). To circumvent this conundrum, installation of UV-cleavable protecting groups for amines
was attempted, in order to facilitate their removal[81]. Given the electronic properties of
nitrobenzyl-protecting groups, it was expected that their incorporation should increase the nucleophilicity of the amine possibly promoting the reaction with the electrophilic dehydroalanine moiety.
Further transformation into a protected amine was attempted with a photocleavable amine (5), which can be deprotected using long wavelength UV light (365 nm). To this end, we employed the Gabriel synthesis that transforms the primary alkyl halide of the commercially available 4,5-dimethoxy-2-nitrobenzylbromide (3) into the primary amine of N-(4,5-dimethoxy-2-Nb)-glycine (5). First, the bromide (3) is N-alkylated with potassium phthalimide to give the corresponding N-pthalamide (4). Next, the primary amine of (5) was liberated via the Ing-Manske procedure, involving reaction with hydrazine[82], affording a 15% overall yield in two steps. (Scheme 1).
Br MeO MeO NO2 NK O O MeOH reflux, 18h DMF, RT, 18h 2 NH MeO MeO NO2 O MeO MeO NO2 N O 3 4 5 N2H4
Scheme 1| A) Synthesis strategy for generating the photocleavable N-(4,5-dimethoxy-2-Nb)-glycine (5) for amination of UCH-L3-Dha.
2
alterations arising from genetic code reassignment[43].
While genetic encoding of L-azidoalanine didn’t yield the desired results, we attempted to achieve this modification by utilizing post-translational chemical modification of the active site with dehydroalanine followed by nucleophilic addition of a photocleavable amine. Although transformation of the active-site cysteine to dehydroalanine in our model protein UCH-L3 proceeded effortlessly and selectively, further installation of a UV-cleavable amine was futile. Given this conundrum, amines compatible with mild reaction conditions but nucleophilic enough to react with the electron-deficient dehydroalanine moiety need to be developed.
Collectively, the evidence provided from our experiments as well the data reported in literature exemplify the impediments associated with amber codon reassignment. Despite the repercussions of genetically encoding L-azidoalanine, understanding the underlying fundamental cellular mechanisms using proteomics, metabolomics, and transcriptomics
might permit incorporation of this unnatural amino acid in the future[93].
Although genetic code expansion has undoubtedly enabled the site-specific installation of
a variety of unnatural amino acids ranging from photo-crosslinkers to fluorophores[11-15],
such as p-benzoyl-phenylalanine (BPA)[94], or even the incorporation of PTMs such as
phosphotyrosine[26], substantial development of optimized aaRS/tRNA pairs and expression
systems is still required to introduce the same utility to other PTMS such as Ubiquitination[95].
Recently, Chin et al., have reported the genetic incorporation of 2,3-diaminopropionic acid
(Dap) protected with a photocleavable group using an evolved PylRS tRNA synthetase[28].
While the authors demonstrate the capability of introducing this unnatural amino acid site-specifically into model proteins including TEV-Ub, introducing this unnatural amino acid to substitute the active site of Ubiquitin conjugating enzymes would be of vested interest. Nonetheless, we demonstrate that Ub-Dha, a chemical probe that is sequentially transferred
through the Ubiquitination cascade, can accurately mimic the E2-Ub thioester complex[96]
(see Chapter 3). In contrast to genetic encoding strategies, which necessitate extensive optimization for every enzyme, the cascading activity-based probe permits the facile generation of any stable E2-Ub conjugate enabling a variety of structural and biochemical
studies[96].
H2N COOH
N3
L-Azidoalanine (1)
Toxic metabolite that can be “neutralized”
by OASS UbcH5c wildtype UbcH5c-AzAla 3 N SH L-Cysteine SH H2N COOH E. coli evolved PylRS GST-UbcH5c TAG85 L-Cysteine Natural amino acid
Ribsosome mRNA growing polypeptide chain O-acetylserine-sulphhydryalse CYSTEINE BIOSYNTHESIS ? TAG L-Azidoalanine
Is PylRS charged with L-azidoalanine? PylRS OASS CysK L-Cysteine↑ A)
Figure 6| Possible mechanism of L-Azidoalanine toxicity in bacteria. A) Scheme depicting metabolic pathway of L-azidoalanine in bacteria based on model reported by Elbetieha et al.[70]. Formation
of L-azidoalanine occurs through metabolic processing of both O-acteylserine and azide by O-acetylsulfhydrylase (OASS)[89]. Simultaneously OASS upregulates the L-cysteine production via the cysteine synthetase A (CysK) to counterbalance the toxic effect of L-azidoalanine metabolites[90-92]. However, the mechanism by which L-azidoalanine affects downstream metabolic processes that induce DNA damage still remains unclear.
