Variant evolution reveals hinge sequence flexibility

To investigate the conformational malleability of mRNA-[tRNA]₂ translocation specifically, and of protein synthesis in general, we sought to evolve head domain hinge variants. Recent computational and structural studies [26] localized the axis of head swivel between helix h28, the single covalent connection between the head and body domains, and the coaxially stacked helices h35 and h36, which form a non-covalent connection between these domains (Fig 2A and 2B). The two hinge points of the head domain about this swivel axis were localized inside h28 (hinge 1) and in the single-stranded linker connecting h34 and h35 (hinge 2). With these hinges identified as critical control elements of translocation, selection of alternative hinge structures that permitted protein synthesis was performed. Two separate libraries were constructed, consisting of degenerate mutations in the general region of each head domain hinge, spanning 16S rRNA positions 1385–1394 (Hinge 1 library) and 1063–1070 (Hinge 2 library) (Fig 2A).

Iterative selection of these libraries was carried out in vitro to bypass the constraints of cell viability, strain competition and library transformation, focusing solely on the synthetic capacity of the ribosome. A cell-free ribosome synthesis and evolution method called RISE (Ribosome Synthesis and Evolution) [28], was employed to perform this selection (Fig 2C). RISE combines an in vitro integrated synthesis assembly and translation (iSAT) system [30] with ribosome display. In this method, a library of rRNA sequence variants (encoded into a pT7rrnB plasmid template) is transcribed and assembled into a ribosome library in (ribosome-free) S150 extract supplemented with total ribosome proteins (TP70). Following translation of a selective peptide (3xFLAG) by functional ribosomes, the ribosome stalls on a stop-less truncated mRNA, and the stalled mRNA-ribosome-peptide complexes are captured using a magnetic resin. The selected rRNA is then purified, reverse transcribed, and either sequenced or assembled into a new library for further RISE cycles.

Previous in vitro ribosome synthesis and evolution approaches have carried out mutation mapping of the 70S ribosome [31], enabled assessment of computationally designed ribosomes [32], and evolved the ribosome for new function [33]. Here, we applied a modified version of RISE to the two constructed hinge libraries, performing 3 rounds of selection for each (see Methods, Ribosome synthesis and evolution). Characterized by iSAT sfGFP translation, most of the libraries activity increase was achieved after a single round of RISE (S1 Table), with minimal improvements from subsequent rounds. After completing the in vitro selection, eight of the top genotypes for each library (ranked by read enrichment after selection relative to the original library, see Methods, Deep sequencing of RISE cDNA, S2 Table) were individually cloned and tested in iSAT by translation of sfGFP (Fig 2D). Surprisingly, most of the variants displayed a level of activity close to wild type (> 70%), with no observed correspondence with enrichment rank. Notably, head domain motions have several known roles outside translocation, such as ribosome biogenesis [34], translation initiation [35] and tRNA selection [36]. In this sense, iSAT production and measurements are indicative of the overall efficiency of ribosome synthesis and reporter translation, not solely of translocation. RISE therefore enriches for variants capable of ribosome biogenesis and selection tag translation, rather than for mutants that are especially adept at performing translocation. Carrying out ribosome synthesis prior to reporter translation (in vitro translation or iT, see Methods, iSAT and iT reactions) generated efficiencies more uniformly comparable to the wild type ribosomes (Fig 2E). This suggests that elongation processes (including translocation) have a similar in vitro efficiency to wild type ribosomes across the top hinge mutants, and most of the variation observed through iSAT (as in Fig 2D) is likely to stem from ribosome assembly.

