Introduction
Ubiquitination is an abundant post-translational modification. 1 Indeed, hundreds of enzymes exist to attach the small, 76 amino acid protein ubiquitin (Ub) mainly to lysine side-chains of particular cellular proteins. 2 Ub itself has multiple lysines, onto which further Ub can be attached. As a result, there are many possible Ub chains of various lengths and linkage types. These Ub chains can differently perturb the tagged protein 3,4 or be recognized by different cellular receptors, allowing them to perform different functions. 5,6 Lys48-linked Ub chains are particularly important, as these direct their tagged proteins for degradation by the 26S proteasome. 7
Cancer cells use the ubiquitin–proteasome system to remove critical proteins and evade programmed cell death. 8 Consequently, inhibitors of the 26S proteasome have proved powerful research tools and successful anticancer therapeutics, 9 although resistance can emerge. 10 An attractive alternative is to target the signal for degradation: find molecules to bind to and interfere with the recognition of Lys48-linked Ub chains. 11–15 However, given the breadth of cellular processes that rely on differently linked Ub chains, specificity is crucial 5 and also a challenge due to the following reasons: Ub chains are assembled from the same monomer, and the differences in structure and dynamics of differently linked-Ub chains can be subtle. 16,17 Moreover, chain length is also important. For example, tetra Ub (Lys48-linked, K48 Ub 4 ) is postulated to be the minimal chain length to label a protein for degradation, 7 although there is some evidence suggesting shorter chains can also perform this function. 18,19 Molecules specific for long ( e.g. ≥3), Lys48-linked Ub chains are required to specifically modulate protein degradation.
Ub chains have proved challenging targets for traditional drug discovery using small molecules. Those discovered to date suffer from weak binding and poor specificity for linkage and length. 12–15 Typically, these are symmetric molecules that bind monomeric Ub weakly, and then achieve what affinity they have for Ub chains through multimeric interactions. However, true recognition of Ub chains with a specific linkage was achieved recently using an alternative modality – de novo cyclic peptides. 11
Nawatha et al. discovered de novo cyclic peptides capable of specifically binding Lys48-linked Ub chains. To achieve this, they first used solid-phase peptide synthesis (SPPS) coupled to Lys analogue native chemical ligation 20,21 to generate Lys48-linked Ub chains. 11 These Ub chains were of sufficient purity for use as targets for the RaPID system, a method that uses reprogrammed ribosome peptide synthesis to synthesize, and then screen, extremely large (>10 12 ) cyclic peptide libraries for binding molecules. 22 Tight-binding, having low nM K D , and linkage-specific cyclic peptides were discovered, and their novel Ub recognition abilities were explained by their binding modes. The cyclic peptide bound in one orientation on the symmetric Ub chain target, simultaneously interacting with two or more Ub in the chains. Importantly, these cyclic peptides inhibited Ub chain recognition in vitro and had activity when added to cells, causing Ub-conjugate accumulation and apoptosis. 11
Unfortunately, these first-generation cyclic peptides did not necessarily exhibit physicochemical properties comparable to known drug-like cyclic peptides. Orally-available, membrane-permeable, macrocyclic and cyclic peptides, such as the natural product cyclosporin, tend to be small (≤1.2 kDa) 23,24 and have few H-bond acceptors, fewer H-bond donors, and no formal charges at pH 7. 25–27 Key to accessing this property space using peptides is the use of non-canonical amino acids, such as amine N -methylation, 28 which have the added benefit of conferring protease-resistance. By contrast, the first-generation Ub chain binding cyclic peptides were large (∼1.7 kDa) and made almost entirely of H-bond-rich, protease-recognizable, canonical amino acids. 11
A better subclass of cyclic peptides with improved bioactivity targeting the ubiquitin–proteasome system is needed to provide research tools and drug leads. A potential route to such molecules is to screen cyclic peptide libraries which include from the beginning non-canonical amino acids known to confer desirable physical properties. 22,29 However, it is not yet known if cyclic peptides from such a library can succeed at a challenging molecular recognition task like specific Ub chain binding. Here, we constructed and screened a highly non-proteinogenic peptide library to discover de novo cyclic peptides capable of modulating specific Lys48-linked ubiquitin chains.
Results
Target Ub chain synthesized by total chemical synthesis
To discover binders specific for the Lys48-linked Ub tetramer, while also avoiding binders that use di- or mono-Ub as the recognition element, high purity K48 Ub 4 and K48 Ub 2 chains were required. Ub chains were assembled by total chemical synthesis, as described previously, 11,20,21 for use as the target and anti-target respectively. Each was biotinylated at the N-terminus for attachment to a streptavidin solid support, enabling screening of pooled libraries of peptides.
