Translational tuning optimizes nascent protein folding in cells

See allHide authors and affiliations

Science  24 Apr 2015:
Vol. 348, Issue 6233, pp. 444-448
DOI: 10.1126/science.aaa3974

Ribosomes help careful protein folding

Protein assembly in vitro is useful for studying small molecules but is problematic for studying the assembly of larger, more complex proteins. Kim et al. analyzed the biogenesis of the mutation-prone nucleotide-binding domain of the cystic fibrosis conductance regulator (CFTR) (see the Perspective by Puglisi). Newly synthesized polypeptides emerged relatively slowly from the ribosome and folded through a modulated pathway that ensured correct protein folding. Some parts of the protein chain folded immediately upon synthesis, whereas other segments did so more slowly. It appears that acquiring the correct conformation for this complex protein is partly guided by the ribosome itself.

Science, this issue p. 444; see also p. 399


In cells, biosynthetic machinery coordinates protein synthesis and folding to optimize efficiency and minimize off-pathway outcomes. However, it has been difficult to delineate experimentally the mechanisms responsible. Using fluorescence resonance energy transfer, we studied cotranslational folding of the first nucleotide-binding domain from the cystic fibrosis transmembrane conductance regulator. During synthesis, folding occurred discretely via sequential compaction of N-terminal, α-helical, and α/β-core subdomains. Moreover, the timing of these events was critical; premature α-subdomain folding prevented subsequent core formation. This process was facilitated by modulating intrinsic folding propensity in three distinct ways: delaying α-subdomain compaction, facilitating β-strand intercalation, and optimizing translation kinetics via codon usage. Thus, de novo folding is translationally tuned by an integrated cellular response that shapes the cotranslational folding landscape at critical stages of synthesis.

Most proteins must acquire a defined three-dimensional structure in order to function. Folding pathways that generate these structures have primarily been characterized by using model substrates that fold rapidly, spontaneously, and reversibly in vitro (1, 2). In cells, however, protein folding is kinetically coupled to synthesis as the nascent polypeptide emerges from the ribosome. Whereas certain small proteins may remain unstructured during synthesis (3), many complex proteins exhibit length-dependent folding intermediates whose structural properties (4) and/or folding efficiencies (5) deviate from those observed in vitro. In such cases, the folding energy landscape, as well as folding outcome, can be influenced by ribosome effects (4, 6, 7), polypeptide elongation rate (810), molecular crowding (11, 12), and cotranslational interactions with cellular chaperones (13, 14). Indeed, cotranslational constraints can bias kinetically competing folding events to generate alternate stable structures with different functional properties (8, 15, 16). Despite improved computational methods, few principles have been established experimentally to explain how biosynthetic parameters influence specific folding events and outcome (3, 4, 1719).

To address this issue, we used fluorescence resonance energy transfer (FRET) to examine structural transitions of ribosome-bound folding intermediates generated through in vitro translation of truncated RNA transcripts. This approach derives from the principle that during folding, certain residues distant in primary structure are brought into close proximity, increasing the FRET efficiency between donor and acceptor fluorophores that are cotranslationally incorporated into the nascent polypeptide (Fig. 1A) (18, 19). Here, the donor fluorophore, cyan fluorescent protein (CFP), was fused to the N terminus of the first nucleotide-binding domain (NBD1) from the cystic fibrosis transmembrane conductance regulator (CFTR), and a small acceptor dye was incorporated at surface-exposed residues (UAG codons) by using a synthetic suppressor tRNA (figs. S1 and S2). FRET measurements obtained at sequential nascent chain lengths thus provide conformational snapshots into the equilibrium ensemble of stably arrested ribosome-bound nascent chains in the context of their native biosynthetic machinery (Fig. 1A) (3, 1720).

Fig. 1 FRET-detected cotranslational compaction of NBD1 subdomains.

