Cotranslational protein folding on the ribosome monitored in real time

See allHide authors and affiliations

Science  27 Nov 2015:
Vol. 350, Issue 6264, pp. 1104-1107
DOI: 10.1126/science.aad0344

Proteins shape up in the ribosome

Proteins consist of linear chains of amino acids. These chains must fold into complex three-dimensional shapes to become functional. Holtkamp et al. “watched” how a small helical protein folds as it is being synthesized by the ribosome. The lengthening polypeptide passes out through the ribosome exit tunnel where folding starts. The initially compact structure quickly rearranges into a native three-dimensional structure as the polypeptide emerges from the tunnel.

Science, this issue p. 1104


Protein domains can fold into stable tertiary structures while they are synthesized on the ribosome. We used a high-performance, reconstituted in vitro translation system to investigate the folding of a small five-helix protein domain—the N-terminal domain of Escherichia coli N5-glutamine methyltransferase HemK—in real time. Our observations show that cotranslational folding of the protein, which folds autonomously and rapidly in solution, proceeds through a compact, non-native conformation that forms within the peptide tunnel of the ribosome. The compact state rearranges into a native-like structure immediately after the full domain sequence has emerged from the ribosome. Both folding transitions are rate-limited by translation, allowing for quasi-equilibrium sampling of the conformational space restricted by the ribosome. Cotranslational folding may be typical of small, intrinsically rapidly folding protein domains.

In living cells, folding of many proteins begins cotranslationally as soon as the N-terminal part of a given protein emerges from the peptide exit tunnel of the ribosome (13); secondary structure elements, such as α helices, can form within the exit tunnel (47). Whereas small- and medium-size protein domains can acquire their native structures in less than a second (8), the bacterial ribosome requires 5 to 10 s to synthesize a protein domain 100 amino acids in length. Cotranslational folding may be attenuated by interactions of the nascent peptide with the ribosome (9) and by auxiliary proteins, such as chaperones or other protein biogenesis factors (1012). Changes in local translational kinetics, such as those caused by rare codons or transfer RNA (tRNA) abundance, can influence the conformation of newly synthesized proteins (13, 14). Little is known about the exact timing of cotranslational protein folding in relation to protein synthesis or the conformation of the polypeptide emerging from the ribosome.

To monitor cotranslational protein folding during ongoing translation, we used a reconstituted in vitro translation system in combination with the selective site-specific labeling of nascent proteins with fluorescent probes (15). We studied the N-terminal domain (NTD) of Escherichia coli N5-glutamine methyltransferase HemK. The HemK NTD consists of five helices (residues 3 to 12, 20 to 29, 35 to 42, 48 to 65, and 67 to 73) (Fig. 1) connected by a linker (residues 72 to 96) to the C-terminal domain (CTD) (16). The isolated HemK NTD (residues 1 to 73) forms a stable α-helical structure independent of the CTD (figs. S1 and S2). Free NTD in solution folds on a (sub)millisecond time scale in a predominantly two-state fashion (fig. S3 and table S1). For in vitro translation, we used an mRNA construct coding for the N-terminal 112 amino acids of HemK (HemK112) comprising the NTD (73 amino acids) plus 39 amino acids from the linker and the CTD. This length should allow the NTD to fully emerge from the ribosome exit tunnel when HemK112 is synthesized (Fig. 1).

Fig. 1 Schematics of the model protein HemK NTD.

(A) Secondary structure elements of HemK; helices H1 to H5 are shown as bars assuming fully α-helical conformation (1.5 Å per residue). Green and red stars indicate, respectively, the positions of the BOF-Met and BOP-Lys dyes in the proteins. Dark and light gray shaded areas indicate the positions of the ribosome exit tunnel and the vestibule relative to the nascent peptide if all secondary structures were formed. Vertical bars delineate boundaries of the exit tunnel assuming a fully extended conformation of the nascent peptide (3.5 Å per residue). Numbers of amino acids (42 to 112 aa) correspond to lengths of the nascent peptides. PTC, peptidyl transferase center. (B) Crystal structure of the HemK NTD (PDB ID 1T43). Secondary structure elements H1 to H5 are color-coded as in (A).

