Post-translational backbone rearrangement produces polyketide-like units in peptides

Post-translational backbone rearrangement produces polyketide-like units in peptides

While peptides expressed by the ribosome are generally composed of twenty proteinogenic α-amino acids, various unique non-canonical building blocks are present in peptidic natural products. For example, some peptides produced in bacteria contain hydroxyhydrocarbon (Hhc) units, which are β-, γ-, or δ-amino acids with hydroxy modifications of the main chain1-3. The representative class of Hhc units is γ-amino-β-hydroxy acids, so-called statine derivatives, and can mimic the tetrahedral transition state in peptide hydrolysis4,5. Furthermore, Hhc units can rigidify the local and global conformation of the peptides via intramolecular hydrogen bonds6. Therefore, the Hhc units have been regarded not only as the prominent components in natural products but also as attractive building blocks for developing artificial peptide agents. 

In nature, the Hhc-containing peptides are biosynthesized by polyketide synthase (PKS)-nonribosomal peptide synthetase (NRPS) hybrid megasynthetases7. They construct a specific sequence of Hhc-containing peptides, which are precisely defined by the sequence of domains in the megasynthetases. Although the PKS-NRPS hybrids are excellent biosynthetic machinery for producing certain natural products, due to their sophisticated mode of action, it is technically difficult to modify them to yield artificial analogs bearing Hhc units. Hence, to date, Hhc-containing artificial peptides have been developed exclusively by chemical synthesis.   

Well then, is it possible to utilize the ribosome-mediated translation reaction to express Hhc-containing peptides? Unfortunately, the answer is “tremendously difficult” because of the following two reasons. First, since the ribosome and translation factors have evolved to elongate peptide chains using α-amino groups, Hhc units bearing a β-/γ-/δ-amino group are inherently poor or inactive substrates in translation. In addition, amino acids need to be charged on tRNA to be used in translation, but Hhc-tRNAs cannot exist stably in the first place – they are spontaneously degraded via an intramolecular ring-closing reaction. Therefore, even though state-of-the-art methods of translation engineering have recently made it possible to utilize diverse non-canonical amino acids in translation, the repertoire of Hhc units directly accessible by ribosomal synthesis is still highly limited.  

To overcome these obstacles, we have designed translation-compatible chemical precursors of Hhc units and devised a peptide backbone-rearrangement reaction for post-translational generation of Hhc units. In this strategy, Hhc units with a masked amino group are incorporated into peptide via an ester linkage, then the deprotection of the amino group followed by spontaneous O-to-N acyl shift yields Hhc units in ribosomally synthesized peptide backbones (Figure 1)8. We adopted an azide as the protecting group because it could be reduced by phosphine reagents in aqueous and mild conditions without affecting to other functional groups in the peptide, known as Staudinger reduction. Thus, we designed a series of azide/hydroxy acids (AzHyA) for the substrates of both translation and acyl shift reaction. 

Figure 1: The post-translational backbone acyl-shift reaction to incorporate Hhc units into ribosomally synthesized peptides. 

As a proof of concept, we first prepared 4-azide-3-hydroxybutanoic acid (γN3βOH) as a model AzHyA residue. Our engineered translation system, so-called flexible in vitro translation (FIT) system, enabled the genetic code reprogramming and yielded a model macrocyclic peptide bearing γN3βOH in the peptide backbone via a β-ester linkage9. Although we initially expected that post-translational chemical reduction of the azide group by a phosphine reagent would smoothly proceed to yield the side chain amino group and induce O-to-N acyl shift to generate the objective Hhc unit, it turned out that the reaction, in this case, competed with several side-reactions, including undesirable conversion of the azide to a hydroxy group and hydrolysis of the β-ester linkage. Nonetheless, after the intensive optimization of the conditions, we overcame these side-reactions by elevating pH and controlling reaction temperature, yielding the desired Hhc-containing peptide with almost quantitative conversion efficiency.  

Under the optimized reaction conditions, we explored the scope and limitations of the acyl shift strategy by testing a series of AzHyA derivatives (Figure 2). In addition to the model γ-peptide type Hhc without side chain substitution derived from γN3βOH, γ-AzHyAs with side chain substitutions (Sta-N3 and PhSta-N3) were also applicable to the post-translational acyl shift, yielding pharmaceutically-relevant statine and phenylstatine units. Furthermore, β- and δ-peptide type Hhc units could also be generated by using a series of AzHyAs (βN3αOH, γN3αOH, and δN3αOH) with the excellent to modest conversion yields. Notably, AzHyAs possessing α,β-dihydroxy group (γN3αOHβOH and δN3αOHβOH) could improve the conversions of long-range acyl shift yielding γ- and δ-peptides, plausibly via kinetically favorable O-to-O-to-N tandem acyl shift (see Figure 1). Altogether, we have demonstrated that by designing appropriate AzHyAs having desirable side chain structures and azide/hydroxy groups at appropriate positions, a variety of Hhc units can be post-translationally generated on the peptide backbone. 

