Sequence-Specific Peptide Synthesis by an Artificial Small-Molecule Machine

See allHide authors and affiliations

Science  11 Jan 2013:
Vol. 339, Issue 6116, pp. 189-193
DOI: 10.1126/science.1229753

Ribosomal Rotaxane?

The ribosome is an extraordinarily sophisticated molecular machine, assembling amino acids into proteins based on the precise sequence dictated by messenger RNA. Lewandowski et al. (p. 189) have now taken a step toward the preparation of a stripped-down synthetic ribosome analog. Their machine comprises a rotaxane—a ring threaded on a rod—in which the ring bears a pendant thiol that can pluck amino acids off the rod; the terminal nitrogen then wraps around to form a peptide bond and liberate the thiol for further reaction. The system was able to link three amino acids in order from the preassembled rod.


The ribosome builds proteins by joining together amino acids in an order determined by messenger RNA. Here, we report on the design, synthesis, and operation of an artificial small-molecule machine that travels along a molecular strand, picking up amino acids that block its path, to synthesize a peptide in a sequence-specific manner. The chemical structure is based on a rotaxane, a molecular ring threaded onto a molecular axle. The ring carries a thiolate group that iteratively removes amino acids in order from the strand and transfers them to a peptide-elongation site through native chemical ligation. The synthesis is demonstrated with ~1018 molecular machines acting in parallel; this process generates milligram quantities of a peptide with a single sequence confirmed by tandem mass spectrometry.

Cells achieve the sequence-specific synthesis of information-rich oligomers and polymers through the operation of complex molecular machines that transcribe information from the genetic code (1). The most extraordinary of these is the ribosome (24), a ~2.6-MD (bacterial) to ~4.3-MD (eukaryotic) molecular machine found in all living cells that assembles amino acids from tRNA building blocks into a peptide chain with an order defined by the sequence of the mRNA strand that it moves along. Artificial small-molecule machines (5) have previously been used to store information (6, 7) and do mechanical work (811); others have been employed in synthesis to processively epoxidize an unsaturated polymer (12, 13), switch “on” and “off” catalytic activity (1417), and change the handedness of a reaction product (18). Large synthetic DNA molecules have been used to guide the formation of bonds between unnatural building blocks (1922) and assemble gold nanoparticles in particular sequences (23). Here, we report on the design, synthesis, and operation of a rotaxane-based small-molecule machine in which a functionalized macrocycle operates on a thread containing building blocks in a predetermined order to achieve sequence-specific peptide synthesis. The design of the artificial molecular machine is based on several elements that have analogs in either ribosomal (24) or nonribosomal (24) protein synthesis: Reactive building blocks (the role played by tRNA-bound amino acids) are delivered in a sequence determined by a molecular strand (the role played by mRNA). A macrocycle ensures processivity during the machine’s operation (reminiscent of the way that subunits of the ribosome clamp the mRNA strand) and bears a catalyst—a tethered thiol group—that detaches the amino acid building blocks from the strand and passes them on to another site at which the resulting peptide oligomer is elongated in a single specific sequence, through chemistry related to nonribosomal peptide synthesis (24).

The chemical structure of the artificial molecular machine, 1, is shown in Fig. 1. Strand 2 bears three amino acids attached to the track by weak phenolic ester linkages (25) and separated from each other by rigid spacers that minimize the possibility of the reactive arm of the machine coming into contact and reacting with a building block out of sequence (26). Macrocycle 3 contains an endotopic pyridine group that directs the threading of strand 2 during the Cu(I)-catalyzed cycloaddition of the terminal alkyne with the azide-bearing stopper group 5, leading to the assembly of rotaxane 4 in 30% yield (Fig. 1, i). This active template (27, 28) strategy ensured that the resulting threaded structure did not have residual attractive intercomponent interactions that would tend to localize the position of the ring rather than allow it to move freely up and down the strand between blocking groups. Once we assembled the macrocycle-strand-stopper conjugate 4, we used reversible hydrazone exchange to introduce a cysteine derivative bearing the reactive arm [a trityl (Trt)–protected thiol group] and the site for peptide elongation [a tert-butoxycarbonyl carbamate (Boc)–protected amine at the end of a glycylglycine residue] (Fig. 1, ii). The fully assembled machine 1 is stable in its protected form, with upfield shifts of the HP1 triazole and nearby HS12-S14 proton signals evident in the 1H nuclear magnetic resonance (NMR) spectrum on account of shielding from the phenyl rings of the macrocycle, confirming that the ring is trapped in the region of the strand between the terminal stopper and the Boc-phenylalanine ester (Fig. 2).

