Display Selection of Exotic Macrocyclic Peptides ... - ACS Publications

Subscriber access provided by EKU Libraries

Communication

Display selection of exotic macrocyclic peptides expressed under a radically reprogrammed 23 amino acid genetic code Toby Passioura, Wenyu Liu, Daniel Lorenz Dunkelmann, Takashi Higuchi, and Hiroaki Suga J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b03367 • Publication Date (Web): 29 Aug 2018 Downloaded from http://pubs.acs.org on August 31, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Display selection of exotic macrocyclic peptides expressed under a radically reprogrammed 23 amino acid genetic code Toby Passioura,* Wenyu Liu, Daniel Dunkelmann,† Takashi Higuchi, and Hiroaki Suga.* Department of Chemistry, Graduate School of Science, The University of Tokyo, 7-3-1, Hongo, Tokyo 113-0033.

Supporting Information Placeholder

ABSTRACT: Bioactive naturally occurring macrocyclic peptides often exhibit a strong bias for hydrophobic residues. Recent advances in in vitro display technologies have made possible the identification of potent macrocyclic peptide ligands to protein targets of interest. However, such approaches have so far been restricted to using libraries composed of peptides containing mixtures of hydrophobic and hydrophilic/charged amino acids encoded by the standard genetic code. In the present study, we have demonstrated ribosomal expression of exotic macrocyclic peptides under a radically reprogrammed, relatively hydrophobic, genetic code, comprising 12 proteinogenic and 11 non-proteinogenic amino acids. Screening of this library for affinity to the interleukin-6 receptor (IL6R) as a case study successfully identified exotic macrocyclic peptide ligands with high affinity, validating the feasibility of this approach for the discovery of relatively hydrophobic exotic macrocyclic peptide ligands.

Macrocyclic peptide ligands are an appealing chemical class for drug discovery. However, compared to traditional small molecule drugs, they typically exhibit modest pharmacokinetic properties (limited oral bioavailability and short plasma half-life, etc.). By contrast, a number of microbial macrocyclic peptide natural products are known which exhibit small-molecule-like pharmacokinetics (e.g. cyclosporin A or CsA). Such compounds have very few hydrophilic/charged residues, and are commonly backbone N-methylated, increasing their overall hydrophobicity, and, in turn, improving their pharmacokinetic profile.1-3 Since peptides may be synthesized by translation, they are amenable to display screening techniques (phage display, mRNA display, etc.), allowing the parallel screening of very diverse (>1012 compound) macrocyclic peptide libraries.4-7 We, and others, have shown that the combination of such techniques with genetic code reprogramming approaches (in which proteingenic amino acids are substituted with non-proteinogenic amino ac-

ids) further allows the display screening of macrocyclic peptide libraries containing diverse structural features (backbone N-methylation, etc.).4-5, 8-12 In particular, RaPID (Random nonstandard Peptides Integrated Discovery) methodology has been used to discover numerous ligands with very high target affinities (dissociation constants in the low nM to pM range) and which include diverse non-canonical moieties, i.e. pseudo-natural peptide ligands. However, even in the case of RaPID, the libraries used for display to date (and therefore the molecules identified from them) have all contained multiple strongly hydrophilic/charged residues, and, as such do not represent the structural features seen in CsA or other natural products. To address this, we sought to construct and screen a high diversity library of exotic macrocyclic peptides synthesized under a radically reprogrammed genetic code in which all strongly hydrophilic/charged proteinogenic residues (Gln, Asn, Arg, Lys, Glu, and Asp) were replaced with structurally diverse, nonproteinogenic, and relatively hydrophobic building blocks. As a first step, we chose 28 relatively hydrophobic (compared to Gly) non-proteinogenic aminoacyl-donors (cyanomethyl or 3,5-dinitrobenzyl esters) bearing noncharged sidechains and/or α-N-methyl substitutions, which were aminoacylated onto a synthetic tRNA bearing a CCC anticodon (tRNAGluE2CCC) using the appropriate flexizymes (eFx or dFx).13-14 To the best of our knowledge, none of these had previously been quantitatively evaluated for flexizyme-mediated aminoacylation and subsequent translation, although at least 18 of the 28 had previously been tested for efficiency of translation using alternative genetic code reprogramming techniques or in non-quantitative flexizyme-mediated studies.9, 15-18 Over 20% aminoacylaton yield was observed for most aminoacyl-donors, with the exception of AdodCME, Gly(tBu)-CME and Phe(5F)-CME, the aminoacylation efficiency of which were approximately 10%, but yet sufficient for testing of peptide expression (Supplementary Figure S1).