Unexpectedly, growth of E. coli BL21(DE3) co-expressing GFP-TAG35 and PylRS-WT or mutant PylRS (KW1) in the presence of high concentrations of L-azidoalanine accelerated bacterial growth (Figure S1 A and C). A plausible explanation for this observation might be that increased unnatural amino acid concentrations (multifold higher than the standard 1mM)
may promote optimal incorporation as suggested by Wandrey et al.[45]. Additionally, factors
such as timing of unnatural amino acid addition and induction, biomass formation, PylRS/ PylCUA concentrations, as well as expression time all determine the optimal incorporation
and require optimization in the context of unnatural amino acid incorporation[45].
However, in the light of this data, it becomes increasingly clear that the genetic incorporation of L-azidoalanine is more challenging than anticipated most likely necessitating the engineering of bacterial strains deficient for key enzymes in these pathways and the
translation apparatus (i.e. ribosome, release factors etc.) using CRISPR Cas[42]. With these
2
Table 2 | Mutagenesis primers used for introducing mutations identified from positive and negativeselections into plasmid pEVOL-PylRS-WT. Corresponding mutated positions are bold and underlined.
Position Sequence N346V(fw) 5’ GTTTACCATGCTGGTTTTTTGTCAAATGGCAAATGGG 3’ N346V(rev) 5’ CCCATTTGCCATTTGACAAAAAACCAGCATGGTAAAC 3’ C348W (fw) 5’ GTTTACCATGCTGAATTTTTGGCAAATGGCAAATGGG 3’ C348W (rev) 5’ CCCATTTGCCATTTGCCAAAAATTCAGCATGGTAAAC 3’ N346V_C348W (fw) 5’ GTTTACCATGCTGGTTTTTTGGCAAATGGCAAATGGG 3 N346V_C348 (rv) 5’ CCCAACCTGAACCCATTTGCCCAAAAAACAGCATGGTAAAC 3’ V401L (fw) 5’ CTGAGCAGCGCGGTGCTTGGCCCGATTCCGCTGG 3’ V401L (rev) 5’ CCAGCGGAATCGGGCCAAGCACCGCTGCTCAG 3’ Uptake assays
Uptake assays were performed with E.coli BL21(DE3), E.coli JX33, or E. coli Rosetta transformed with pBAD-GFP_35TAG and pEVOL pYLRS_WT or pEVOL pYLRS_KW29. An overnight culture was diluted 1:100 in a volume of 200 μL LB media (10 g tryptone, 10 g NaCl, 5 g yeast extract, pH 7.5), 2YT (16 g tryptone, 10 g yeast extract, 5 g NaCl pH 7.0) or M9 (Gibco) media supplemented with 100 μg/μL ampilicillin, 34 μg/μL chloramphenicol, as well as 1 mM Boc-Lysine or increasing concentrations (1-10 mM) L-azidoalanine (Cat# 11022, IRIS Biotech) in 96-well plates (Corning, flat bottom, clear). Samples were measured every 20 min at 37°C for 12 hours at 600 nm while shaking using a Clariostar plate reader (BMG Biotech). Growth curves were plotted and analyzed using Graphpad PRSIM.
Protein Expression and Purification
Modified GFP
E. Coli BL21(DE3) harboring pBAD-GFP35_TAG and pEVOL-pYLRS-WT or pEVOL-pYLRS-KW5
were cultured at 37°C in LB media supplemented with 100 μg/μL ampilicillin, 34 μg/μL chloramphenicol. For protein expression, saturated overnight cultures were diluted 1:100 in LB media supplemented with antibiotics and grown until OD600 0.6 was reached. Cultures were induced with 10% w/w L-arabinose (Sigma Aldrich) and were supplemented with 1 mM
Boc-Lysine or 1 mM L-azidoalanine and incubated for 2h at 30°C as previously described[97].
Purification of the modified GFP-proteins were harvested by centrifugation at (4000 rpm, 20 min, 4°C) and were lysed in lysis buffer (50 mM NaHPO4, pH 8.0, 200 mM NaCl and a complete protease inhibitor tablet (Roche)), followed by sonication on ice. Cell lysates were clarified by centrifugation (20.000 rpm, 4°C, 30 min) and incubated with 50 µL equilibrated TALON His-tag purification resin (Clontech) at room temperature for 1h while rotating.