Construction of a highly non-proteinogenic cyclic peptide library
We sought to create a large cyclic peptide library containing multiple non-canonical amino acids, to emulate the properties of drug-like, highly non-proteinogenic cyclic peptides. To form cyclic peptides, the genetic code was reprogrammed so that ribosomal peptide synthesis was initiated using a chloroacetyled amino acid. 30 We withdrew the natural initiator amino acid Met and its cognate aaRS during the construction of a custom reconstituted in vitro translation system, 31 and supplemented this with initiator tRNA loaded with ClAc- d -Phe, 22 using an artificial ribozyme flexizyme. 32 We then further reprogrammed the genetic code to remove the charged amino acids Glu, Asp, and Arg, and add the following non-canonical amino acids: 22,29,33 three N -methylated: MeGly, MeAla and MePhe; two d -stereochemistry: d -Phe and d -Ala, and the linear hydrophobic side-chain Aoc ( Fig. 1A ). Each was flexizyme-loaded onto tRNA and supplied to the translation system. The two d -amino acids were loaded onto an optimized tRNA Pro and the translation system supplemented with the E. coli translation factor EF-P to enhance translation efficiency. 34,35 To test the fidelity of this extensively reprogrammed 22 amino-acid genetic code (Fig. S1A, ESI † ), we confirmed the ribosomal synthesis of peptides containing these non-canonical amino acids, with the major products detected by MALDI-TOF MS having the expected masses (Fig. S1B, ESI † ).
Non-proteinogenic cyclic peptide library construction and selection
Following the RaPID protocol, 22 we assembled a greater than trillion-member cyclic peptide library using the ribosome and our reprogrammed genetic code, using degenerate DNA codons to encode cyclic peptides with 6–12 randomized amino acids each ( Fig. 1B ). mRNA display 36 was used to covalently tag each unique cyclic peptide with its encoding mRNA/DNA, for later resynthesis of the library, sequencing, or both. In the first round of screening, the cyclic peptide library was incubated with the target K48 Ub 4 , the bound mRNA/DNA amplified and the enriched library resynthesized. Subsequent rounds included a ‘negative selection’ step where K48 Ub 2 and streptavidin binders were discarded before incubation with K48 Ub 4 . This was to avoid a problem encountered in the first cyclic peptide screenings effort, 11 where the majority of the cyclic peptides recovered used K48 Ub 2 as the main recognition element, then using multimeric interaction to bind the longer chain target K48 Ub 4 .
After each round, the fraction of mRNA/DNA recovered increased, suggesting an enrichment for desired K48 Ub 4 -binding peptides (Fig. S2, ESI † ). Next-generation sequencing (NGS) was then used to assess the library composition after each round of selection, and this showed the progressive enrichment of particular, putative K48 Ub 4 -binding, peptide sequences ( Fig. 1C ). The top DNA sequence, and that of an analogue, were resubmitted to the in vitro translation system to confirm the integrity of the reprogrammed genetic code. The expected cyclic peptides, named Ub4a and Ub4e respectively ( Fig. 1C ), were detected by MALDI-TOF MS (Fig. S3, ESI † ). Ub4a and Ub4e each contain an 11 amino acid macrocycle, with two N -methyl amino acids, a non-canonical side-chain as well as a d -amino acid. Ub4e contains a His, which may be partially protonated at pH 7. Otherwise, as hoped, Ub4a and Ub4e are uncharged peptides with no formally charged amino acids at neutral pH.
Cyclic peptides preferentially bind longer K48-linked Ub chains
Cyclic peptides Ub4a and Ub4e were both synthesized by SPPS and tested for binding to the target K48 Ub 4 by SPR, and both displayed high affinity, nM K D , for the Ub chain ( Fig. 1D and Table S1, ESI † ). As expected, binding to the anti-target K48 Ub 2 was considerably weaker, with >100 fold higher K D , ( Fig. 1D and Fig. S4 and Table S2, ESI † ). To further confirm binding, we synthesized Ub4a labelled with a fluorescein dye (Fig. S5, ESI † ) and used this as a probe for Ub-chains transferred onto nitrocellulose membranes (Fig. S6A and B, ESI † ). Under these denaturing conditions, K48 Ub 2 and K48 Ub 4 bound Ub4a -fluorescein, with greater intensity then alternatively linked dimers K63 Ub 2 and K11 Ub 2 .