(A) Schematic of ribosome-bound nascent chain at sequential stages of synthesis, showing change in distance (d) between donor and acceptor probes as polypeptide transitions from unfolded (middle) to folded (right) conformation. (B) Location of acceptor probes (Arg450, Arg487, Asp567, and Asn597) in NBD1 primary structure. Helices and β-strands are drawn as cylinders and filled arrows, respectively. (C) Acceptor probe sites in NBD1 crystal structure [Protein Data Bank (PDB) 2BBO]. (D) FRET efficiency (EFRET) plotted as a function of nascent chain length for indicated acceptor probe (n ≥ 3 ± SEM or average of n = 2 replicates). Dotted line indicates synthetic stage needed to achieve maximal FRET. Shown on right is the minimal polypeptide outside the ribosome required to optimally position S3 (Arg450), S6 (Arg487), S7 (Asp567), and S8 (Asn597).

Using this system, we defined the cotranslational folding pathway of CFTR NBD1, whose defective folding causes cystic fibrosis (2124). NBD1 contains three subdomains (N-terminal, α-helical, and parallel-F1-type-β-sheet core) and exhibits a complex vectoral topography that limits CFTR maturation (22, 25) and prevents reversible folding in vitro. To examine its cotranslational folding pathway, FRET acceptor sites were chosen within 4 to 9 Å of the CFP fusion (Thr389) (26) to report on the positioning of strands S3, S6, S7, and S8 in the β-sheet core (Fig. 1, B and C). Analysis of sequentially stalled polypeptides yielded a characteristic length-dependent rise and plateau in FRET for each acceptor site (Fig. 1D). This rise in FRET reports on acquisition of a native-like fold (19) and reflects the earliest biosynthetic stage at which the acceptor dye and its corresponding β-strand are optimally positioned within NBD1. Results show that S3, S6, S7, and S8 could therefore reach a native-like structure when the ribosome has synthesized residues 550, 624, 654, and 674, respectively (Fig. 1D), although actual folding intermediates will depend on relative folding kinetics and translation elongation rate.

Despite their proximity, S6 exhibited a more gradual rise in FRET and was optimally positioned at a later stage of synthesis than was S3 (Fig. 1D). Because the ribosome exit tunnel sequesters approximately 40 residues, optimal S6 positioning was therefore delayed until the entire α-subdomain and S7 emerged into the cytosol (Fig. 1D). To understand this delay, we compared the conformation of S3 and S6 in ribosome-bound versus ribosome-released polypeptides, where each construct contained equivalent cytosolically exposed residues (Fig. 2) (19, 27). Under these conditions, ribosome attachment had no detectable impact on N-terminal subdomain folding (Fig. 2, A and B, and fig. S3A). In contrast, S6 exhibited a higher FRET efficiency and achieved a native-like conformation at shorter chain lengths (by nearly 50 residues) in the ribosome-free state (Fig. 2, C and D, and fig. S3B). Ribosome attachment therefore delayed S6 positioning during α-subdomain synthesis.

Fig. 2 Ribosome delays α-subdomain folding and facilitates β-sheet core formation.

(A, C, and E) Illustration showing hypothetical folding outcomes of ribosome-bound and released polypeptides with equivalent cytosolically exposed residues. (B, D, and F) EFRET obtained for ribosome released (black solid line) and ribosome-bound (red dotted line from Fig. 1D) polypeptides plotted against number of cytosolic residues (#) C-terminal to the acceptor site (n ≥ 3 ± SEM or average of n = 2 replicates) located at residues 450, 487, 567, in (B), (D), and (F), respectively. (G) Illustration depicting (i) role of ribosome in delaying S6 and α-subdomain folding to facilitate S7 and S8 insertion and (ii) resultant NBD1 misfolding due to premature S6/α-subdomain folding in the absence of ribosome.