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 1. Ribosomal compatibility of hydrophobic non-canonical amino acids. Each of the indicated amino acids was synthesised as either a cyanomethyl or dinitrobenzyl ester and aminoacylated onto the 3’ end of a synthetic tRNA (bearing the CCC anti-codon cognate to the glycine GGG codon). (a) Translation efficiency of a model peptide sequence (P1-G5X) was assessed in an in vitro reaction. (b) Translation efficiency was determined through incorporation of 14C-labelled Asp, separation of the resulting products by polyacrylamide gel electrophoresis and autoradiography. Representative gel images for each amino acid are shown, with the quantified intensities (mean of triplicate experiments, error bars indicate standard deviation) normalised to canonical translation (employing an AARS enzyme for aminoacylation of the Gly tRNA) of glycine (Gly (AARS)) at the same codon shown in the bar graph below. (c) The fidelity of translation was confirmed in each case by MALDI-TOF MS. C. = calculated mass of [M+H] ion, O. = observed mass. Secondary peaks at +38 m/z are potassium adducts formed during mass spectroscopic analysis.


Page 2 of 8

Page 3 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2. Ribosomal synthesis and screening of a very high diversity reduced polarity macrocyclic peptide library for ligands to IL6R. (a) The codon table (NNS codons only, N = U, A, C or G; S = C or G) for reprogrammed translation (right) with the canonical E.Coli codon table (left) for comparison. Color coding indicates hydrophobicity (∆miLogP) relative to Gly. Structures of non-proteinogenic amino acids as per Figure 1c. ClAc-Tyr = N-chloroacetyl-L-tyrosine, *indicates that ClAc-Tyr is only translated as the initiating N-terminal residue. (b) Translation of a semi-randomized mRNA template comprised of an AUG start codon (reprogrammed to ClAc-Tyr an 8 or 9 NNS codon sequence, a Cys encoding UGC codon and a (Gly-Thr)3 spacer (followed by spontaneous cyclisation through reaction of the N-chloroacetyl moiety with the downstream Cys thiol to form a thioether) was then used to synthesize a library of in excess of 1 x 1012 macrocyclic peptides, which was screened for affinity to IL6R. (c) and (d) αIL6R-2 and αIL6R-2 were synthesised by solid phase peptide synthesis as shown. Noncanonical structural features are highlighted in red. Association rate (ka) dissociation rate (kd) and dissociation constant (KD) for the interaction of each peptide with IL6R are reported as mean +/- standard deviation. Next, the expression efficiency and translational fidelity of each amino acid was assessed in a Gly-deficient FIT (Flexible In vitro Translation) reaction.13, 19 A P1-G5X peptide was expressed in the presence of aminoacyl-tRNAGluE2CCC and [14C]-Asp, such that the Gly GGG codon was substituted with the respective non-proteinogenic amino acid (Figure 1a). The band intensity of the radiolabeled P1-G5X on tricine-SDS PAGE was quantified and normalized to the level of expression of the parental peptide, P1G5G(GARS), expressed using the E.Coli Gly acyl-tRNA synthetase (Figure 1b). Most of P1-G5X peptides were expressed with moderate to high yield, i.e. over 50% relative to P1-G5G(GARS), with two notable exceptions being Gly(tBu) and Phe(5F), where no and weak expression, respectively, were observed (Figure 1B). Although the poor incorporation of these amino acids could be attributed to low aminoacylation efficiencies, the fact that Adod was incorporated into peptides with greater efficiency than

Gly(tBu) and Phe(5F), suggests that these latter two amino acids may be intrinsically poor substrates for the ribosome. Translational fidelity of all peptides was confirmed by MALDI-TOF MS (Figure 1c), indicating that the P1-G5X peptides were expressed correctly, except in the case of Gly(tBu) which was not detected.