To assess the viability of selected variants in vivo, the top genotypes were cloned in the ampicillin-resistant pAM552G backbone, containing the rrnB operon under the temperature-sensitive phage lambda (pL) promoter. The constructs were then transformed into the Squires strain [37] SQ171fg, lacking chromosomal rDNA and carrying a pCSacB plasmid with the RiboT-v2 sequence [38]. The sucrose sensitivity of pCSacB allows the native tethered ribosome to be replaced by the specific pAM552G ribosome mutant plasmid, and total RNA was isolated from the strains and amplified to confirm the presence of the expected mutations in the 16S rRNA (see Methods, Plasmid replacement and selection in SQ171fg cells). Doubling rates were derived from the transformed variant strains and compared to wild type (Fig 2F). The mutant cells had varied growth, as demonstrated by doubling rates ranging from ~50% to 130% of wild type. As with in vitro activity, no relationship between enrichment rank and doubling rate was apparent. A classification of variant activities in vivo (S3 Table) suggests that most mutant deficiencies stem from ribosome biogenesis rather than translation activity. The viability of most of these strains (15 of the 16 variants have growth rate greater than 70% of the wild type) indicates the capacity of the alternate head domain hinges to carry out protein synthesis in general, and translocation specifically, in the multiple contexts necessary to produce a working bacterial proteome. A subset of the variants had growth rates comparable to (H1_2, H1_4, H2_1, H2_6, H2_8) or larger (H1_5, H2_10) than wild type. This is consistent with the bases involved in these mutations not being conserved across bacteria and archaea [27].

Molecular dynamics simulations of forward and reverse head swivel

To explore the mechanics of translocation for the hinge variant ribosomes, we performed all non-hydrogen atom structure-based simulations of EF-G-bound ribosomes. Models for the various translocation states were constructed using PDB ID: 4V9D (PRE-Classical) [39], PDB ID: 4V7B (INT2) [9], modeled hyper-swiveled state (INT3) [20], and PDB-ID:4V5F (POST) [11] as starting structures. The energy of each model was minimized, and the resulting models were used as either a starting structure or energy minima in translocation simulations. These simulations are performed on a minimally frustrated landscape where the barriers are defined by the ribosome structure, describing the steric and entropic limits to translocation motions. Forward and reverse head-swivel motions of the ribosome were simulated as previously described [20,23]. Notably, these simulations do not cover the entire translocation stage, as a repetitive exchange between INT2 and INT3 prior to substrate release is an important phase of the process [15,16]. Forward swivel structure-based simulations were initiated from the PRE conformation of the ribosome with tRNA in the A/P and P/E positions. Stabilizing contacts from the INT3 ribosome conformation were included to define INT3 as an energetic minimum in the simulation. In these structure-based simulations only the endpoint (INT3) is defined as the energetic minima; however, the contact weights were rescaled (Methods, All atom structure-based simulations) to enable reversible fluctuations of the head domain during progression towards the end state. Using this approach, the ribosome head and tRNA were allowed to sample different configurational space until contacts of the INT3 conformation were adopted. Similarly, reverse swivel simulations, capturing structural dynamics required for the head domain to return to its classical position (POST state) were performed. Here, simulations were initiated in the INT3 state with a head-swivel of ~26° and were allowed to sample configurational space until the POST state was achieved and stabilized by setting the corresponding contacts as a dominant energy minimum. Reverse head-swivel simulations were performed in the absence of deacylated tRNA to reflect the potential of dissociation of tRNA from the E-site once the INT3 state has been adopted [20].

Variant core motion preserves hinge points

To investigate if hinge variants introduced structural deviations of the hinges governing head swivel motions, their helical axis was deduced. Specifically, localization of the head domain hinges was performed by measuring position differences in the helical axis of the 16S rRNA between classical and swiveled ribosomes [26]. A series of helices leads from the ribosome body and forms the head domain. On one end, h28 (the ‘neck’) makes a covalent connection from the head to the body domain of the SSU, where it stacks coaxially with h1, h2 and h3. At the other end, h36 forms a non-covalent bridge with h2 via A-minor interactions. The core helices h28-h29-h30-h32-h34-h35-h36 lead from and to the body, forming the head domain (Fig 3A). Based on these helices, a continuous helical axis for the head has been calculated, and the motion across the head due to swivel can be quantified [26] (Methods, Calculation of core helical axis motion). The displacement of the helical axis between the simulated (16° swivel) swiveled structures relative to the classical configuration was calculated for the wild type and top variants. This quantification showed that the wild type hinge points are preserved for all variants, localizing to inflection points near 16S rRNA positions 928 and 1065 (Fig 3B and 3C, individual variant displacements displayed in S1 and S2 Figs). The hinge points were preserved for the reverse swivel motion as well (S3C and S3D Fig.). Intriguingly, we observed a broad distribution of helical axis displacements, deviating over 5 Å in some trajectories relative to wild type ribosomes. These distinctions indicate that although the head-swivel angle is maintained, the helices themselves can be arranged in different configurations to achieve these angles.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Variant helical axis and hinge motions.

