Frameshifting dynamics

Abstract

Translation of messenger RNA by a ribosome occurs three nucleotides at a time from start signal to stop. However, a frameshift means that some nucleotides are read twice or some are skipped, and the following sequence of amino acids is completely different from the sequence in the original frame. In some messenger RNAs, including viral RNAs, frameshifting is programmed with RNA signals to produce specific ratios of proteins vital to the replication of the organism. The mechanisms that cause frameshifting have been studied for many years, but there are no definitive conclusions. We review ribosome structure and dynamics in relation to frameshifting dynamics provided by classical ensemble studies, and by new single-molecule methods using optical tweezers and FRET.

Keywords: FRET; fluorescence; optical tweezers; programmed frameshifting; ribosome structure; single molecule; translation.

Figure 1

Overall architecture of the 70S bacterial ribosome, in its classical, unrotated state (PDB: 3I8G,36 3I8F,36 2WRJ37) (A) The 70S bacterial ribosome is made of two subunits, 30S (blue, also in B) and 50S (brown, also in C). The two subunits stack together with their A- (aminoacyl), P- (peptidyl), and E- (exit) sites facing inward toward each other, thus defining an intersubunit interface where cycles of translation occur. Ribosomal proteins are shown as ribbons; ribosomal RNAs are lighter color, and the mRNA is space filling in red. The tRNAs are violet for A-, green for P-, and yellow for E-tRNA; the same color-coding for these classical state tRNAs is used in all figures. Ribosomal proteins L1 and L9 are identified on the left side of the 50S subunit (L designates proteins from the 50S subunit, S from the 30S). The decoded mRNA lies between the 30S head and body domains, wrapped in an inverted U-shape around the 30S neck (shown in panel B). (B) 30S, view from the mRNA entrance (3'-end shown) with the 50S removed. A trench-like region on the 30S is its neck, which serves as a saddle to display the decoded mRNA (red) codons. The appropriate interactions between the anticodon stem-loop of the tRNAs and the decoded codons on the mRNA are probed by highly conserved resides (detailed in Figure 4), which are adjacent to the neck and come both sides from the 30S head and body. The L-shape tRNAs have their aminoacyl-acceptor ends oriented toward the upper-right corner, revealing how their corresponding binding-sites on the 50S are configured parallel in space. (C) 50S, view of its tRNA-binding/intersubunit interface. The codon-anticodon duplex (red) is out from the paper plane toward us, whereas the aminoacyl-acceptor ends of the tRNAs are bound tightly within each of their designated cavities, clustering around the center of the 50S. Note that the A- and PtRNA acceptorends are positioned relatively close to each other and to the nearby peptidyl-transferase center (PTC). This is where elongations of the polypeptide chain take place.

Figure 2

A bent A/T state adopted by an aa-tRNA during its accommodation (PDB: 2XQD, 2XQE)104 The first step in each translation elongation cycle is the binding of the correct aminoacyltRNA (aa-tRNA, space-filling in purple), specified by the mRNA (space-filling in red) codon displayed in the 30S A-decoding site. The delivery of aa-tRNAs comes in the form of ternary complexes (TC: EF-Tu•GTP•aatRNA, in cyan, pink, and purple respectively), with the elongation factor protein EF-Tu (ribbons in cyan) bound at the amino-acid end of the aa-tRNA, thus preventing deacylation. The use of a non-hydrolysable GTP analog, GDPCP (spheres/sticks in bright pink) allowed the A/T state of an A-tRNA during accommodation to be determined by X-ray diffraction. A zoom-in view on the right superimposes the A/T state aa-tRNA (thin helical cartoon in purple) to its final fully-bound A/A state conformation (space-filling in violet). While the tRNA anticodon stem loop has mostly configured within the 30S A-decoding site, upon GTP hydrolysis and EF-Tu release, its amino acid acceptor end will dock into the 50S A-site pocket. Although the aa-tRNA undergoes significant rearrangement from its A/T to A/A state, the 70S ribosome remains in the classical and unrotated state, evidenced by the P/P (space-filling in green) and E/E (space-filling in yellow) t-RNAs bound within the same ribosomal complex.

Figure 3

Three sequence signals that are found in programmed frameshifting mRNAs from bacteria. The internal Shine-Dalgarno and the hairpin position the slippery sequence at the 30S codon•anticodon binding sites where the mRNA•tRNA base-pairing occurs. The example sequences and the spacing shown are from the dnaX gene in E. coli.