NMR reveals unit- and residue-specific interactions between Ub chains and cyclic peptide Ub4a
We used NMR to confirm the physical interaction between cyclic peptide Ub4a and Ub chains as well as to map the Ub units and residues involved in binding. The unlabeled peptide was titrated into a solution of selectively 15 N-labeled Ub chains, and the binding was monitored by NMR. For this purpose, K48 Ub 4 , K48 Ub 3 , and K48 Ub 2 with specific Ub units uniformly 15 N-enriched were assembled from respective recombinant Ub monomers using controlled enzymatic chain assembly methodology. 11,37,38 The addition of Ub4a to K48 Ub 4 caused strong residue-specific perturbations in the 1 H– 15 N NMR spectra of the proximal Ub unit (the one bearing free C terminus) ( Fig. 2A ). We observed a gradual decrease in intensity of the ‘unbound’ signals accompanied by the appearance and increase of new (‘bound’) signals at a different position in the spectrum, a textbook example of slow exchange, consistent with the slow dissociation rates (∼1 per minute) detected by SPR (Table S1, ESI † ). The perturbed residues are located primarily in and around the L8-I44-V70 hydrophobic patch on Ub surface, as well as the α-helix, and for most residues, the spectral changes saturated at about 1 : 1 peptide : K48 Ub 4 molar ratio. However, substantially weaker changes in the NMR spectra were observed for the distal Ub (the opposite end of the chain) of K48 Ub 4 , where the signal shifts were substantially smaller. Furthermore, most residues exhibited gradual signal shifts characteristic for fast exchange on the NMR time scale. The noncovalent Ub–Ub interactions in K48-linked chains are sufficiently weak to allow fast interconversion between closed (possessing Ub–Ub interface) and open conformations. 38–40 Thus, it is possible that Ub4a binding disrupts the interactions between distal Ub and the rest of the Ub chain ( Fig. 2B ). These results indicate that interactions with the distal Ub in the tetramer do not significantly contribute to Ub4a binding and suggests that K48 Ub 3 is the actual peptide-binding element.
To test this hypothesis, we analyzed Ub4a binding to the proximal, middle (endo), and distal Ub units of K48 Ub 3 . All three Ub units exhibited slow-exchange behavior upon Ub4a titration indicative of tight binding. The observed spectral changes in the proximal Ub of K48 Ub 3 were almost indistinguishable from those in the proximal Ub of K48 Ub 4 ( Fig. 2C and Fig. S7, ESI † ), pointing to strong similarity in the peptide's contacts with the proximal Ub in both chains. Also, the distal Ub of K48 Ub 3 exhibited strong NMR signal shifts ( Fig. 2D ), in stark contrast with the distal Ub of K48 Ub 4 (see also Fig. S8, ESI † ). Similar strong NMR signal shifts were observed for the middle Ub of K48 Ub 3 (Fig. S9, ESI † ). While the structure of K48 Ub 3 : Ub4a complex awaits determination, these results indicate that all three Ub units in K48 Ub 3 are engaged in tight interactions with Ub4a ( Fig. 2E and F ).
To further verify Ub4a selectivity for Ub chains longer than the dimer, Ub4a peptide was titrated into K48 Ub 2 with 15 N-labeled proximal Ub. In stark contrast with the proximal Ub in K48 Ub 4 and K48 Ub 3 , here we observed little or no signal shifts in the NMR spectra as well as strong signal broadening/disappearance for several residues located on the hydrophobic patch surface of Ub (Fig. S10, ESI † ). This behavior is indicative of intermediate exchange on the NMR time scale, consistent with faster dissociation rates hence weaker binding to K48 Ub 2 detected by SPR (Fig. S4 and Table S2, ESI † ).
To summarize, the NMR results corroborate tight Ub4a binding to K48-linked Ub chains of length ≥3 and revealed differential and residue-specific physical interactions between the peptide and various Ub units in these chains. The observed spectral perturbations generally map Ub4a binding sites to the hydrophobic patch surfaces of the Ub units involved. However, there are clear differences in the strength, direction, and exchange regime of these perturbations depending on the chain length and the position of the Ub unit within each chain. In particular, the strikingly weaker changes in NMR spectra observed for the distal Ub of K48 Ub 4 suggest that it is the trimer (or the first three Ub units in the tetramer) that is the Ub4a recognition element in K48-linked chains.
Cyclic peptides prevent recognition of Ub chains by specific proteases and the proteasome
We chose to investigate whether binding of the discovered cyclic peptides can prevent their recognition of Ub-chains by interacting proteins. For example, deubiquitinating enzymes (DUBs) recognize Ub-chains and cleave the connecting isopeptide bond. The DUB OTUB1 is specific for K48-linked Ub-chains 41 and can digest K48 Ub 4 to shorter chains and monomer in vitro ( Fig. 3A ). However, in the presence of equimolar Ub4a , K48 Ub 4 is protected from the activity of OTUB1 ( Fig. 3A ). This suggests that binding to the Ub chain by the cyclic peptide can interfere with recognition by the DUB, protecting against DUB cleavage. Ub4a also offers some protection to K48 Ub 2 from cleavage by OTUB1 (Fig. S11A, ESI † ). However, the cyclic peptide was not able to protect K11 Ub 2 or K63 Ub 2 from cleavage by the K11 linkage-specific DUB Cezanne 42 and the linkage non-specific USP2, respectively (Fig. S11B and C, ESI † ). Likely this is due to the specificity of these cyclic peptides binding to Ub-chains with the K48-linkage.