In adenosine 5′-triphosphate–binding cassette transporters, S7 and S8 are synthesized after the α-subdomain and must intercalate between S3 and S6 to form the four-strand hydrophobic β-sheet core (Fig. 1, B and C). Yet, S6, S7, and S8 are all positioned into the native NBD fold at a similar stage of synthesis (Fig. 1D). Because optimal S7 positioning requires S8 and is a prerequisite for CFTR folding (fig. S4), we tested whether ribosome–mediated delay of S6 positioning might facilitate β-sheet core formation. Indeed, ribosome release before core folding prevented S7 insertion, as evidenced by a reduction in FRET efficiency (Fig. 2, E and F, and fig. S3, C and D). The ribosome therefore exhibits two opposite but interrelated effects on nascent chain conformation that directly influence folding outcome: delaying spontaneous folding at one stage of synthesis (S6 positioning), while enabling folding at another (S7/S8 intercalation). Because premature polypeptide release uncouples these events, the ribosome appears to facilitate NBD1 folding, potentially in concert with associated chaperones, by maintaining a relatively open α-subdomain conformation to ensure orderly and sequential insertion of S6, S7, and S8 during a critical translational window (Fig. 2G).

These results predict that β-sheet core formation depends on the timing and/or kinetics of α-subdomain folding. Because the predicted translation elongation rate (10) slows dramatically during α-subdomain synthesis (Fig. 3A and table S1), we introduced synonymous codon substitutions predicted to either maximize or minimize the translation rate within this region (Fig. 3A and fig. S5). Metabolic pulse labeling verified that Fast(525-593) synonymous codons did indeed increase the rate of NBD1 synthesis in cells (Fig. 3B and fig. S6). Remarkably, despite generating identical amino acid sequences (fig. S7 and table S2), Fast(525-593) codon substitutions resulted in aggregation of full-length CFTR (Band B) (Fig. 3C) as well as isolated NBD1 (fig. S8) 1 to 4 hours after synthesis, suggesting either delayed conversion into the insoluble fraction or less efficient degradation (Fig. 3, D and E). In contrast, Slow(525-593) codon substitutions had no detectable effect on either NBD1 or full-length wild-type (WT) or ΔF508 CFTR (Fig. 3, B to D, and figs. S8 and S9). Similarly, Fast substitutions within codons 501 to 540 had no effect on CFTR aggregation (fig. S10).

Fig. 3 CFTR folding dependence on NBD1 translation elongation rate.

(A) Predicted translation elongation rate (10) calculated as a 15–amino acid moving window average, for WT NBD1 (black), Slow(525-593) (red), and Fast(525-593) (blue) aligned with NBD1 secondary structural elements as they emerge from the ribosome. (B) Autoradiogram of [35S]Met-labeled WT, Slow(525-593), and Fast(525-593) NBD1. Graph shows fold change in protein compared with that of WT at 5 min (n = 3 replicates ± SEM). (C) CFTR immunoblot from human embryonic kidney (HEK) 293T cell lysate showing core glycosylated (band B) and mature (band C) CFTR from total, radioimmunoprecipitation assay (RIPA)–soluble, and insoluble lysate fractions. Graph shows fold increase over that of WT (n ≥ 5 replicates ± SEM). (D) [35S]Met-labeled CFTR immunoprecipitated from RIPA soluble and insoluble fractions. Graphs show band C (left) and band B (right) CFTR as percentage of total protein recovered at time (T) = 0 (n = 3 replicates ± SEM). (E) Schematic showing proposed effect of Fast(525-593) substitutions on CFTR processing (blue arrows).