We next designed a reprogrammed genetic code comprised only of non-charged and/or hydrophobic amino acids in which 12 proteinogenic and 11 nonproteinogenic amino acids were assigned to a 32 codon genetic code utilizing NNS (with N = any nucleotide, S = G or C) codons (Figure 2a). This library included 6 amino acids for which flexizyme-mediated translation had been tested in the present study (Ala(tBu), HseMe, Glu(Me), Aoc, Ala(2-Thi), and Thr(Me) and 5 which had been assessed previously (MeGly, D-Ala, MeLeu, MeTyr(Me) and ClAcTyr), although it should be noted that only 4 (MeGly, Ala(tBu), D-Ala and ClAcTyr) had



previously been used for ribosomal peptide library synthesis, with the other 7 (HseMe, MeLeu, Glu(Me), Aoc, Ala(2-Thi), Thr(Me), and MeTyr(Me)) used for the first time in the present study.9, 11, 20-22 The 5 amino acids from previous studies were included so as to generate a final library with multiple backbone N-methylations and more structural diversity with respect to stereochemistry and amino acid side chains than could have been achieved using the amino acids shown in Figure 1 alone. In this radically reprogrammed genetic code, 7 of the most hydrophilic proteinogenic amino acids (His, Asn, Gln, Lys, Asp, Glu and Arg) were substituted with hydrophobic and/or non-charged) non-proteinogenic alternatives (Figure 2a). The Met AUG codon was reprogrammed to ClAcTyr as an initiator and Ala(tBu) as an elongator; with the former promoting macrocyclization via spontaneous reaction with a downstream Cys designated by a UGC codon (Figure 2b). To improve overall chemical diversity, this library was designed so as to “split codon boxes”23 (in this case, encoding 3 different amino acids at the 3 NNS codons that encode Arg in the canonical genetic code), thereby increasing the total number of amino acids encoded beyond the canonical limit of 20 to 23, and marking the first use of this technique for peptide library synthesis. Fidelity of incorporation at these 3 Arg codons was confirmed by translation of appropriate test templates in the presence of all 3 cognate aminoacylated tRNAs (Supplementary Figure S2, and the overall genetic code was validated by translation of appropriate test peptides, also demonstrating high fidelity (Supplementary Figure S3). An mRNA displayed macrocyclic peptide library of ~1012 molecules was then synthesized in an in vitro translation reaction under this code, such that each peptide was comprised of 8–9 random residues flanked by the cyclizing ClAcY and Cys residues (Figure 2b) To validate the synthesized peptide library, we screened it against a known cancer-related target, the interleukin6 receptor (IL6R), using RaPID display (Supplementary Figure S4). Five rounds of affinity-based selection followed by deep sequencing revealed that the final library was highly enriched for two families of peptides sharing a Glu(Me)-MeTyr(Me)-MeGly-Hse(Me) motif (Supplementary Figure S5). These peptides contained numerous hydrophobic non-canonical residues, comprising 50% or more of the overall peptide in each case. Notably, comparable selection experiments using a macrocyclic peptide library comprised of canonical amino acids (excluding methionine), yielded more hydrophilic ligands with no obvious bias for hydrophobic residues, demonstrating that our library design lead to the identification of ligands with greater hydrophobicity (Supplementary Figures S6 and S7). To assess the affinity of the identified thioethermacrocyclic peptides for IL6R, two (αIL6R-1 and αIL6R-2) were chemically synthesized and purified by standard methods. Their binding kinetics to IL6R were

assessed by surface plasmon resonance (Figure 2c and d, and Supplementary Figure S8), revealing that both αIL6R-1 and αIL6R-2 exhibited potent affinity (KD = 44.0 and 357 nM, respectively). Despite the relatively hydrophobic nature of the peptides tested, this interaction was highly specific, since neither αIL6R-1 nor αIL6R-2 exhibited strong affinity to murine IL6R, and substitution of the conserved N-methyl-O-methyltyrosine or N-methyl-glycine residues of αIL6R-1 (positions 5 and 6 respectively) to alanine completely abolished binding to human IL6R (Supplementary Figure S9). This data validates the use of such reduced polarity macrocyclic peptide libraries for the discovery of “hit” compounds with high target affinities. This is, to the best of our knowledge, the first report of high affinity macrocyclic peptides isolated from a library completely devoid of charged and highly polar moieties, and is a proof-ofconcept that high binding affinity can be achieved by such pseudo-natural macrocyclic peptides in the absence of charge interactions. Comparing the peptides obtained from the hydrophobic and canonical selection experiments, it is interesting to note that the peptides from the more hydrophobic library clustered into two closely-related families, whereas the library comprised of canonical amino acids yielded at least seven distinct families of peptides (compare Supplementary Figures S5 and S6). This may reflect the more limited binding modes (i.e. abrogation of charge interactions and diminished potential for polar interactions) available to peptides in the absence of charged/highly polar side chain residues. Nonetheless, the binding affinities of the peptides obtained from each library appeared comparable, demonstrating that high affinity ligands could be obtained despite the more limited binding modes available. It is worth noting that although we only directly measured the binding kinetics for 5 peptides in this study, the fact that the overall recovery rates from both the hydrophobic and canonical libraries were similar after 5 rounds of selection (0.15% and 0.26% respectively), suggesting that global IL6R binding affinities were comparable in both selected libraries (Supplementary Figure S10). In summary, we have quantitatively determined the compatibility of 28 non-charged non-proteinogenic amino acids with ribosomal synthesis. Using a subset of these, we devised a radically reprogrammed genetic code which excluded the 7 most hydrophilic proteinogenic amino acids, but expanded the genetic code overall to 23 amino acids by inclusion of 11 nonproteinogenic alternatives. Under this code, we assembled a unique, relatively hydrophobic, macrocyclic peptide library. To the best of our knowledge, this represents the most “natural product-like” and diverse (in terms of potential amino acids at each codon) macrocyclic peptide library reported, to date. Screening this library for ligands of IL6R lead to the isolation of at least 2 high affinity ligands with dissociation constants in the nano-