A critical Ub-chain interacting protein is the 26S proteasome, which recognizes K48-linked Ub-chains in order to degrade the tagged protein. It is possible that the cyclic peptides, by binding the Ub-chain, can inhibit this recognition and degradation. To test this, we added synthetic α-globin- K48 Ub 4 43 to purified 26S proteasome in vitro , with and without Ub4a or Ub4e . Similar to the addition of the direct proteasome inhibitor MG132, 44 the cyclic peptides were able to protect α-globin- K48 Ub 4 43 from degradation by the 26S proteasome ( Fig. 3B ). Presumably, these cyclic peptides bind K48-linked Ub chains in a manner that prevents recognition by the 26S proteasome ( Fig. 3B ).
The NMR finding that only first three Ub units (counting from the proximal end) in K48 Ub 4 bind Ub4a tightly, was independently verified by comparing the ability of the peptide to protect K48 Ub 4 and K48 Ub 3 against disassembly by OTUB1 that preferentially cleaves from the distal end of a poly-Ub chain. The results (Fig. S12A, ESI † ) show that Ub4a effectively shields K48 Ub 3 from disassembly by OTUB1, consistent with the NMR data showing tight peptide binding to all Ub units in the trimer. Interestingly, in the presence of Ub4a , OTUB1 cleaves only one Ub molecule from K48 Ub 4 resulting in visible and sustained accumulation of a trimer, thus supporting the conclusion that the distal Ub unit in the tetramer binds Ub4a loosely. A slightly different trend was observed for Ub4a protection against proteasome-associated DUB Ubp6, where some amount of K48 Ub 4 was reduced to a trimer but a significant portion of it remained unaffected (Fig. S12B, ESI † ), likely reflecting Ubp6's preference for cleaving the endo isopeptide bond (between two middle Ubs) in K48 Ub 4 . 45 By contrast, Ub4a effectively blocked K48 Ub 4 disassembly by a general linkage-nonspecific DUB USP2 (Fig. S12C, ESI † ).
Cyclic peptides can enter cells
Having shown that cyclic peptides can modulate the function of Ub-chains in vitro , we next sought to test their effects in live cells. However, as the ubiquitin system is largely intracellular, it was important to check their ability to enter these cells. We used fluorescein-labelled cyclic peptides (Fig. S5, ESI † ) and live-cell imaging to monitor their entrance into HeLa cells. Uptake of Ub4a -fluorescein and Ub4e -fluorescein peptides could be observed after incubation for 2 hours with 3 nM peptide, whereas fluorescein alone showed negligible uptake even at 1000 nM ( Fig. 3C ). To quantify the uptake, we incubated HeLa cells with Ub4a -fluorescein for 16 h, the cells were washed and the residual fluorescence indicates significant retention of peptide-fluorescein relative to fluorescein alone (Fig. S13, ESI † ).
Cyclic peptides lead to Ub-conjugates accumulation
We have shown using purified proteins that cyclic peptide binding can inhibit Ub-chain cleavage by DUB activity, and therefore might be expected to cause an accumulation of Ub-chains and proteins tagged with Ub-chains in the cell. We incubated HeLa cells with cyclic peptides Ub4a or Ub4e , as well as MG132, lysed the cells, separated the proteins and stained using anti-Ub western blot to detect Ub-conjugates. In a dose-dependent manner, the cyclic peptides Ub4a and Ub4e caused accumulation of intracellular Ub-conjugates, producing a similar effect to the proteasome inhibitor MG132 ( Fig. 4A ). Incidentally, cyclic peptide Ub4a could be used to detect these Ub-conjugates on the western blot membrane (Fig. S6C and D, ESI † ).
Ub4a can induce apoptosis in cancer cell lines
Inhibition of the ubiquitin system using direct proteasome inhibitors can induce apoptosis, enabling these inhibitors to be used as cancer therapeutics. 46 We wanted to examine whether our cyclic peptides, through their Ub-chain binding, could also induce apoptosis. 47,48 We incubated Ub4a with the cancer-derived U87 cells, which are a human primary glioblastoma cell line (brain cancer), and cells sorted for their ability to bind fluorescently labelled annexin-V using Fluorescence Activated Cell Sorting (FACS), detecting both early and late apoptotic cells. Ub4a at 100 nM exhibited a clear increase in apoptosis after 24 h treatment, similar to the direct proteasome inhibitors: bortezomib, ixazomib, and carfilzomib 46 ( Fig. 4B ). Furthermore, we tested the effects of the cyclic peptides on three other cancer cell lines, SH-SY5Y cells (neuroblastoma), MDA-MB-231 cells (epithelial human breast cancer) and HeLa cells (cervical cancer) which all showed similar induction of apoptosis upon treatment with cyclic peptides (Fig. S14, ESI † ). Interestingly, under these conditions the Ub4a cyclic peptide did not induce apoptosis in the HEK-293 cell line (embryonic kidney derived), whereas treatment with the direct proteasome inhibitor MG132 did induce apoptosis ( Fig. 4C ).