We rapidly mixed synchronized initiation complexes containing BodipyFL (BOF)–Met-tRNAfMet with elongation factors (EF-Tu, EF-Ts, and EF-G) and purified aminoacyl-tRNAs. Instead of Lys-tRNA, we added εNH2-Bodipy576/589 (BOP)–Lys-tRNALys to introduce the second fluorescence label at position 34 of the nascent peptide (Fig. 1). The two reporters provide a donor-acceptor pair to monitor Förster resonance energy transfer (FRET). Donor- and acceptor-only controls served to correct the time courses measured in the presence of both donor and acceptor (fig. S4A). The appearance of FRET after 10 s (Fig. 2A and fig. S4C) implies the formation of a structure in which positions 1 and 34 of HemK112 come into close proximity to one another. When compared to the time course of translation, the band representing the full-length protein appears after 40 s (Fig. 2B and fig. S5). Thus, chain compaction begins earlier than the full-length domain emerges from the ribosome.

Fig. 2 Cotranslational folding monitored by FRET.

(A) Changes in BOF-Met1 (donor, green) and BOP-Lys34 (acceptor, red) fluorescence corrected for bleed-through and fluorescence changes unrelated to folding (see supplementary materials). Translation was carried out in HiFi buffer at 37°C. (B) Time courses of translation, derived from SDS–polyacrylamide gel electrophoresis experiments (circles) and normalized FRET changes (acceptor; red lines), for HemK constructs of different lengths. FRET traces are normalized to the maximum fluorescence change (endpoint) measured for HemK112; a.u., arbitrary units. (C) Endpoints of FRET change (acceptor; light gray bars) for HemK nascent peptides of different lengths. +Pmn, FRET in HemK70 peptide released from the ribosome by puromycin treatment; dark gray bars, nascent peptides (chain lengths 70 or 112) with replacement Y3F. Error bars indicate SD (n = 12 replicates).

We translated HemK peptides of different lengths ranging from 42 to 112 residues (Figs. 1A and 2B) and derived a translation velocity of 3.6 ± 0.1 amino acids/s (Fig. 2B and fig. S5B). Translation of HemK42 results in a low final FRET efficiency, reflecting an extended conformation of short nascent chains within the exit tunnel (Fig. 2, B and C). Synthesis of HemK56 leads to the formation of a high-FRET state, which suggests that the two probes come close—presumably by compaction of the nascent peptide—even though BOP-Lys on the nascent HemK56 chain should be occluded inside the tunnel, even when the peptide is in a fully extended conformation (Fig. 1A). Thus, the structure manifested by high FRET must form largely inside the ribosome, close to the exit of the peptide tunnel. With the HemK70 construct, the FRET efficiency is even higher, and the time of FRET appearance coincides with the synthesis of the 70–amino acid product (Fig. 2B). For all constructs, FRET rapidly increases after ~10 s of translation, whereas the formation of full-length HemK84, 98, and 112 is markedly slower than the appearance of the high-FRET state.

The final FRET efficiency is somewhat lower for HemK84, 98, and 112 relative to HemK70, indicating that there are structural rearrangements when the peptide chain becomes longer than 70 amino acids (Fig. 2C). With fully translated HemK112, the entire 73–amino acid NTD is likely to emerge from the exit tunnel, and thus the lower FRET level reflects the end state of folding. In contrast, the conformation captured by using the HemK70 construct does not represent a native-like fold, but rather a compact state that is stabilized by the ribosome until the next part of the protein sequence emerges. When we released the nascent HemK70 peptide from the ribosome by puromycin treatment, the FRET efficiency decreased to the level measured with the longer constructs (Fig. 2C and fig. S6), reflecting spontaneous domain folding (fig. S3). The decrease of FRET in HemK112 is attributable to the stabilization of the N terminus: BOP-Lys34 is in close proximity to Tyr3, which can quench BOP fluorescence (17, 18). When we substituted Tyr3 with Phe, we recovered the high FRET signal (Fig. 2C). These data indicate the existence of two states along the folding pathway of the NTD: a compact state, formed early during translation, which rearranges into a near-native fold upon further translation.