(a) Calculated helical axis (red) for the structural core of the 30S head in 16S rRNA. The head (domain III) is shown in cyan, and the body (domains I, II and IV) is shown in green (PDB ID 4V7B, Reference 9). (b) Displacement at each position along the calculated helical axis from the classical PRE state to the simulated conformation at 16° swivel for wild type and top selected hinge 1 variants. (c) Same as (b) for wild type and top selected hinge 2 variants. (d) Hinge 1 motion in the wild type ribosome. Hinge 1 is positioned in a weak point in h28 and its motion results from the kinked helix h28 being straightened during swivel. Cylinders show helical axes in the 16S rRNA core for the classical PRE (tan) and swiveled INT2 (magenta) conformations. (e) Root-mean-square-fluctuations (RMSF) of hinge 1 during forward swivel for wild type and top selected hinge 1 variants. (f) Same as (e) for wild type and top selected hinge 2 variants. Symbol keys are ordered according to their RISE selection enrichment.

https://doi.org/10.1371/journal.pcbi.1012319.g003

Reduced hinge fluctuations allow effective translocation

To quantify the motion of the hinges throughout forward head swivel, the root mean square fluctuation (RMSF) across the hinge regions was measured (Methods, Hinge fluctuations). Interestingly, hinge 1 fluctuations (Fig 3D) were reduced relative to wild type for all hinge variants (including hinge 2 variants) during forward head swivel (Fig 3E and 3F). In contrast to these results, hinge 2 fluctuations during forward swivel and hinge 1 and 2 fluctuations during reverse swivel were not significantly altered in the variants (S3 Fig). This suggests that reduced flexibility across h28 during forward swivel is a selectable feature and that hinge 1 may exhibit more conformational plasticity during forward swivel motions, whereas fluctuations for reverse swivel are more limited. To understand if the change in fluctuations can be reflected in the global motions of hinge 1, a principal component analysis was performed for hinge 1 forward head-swivel motions. Subtle alterations were observed for the first principal component describing the movement of hinge 1 for representative variants (S4A–S4C Fig). Projection of these motions onto the structure of hinge 1 indicates that variants undergo motions parallel to the helical axis of hinge 1 (S4D–S4F Fig). This indicates that the dominant motion of hinge 1 during forward swivel is not altered by its flexibility.

Variant translocation displays flexible head domain trajectories

Following EF-G binding, the head domain is thought to swivel and tilt to translocate the mRNA-[tRNA]₂ substrate [7,15,20,23]. This swivel-tilt coordination controls the opening of the SSU corridor that the translocation substrate moves through [9,23]. To study the global conformational flexibility of translocation in the hinge variants, multiple degrees of freedom of the head domain were considered. Specifically, the trajectory of the head was described by both the head swivel and the orthogonal head tilt (Fig 4A). Using the head swivel and tilt angles, the global trajectories of the head domain relative to the SSU body were described. To compare the variant trajectories to that of wild type, heat maps showing the probability of conformations during forward swivel were derived. Simulations for the variants were carried out for the steps required for the wild type ribosome to reach maximum swivel and final tRNA positioning (INT3). Notably all the variants studied reached this state faster than wild type (Figs 4B, S5 and S6), completing forward swivel (78 ± 5) % faster (S4 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Alternate translocation trajectories of head domain motion.