Finally, we compared the apoptotic induction abilities of the non-proteinogenic cyclic peptide Ub4a with two Ub-binding largely proteinogenic cyclic peptides from previous studies Ub4ix 11 and mJ08-L8W . 49 Ub4a was able to induce apoptosis in a greater proportion of U87 ( Fig. 4D ), SH-SY5Y, MDA-MB-231 and HeLa cells (Fig. S14, ESI † ). One possible reason for the efficacy of Ub4a is that it is resistant to proteases. We tested the stability of Ub4a in human plasma and found that the non-proteinogenic cyclic peptide is very stable in human plasma, with a half-life of multiple days (∼60 h) (Fig. S15, ESI † ).
Ub4a shows anti-cancer activity in vivo
The plasma stability of Ub4a combined with its ability to induce apoptosis in vitro in cancer cells, led us to investigate if this de novo cyclic peptide might show anti-cancer activity in vivo . We sought to investigate whether Ub4a could inhibit the growth of human tumors in mice. 50 Briefly, SCID/NOD mice were injected intravenously with luciferase-expressing human CAG myeloma cells, 51 and after a week of growth, Ub4a (1 mg kg −1 twice a week) or the approved anti-cancer medication bortezomib 52 (0.5 mg kg −1 twice a week) were injected intraperitoneally. Both Ub4a and bortezomib treated mice showed a marked reduction in tumor progression relative to untreated mice ( Fig. 4E, F and Fig. S16, ESI † ). Even in this first test of in vivo activity, 15 days of Ub4a treatment resulted in a reduction in tumor growth similar to the bortezomib positive control ( Fig. 4E and F ).
Discussion
Ubiquitin chains are attractive targets for therapeutic intervention, given their central role in cellular activities and their frequent dysregulation in pathology. 53 Here, we report the discovery of non-proteinogenic cyclic peptides able to specifically bind and modulate long (>3) K48-linked Ub chains: the tag for protein degradation by the proteasome. In contrast to previous efforts, 11 the cyclic peptide library was built to include amino acids known to confer drug-like properties onto peptides, with the aim of generating Ub chain modulating molecules which are proteolytically stable and able to engage their intracellular target in vivo .
Using the RaPID system methodology, a cyclic peptide library was constructed to include d -stereocenteres, N -methylated backbones, non-canonical side chains, while being depleted in canonical charged amino acids, This library was screened against K48 Ub 4 , and two hit non-proteinogenic de novo cyclic peptides were chosen for characterization in detail: Ub4a and Ub4e . Both tightly bound K48 Ub 4 , and showed significantly weaker binding for shorter chain K48 Ub 2 . NMR studies revealed that Ub4a uses K48 Ub 3 as the main recognition element. Interestingly, while K48 Ub 4 contains two overlapping K48 Ub 3 units, 1–3 and 2–4, it is specifically the proximal trimer, 1–3, that engages the cyclic peptide. Chemical shift perturbations upon binding map the binding interactions to residues in and around the surface hydrophobic patch on the Ub units involved ( Fig. 2E and F ). Overlaying these CSPs on the “closed” K48 Ub 4 structure, which is the predominant conformation in the apo form, suggests that cyclic peptide binding involves intercalation into the Ub–Ub interface. The binding mode of Ub4a is broadly similar to that of Ub4ix , a cyclic peptide from the previous study, which also used the proximal K48 Ub 3 . 11 However, given the dissimilar chemical structures of Ub4a and Ub4ix (Fig. S17, ESI † ), their interfaces with K48 Ub 3 are likely very different and await further characterization.
Natural product bioactive cyclic peptides are often highly non-proteinogenic: they are rich in non-canonical amino acids which confer protease resistance and passive membrane permeability. Moreover, these peptides tend to be small (<1000 Da), and lack amino acids that are charged at neutral pH. An archetype for this class is the orally available immunosuppressant cyclosporin. 33 Previously characterized Ub chain binding cyclic peptides, like Ub4ix , were quite unlike cyclosporin: they were highly proteinogenic, larger and rich in charged amino acids at neutral pH (Fig. S17, ESI † ). Promisingly, the cyclic peptides characterized here, Ub4a and Ub4e , are highly non-proteinogenic, containing two N -methyl, a d -stereochemistry and a non-canonical aliphatic side-chain amino acid. They are small, with just 11-amino acids in the macrocycle, making them undecamer cyclic peptides similar to cyclosporin. Lastly, similar to cyclosporin, these peptides lack formal charges, having zero, or close to zero net charge in the case of Ub4e , at neutral pH (Fig. S17, ESI † ).
Importantly, the physicochemical make-up of the peptide library, the non-canonical amino acids added and the charged canonical amino acids removed, did not preclude the discovery of cyclic peptides with impressive binding strength and specificity for the challenging K48 Ub 4 target. Indeed, the non-canonical rich cyclic peptides described here have similar binding affinities to previously discovered, largely canonical, cyclic peptides. 11 Yet the former have smaller macrocycles and had appreciably lower molecular mass, suggesting that the non-canonical amino acid collection is in fact superior at producing high-affinity, selective cyclic peptides.