Subsequently, the beads were washed four times with wash buffer (50 mM NaHPO4, pH
8.0, 100 mM NaCl, 10 mM imidazole (pH 8.0)) before elution with elution buffer (50 mM
Material and Methods
Bacterial strains
For expressions in this study, either BL21(DE3) E. coli or RF-1 deficient E. coli strain JX33[30],
obtained from Lei Wang (The Salk Institute for Biological Studies, La Jolla, California, USA) were used, as indicated. In case of UbcH5c, expression was performed in Rosetta E. coli.
Constructs
Both the pEVOL and the pREP vectors containing M. mazei PylRS[97] were obtained from the
Peter Schultz lab and were used for generating the mutagenesis libraries. The pEVOL-PYLRS-WT used throughout this study already contains the fixed A320T and Y384F mutations, as this has been shown to increase the incorporation fidelity. The pBAD GFP-TAG35 vector was obtained from the lab of Edward Lemke (EMBL, Heidelberg, Germany) and the pGEX-GST-UbcH5c vector was a gift from the lab of Titia Sixma (Netherlands Cancer Institute, Amsterdam, The Netherlands).
Library construction and selection
Mutagenesis libraries were generated by site-directed mutagenesis of the corresponding position using primers containing randomized codons. All other point mutations were introduced in the same manner, but with primers encoding the amber codon at a permissive site. The PCR reaction consisted of 10x Pfu Reaction Buffer (Agilent), 50 ng template DNA, 125 ng primer, 10 mM dNTPs and 1µL Pfu Ultra Enzyme (Agilent) and were performed as described in the Quik Change Mutagenesis Kit (Agilent). All clones were verified by sequencing.
Table 1 | Mutagenesis primers used for saturation mutagenesis and for introducing the amber codon into UbcH5c. Randomized positions are bold and underlined.
Position Sequence
Y384F (fw) 5’ CGATAGCTGCATGGTGTTTGGCGATACCCTGGATG 3’
Y384F (rev.) 5’ CATCCAGGGTATCGCCAAACACCATGCAGCTATCG 3’
A302T (fw) 5’ GCCTGCGCCCGATGCTGNNKCCGAACCTGTATAAC 3’
A302T (rev) 5’ GTTATACAGGTTCGGMNNCAGCATCGGGCGCAGGC 3’
N346NNK_C348NNK (fw) 5’ GTTATACAGGTTCGGMNNCAGCATCGGGCGCAGGC 3’ N346NNK_C348NNK (rev) 5’ CTGAGCAGCGCGGTGNNKGGCCCGATTCCGCTGG 3’
V401NNK (fw) 5’ CCAGCGGAATCGGGCCMNNCACCGCTGCTCAG 3’
V401NNK (rev) 5’ CCAGCGGAATCGGGCCMNNCACCGCTGCTCAG 3’
2
Characterization of mutant proteins
GFP-fluorescence measurements
To measure the GFP-fluorescence for determining the incorporation efficiency and fidelity of the evolved PylRS, BL21(DE3) E. coli or JX33 E. Coli were transformed with pBAD-GFP35_ TAG and pEVOL- pYLRS-WT or a pEVOL-pYLRS-mutants (KW1-5). Expression was performed as described above, but in a volume of 2 mL. After expression, cells were pelleted by centrifugation at (4000 rpm, 10 min, 4°C) and resuspended in a volume of 100 µL PBS. GFP-fluorescence was assayed in a 96-well plate (Corning, flat bottom, clear) at 480 nm using a Clariostar plate reader (BMG Biotech). Data analysis was performed using Graphpad PRSIM.
LC-MS-MS measurements
For LC-MS/MS analysis, purified UbcH5c-AzAla was separated by SDS-PAGE gel electrophoresis, stained with Coomassie, destained with water and the bands to be analyzed excised. Following, further destaining with 30% acetonitrile, dehydration with 100% acetonitrile and subsequent drying in a vacuum concentrator (Eppendorf), samples were digested with 25 ng/uL trypsin (Promega) overnight at 37°C in 100 mM NH4HCO3, pH 8. Peptides were extracted using 5% formic acid and measured on an LTQ-Orbitrap Velos Pro (Thermo Scientific) and separated on an EASY-Spray column (25 cm × 75 μm ID, PepMap C18 2 μm particles, 100 Å pore size) with a 30 min linear gradient from 3–30 % ACN and 0.1 % formic acid. MS scans were acquired in the Orbitrap analyzer.