Figure 2: Structure of the tested AzHyA derivatives and the corresponding Hhc units after the acyl shifts. 

The devised method has allowed for the ribosomal expression of an artificial statine-containing peptide drug, which was originally developed by chemical synthesis. We objected to synthesize P10-P4'statV, a potent β-secretase 1 inhibitor designed by replacing the cleavage site of a β-secretase 1 substrate with a statine residue10. The precursor peptide was expressed by co-reprogrammed translation with Sta-N3 and an N-terminal-protected amino acid, and post-translational deprotections induced liberation of the N-terminal amino group and the objective acyl shift. The ribosomally synthesized P10-P4'statV was identified with the comparison with the authentic synthetic peptide. This result demonstrate the utility of the post-translational backbone acyl shift strategy for the expression of bioactive peptides bearing Hhc units.  

This study demonstrated the expansion of the scope of in vitro engineered translation systems to previously inaccessible Hhc units. The mRNA template-dependent peptide expression offers facile synthesis of designer peptides with variable sequences. The most significant feature of in vitro translation is that combinatorial peptide libraries with a diversity of over 1012 can be readily constructed and applied to selection-based screening for identifying de novo peptide ligands against protein targets of interest11. We believe this method will accelerate the development of artificial peptide agents taking advantage of unique building blocks derived from natural products, which we refer to as “pseudo-natural peptides”. 

The full story of “Post-translational backbone-acyl shift yields natural product-like peptides bearing hydroxyhydrocarbon units” is available at here: 


  1. Umezawa, H., Aoyagi, T., Morishima, H., Matsuzaki, M. & Hamada, M. Pepstatin, a new pepsin inhibitor produced by Actinomycetes. J. Antibiot. 23, 259–262 (1970). 
  2. Nyfeler, R. & Keller-Schierlein, W.  Metabolites of microorganisms. 143. Echinocandin B, a novel polypeptide-antibiotic from Aspergillus nidulans var. echinulatus: isolation and structural components. Helv. Chim. Acta 57, 2459–2477 (1974). 
  3. Plaza, A. et al. Celebesides A-C and theopapuamides B-D, depsipeptides from an Indonesian sponge that inhibit HIV-1 entry. J. Org. Chem. 74, 504–512 (2009). 
  4. Bott, R., Subramanian, E. & Davies, D. R. Three-dimensional structure of the complex of the Rhizopus chinensis carboxyl proteinase and pepstatin at 2.5 Å resolution. Biochemistry 21, 6956–6962 (1982).
  5. Rich, D. H. et al. Inhibition of aspartic proteases by pepstatin and 3-methylstatine derivatives of pepstatin. Evidence for collected-substrate enzyme inhibition. Biochemistry 24, 3165–3173 (1985). 
  6. Radics, G., Koksch, B., El-Kousy, S. M., Spengler, J. & Burger, K. L-α-methylhomoisoserine: a new versatile building block for peptide and depsipeptide modification. Synlett 2003, 1826–1829 (2003). 
  7. Walsh, C. T., O’Brien, R. V. & Khosla, C. Nonproteinogenic amino acid building blocks for nonribosomal peptide and hybrid polyketide scaffolds. Angew. Chem. Int. Ed. 52, 7098–7124 (2013). 
  8. Sohma, Y. & Kiso, Y. Synthesis of O-acyl isopeptides. Chem.Rec. 13, 218–223 (2013).
  9. Goto, Y., Katoh, T. & Suga, H. Flexizymes for genetic code reprogramming. Nat. Protoc. 6, 779–790 (2011).
  10. Sinha, S. et al. Purification and cloning of amyloid precursor protein β-secretase from human brain. Nature 402, 537–540 (1999). 
  11. Goto, Y. & Suga, H. The RaPID platform for the discovery of pseudo-natural macrocyclic peptides. Acc. Chem. Res. 54, 3604–3617 (2021).

Please sign in or register for FREE

If you are a registered user on Nature Portfolio Chemistry Community, please sign in