We showed that Ub4a binding to Lys48-chain Ub chains prevents their recognition by deubiquitinates and the proteasome. Moreover, Ub4a can induce apoptosis in cancer cell lines in vitro , and attenuate the growth of tumor cells in vivo . These anti-cancer effects are likely due to disruption of Ub chain recognition processes which are part of the ubiquitin–proteasome system. Cancer cells are particularly dependent on the proteasome activity to remove certain proteins and evade programmed cell death, and this is the rationalization behind the anti-cancer effects observed for direct proteasome inhibitors such as bortezomib, carfilzomib, and ixazomib, which have been approved for the treatment of multiple myeloma. 54 Indeed, throughout our in vitro and in vivo trials, Ub4a and direct proteasome inhibitors show broadly similar effects and comparable efficacy, the latter is striking considering Ub4a is a hit from a screen and the proteasome inhibitors are the product of extensive optimization.
While Ub4a showed similar effects to the direct proteasome inhibitors, it is important to emphasize that these cyclic peptides act through a different mechanism, targeting a different component of the ubiquitin–proteasome system. Ub-chain binders might therefore overcome the shortcomings of current proteasome inhibitors, the acquired resistance and adverse side-effects which decrease patients’ quality of life. 55,56 Non-proteinogenic cyclic peptide targeting of Ub chains, therefore, have great potential for the development of new anti-cancer therapeutics.
Author contributions
J. M. R. designed and performed the cyclic peptide discovery, carried out SPR assays, analysed the data, and co-wrote the manuscript and the ESI. † M. N. assisted in the chemical synthesis of cyclic peptides, carried out in vitro and cellular assays and co-wrote the manuscript and the ESI. † B. L. synthesized isotope-labelled Ub chains, conducted the NMR studies and assisted with writing the paper. G. B. V. assisted in the synthesis of cyclic peptides. I. L. carried out the confocal microscopy assay and assisted with cellular studies. U. B. carried out the in vivo assays and I. V. assisted in the design of the in vivo assays. A. C. assisted in the design of the confocal microscopy assay and in vitro and cellular studies. D. F. designed and supervised the NMR studies, carried out data analysis and assisted with writing the manuscript and the ESI. † Hi. S. supervised the RaPID study and assisted in the writing of the paper. A. B. designed and supervised the entire project and the writing of the paper.
Data availability
The data that support the findings of this study are available from the corresponding authors upon reasonable request.
Conflicts of interest
There are no conflicts to declare.
Supplementary Material
A. B. holds the Jordan and Irene Tark Academic Chair. This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement no. [831783]). This work was also supported by Japan Agency for Medical Research and Development (AMED), Platform Project for Supporting Drug Discovery and Life Science Research (Basis for Supporting Innovative Drug Discovery and Life Science Research) under JP20am0101090 to H. S. J. M. R. was supported by Grants-in-aid for JSPS Fellows (P13766), a joint ANR-JST grant (ANR-14-JITC-2014-003 and JST-SICORP) and the Novo Nordisk Foundation (0054441).
Hershko A. Ciechanover A. Annu. Rev. Biochem. 1998 67 425 479 10.1146/annurev.biochem.67.1.425 9759494
Berndsen C. E. Wolberger C. Nat. Struct. Mol. Biol. 2014 21 301 307 10.1038/nsmb.2780 24699078
Morimoto D. Walinda E. Fukada H. Sugase K. Shirakawa M. Sci. Rep. 2016 6 39453 10.1038/srep39453 27991582 PMC5172356