Labeling with Rho-Ub-PA
To assess whether the active site of UCH-L3 had been transformed into a dehydroalanine, its reactivity toward the Rho-Ub-PA was evaluated. To this end, 1 µM enzyme was reacted with 0.5 µg of Rho-Ub-PA probe in labeling buffer (50 mM Tris, pH 7.5, 100 mM NaCl, and 1 mM DTT) for 30 minutes at 37°C. The reaction was quenched by the addition of 3x SDS-PAGE loading buffer and the samples subsequently resolved using SDS-PAGE electrophoresis. The probe labeled enzymes were visualized by in-gel fluorescence (λem / λex = 480/530 nm) followed by Coomassie staining.
Synthetic procedures
Dha-modification of UCH-L3
For the modification, purified UCH-L3 (1 µM) was added to 50mM sodium phosphate buffer, pH 8.0 to a final volume of 150 µL. Next 13.4 µL of a 2,5-dibromohexandiamide stock solution (10 mg dissolved in 120 µL DMF, 400 µM) was dropwise to the enzyme solution and incubated for 30 minutes at 37°C, while gently shaking. For LC-MS analysis, the reaction mixture was spun down (max. speed, 1 min, 4°C) to remove the precipitated dibromide.
NaHPO4, pH 8.0, 100 mM NaCl, 200 mM imidazole (pH 8.0)). Fractions were separated and
analyzed on a 10% SDS-PAGE gel (NU-PAGE, Invitrogen) and GFP-containing fractions were pooled, concentrated and flash frozen in liquid nitrogen for further storage at -80°C.
Modified UbcH5c
In the case of the GST-UbcH5c, E. Coli Rosetta cells were transformed with GST-Ubch5c_85TAG and either pEVOL pylRS-WT or pEVOL pylRS-KW5. Saturated overnight cultures were diluted 1:100 and cultured at 37°C until OD600 0.5 was reached and induced with 10% w/w L-arabinose (Sigma Aldrich), 200 mM IPTG (Sigma Aldrich) and supplemented with either 1 mM Boc-Lysine or 1 mM L-azidoalanine and incubated overnight at 18°C. Purification of the modified GST-UBCH5c proteins were harvested by centrifugation at (4000 rpm, 20 min, 4°C) and were lysed in lysis buffer (20 mM Tris (pH 8.0), 25 mM NaCl, 1 mM DTT and complete protease inhibitors (Roche)), followed by sonication on ice. Cell lysates were clarified by centrifugation (max speed, 4°C, 30 min) and incubated with equilibrated Glutathione Sepharose 4B (GE Healthcare) for 3 hours at 4°C. After washing the column with 5 column volumes of wash buffer (50 mM Tris (pH 7.5), 100 mM NaCl, and 2 mM DTT), the protein was eluted with elution buffer (20 mM Tris (pH 7.4), 100 mM NaCl, 75 mM GSH (pH 8.0)) and further purified by size exclusion chromatography (S75 16/600) in a buffer containing 20 mM Tris (pH 7.4), 100 mM NaCl, 2 mM DTT). Appropriate fractions were analyzed on a 10% SDS-PAGE gel (NU-PAGE, Invitrogen) and GST-UbcH5c containing fractions were pooled, concentrated, and flash frozen in liquid nitrogen for further storage at -80°C.
UCH-L3
UCH-L3 was expressed from a pRSET vector as previously described[98]. In brief, BL21(DE3)
E. coli transformed with pRSET-UCH-L3 were cultured at 37°C in LB media supplemented
with 100 µg/ul ampicillin. Upon reaching an OD600 of 0.8, cultures were induced with 0.4
mM IPTG and incubated for another 1.5 hours. Subsequently, cells were harvested by centrifugation, resuspended in lysis buffer (50 mM Tris-HCl, pH 7.5; 1 mM EDTA, 10 mM
MgCl2, 10 mM DTT, and 50 µM PMSF) and incubated with lysozyme (10.000 units/mL) for
2
126.31, 123.68, 110.14, 108.35, 56.32, 38.84.
Amination of UCH-L3-Dha with photocleavable amine (5)
To aminate UCH-L3-Dha, 4 µL of a 500 µM stock solution of the photocleavable amine (5) in DMSO was added to 1 µM UCH-L3-Dha in 50 mM sodium phosphate buffer pH 8.0 in a total reaction volume of 200 µL. After overnight incubation at 37°C, reaction mixture was spun down to remove precipitates and supernatant analyzed by LC-MS. However, since the enzyme precipitated, no signal corresponding to UCH-L3-Dha or its aminated form was identified.