Page 4 of 8

Page 5 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


molar range, demonstrating the applicability of such libraries to the identification of novel pseudo-natural macrocyclic peptide ligands to proteins of interest.

ASSOCIATED CONTENT The Supporting Information is available free of charge on the ACS Publications website. Supplementary data and experimental details (PDF). AUTHOR INFORMATION Corresponding Authors

*[email protected] *[email protected] Present Addresses

† MRC Laboratory of Molecular Biology, Cambridge, UK Funding Sources

No competing financial interests have been declared. This work was supported by Japan Society for the Promotion of Science Grants-in-Aid for Young Scientists (15K16558) to T.P., and partly supported by the Japan Agency for Medical Research and Development, Basic Science and Platform Technology Program for Innovative Biological Medicine (JP18am0301001).

REFERENCES (1) Ahlbach, C. L.; Lexa, K. W.; Bockus, A. T.; Chen, V.; Crews, P.; Jacobson, M. P.; Lokey, R. S., Future medicinal chemistry 2015, 7, 2121-30. (2) Bockus, A. T.; Schwochert, J. A.; Pye, C. R.; Townsend, C. E.; Sok, V.; Bednarek, M. A.; Lokey, R. S., J. Med. Chem. 2015, 58, 7409-18. (3) Nielsen, D. S.; Shepherd, N. E.; Xu, W.; Lucke, A. J.; Stoermer, M. J.; Fairlie, D. P., Chem. Rev. 2017, 117, 8094-8128. (4) Kawamura, A.; Munzel, M.; Kojima, T.; Yapp, C.; Bhushan, B.; Goto, Y.; Tumber, A.; Katoh, T.; King, O. N.; Passioura, T.; Walport, L. J.; Hatch, S. B.; Madden, S.; Muller, S.; Brennan, P. E.;