We probed the folding of HemK nascent peptides of different lengths by limited proteolysis with thermolysin, which cleaves at sites with bulky and aromatic residues (19), and monitored the cleavage of the BOF-labeled N-terminal peptide. In addition to HemK wild-type constructs, we used a mutated NTD in which four conserved Leu residues, comprising the hydrophobic core of the NTD, were simultaneously replaced with Ala (4×A) (Fig. 3A). The NTD with even a single Ala mutation was almost completely unfolded at 37°C (Fig. 3B). HemK70 nascent chains (wild-type or 4×A) were largely protected from protease digestion (Fig. 3C and fig. S7), because the N-terminal BOF resides inside or in close proximity to the peptide exit tunnel, which protects it from proteolytic cleavage. When released from the ribosome by puromycin treatment, HemK70 4×A peptides were rapidly digested, whereas wild-type peptides were more resistant to proteolysis (Fig. 3C and fig. S7B). HemK84 nascent peptides (wild-type or 4×A) were rapidly digested, indicating that the N-terminal parts of the peptides are exposed and do not adopt a protease-resistant conformation. In contrast, longer wild-type nascent peptides were significantly more protease-resistant than the respective 4×A mutants (figs. S7 and S8). The kinetics of proteolysis was similar for HemK112 nascent chains and HemK70 peptides released by puromycin (Fig. 3C and fig. S7), supporting the notion that HemK112 attains a native-like structure similar to the one that forms after release of nascent chains from the ribosome.

Fig. 3 Probing the folding status by proteolysis.

(A) Positions of the Leu → Ala replacements in the protein core of the HemK 4×A construct. (B) The stability of the isolated NTD in solution measured by far-ultraviolet circular dichroism spectra at different temperatures. (C) Relative stability against thermolysin digestion of the native (τwt) and 4×A (τ4×A) NTD on the ribosome and after release into solution by puromycin treatment (+Pmn). Error bars indicate SEM of the fits.

Together, the results of time-resolved FRET and limited proteolysis suggest that cotranslational folding proceeds through a compact state that is formed early during peptide elongation, when the nascent peptides are still confined in the exit tunnel. The latter intermediate state converts to the native-like fold upon emergence of the entire domain from the peptide exit tunnel or after the release of the nascent chain from the ribosome.

We used photoinduced electron transfer (PET) between the N-terminal BOF-Met and Trp residues within the nascent chain, which is ideally suited to study conformational rearrangements on short length scales (20). We introduced single Trp residues in different positions of HemK (6W, F38W, D49W) (Fig. 4A) and monitored the change in BOF-Met fluorescence during translation and movement of the nascent chain through the exit tunnel. We extracted the changes reflecting nascent chain folding by correcting for the fluorescence change of BOF-Met in a HemK construct lacking any intrinsic Trp residue (fig. S9).

Fig. 4 Cotranslational folding monitored by PET.

(A) Positions of PET pairs in the HemK NTD. BOF-Met (green sphere) is shown relative to the intrinsic Trp residue in H1 (native sequence 6W) or engineered into H3 (F38W) or H4 (D49W); in both latter cases, the native Trp residue was replaced with Phe (W6F). Secondary structure elements H1 to H4 are color-coded as in Fig. 1A. (B) Time-resolved PET changes upon translation of HemK70 and HemK112 wild-type or 4×A constructs, corrected for the PET events unrelated to folding (see supplementary materials). (C) The endpoints of PET for the wild-type (gray bars) and 4×A (red bars) HemK NTD nascent chains of different lengths as indicated. Error bars indicate SD (n = 12 replicates). (D) PET time courses with Trp at positions 38 and 49. The time course of HemK112(49W) translation is shown for comparison (black circles). (E) PET endpoints for the 38W (orange bars) and 49W (blue bars) nascent chains of different lengths. Error bars indicate SD (n = 12 replicates).

Upon synthesis of the peptide that contains Trp at position 6 (6W), the PET efficiency changes in a multiphasic way (Fig. 4B). The initial phases are independent of the chain length or mutations in the protein core (4×A). After about 20 s of translation, the PET efficiency for the various constructs starts to deviate. Nascent HemK70 (wild-type and 4×A) or the 4×A variants of HemK112 adopt a high-PET conformation. In contrast, native HemK112 adopts a low-PET state. Comparison of the fluorescence endpoints for different peptide lengths (Fig. 4C) shows that the PET efficiency is low when the nascent chain is short (35 and 42 amino acids), probably because the peptide exit tunnel restricts chain dynamics, thereby inhibiting BOF-Trp interactions. Upon arrival at the end of the exit tunnel (56 to 84 amino acids), the PET efficiency peaks at 13 to 18% and then decreases with increasing peptide length. With the further increase in length and extrusion of the folded domain from the exit tunnel, BOF-Met becomes shielded from fluorescence-quenching interactions with the Trp residue. In contrast, in the intermediate or unfolded state, Trp is solvent-exposed and accessible, consistent with the high PET observed with the 4×A mutants of any length (Fig. 4C). These data demonstrate the existence of a transient compact state that rearranges into the native state when the nascent chain reaches an appropriate length.