Samples were then concentrated, aliquoted, and flash-frozen in liquid nitrogen for storage.
LC-MS measurements
LC-MS measurements were performed using a Waters 2795 Separation Module (Alliance HT) equipped with a Phenomenex Kinetex C18-column (2.1x50, 2.6 μm), a Waters 2996 Photodiode Array Detector (190-750 nm) and LCTTM ESI-Mass Spectrometer. Samples were run using 2 mobile phases: A = 1% CH3CN, 0.1% formic acid in water and B = 1% water and 0.1% formic acid in CH3CN. Flow rate= 0.8 mL/min, runtime = 6 min, column T= 40°C. Gradient: 0 - 95% B. Data processing was performed using Waters MassLynx Mass Spectrometry Software 4.1 (deconvolution with MaxEnt1 function).
LC-MS masses found: UCH-L3 (WT): 26181.2520 and 26213.2520; UCH-L3-Dha: 26147.5020,
26179.5020, and 26367.0020.
HRMS-measurements:
High resolution mass spectra were recorded on a Waters Acquity H-class UPLC with XEVO-G2 XS Q-TOF mass spectrometer equipped with an electrospray ion source in positive mode (source voltage 1.5 kV, desolvation gas flow 900 L/hr, temperature 500°C) with resolution R = 22000 (mass range m/z = 50-2000) and 200 pg/uL Leu-Enk (m/z = 556.2771) as a lock mass.
Synthesis of photo-cleavable amine (4,5-dimethoxy-2-nitrobenzylamine)
To a solution of potassium phthalimide (338.23 mg, 183 mmol) in 7 mL DMF 4,5-dimethoxy-2-nitrobenzyl bromide (495.2 mg, 1.79 mmol) was added and the solution stirred at room temperature for 18 hours. Next, the product was extracted using chloroform, the organic layers combined, washed with 0.2 M NaOH solution, water, and brine followed by drying over MgSO4, and concentration. Subsequently, the precipitated product was triturated with diethyl ether and dried. Next, 4,5-dimethoxy-2-nitrobenzyl phthalimide (568 mg, 1,66 mmol), was dissolved in 10 mL MeOH and 120 µL hydrazine hydrate (2.49 mmol) and refluxed for 16h. After slowing the solution to cool to room temperature, it was acidified with 5M HCl and extracted with EtOAc and the organic layers combined and washed with 10 mL of 5M HCl. The aqueous layers from this extraction were filtered, the pH adjusted with pH 9 with 6M NaOH and extracted with chloroform. The combined organic layers were washed with brine, dried over NaSO4, and concentrated to yield a dark yellow oil 60 mg (0.282 mmol, 15.8 % overall yield over two steps). The final product was characterized both with 1H- and 13C-APT NMR as well as by 2D HSQC NMR.
1H-NMR (300 MHz, Chloroform-d): δ 7.89 (dd, J = 5.4, 3.1 Hz, 2H), 7.76 (dd, J = 5.5, 3.1 Hz,
2H), 7.66 (s, 1H), 6.68 (s, 1H), 5.29 (s, 2H), 3.93 (s, 3H), 3.80 (s, 3H).
2
A) D) B) C) time (min) TB-media No UAA 1mM AzAla 2mM AzAla 5mM AzAla 10mM AzAla BocLysine PylRS-WT 0 100 200 300 0.0 0.2 0.4 0.6 0.8 1.0 O D60 0 (n m ) PylRS-KW5 No UAA 1mM AzAla 2mM AzAla 5mM AzAla 10mM AzAla BocLysine TB-media 0 100 200 300 0.0 0.5 1.0 time (min) O D60 0 (n m ) No UAA 1mM AzAla 2mM AzAla 5mM AzAla 10mM AzAla BocLysine PylRS-KW5M9-media 0 100 200 300 -0.5 0.0 0.5 1.0 time (min) O D60 0 (n m ) No UAA 1mM AzAla 2mM AzAla BocLysine 5mM AzAla 10mM AzAla PylRS-WT M9-media 0 100 200 300 0.0 0.5 1.0 O D60 0 (n m ) time (min)Figure S2 | Bacterial growth assays to determine toxicity of UAAs using rich and minimal media. Bacteria were co-transformed with PylRS (WT or mutant KW1) and GFP-TAG35 and expressed in the presence or absence of the unnatural amino acid using different expression media. A) Growth curves of wild type PylRS in E. coli BL21(DE3) grown in TB-media or B) in M9 minimal media and growth curves of C) evolved mutant (KW5) recognizing L-Azidoalanine in E. coli BL21(DE3) cultured in TB media or D) in minimal media.