Chowdhury, R.; Hopkinson, R. J.; Suga, H.; Schofield, C. J., Nature communications 2017, 8, 14773. (5) Passioura, T.; Katoh, T.; Goto, Y.; Suga, H., Annu. Rev. Biochem. 2014, 83, 727-52. (6) Heinis, C.; Rutherford, T.; Freund, S.; Winter, G., Nat. Chem. Biol. 2009, 5, 502-7. (7) He, B.; Tjhung, K. F.; Bennett, N. J.; Chou, Y.; Rau, A.; Huang, J.; Derda, R., Scientific reports 2018, 8, 1214. (8) Hacker, D. E.; Hoinka, J.; Iqbal, E. S.; Przytycka, T. M.; Hartman, M. C., ACS Chem. Biol. 2017, 12, 795-804. (9) White, E. R.; Sun, L.; Ma, Z.; Beckta, J. M.; Danzig, B. A.; Hacker, D. E.; Huie, M.; Williams, D. C.; Edwards, R. A.; Valerie, K.; Glover, J. N.; Hartman, M. C., ACS Chem. Biol. 2015, 10, 1198208. (10) Schlippe, Y. V.; Hartman, M. C.; Josephson, K.; Szostak, J. W., J. Am. Chem. Soc. 2012, 134, 10469-77. (11) Yamagishi, Y.; Shoji, I.; Miyagawa, S.; Kawakami, T.; Katoh, T.; Goto, Y.; Suga, H., Chem. Biol. 2011, 18, 1562-70. (12) Morimoto, J.; Hayashi, Y.; Suga, H., Angew. Chem. Int. Ed. Engl. 2012, 51, 3423-7. (13) Goto, Y.; Katoh, T.; Suga, H., Nat. Protoc. 2011, 6, 779-90. (14) Murakami, H.; Ohta, A.; Ashigai, H.; Suga, H., Nat. Methods 2006, 3, 357-9. (15) Ojemalm, K.; Higuchi, T.; Jiang, Y.; Langel, U.; Nilsson, I.; White, S. H.; Suga, H.; von Heijne, G., Proc. Natl. Acad. Sci. U. S. A. 2011, 108, E359-64. (16) Xiao, H.; Schultz, P. G., Cold Spring Harbor perspectives in biology 2016, 8. (17) Wolschner, C.; Giese, A.; Kretzschmar, H. A.; Huber, R.; Moroder, L.; Budisa, N., Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 7756-61. (18) Hartman, M. C.; Josephson, K.; Lin, C. W.; Szostak, J. W., PLoS ONE 2007, 2, e972. (19) Passioura, T.; Suga, H., Chemical communications (Cambridge, England) 2017, 53, 1931-1940. (20) Kawakami, T.; Murakami, H.; Suga, H., Chem. Biol. 2008, 15, 32-42. (21) Passioura, T.; Watashi, K.; Fukano, K.; Shimura, S.; Saso, W.; Morishita, R.; Ogasawara, Y.; Tanaka, Y.; Mizokami, M.; Sureau, C.; Suga, H.; Wakita, T., Cell chemical biology 2018. (22) Fujino, T.; Goto, Y.; Suga, H.; Murakami, H., J. Am. Chem. Soc. 2013, 135, 1830-7. (23) Iwane, Y.; Hitomi, A.; Murakami, H.; Katoh, T.; Goto, Y.; Suga, H., Nature chemistry 2016, 8, 317-25.




Page 6 of 8

Page 7 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. Ribosomal compatibility of hydrophobic non-canonical amino acids. Each of the indicated amino acids was synthesised as either a cyanomethyl or dinitrobenzyl ester and aminoacylated onto the 3’ end of a synthetic tRNA (bearing the CCC anti-codon cognate to the glycine GGG codon). (a) Translation efficiency of a model peptide sequence (P1-G5X) was assessed in an in vitro reaction. (b) Translation efficiency was determined through incorporation of 14C-labelled Asp, separation of the resulting products by polyacrylamide gel electrophoresis and autoradiography. Representative gel images for each amino acid are shown, with the quantified intensities (mean of triplicate experiments, error bars indicate standard deviation) normalised to canonical translation (employing an AARS enzyme for aminoacylation of the Gly tRNA) of glycine (Gly (AARS)) at the same codon shown in the bar graph below. (c) The fidelity of translation was confirmed in each case by MALDI-TOF MS. C. = calculated mass of [M+H] ion, O. = observed mass. Secondary peaks at +38 m/z are potassium adducts formed during mass spectroscopic analysis. 1411x1411mm (72 x 72 DPI)



Figure 2. Ribosomal synthesis and screening of a very high diversity reduced polarity macrocyclic peptide library for ligands to IL6R. (a) The codon table (NNS codons only, N = U, A, C or G; S = C or G) for reprogrammed translation (right) with the canonical E.Coli codon table (left) for comparison. Color coding indicates hydrophobicity (∆miLogP) relative to Gly. Structures of non-proteinogenic amino acids as per Figure 1c. ClAc-Tyr = N-chloroacetyl-L-tyrosine, *indicates that ClAc-Tyr is only translated as the initiating N-terminal residue. (b) Translation of a semi-randomized mRNA template comprised of an AUG start codon (reprogrammed to ClAc-Tyr an 8 or 9 NNS codon sequence, a Cys encoding UGC codon and a (Gly-Thr)3 spacer (followed by spontaneous cyclisation through reaction of the N-chloroacetyl moiety with the downstream Cys thiol to form a thioether) was then used to synthesize a library of in excess of 1 x 1012 macrocyclic peptides, which was screened for affinity to IL6R. (c) and (d) αIL6R-2 and αIL6R-2 were synthesised by solid phase peptide synthesis as shown. Non-canonical structural features are highlighted in red. Association rate (ka) dissociation rate (kd) and dissociation constant (KD) for the interaction of each peptide with IL6R are reported as mean +/- standard deviation. 1334x909mm (72 x 72 DPI)


Page 8 of 8

Display Selection of Exotic Macrocyclic Peptides ... - ACS Publications

Recommend Documents