Thus, the translation elongation rate appears to be tailored to folding needs of specific peptide regions and, in this case, kinetically couples α-subdomain and β-sheet core folding as H3 and H4 emerge from the ribosome (Fig. 3A). This region is home to numerous suppressor mutations that improve NBD1 stability, CFTR folding efficiency, and ΔF508 CFTR processing (24, 28). One such variant, D529F, improves NBD1 and full-length CFTR folding without affecting NBD1 thermal stability (25), suggesting that it may act along the folding pathway. (Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. In the mutants, other amino acids were substituted at certain locations; for example, D529F indicates that aspartic acid at position 529 was replaced by phenylalanine.) Asp529 makes several polar interactions with helix H5 (Gln552 and Arg555) (Fig. 4A) that are adjacent to the hydrophobic core of the α-subdomain (Ile502, Cys524, Leu526, Leu541, and Ile556). Phe substitution at this site might therefore affect timing of local α-subdomain collapse. Consistent with this hypothesis, D529F resulted in S7 positioning at an earlier stage of synthesis without affecting S6 (Fig. 4B and fig. S11A). D529F also eliminated ribosome dependence for β-sheet core folding (Fig. 4C) and suppressed aggregation of full-length CFTR caused by Fast(553-593) codon substitutions (Fig. 4D), thus restoring kinetic coupling between α-helical subdomain and β-sheet core.

Fig. 4 Codon usage is integrated with sequence specific folding constraints.

(A) NBD1 structure (PDB 1XMI) showing location of Asp529 (left) and predicted interactions with Gln552 and Arg555 in helix H5 (right). (B) EFRET profile for D529F and WT (from Fig. 1D) NBD1. (C) Ribosome release assay of D529F CFP-NBD1-D567UAG truncated at residue 614. (D) Immunoblot of CFTR constructs (left) and fold increase over WT (right) (n ≥ 4 replicates ± SEM) as in Fig. 3C. (E) Illustration showing that NBD1 cotranslational folding is both delayed (upward arrows) and enhanced (downward arrows) by cellular biosynthetic machinery to optimize folding outcome.

This study delineates cotranslational folding of a topologically complex protein as a series of dynamically modulated folding events that can be viewed as a function of chain length (Fig. 4E). As the nascent polypeptide emerges from the ribosome, formation of low FRET-associated open conformers is interrupted by discrete intervals of peptide compaction. Although each of these folding events could be theoretically analogous to in vitro folding of an equivalent peptide domain (3), this does not appear to be the case. Rather, folding occurs in sequential, coupled steps, the timing of which is both positively and negatively influenced by biosynthetic machinery. Rapid cotranslational folding of the N-terminal subdomain likely provides a template for subsequent interfacial interactions that assist domain assembly (29). In contrast, optimal folding of the noncontiguously synthesized β-strands is achieved by actively delaying placement of a presynthesized N-terminal strand (S6) until C-terminal strands (S7 and S8) are available. This process is coordinated by maintaining the nascent polypeptide in a folding-competent conformation (4, 27) both by direct ribosome effects (4, 7) and the translation rate as dictated by codon usage (11, 12). Cotranslationally recruited chaperones (30), not examined here, may also contribute to the delay in S6 placement and α-subdomain collapse (13). We refer to this overall process as “translational tuning,” in which multiple mechanisms are simultaneously integrated during synthesis to modulate intrinsic folding properties of the nascent chain. Translational tuning also integrates conserved codon usage with biophysical properties imposed by amino acid sequence, both of which are tailored to optimize outcome based on specific folding requirements (31, 32).

Supplementary Materials

Material and Methods

Figs. S1 to S11

Table S1 and S2

References (3344)

References and Notes

  1. Acknowledgments: We thank L. David, B. Conti, and members of the Skach laboratory for helpful discussions; CFTR Folding Consortium for providing 3G11 CFTR antibody. This research was funded by NIH grants R01GM53457 and R01DK51818 (to W.R.S.) and Cystic Fibrosis Foundation Therapeutics grants KIM10F0 (to S.J.K.) and SKACH05X0 (to W.R.S.). Mass spectrometric analysis was performed by the OHSU Proteomics Shared Resource with support from P30EYE010572, P30CA069533, S10OD012246, and S10RR025571. U.S. patent application 13/664,252 covers the methodology to assess and redesign the NBD1 coding sequences, claiming priority to provisional application 61/553,861, filed on 31 October 2011. No other counterpart applications have been filed.

Stay Connected to Science


Navigate This Article