Fig. 1

Synthesis of rotaxane-based molecular machine 1, incorporating a strand bearing amino acid building blocks (2), a macrocycle (3) with a site for attachment of the reactive arm, and a terminal blocking group (5) that prevents the threaded macrocycle from coming off the strand until all of the amino acid groups have been cleaved. Structure 1 is shown as the major diastereomer; a small amount of epimerization (<5%) of some of the acyl amino units occurs during incorporation into the track. Boc, CO2C(CH3)3; Piv, COC(CH3)3; Trt, CPh3; Ph, phenyl. Reaction conditions: (i) Cu(CH3CN)4PF6 in dichloromethane:t-butanol (2:1), room temperature, 4 days, 30%. (ii) PhNH2 (catalyst), BocGlyGlyCys(S-Trt)NHN=CHC6H4OCH3 in 3:1 dimethylsulfoxide: aqueous 2-(N-morpholino)ethanesulfonic acid buffer (pH 6.0), 60°C, 2 days, 90%. The italicized letters indicate key signals in the 1H NMR spectrum shown in Fig. 2B. For the full lettering scheme and assignments, see the supplementary materials, page S28 (26).

Fig. 2

Proton NMR spectrum of (A) the noninterlocked thread and (B) rotaxane 1, in d6-dimethylsulfoxide (500 MHz, 298 K). Rotaxane 1 exists in both E- and Z-hydrazone forms. The assignments correspond to the lettering shown in Fig. 1. C′, S-Trt-cysteine; G′, N-Boc-glycine; F′, N-Boc-phenylalanine; L′, N-Boc-leucine; A′, N-Piv-alanine. ppm, parts per million.

We used acid-catalyzed cleavage of the Boc and trityl protecting groups (Fig. 3, i) to activate the molecular machine and then allowed it to operate (Fig. 3, ii) at 60°C under microwave heating in a 3:1 acetonitrile:dimethylformamide solution in the presence of N,N-diisopropylethylamine (a non-nucleophilic base) and tris(2-carboxyethyl)phosphine (a reducing agent that cleaves any disulfide bonds formed through thiol oxidation). The design of the machine is such that once the thiolate residue of the cysteine group (6a) is deprotected, it is poised to undergo a transacylation reaction with the first amino acid phenolic ester that blocks the macrocycle’s path on the track (6b). We hypothesized that the subsequently formed phenylalanine thioester (6c) would be able to react further, transferring the amino acid by native chemical ligation (29) to the glycylglycine amine group by an 11-membered-ring transition state [the dipeptide spacer between the cysteine residue and the amine of the peptide-elongation site was introduced because native chemical ligation is reported to be very slow via 8-membered-ring transition states (30)]. This sequence simultaneously transfers the amino acid to the end of the growing peptide (6d) and regenerates the catalytic thiolate group, ready for the cleavage and transfer of further building blocks. S-N acyl transfer is a key feature of nonribosomal peptide synthesis (24).

Fig. 3

Proposed mechanism for sequence-specific peptide synthesis by molecular machine 1. After activation of the machine by acidic cleavage of the Boc and Trt protecting groups, under basic conditions successive native chemical ligation reactions transfer the amino acid building blocks to the peptide-elongation site on the macrocycle in the order they appear on the thread. Once the final amino acid is cleaved, the macrocycle bearing the synthesized oligopeptide 7 dethreads from the strand. The hydrazide peptide 9 is subsequently released from the macrocycle by hydrolysis. Reaction conditions: (i) 20% CF3CO2H in dichloromethane, room temperature, 2 hours, 100%. (ii) ((CH3)2CH)2NEt, (HO2CCH2CH2)3P in 3:1 acetonitrile:dimethylformamide, 60°C, 36 hours. Et, ethyl. (iii) 30% CF3CO2H in 3:1 dichloromethane:water, room temperature, 18 hours.

Once the covalent bond connecting an amino acid to the strand is broken, the macrocycle is able to move further along the track until its path is blocked by the next amino acid group (6d). The O-S acyl transfer/S-N acyl transfer/catalyst regeneration/ring movement process continues (6e to 6h) until the last amino acid on the track is cleaved (6h) and the macrocycle detaches from the strand (8) with the newly formed, full length, peptide attached (7). The artificial molecular machine synthesizes the peptide from the C terminus to the N terminus, the opposite direction of ribosomal translation (24).

After a 36-hour operation of 6a at 60°C, no starting material remained, as shown by high-performance liquid chromatography (HPLC), and two major products were isolated from the reaction mixture (26). We used 1H NMR spectroscopy and mass spectrometry to identify one product as the completely deacylated thread, 8. The other product had a 1H NMR spectrum and molecular weight (Fig. 4) consistent with the macrocycle bearing the hydrazone linked to the hexapeptide (Piv)AlaLeuPheGlyGlyCys, 7 [Piv, COC(CH3)3].