(a) The main head domain motions during translocation. Solvent view of the 30S subunit (Left) showing head swivel in the direction of translocation. 70S particle (Right) showing the direction of head tilt, orthogonal to head swivel. (b) Relative free energy (ΔG*) landscapes of the forward swivel (PRE to INT3) stage of translocation with head domain swivel and tilt angles as reaction coordinates. Heat maps are obtained by normalizing the number of frames in the targeted MD simulation, where the frame count is divided by the maximum frame count. Shown are the maps for the wild type and seven representative hinge variants active in vitro and in vivo. Maps correspond to ten trajectories per genotype. (c) Interactions between the small subunit and tRNAs for simulated translocation intermediates. Subunit interface view of the simulated wild type and representative variant conformations at 16° of forward swivel showing tRNA contacts in the A (yellow), P (magenta) and E (teal) sites.

https://doi.org/10.1371/journal.pcbi.1012319.g004

The wild type ribosome trajectory of the head domain describes an arc-like tilt throughout the forward swivel (S7 Fig). The tilt angle rises from ~2° at initial (~0°) head swivel to a maximum tilt of ~6° as the head swivel nears ~17°, as observed in previous intermediate structures [6,40] (Fig 4B). Tilt then reverts to ~1° at maximum (~23°) swivel. The state observable at ~4°-6° tilt and ~15°-18° swivel corresponds to the INT2 translocation intermediate [6,7,15,20], displaying a large degree of swivel and tRNA motion and leading to complete substrate repositioning at INT3. Importantly, the selected variants displayed a variety of head domain trajectories. While certain trajectories were closer to the wild type ribosome (Fig 4B, H2_5), some had a substantially reduced head tilt (Fig 4B, H1_4, H2_8) and others showed alternative intermediate conformations (Fig 4B, H1_3, H1_8, H2_2, H2_10).

An important consequence of this variation in the degree of head tilt during forward swivel is the flexible extent of the platform opening and of the small subunit ‘grip’ on the tRNAs. Forward swivel simulation snapshots (at 16° swivel) show the extent of this variability (Fig 4C). As established previously, these structures show that during forward swivel specific head-tRNA contacts are maintained, while body-tRNA contacts are disrupted [6,8,9]. The different tilt profiles, however, result in variation of the proximity and extent of the 30S-tRNA contacts. As small tilt angles (< 4°) can achieve the fully translocated tRNA positions at INT3, this indicates that the forward tilt motions and opening of the platform is not a requisite feature for tRNA translocation.

In contrast to forward swivel, the reverse swivel for the wild type was completed faster, with transit time being (55 ± 10) % that of the variants (S4 Table). As in the forward swivel, the wild type ribosome tilt trajectory for the head described an arc, and the selected mutant trajectories were variable. Unlike forward swivel, all the variant reverse swivels consistently achieve significant (~7°-12°) degrees of tilt (S8 and S9 Figs). Under this framework, reverse head swivel requires more tilt than forward swivel, but large tilt angles are not essential to achieve the POST position.

Variant translocation displays flexible tRNA trajectories

To describe the observed conformational flexibility of translocation from the perspective of substrate motion, the trajectory of the incoming deacyl tRNA into the E-site was monitored. This was done by measuring the distance between the deacyl tRNA and the E loop in the SSU body (h23, G691-A695) (Fig 5A). Using this distance and the extent of head swivel, the tRNA trajectory was described in heat maps for wild type and variants (Figs 5B, S10 and S11). The wild type ribosome motion of deacyl tRNA occurs through two main stages. The first stage displays fast tRNA motion, going from an initial distance to the E loop of ~32 Å (at ~5° swivel) to ~23 Å (at ~10° swivel). This stage takes only (25 ± 3) % of the time required to reach maximum swivel (S4 Table). The second stage consists of slower tRNA motion from ~23 Å (at ~10° swivel) to ~20 Å (at ~25° swivel). This latter stage includes the transition from INT2 (~15° swivel) to INT3. Exchange between these states is thought to consist of an accurate final positioning of the translocation substrate [7,15,20].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Alternate translocation trajectories of tRNA motion.