Hagai T. Levy Y. Proc. Natl. Acad. Sci. U. S. A. 2010 107 2001 2006 10.1073/pnas.0912335107 20080694 PMC2836630
Komander D. Rape M. Annu. Rev. Biochem. 2012 81 203 229 10.1146/annurev-biochem-060310-170328 22524316
Yau R. Rape M. Nat. Cell Biol. 2016 18 579 586 10.1038/ncb3358 27230526
Thrower J. S. Hoffman L. Rechsteiner M. Pickart C. M. EMBO J. 2000 19 94 102 10.1093/emboj/19.1.94 10619848 PMC1171781
Adams J. Cancer Cell 2004 5 417 421 10.1016/S1535-6108(04)00120-5 15144949
Huang X. Dixit V. M. Cell Res. 2016 26 484 498 10.1038/cr.2016.31 27002218 PMC4822129
Lu S. Wang J. Biomark Res. 2013 1 13 10.1186/2050-7771-1-13 24252210 PMC4177604
Nawatha M. Rogers J. M. Bonn S. M. Livneh I. Lemma B. Mali S. M. Vamisetti G. B. Sun H. Bercovich B. Huang Y. Ciechanover A. Fushman D. Suga H. Brik A. Nat. Chem. 2019 11 644 652 10.1038/s41557-019-0278-x 31182821 PMC7341950
Nguyen T. Ho M. Kim K. Yun S. I. Mizar P. Easton J. W. Lee S. S. Kim K. K. Molecules 2019 24 1073 10.3390/molecules24061073 30893775 PMC6471062
Nguyen T. Ho M. Ghosh A. Kim T. Yun S. I. Lee S. S. Kim K. K. Biochem. Biophys. Res. Commun. 2016 479 33 39 10.1016/j.bbrc.2016.09.010 27613091
Nakasone M. A. Lewis T. A. Walker O. Thakur A. Mansour W. Castaneda C. A. Goeckeler-Fried J. L. Parlati F. Chou T. F. Hayat O. Zhang D. Camara C. M. Bonn S. M. Nowicka U. K. Krueger S. Glickman M. H. Brodsky J. L. Deshaies R. J. Fushman D. Structure 2017 25 1839–1855 e1811 10.1016/j.str.2017.10.007 PMC5731780 29153505
Verma R. Peters N. R. D'Onofrio M. Tochtrop G. P. Sakamoto K. M. Varadan R. Zhang M. Coffino P. Fushman D. Deshaies R. J. King R. W. Science 2004 306 117 120 10.1126/science.1100946 15459393
Ye Y. Blaser G. Horrocks M. H. Ruedas-Rama M. J. Ibrahim S. Zhukov A. A. Orte A. Klenerman D. Jackson S. E. Komander D. Nature 2012 492 266 270 10.1038/nature11722 23201676 PMC3605796
Castaneda C. A. Chaturvedi A. Camara C. M. Curtis J. E. Krueger S. Fushman D. Phys. Chem. Chem. Phys. 2016 18 5771 5788 10.1039/C5CP04601G 26422168 PMC4758893
Lu Y. Lee B. H. King R. W. Finley D. Kirschner M. W. Science 2015 348 1250834 10.1126/science.1250834 25859050 PMC4450770
Livneh I. Kravtsova-Ivantsiv Y. Braten O. Kwon Y. T. Ciechanover A. BioEssays 2017 39 1700027 10.1002/bies.201700027 28493408
Ajish Kumar K. S. Haj-Yahya M. Olschewski D. Lashuel H. A. Brik A. Angew. Chem., Int. Ed. 2009 48 8090 8094 10.1002/anie.200902936 19780082
Bavikar S. N. Spasser L. Haj-Yahya M. Karthikeyan S. V. Moyal T. Kumar K. S. Brik A. Angew. Chem., Int. Ed. 2012 51 758 763 10.1002/anie.201106430 22131237
Yamagishi Y. Shoji I. Miyagawa S. Kawakami T. Katoh T. Goto Y. Suga H. Chem. Biol. 2011 18 1562 1570 10.1016/j.chembiol.2011.09.013 22195558
Pye C. R. Hewitt W. M. Schwochert J. Haddad T. D. Townsend C. E. Etienne L. Lao Y. Limberakis C. Furukawa A. Mathiowetz A. M. Price D. A. Liras S. Lokey R. S. J. Med. Chem. 2017 60 1665 1672 10.1021/acs.jmedchem.6b01483 28059508 PMC5677520
Matsson P. Kihlberg J. J. Med. Chem. 2017 60 1662 1664 10.1021/acs.jmedchem.7b00237 28234469
Doak B. C. Over B. Giordanetto F. Kihlberg J. Chem. Biol. 2014 21 1115 1142 10.1016/j.chembiol.2014.08.013 25237858
Over B. Matsson P. Tyrchan C. Artursson P. Doak B. C. Foley M. A. Hilgendorf C. Johnston S. E. Lee M. D. T. Lewis R. J. McCarren P. Muncipinto G. Norinder U. Perry M. W. Duvall J. R. Kihlberg J. Nat. Chem. Biol. 2016 12 1065 1074 10.1038/nchembio.2203 27748751
Naylor M. R. Bockus A. T. Blanco M. J. Lokey R. S. Curr. Opin. Chem. Biol. 2017 38 141 147 10.1016/j.cbpa.2017.04.012 28570865
Chatterjee J. Rechenmacher F. Kessler H. Angew. Chem., Int. Ed. 2013 52 254 269 10.1002/anie.201205674 23161799