Fig. 4

(A) Proton NMR spectrum of molecular machine operation product 7 in d6-dimethylsulfoxide (500 MHz, 298 K). (B) Tandem mass spectrum of a single isotope of the 2+ ion [mass/charge ratio (m/z) = 876.64] of S,N-acetal–derivatized 7. (C) Superimposed tandem mass spectra of 2+ ions of S,N-acetal–derivatized macrocycles bearing the peptide sequences (Piv)AlaPheLeuGlyGlyCys (red; 2+ ion isotope-selected m/z = 876.73) and (Piv)AlaLeuPheGlyGlyCys (blue; 2+ ion isotope-selected m/z = 876.92), each prepared unambiguously by conventional peptide synthesis.

To confirm that the product of the molecular machine’s operation had the amino acids assembled in the correct order, we used tandem mass spectrometry (MS/MS) to determine the peptide sequence. We then validated the sequence by comparing it with an authentic sample and an isomer in which the order of the Phe and Leu residues was reversed, each prepared unambiguously by conventional peptide synthesis (26). Figure 4A shows the 1H NMR spectrum of the molecular machine product 7, and Fig. 4B shows the MS/MS spectrum of one isotope of a 2+ ion of 7 derivatized through the cysteine as an S,N-acetal, a species that gave a sufficient signal-to-noise ratio for the MS/MS experiment. Figure 4C shows the superimposition of the MS/MS spectra of a similar isotope and ion from the authentic samples of the macrocycle bearing the sequence (Piv)AlaPheLeuGlyGlyCys (red peaks) and (Piv)AlaLeuPheGlyGlyCys (blue peaks). The difference in the fragmentation masses of the two sequence isomers is apparent in Fig. 4C (725.73 for the LeuGlyGlyCys-macrocycle, 742.92 for the PheGlyGlyCys-macrocycle), and product 7 was confirmed as corresponding to the intended sequence isomer.

With the use of HPLC-MS analysis of the reaction mixture from the operation of 1, we did not detect any products corresponding to other peptide compositions (neither different sequences nor peptides with more or less than one Phe, Leu, or Ala residue), indicating that the peptide synthesis occurs overwhelmingly within the confines of the molecular machine. In contrast, a control reaction carried out under identical conditions but using the nonthreaded strand and macrocycle yielded several products, including strands with one or more amino acid groups cleaved, but there was no evidence for the formation of 7 under these conditions. Thus, the threaded architecture of the molecular machine—encompassing the catalytic site, elongation site, and the building block strand—is essential for the sequential peptide synthesis, and the mode of operation of the molecular machine is consistent with the mechanism shown in Fig. 3. The peptide (9), still bearing the GlyGlyCys unit at the C terminus, could subsequently be cleaved from the macrocycle by hydrolysis (Fig. 3, iii).

On a scale of tens of milligrams, we performed the synthesis of the small peptide through autonomous multistep production by artificial small-molecule machine 1, corresponding to parallel synthesis by ~1018 machines. Once operation is initiated, the synthetic tasks performed by 1 proceed automatically, requiring no further intervention. As the catalytic thiolate is constrained by the threaded architecture of the machine from reacting with building blocks out of sequence, the act of balancing the rate of reactions with the speed that templates rearrange (1922) is unnecessary for a rotaxane-based machine.

Rotaxane 1 is a (very) primitive analog of the ribosome. Limitations of the first-generation artificial system include slow kinetics (1 takes ~12 hours to make each amide bond, compared to the 15 to 20 amide bonds synthesized per second by a ribosome) and loss of the sequence information on the strand as it is translated into the product. Furthermore, the size of oligopeptide that can be produced may ultimately be restricted by the size of the cyclic transition states involved in S-to-N acyl transfer [although peptide ligation has been successfully used with up to 29-membered cyclic transition states (31)]. Nevertheless, 1 demonstrates that relatively small, highly modular, artificial molecular machines can be designed to autonomously perform iterative tasks in synthesis. The principles employed in the design and operation of 1 should be broadly applicable to other types of monomer and chemical reactions (32).

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S37

References (3335)

References and Notes

  1. S,S-acyl transfer is readily reversible, so employing thioesters (the most commonly used acyl source for native chemical ligation) to attach the building blocks to the strand would risk an amino acid being returned to the track after the macrocycle had passed by, potentially reducing the sequence integrity of the peptide synthesis.
  2. Materials and methods are available as supplementary materials on Science Online.
  3. Acknowledgments: We thank V. Aucagne for useful suggestions during early versions of the molecular machine design. This research was funded by the Engineering and Physical Sciences Research Council (UK). We are grateful to the following organizations for postdoctoral fellowships: Fundacja na Rzecz Nauki Polskiej (to B.L.), Fonds de la Recherche Scientifique and Wallonie-Bruxelles International (to G.D.B.), the European Union 7th Framework Marie Curie Intra European Fellowship Program (to M.J.A.), Deutscher Akademischer Austausch Dienst (to P.M.E.G. and D.H.), and Deutsche Akademia der Naturforscher Leopoldina and Peter und Traudl Engelhorn–Stiftung (to D.M.D.).
View Abstract

Navigate This Article