(a) Deacyl tRNA motion during forward swivel was tracked by its distance to the E loop (h23, G691-A695). (b) Relative free energy (ΔG*) landscapes of the forward swivel (PRE to INT3) stage of translocation with head domain swivel and E loop-Deacyl tRNA distance as reaction coordinates. Heat maps are obtained by normalizing the number of frames in the targeted MD simulation, where the frame count is divided by the maximum frame count. Shown are the maps for the wild type and seven representative hinge variants active in vitro and in vivo. Maps correspond to ten trajectories per genotype.

https://doi.org/10.1371/journal.pcbi.1012319.g005

Interestingly, tRNA motion for the selected mutants varied widely in the initial, fast phase of translocation (Fig 5B). This initial phase displays varying transit of the previously observed, short-lived intermediate INT1 (~30 Å, ~6° swivel) [6,7,15,20], with some clear INT1 dwells (H1_2, H1_3, H2_10, H2_6, H2_5) and some nearly absent dwells (H1_4, H2_8). The following tRNA motion phase (~10° to ~25° swivel), though still faster than wild type for all variants, is uniformly the slower stage of forward swivel (S4 Table). Notably, the dynamics of this phase are considerably uniform across variants.

Intersubunit bridge conformations in variant translocation

Intersubunit bridges maintain the stable association between the large and small ribosome subunits. Of particular importance for translocation are bridges B1a (between protein uS13 and helix H38) and B1b (between uS13 and uL5 and uL31), which connect the head domain with the central protuberance of the large subunit and are disrupted during head swivel [41]. To investigate the variant bridge trajectories, heat maps of the forward and reverse stages of translocation were developed, with head swivel and bridge component separation as coordinates [42] (Methods, Intersubunit bridge displacement measurements) (S12–S19 Figs). Bridge trajectories are mostly conserved during the forward swivel stage of translocation (S12, S13, S16 and S17 Figs). However, the bridge trajectories are more diverse during reverse head swivel, particularly for B1b (S18 and S19 Figs).

Plasmid and library construction

The plasmids pT7rrnB (containing the rDNA operon rrnB under a T7 promoter), pJL1sfGFP (encoding the reporter sfGFP under a T7 promoter, Addgene accession #102634) and pRDV3xFLAG-TolA-HH (encoding the 3xFLAG tag for ribosome selection under a T7 promoter) were used in in vitro integrated synthesis assembly and translation (iSAT) reactions [30,50,51]. The plasmid backbone pAM552G [38,52] (Addgene accession #154131) was used for testing ribosome variants from RISE in vivo (see below). pAM552G carries Ampicillin resistance and a copy of the rrnB operon under the temperature-sensitive phage lambda (pL) promoter. For library creation and cloning, a pT7rrnB variant with a 438 bp deletion in the 16S gene (pT7rrnBΔ438) was created by inverse PCR, including unique cut sites for the off-site type IIs restriction enzyme SapI. Libraries of rDNA were generated by cloning PCR amplified (with SapI sites) library amplicons ordered from Epoch Life Science and ligating them into the SapI-digested pT7rrnBΔ438 variant. The reconstituted library was then transformed into MegaX DH10B T1 R electrocompetent cells (ThermoFisher), cultured and purified for the initial RISE cycle. The top rDNA variants enriched from RISE were constructed on pT7rrnB and pAM552G backbones for in vitro and in vivo testing, respectively. The variants were created from the wild type plasmids using the Q5 Site-Directed Mutagenesis Kit (NEB) and transformed into MegaX cells made chemically competent (Mix & Go E. Coli Transformation Kit, Zymo Research). The top 8 variants of each library selection were constructed, though H2_3 and H2_4 could not be cloned. pAM552G variants were incubated at 30°C to prevent plasmid rRNA expression. Full plasmid nanopore sequencing was carried out to confirm the expected variant genotypes.