Figure 4

The tightly structured codon•anticodon binding sites in a classical state 70S ribosome (PDB: 3I8G, 3I8F)36 (A) As the mRNA template (cartoon in gray) wraps around the 30S neck (not shown) into an inverted Ushape, the three decoded codons are distinctively kinked into segments specifically oriented toward their corresponding base-pairing anticodons (nucleotide residue 34–36 highlighted in red) from the tRNAs. A Shine-Dalgarno sequence located ~9-nt upstream from the P-site codon is also highlighted in red. (B) The kinked conformation of the mRNA and the spacing between the three base-pairing tRNAs reveals how the translation reading frame is kept in registry. Highly conserved nucleotide residues from the 16S rRNA (shown in sticks and color-coded for their corresponding interacting codon•anticodon binding sites) help to stabilize the interactions around the codon•anticodon duplex, as shown in C. (C) Close-up views at each partitioning interface along the three codon•anticodon binding sites. From left to right: 5'-end of E-site codon, between P-E sites (mRNA 5'-end facing us), between A-P sites (mRNA 3'-end facing us), and 3'-end of A-site codon. It is proposed that the 45°-kink between A- and P-site codons is stabilized by the nearby A1400 and A1401 (middle right figure). Also, shown in middle left, G1338 to U1341 and A790 collectively form a block between the P- and E-sites, with a gap of ~14Å at the narrowest. This effectively prevents movement or slippage of the P-tRNA, particularly by its anticodon stemloop portion. Another closing “latch” securing the A-tRNA is between G530 and C1054. Many of these residues have to move apart in order to permit mRNA•tRNA translocation. This is a static view of the

Figure 5

Ribosome rotation dynamics (PDB: 3I8G,36 classical; 4GD1,10 rotated complex 1; 3SFS,35 rotated complex 2) Depending on the protein factors and tRNAs engaged, the ribosome has been captured in different rotation states. For instance, rotated complex 1 is crystallized with a P/E-hybrid tRNA (thin helical in orange) and recycling factor protein, RRF (not shown), while complex 2 is bound with a release factor protein 3, RF3 (not shown). (A) Overlays of 30S head domain (shown as 16S rRNA residue 930-1390): rotated state (thin helical in cyan for complex 1; dark blue for complex 2) versus classical, unrotated state (space-filling in light gray; classical state tRNAs shown in spacing-filling for comparison). The 30S head can rotate forward (counterclockwise) as large as ~15° (complex 2, right), while a smaller 4°-rotation in the same direction (complex 1, left) is enough to accompany the P-tRNA transition into its P/E-hybrid state (thin helical in orange). The fact that the P-tRNA acceptor end can swing by ~37° and dock into the 50S E-site indicates that the P-E gating residues, e.g. A790 and A1339, have moved apart upon 30S head forward rotation (detailed in panel C).

Figure 6

The springy Shine-Dalgarno•anti Shine-Dalgarno mini-helix (PDB: 2HGR,54 2QNH,52 6-nt; 3I8G,36 9-nt; 2HGP,54 11-nt) (A) Left-most: Around the mRNA 5’-end exit pore, a ~6-bp long SD•antiSD mini-helix (thin helical cartoon in magenta) resides within a pocket, surrounded by ribosomal proteins S11 and S18 (semitransparent space filling, with ribbon; in gray and magenta, respectively) from the 30S. Left two figures: The mini-helix (in magenta and purple) readily forms while the ribosome translates only 6-nt downstream from it (P-site codon colored in green, A-site codon in violet). Under such short spacing, the backbone of E-site codon (highlighted in yellow) is stretched and distinctively distorted from proper basepairing toward the E-tRNA anticodon (residue 34–36 highlighted in red, the rest in yellow); meanwhile, the center of the mini-helix roughly sits on top of the highly conserved residue, U723 (semitransparent spacefilling, with sticks; same color as the mini-helix), though the detailed orientation of the helix may vary slightly. Right two figures: As the ribosome translates forward (seen as the mini-helix sliding away from the residue U723), the nucleotide spacing increases (9-nt, middle right; 11-nt, right-most), thus relaxing the E-site codon (yellow) to resume hydrogen bonding with the anticodon (red). This illustrates how the formation of a SD•antiSD mini-helix, as well as its spacing between the codon•anticodon binding sites, can tune the availability of E-site condon•anticodon basepairing, which directly interferes with the mRNA registry and reading frame maintenance during translation. Right-most: The mRNA entry tunnel is composed of ribosomal proteins S3, S4, and S5 (semi-transparent space-filling, with ribbon; in cyan, dark pink, and slate, respectively). Counting from the first base in the Psite codon, nucleotide #13–15 would be located at the mRNA entry pore. This suggests a minimal spacing of 7 to 9-nt between a slippery sequence placed at the P- and A-site and a downstream hairpin at the entry pore. (B) Overlays of the decoded mRNA: various stretched forms of the E-site codon (at spacing of 6-nt and 9- nt) versus the relaxed form (11-nt).