Passioura T. Liu W. Dunkelmann D. Higuchi T. Suga H. J. Am. Chem. Soc. 2018 140 11551 11555 10.1021/jacs.8b03367 30157372
Goto Y. Ohta A. Sako Y. Yamagishi Y. Murakami H. Suga H. ACS Chem. Biol. 2008 3 120 129 10.1021/cb700233t 18215017
Shimizu Y. Inoue A. Tomari Y. Suzuki T. Yokogawa T. Nishikawa K. Ueda T. Nat. Biotechnol. 2001 19 751 755 10.1038/90802 11479568
Goto Y. Katoh T. Suga H. Nat. Protoc. 2011 6 779 790 10.1038/nprot.2011.331 21637198
Ahlbach C. L. Lexa K. W. Bockus A. T. Chen V. Crews P. Jacobson M. P. Lokey R. S. Future Med. Chem. 2015 7 2121 2130 10.4155/fmc.15.78 26067057 PMC5186412
Katoh T. Iwane Y. Suga H. Nucleic Acids Res. 2017 45 12601 12610 10.1093/nar/gkx1129 29155943 PMC5728406
Katoh T. Wohlgemuth I. Nagano M. Rodnina M. V. Suga H. Nat. Commun. 2016 7 11657 10.1038/ncomms11657 27216360 PMC4890201
Roberts R. W. Szostak J. W. Proc. Natl. Acad. Sci. U. S. A. 1997 94 12297 12302 10.1073/pnas.94.23.12297 9356443 PMC24913
Ranjani V. Assfalg M. Fushman D. Methods Enzymol. 2005 399 177 192 16338356 10.1016/S0076-6879(05)99012-5
Varadan R. Walker O. Pickart C. Fushman D. J. Mol. Biol. 2002 324 637 647 10.1016/S0022-2836(02)01198-1 12460567
Eddins M. J. Varadan R. Fushman D. Pickart C. M. Wolberger C. J. Mol. Biol. 2007 367 204 211 10.1016/j.jmb.2006.12.065 17240395
Ryabov Y. E. Fushman D. J. Am. Chem. Soc. 2007 129 3315 3327 10.1021/ja067667r 17319663 PMC2586837
Mevissen T. E. Hospenthal M. K. Geurink P. P. Elliott P. R. Akutsu M. Arnaudo N. Ekkebus R. Kulathu Y. Wauer T. El Oualid F. Freund S. M. Ovaa H. Komander D. Cell 2013 154 169 184 10.1016/j.cell.2013.05.046 23827681 PMC3705208
Bremm A. Freund S. M. Komander D. Nat. Struct. Mol. Biol. 2010 17 939 947 10.1038/nsmb.1873 20622874 PMC2917782
Sun H. Mali S. M. Singh S. K. Meledin R. Brik A. Kwon Y. T. Kravtsova-Ivantsiv Y. Bercovich B. Ciechanover A. Proc. Natl. Acad. Sci. U. S. A. 2019 116 7805 7812 10.1073/pnas.1822148116 30867293 PMC6475391
Goldberg A. L. J. Cell Biol. 2012 199 583 588 10.1083/jcb.201210077 23148232 PMC3494858
Mansour W. Nakasone M. A. von Delbruck M. Yu Z. Krutauz D. Reis N. Kleifeld O. Sommer T. Fushman D. Glickman M. H. J. Biol. Chem. 2015 290 4688 4704 10.1074/jbc.M114.568295 25389291 PMC4335208
Deshaies R. J. BMC Biol. 2014 12 94 10.1186/s12915-014-0094-0 25385277 PMC4226866
Crawford L. J. Walker B. Irvine A. E. J. Cell Commun. Signaling 2011 5 101 110 10.1007/s12079-011-0121-7 21484190 PMC3088792
Manasanch E. E. Orlowski R. Z. Nat. Rev. Clin. Oncol. 2017 14 417 433 10.1038/nrclinonc.2016.206 28117417 PMC5828026
Huang Y. Nawatha M. Livneh I. Rogers J. M. Sun H. Singh S. K. Ciechanover A. Brik A. Suga H. Chemistry 2020 26 8022 8027 10.1002/chem.202000273 32105365
Barash U. Lapidot M. Zohar Y. Loomis C. Moreira A. Feld S. Goparaju C. Yang H. Hammond E. Zhang G. Li J. P. Ilan N. Nagler A. Pass H. I. Vlodavsky I. J. Natl. Cancer Inst. 2018 110 1102 1114 10.1093/jnci/djy032 29579286 PMC6186523
Weissmann M. Arvatz G. Horowitz N. Feld S. Naroditsky I. Zhang Y. Ng M. Hammond E. Nevo E. Vlodavsky I. Ilan N. Proc. Natl. Acad. Sci. U. S. A. 2016 113 704 709 10.1073/pnas.1519453113 26729870 PMC4725485
Adams J. Palombella V. J. Sausville E. A. Johnson J. Destree A. Lazarus D. D. Maas J. Pien C. S. Prakash S. Elliott P. J. Cancer Res. 1999 59 2615 2622 10363983
Vilchez D. Saez I. Dillin A. Nat. Commun. 2014 5 5659 10.1038/ncomms6659 25482515
Fricker L. D. Annu. Rev. Pharmacol. Toxicol. 2020 60 457 476 10.1146/annurev-pharmtox-010919-023603 31479618
Kaplan G. S. Torcun C. C. Grune T. Ozer N. K. Karademir B. Free Radical Biol. Med. 2017 103 1 13 10.1016/j.freeradbiomed.2016.12.007 27940347
Liu F. S. Taiwan. J. Obstet. Gynecol. 2009 48 239 244 10.1016/S1028-4559(09)60296-5 19797012