iSAT and iT reactions

iSAT reactions were performed by mixing a master mix containing salts, substrates, rDNA plasmids/libraries, reporter/tag plasmids, and cofactors with ribosome-free MRE600 S150 extract and total proteins of 70S ribosomes, TP70 [28]. The final concentrations of the main additions were ~2 mg/mL for the S150 protein content, 100 nM TP70, 1.5 μM T7 RNA polymerase, 2.2 nM (11 ng/μL) rDNA encoding plasmids or 0.36 nM (1.8 ng/μL) rDNA libraries, and 2.6 nM (3.9 ng/μL) pJL1sfGFP, pRDV 3xFLAG. Reactions were mixed and incubated at 37°C for 3 hours (for sfGFP expression) or 90 minutes (for RISE). To monitor protein synthesis alone from pre-assembled ribosomes (in vitro translation or iT), reactions with no reporter plasmid were incubated at 37°C for 1h to mature the ribosomes, followed by addition of the pJL1sfGFP plasmid. sfGFP expression was monitored using a Bio Rad CFX Touch Real-Time PCR Detection System with a 96 well plate reader.

Ribosome synthesis and evolution

RISE selections were performed as in a previous study [28] with some modifications. iSAT reactions of 12 μL were carried out with the different rDNA libraries and pRDV 3xFLAG as a ribosome display tag. The anti-ssRA oligonucleotide (5’-TTA AGC TGC TAA AGC GTA GTT TTC GTC GTT TGC GAC TA−3’) was included at 5 μM to hinder stalled ribosome dissociation. Reactions were incubated for 90 minutes at 37°C. Following incubation, reactions were kept at room temperature throughout the protocol and diluted with 15 volumes (180 μL) of wash buffer with Tween, WBT (50 mM Tris-acetate (pH 7.5 at 4°C), 50 mM magnesium acetate, 150 mM NaCl and 0.05% Tween 20). For each RISE reaction 10 μL packed gel volume of ANTI-FLAG M2 magnetic beads (Sigma-Aldrich) were washed three times with 200 μL wash buffer, WB (50 mM Tris-acetate (pH 7.5 at 4°C), 50 mM magnesium acetate and 150 mM NaCl). The diluted iSAT reactions were added to the washed beads and incubated with gentle rotation for 20 minutes. Reactions were then washed 5 times with WBT buffer, 2 times with WB buffer and resuspended in elution buffer, EB (50 mM Tris-acetate (pH 7.5 at 4°C), 150 mM NaCl and 50 mM EDTA (Ambion)), incubating the suspension with gentle rotation for 30 minutes. The supernatant was recovered from the beads and the rRNA was purified using a Qiagen RNeasy MinElute Cleanup Kit. RT-PCR of this rRNA was done using SuperScript III One-Step RT-PCR System with Platinum Taq DNA Polymerase (Invitrogen) in 100 μL reactions according to product literature. The RT-PCR was performed with primers that amplified the bases 1026–1476 of the 16S rRNA gene and included SapI cut sites. This amplicon was then used for next-generation sequencing or cloned into the pT7rrnBΔ438 variant and amplified for further RISE cycles.

Deep sequencing of RISE cDNA

Libraries with one variable region were sequenced with 250 bp paired-end, non-overlapping Illumina NGS at the University of Minnesota Genomics Center. Since the regions of interest for each of the libraries studied (Hinges 1 and 2) are located close to the ends of the primer regions, no alignment was necessary. Sequences were extracted from the raw reads using Python, searching for up- and downstream sequences of N+2 (e.g. for a variable region of 8 nt, 10 nt up stream and 10 nt downstream were queried). No mismatches were allowed in the search sequences. Sequences containing nucleotides with phred-score of less than 30 (p>0.001) were discarded. Extracted sequences were trimmed and counted for both the original input library and the sequence pool after 3 rounds of RISE selection. Variant abundance was calculated as counts of each sequence relative to all sequence counts of the library. Enrichment was calculated as variant abundance in output library relative to abundance in the original library. Winners were determined via ranking by highest enrichment.