Figure 7

Real-time step-by-step translation in an optical-tweezers experiment (A) Schematic drawing of the hairpin helicase assay used to follow single-ribosome translation. A selfcomplementary mRNA template (black line), which forms a hairpin, is held by its two ends. The tethering is done by attaching the RNA ends hybridized with either biotin- or digoxiginin- labeled DNA oligos, to either streptavidin- or anti-digoxigin antibody- coated polystyrene beads. While the bottom bead is fixed by suction onto the micropipette tip, the other is controlled by the focused laser trap. This allows us to adjust the force applied to the mRNA, thus tuning the mechanical stability of the hairpin that the ribosome will translate and unwind in 3-bp per codon-step. Note that no modifications on the ribosome, or other translation-essential components, are made. At the end of translation, the residual mRNA hairpin size provides a crosscheck confirming which codon the ribosome has reached, e.g. either the in- or out-of-frame stop codon on a frameshift-promoting mRNA. (B) Example of a step-pause-step translation trajectory from a single ribosome recorded as a function of time. This is captured as a series of fixed-size step-wise mRNA lengthening, i.e. extension, over time (top panel, ~2.7 nm per codon translated at ~20 pN). Here, a total of ten consecutive translation elongation cycles (arrow-highlighted) has been successfully followed in real-time. Each increment (vertical step) signifies the EF-G•GTP catalyzed ribosome translocation. If a frameshift occurs in parallel with the translocation step, a different extension step size will be observed. (Panel B is reproduced with permission from Figure 2A in reference .)

Figure 8

Mass spectrometry analysis to identify mis-incorporation during recombinant protein overexpression in mammalian cells In addition to programmed frameshifting mRNA signals, it is known that hungry or rare codons can also induce frameshifting and miscoding with appreciable efficiencies.^,, A nutritional stress was experienced when Chinese hamster ovary cells (CHO) were transfected to over-express a humanized monoclonal antibody for clinical use. This results in unexpected yet specific mis-incorporations of serine (Ser, S) for asparagine (Asn, N) encoded within the antibody, at a rate of 1–2 % per Asn residue. (A) Deconvoluted mass spectra of the synthesized antibody light chain, examined through LC/MS intact peptide detection. In low-expressing cell lines, only the predicted mass for the antibody light chain (top panel, ~23909 Da) was observed. Upon over-expression, a lighter, “-27 Da,” species (bottom panel; peak b, 23882 Da) had emerged unexpectedly and was persistently seen, suggesting one of four possible amino acid substitutions was made (Arg→Glu, Gln/Lys→Thr, or Asn→Ser) during protein overexpression. (The 23930- Da species was not assigned in this study.) (B) Tandem MS polypeptide sequencing to resolve the identity and position of the unexpected amino acid substitution. The two species detected (bottom panel A, peak a and b) were selected (based on their molecular mass and LC elution time), fragmented, and further subject to a second run of MS. As the fragment ions are generated by breaking amide bonds along the polypeptide backbone (corner insets), thus producing C-terminal bearing y-ions and N-terminal bearing b-ions of various residue length, the identity of an amino acid at each position can be determined by the difference in m/z values of two adjacent y- or bions. The peptide sequence for the expected antibody light chain is assigned (top panel inset) and confirmed to bear two Asn residues as predicted (@m/z = 114.04 for N at both b3-b2 and y12-y11). For the unexpected lighter species (bottom panel), its “-27 Da” lower mass was revealed as an Asn → Ser substitution at one of the two Asn positions (@m/z = 87.03 for S at b3-b2). Ser substitution at the other Asn position (@m/z = 87.03 for S at y12-y11) also exists and was resolved (eluted differently in LC) and assigned. (Panels are obtained with permission from Figure 1A, C and Figure 4 in reference .)