To better define the timing of native-state formation, we carried out PET experiments using constructs carrying Trp engineered at positions 38 or 49 of the HemK NTD (Fig. 4, A and D, and fig. S2). With Trp at position 38, a high-PET intermediate emerges as the nascent chain length reaches 98 amino acids, somewhat later than with Trp at position 6 (Fig. 4, C and E). Further translation results in a rearrangement (PET decrease) that leads to the final state. With Trp at position 49, the transient folding state is not observed; rather, PET reports the formation of the final native-like structure (Fig. 4E). In that case, the time course of folding coincides with the synthesis of the full-length HemK112 (Fig. 4D), which should be sufficient to extrude all helices required to fold into the NTD just outside the ribosome exit tunnel, and does not change further with the increase of the chain length up to 154 amino acids.

Assuming that residues 74 to 112 that do not belong to the NTD adopt a fully extended conformation (~3.5 Å per residue) within the ribosome exit tunnel (length ~100 Å), the folded nascent domain would reside ~33 Å away from the ribosome exit tunnel. However, for HemK98, which shows a fold similar to that of nascent HemK112, the same calculation suggests that the native-like fold may assemble even before the domain is fully released from the tunnel, as the distance within the ribosome covered by residues 74 to 98 is only 84 Å—shorter than the tunnel length, but within the area attributed to the tunnel vestibule (21, 22) (Fig. 1A). The difference in the environment or local folding of HemK98 and HemK112 is reported by PET from BOF to Trp38 (Fig. 4E). Thus, the compact state may rearrange to the native-like fold already in the tunnel vestibule, just before the full domain emerges from the exit tunnel. When HemK70 folding was followed after the release of the compacted nascent chains into solution by puromycin treatment, PET changed in exactly the same way as upon protein folding on the ribosome (Fig. 4 and fig. S10). In contrast, release of HemK112 did not lead to any additional PET changes, which suggests that the final conformation of HemK112 on the ribosome and in solution after folding is very similar.

Our results provide an insight into nascent protein folding on the ribosome in real time (fig. S11). When the peptide reaches a length of about 56 to 70 amino acids, the nascent chain becomes compact, at a stage of translation when parts of the nascent peptide are still enclosed in the exit tunnel of the ribosome—possibly in the exit tunnel vestibule, which can accommodate a substantial degree of protein structure (e.g., tertiary hairpins) (7, 22). If such a state formed off the ribosome, it was too short-lived to be captured by stopped-flow experiments. The ribosome may induce an alternative folding pathway, or it may stabilize an arrangement that is hardly sampled in solution.

Retention of compact or intermediate states may represent a fundamental feature of cotranslational folding that acts to prevent the chain from falling into kinetic traps, such as stably misfolded non-native conformations that may appear when only part of a protein has been synthesized. Owing to the slow pace of translation, folding of intrinsically rapidly folding domains appears to exhibit equilibrium-like properties (21) with a landscape of accessible conformations restricted by the environment of the exit tunnel. Our findings show how the ribosome can, in principle, define the pathway for cotranslational folding.

Supplementary Materials

Materials and Methods

Figs. S1 to S11

Table S1

References (2328)

References and Notes

  1. Acknowledgments: We thank W. Wintermeyer for critically reading the manuscript, and O. Geintzer, S. Kappler, T. Wiles, M. Zimmermann, T. Uhlendorf, and A. Bursy for expert technical assistance. Supported by the Max Planck Society, Deutsche Forschungsgemeinschaft grant FOR1805 (M.V.R.), and Human Frontier Science Program grant RGP0024-2010 (M.V.R. and A.A.K.). Data described in this manuscript can be found in the supplementary materials. A.A.K. acknowledges the support of Max Planck Institute of Biophysical Chemistry (Göttingen) during his sabbatical stay in fall 2013.
View Abstract

Stay Connected to Science

Navigate This Article