Plasmid replacement and selection in SQ171fg cells

Electrocompetent E. coli SQ171fg cells containing a pCSacB plasmid (granting sucrose sensitivity) with kanamycin resistance (KanR) were prepared. The pCSacB/KanR plasmid carries the sequence for the engineered, subunit-tethered ribosome Ribo-Tv2 [38], serving as the only rRNA operon in the cell. Variant ribosomes of interest in the ampicillin-resistant pAM552G backbone can be transformed into the cells, and the original pCSacB-RiboTv2-KanR plasmid can be replaced by culture with sucrose and carbenicillin. This replacement and cell viability based on the variant ribosomes is confirmed by susceptibility to Kan [32]. 150 to 300 ng of each variant plasmid in the pAM552G backbone was transformed into 50 μL of electrocompetent SQ171fg cells. Cells were recovered in 800 μl of SOC medium in 1.5 ml tubes at 37°C with shaking at 250 rpm for one hour. After recovery, 1.2 ml of SOC with 0.42% sucrose and 100 μg/mL were added to the cells (60 μg/ml carb, 0.25% sucrose final concentrations). These 2 ml cultures were further grown in 50 mL falcon tubes at 37°C overnight. The overnight cultures were pelleted, resuspended in 150 μL LB medium and plated in LB agar plates with 5% sucrose and 100 μg/mL carb.

The plates were incubated until colonies were clearly visible (24–72 hours). If no colonies were observed for a specific variant, the transformation was repeated twice more. Colonies were picked from each plate and grown overnight in LB medium with Carb₁₀₀. These cultures were then preserved as glycerol stocks and spotted (3 μL) in both Carb₁₀₀ and Kan₅₀ plates. Colonies that grew in carb plates only were grown from glycerol stocks in LB with Carb₁₀₀ and further analyzed. Total RNA was isolated from these cultures using Ambion Trizol LS. RT-PCRs of this total RNA were performed amplifying a 635nt region of the 16S rRNA containing both hinge library sites. These amplicons were then nanopore sequenced, and the presence of the expected mutations was confirmed. For all hinge variants capable of supporting life, cell doubling rates were measured. To do this, overnight cultures were grown in LB with Carb₁₀₀ from glycerol stocks. We inoculated 200 μL of LB with 5 μL of each overnight culture and recorded the OD₆₀₀ every 5 minutes over 5 hours with 15-second mixing in between reads in a SPECTRAmax 340PC Microplate Spectrophotometer with a qPCR seal.

Model generation for molecular simulations

Molecular simulations were performed using previously derived models of the PRE, INT3, and POST E. coli ribosome conformations [20]. The PRE conformation was modeled from PDB ID: 4V9D [39] with tRNA in the A/P and P/E positions. The INT3 conformation model was generated previously [20]. In brief, the INT3 conformation was modeled by restraining the coordinates of the SSU body from PDB ID: 4V7B and rotating the head domain to a 22.6° head swivel angle along the same helical axis as observed between the INT1 and INT2 states [9]. RMSD restraints were used to treat the SSU body, EF-G, LSU, L1, tRNA, and mRNA as rigid bodies to achieve the INT3 conformation. NAMD 2.12 [53] with CHARMM C36 force fields [54] was implemented to minimize and equilibrate the INT3 conformation. Lastly, the POST conformation was modeled from PDB ID: 4V5F [11] with tRNA in the P/P and E/E positions. The mRNA bound to each ribosome model was converted to 5’-GGAAAAAUGUUUAAAAAA-3’, encoding for tRNA^fMet and tRNA^Phe.

Hinge variants were introduced into the 16S rRNA using the Swapna package in Chimera [55]. To each model a water box of SPC/E model waters were added extending 15 Å beyond the extremities of the ribosome in addition to KCl and MgCl₂ ions to a concentration of 30 mM and 7 mM, respectively with GROMACS 2022 [56,57]. Each hinge variant was minimized in GROMACS 2022 using an iterative steepest descent approach by minimizing the potential energy of the water proceeded by the ribosome sidechains and backbone.