Although growing the bacteria in a rich media such as TB media promoted bacterial growth, uptake and toxicity followed a similar trend as previously observed (Figure S2 A and C). However, culturing bacteria in M9 minimal media supplemented with the unnatural amino acids, attenuated bacterial growth (Figure S2 B and D) suggesting that minimal media is unsuitable for this application.
Supplementary Information
Toxicity of the unnatural amino acids used in this study were assessed by monitoring bacterial growth over time in the presence or absence of the unnatural amino acid. Different bacterial strains E. coli (BL21(DE3) and the E. coli RF-1 knockout strain co-expressing PylRS-WT or PylRS-KW1 and GFP-TAG35 were assayed (Figure S1).
A) D) B) BL21(DE3) C) PylRS-KW1 No UAA 1mM AzAla 2mM AzAla 5mM AzAla 10mM AzAla BocLysine BL21(DE3) No UAA 1mM AzAla 2mM AzAla 5mM AzAla 10mM AzAla BocLysine time (min) 0 200 400 600 0.0 0.3 0.6 0.9 1.2 O D60 0 (n m ) PylRS-WT 0 200 400 600 0.0 0.3 0.6 0.9 1.2 time (min) O D60 0 (n m ) No UAA 1mM AzAla 2mM AzAla BocLysine 5mM AzAla 10mM AzAla PylRS-WTJX33 0 200 400 600 0.0 0.5 1.0 1.5 time (min) O D60 0 (n m ) No UAA 1mM AzAla 2mM AzAla 5mM AzAla 10mM AzAla BocLysine PylRS-KW1 JX33 0 200 400 600 0.0 0.5 1.0 1.5 2.0 time (min) O D60 0 (n m )
Figure S1| Bacterial growth assays to determine toxicity of UAAs. Bacteria were co-transformed with PylRS (WT or mutant KW1) and GFP-TAG35 and expressed in the presence or absence of the unnatural amino acid. A) Growth curves of wild type PylRS in E. coli BL21(DE3) or B) E. coli JX33 and growth curves of C) evolved mutant (KW1) recognizing L-Azidoalanine in E. coli BL21(DE3) or D) in E. coli JX33.
Interestingly, bacterial growth of E. coli BL21(DE3) in the presence of up to 10mM L-azidoalanine was comparable to the positive control of 1 mM Boc-Lysine (Figure S1 A). Interestingly, this high unnatural amino acid concentration slightly accelerated bacterial growth in the E. coli BL21(DE3) co-expressing PylRS-KW1 (Figure S1C). In contrast, the RF-1 knockout E. coli strain JX33, preferred lower unnatural amino acid concentrations (1 mM) for optimal bacterial growth (Figure S1 B and D).
2
A. B. C. Fluorescence scan UCH-L3-UbPA UCHL3 UCHL3 (C90A) UCHL3 (Dha) Rho-UbPA + + + + + + + - -- -kDa Coomassie Rho-UbPA 37 UCH-L3-UbPA UCH-L3 37 27 * * 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 0 100 % 5000 10000 15000 20000 25000 30000 0 0 100 % 26147 26367 UCH-L3-DhaFigure S4 | A) Structure of UCH-L3 (PDB: 1XD3) with the active-site cysteine highlighted in magenta and the other cysteines highlighted cyan. B) LC-MS of UCH-L3 Dha after 24h reaction time revealing that only one cysteine is modified. C) Reactivity of UCH-L3-Dha with Rho-Ub-PA. In-gel fluorescence scanning (upper panel) as well as subsequent Coomassie staining reveal that UCH-L3 is inert towards Rho-Ub-PA strongly that the active-site cysteine is the one modified.
A) B)
Residue Role in Pyrrolysine binding Y384
A302 N346 C348 V401
Hydrogen bond formation with α-amino of Pyl Hydrophobic interaction with α-amino group of Pyl Binds to Nε-carbonyl of Pyrrolysine “gate-keeper’’ for Pyl binding Accomodates Methyl-pyrroline ring of Pyl
Mutations and their functions
Y384F: Increases aminoacylation rate A302T: Increases aminoacylation rate N346V: decreases binding cavity size,
permitting recognition of shoter substrates
C348W: Dramatically decreases binding
cavity size
V401L: Mutation accommodates shorter
side chains Stabilization of substrate A302 N346 C348 V401 Boc-Lysine