Figure 9

smFRET observations on tRNA dynamics during aa-tRNA selection and accommodation into Asite. ^, (A) Schematic drawings of proposed molecular basis for the observed FRET fluctuations during initial selection and proofreading. The 50S and 30S subunits are depicted in blue and purple rectangles, respectively, with mRNA in a black line, and tRNAs in brown curves. Green and purple round squares denote EF-Tu•GTP and EF-Tu•GDP+Phosphate, respectively. Green and pink stars are donor and acceptor fluorophores, respectively. Stepwise and fluctuating FRET evolved upon delivery of TC(Phe- (Cy5)tRNAPhe•EFTu• GTP) to a ribosomal complex bearing fMet-(Cy3)tRNAfMet in the P-site. Contour plots generated by superimposing smFRET trajectories, post-synchronized to the first observation of FRET above a noise threshold, and FRET histograms with (B) cognate (C) near-cognate ribosomal complexes. FRET histograms were fitted with three Gaussian functions. Cognate complexes fluctuated between three FRET states assigned as codon recognition (CR), GTPase activated (GA), accommodated (AC) states. The assignments were confirmed by control experiments using a non-hydrolyzable GTP analogue (GDPNP), and the antibiotic kirromycin, which traps ternary complex (TC) right after GTP hydrolysis. Figure A is reproduced with permission from Figure 6C in reference ; Figure B and C are modified with permission from Figure 3A in reference .

Figure 10

Single molecule studies on dynamics of pretranslocation (PRE) complexes bearing deacylated PtRNA and peptidyl/analogue of petidyl A-tRNA. (A) A x-ray crystal structure of a hybrid state ribosomal complex (PDB: 4GD1, 3R8S). Comparing to the classical state (Figure 1), the two subunits have rotated relative to each other; the hybrid P/E-tRNA (orange) interacts closely with the L1 stalk (magenta), which has moved inward toward the 50S E-site. For illustration purposes, an A/P-tRNA mimic (cyan) is manually built in to better depict the doubly-occupied hybrid state ribosome. The structural and positioning reference for such reconstruction is based on a crystallized A-site anticodon stem-loop (17-nucleotide long) inside a rotated state ribosomal complex (PDB: 3I1Z). Single molecule FRET studies investigated the dynamics of the conformational changes as indicated in the arrows. (B) Intersubunit rotation dynamics were probed with (Cy3)L9 in the 50S subunit and (Cy5)S6 in the 30S subunit. Pretranslocation (PRE) complex contained tRNAfMet in the P site and N-Ac-Phe-tRNAPhe (Ac) in the A site. Fluorescence time traces displayed high (0.58) and low (0.4) FRET states, assigned as unrotated and rotated states, respectively. (C) L1 stalk interactions with P/E-tRNA were studied with a FRET pair on L1 protein in the 50S subunit and tRNAPhe in the P site. Ribosomal complex with (Cy5)L1 enzymatically formed a posttranslocation (POST) complex bearing fMet-Phe-(Cy3)tRNAPhe in the P site. FRET between L1 and tRNAPhe presented steady low (0.21) FRET state. Real time stopped-flow delivery of a cognate tRNA in the form of TC (Lys-tRNALys •EFTu• GTP) in the absence of EF-G to the POST complex triggered fluctuating FRET states between low (0.21) and high (0.84), presumably upon peptidyl-transfer reaction forming pretranslocation (PRE) complex (left fluorescence time traces). Low and high FRET states were assigned as an L1 stalk open state with classical P/P tRNA and an L1 stalk closed state interacting with hybrid P/E tRNA, respectively. Co-delivery of LystRNALys• EF-Tu•GTP and EF-G•GTP evolved a steady high (0.84) FRET state, suggesting that EF-G stabilizes the L1 stalk closed state and drives translocation forward from the state (see right fluorescence time traces). Fluorescence time traces in Figure B is modified with permission from Figure 2A in reference ; Fluorescence time traces in Figure C is modified with permission from Figure 5 in reference .

Figure 11

Real-time observation of translation using Zero-mode waveguides (ZMW). (A) Ribosomal initiation complex bearing fMet-(Cy3)tRNAfMet in the P-site was immobilized in a ZMW well, cylindrical aluminum nano-well (50–200 nm in diameter), enabling imaging in the presence of high concentrations of fluorophore-labeled molecules. (B) A representative trace of fluorescence intensities from a ribosomal complex translating six repeating Phe and Lys codons after initiation codon: M(FK)6. Ternary complexes (TC) with fluorescently labeled tRNAs, 200 nM Phe-(Cy5)tRNAPhe and 200 nM Lys- (Cy2)tRNALys, in the presence of 500 nM EF-G•GTP were delivered to the ribosomal initiation complex. Simultaneous illumination and detection of multi-color fluorophores Cy2 (blue), Cy3 (green), and Cy5 (red) allowed them to image tRNAs binding and release from the ribosomal complex during active translation elongation, thereby following translation codon by codon. Figure A and B are modified with permission from Figure 1 and 3A,