All-atom structure-based simulations

All-atom structure-based simulations were performed using GROMACS 2022 software with forcefields generated by the SMOG-2.4.2 software [58]. Simulations were performed at a temperature of 0.5ε, where ε is reduced units. The functional form of the potential used in the simulations is defined by: (1) where ε_r = 50 ε₀, ε_θ = 40 ε₀, ε_χi = 10 ε₀, ε_χp = 40 ε₀,ε_NC = 0.1 ε₀, σ_NC = 2.5 Å, and ε₀ = 1 and . Contacts were determined using a cutoff distance of 4 Å between heavy atoms. The strength of the contributions between the mRNA and tRNA substrate with the ribosome were rescaled by 0.6 to allow for translocation of the substrate. Additionally, the interface contacts between the SSU head and body were scaled by 0.3 to allow for head swivel motions. Contacts between the SSU and LSU were scaled by 0.3 for intersubunit contacts within 20 Å of core and by 0.1 for those outside this radius. This contact scaling approach has been utilized to accurately simulate translocation from the PRE to POST conformation of the ribosome previously [23]. Finally, interactions involving EF-G with ribosomes or tRNA were adjusted by a factor of 0.6 to account for the temporary nature of EF-G binding. These scaling factors are determined from previously established protocols of simulating head-domain movements, enabling coupled head-swivel and substrate translocation. Additionally, these values enable wild type trajectories to match the translocation dynamics observed in single-molecule fluorescence Förster resonance energy transfer (smFRET) experiments [20,23]. Electrostatics from ions such as Mg²⁺ were not included in the simulations; however, in the absence of ions structure-based simulations capture the conformational sampling of ribosomes from biochemical experiments at 2–7 mM Mg²⁺ [20,59,60]. Simulations were performed for 1.5 x 10⁷ timesteps with a step size of 0.002, as previous simulations of this type have indicated that 1 reduced time unit (t_ru) is ~1 ns [59,61], each simulation is ~30 μs. At minimum 10 simulations were performed for wild type ribosomes and each variant, totaling ~6.9 ms of simulations time.

Calculation of core helical axis motion

The helical axis of SSU head swivel motions were determined using the Curves+ software [62,63]. Specifically, the helical axis of h28, h29, h30, h32, h34, h35, and h36 were determined using snapshots of molecular simulation trajectories of ribosomes when they had reached the PRE (0°), INT2 (16°), INT3 (22.6°), or POST (0°) states defined by specific head swivel angles. Additionally, an intermediate in the transition between INT3 and POST was used to investigate the helical axis of reverse head-swivel at an angle of 13°. This angle was chosen as an intermediate from free-energy landscape approximations.

Hinge fluctuations

Backbone root mean squared fluctuation (RMSF) of hinges was measured using GROMACS 2022 [56,57]. RMSF was calculated with: (2) where r_i are the coordinates of atom i and 〈r_i〉 are the time averaged coordinates.

Head swivel and tilt measurements

Head swivel and tilt measurements were performed as previously described [20,23]. In brief, a set of core residues were utilized to align residues of the 16S body to a structural reference PRE configuration and the vector of the reference is measured from a plane derived from three fixed atoms of A977 and A1374 P and U1298 O2P. Subsequently core residues of the head domain are used to align the reference head to those of the trajectory and the vector is measured again from the same vector. Comparing the vectors from the simulation to the reference allows for quantification of the tilt and head swivel angles.

Body rotation measurements

Similar for the measurements of head tilt and swivel, rotations of the SSU body were measured using a plane derived from atoms P of C237 and C811 and O6 from G1488. The model from the simulations in each frame was aligned to a reference using core residues in the 23S, then the vector for the plane is measured. Subsequently, the SSU body reference structure is aligned to the model from the simulation using core residues of the 16S from the body and the vector of the plane is determined. The angle between the two vectors is then used to measure the body rotation during translocation.

tRNA movement measurements

tRNA movement was assessed by measuring the distance between the O3’ atom of G768 of the 16S rRNA and the O3’ atom of A35 of the peptidyl tRNA. Distance measurements were performed in GROMACS 2022. This reaction coordinate was selected to describe the barrier of tRNA movement from the P/E to E/E position during translocation.

Intersubunit bridge displacement measurements

To measure the displacement of the intersubunit bridges B1a and B1b during each simulation, the center of mass of residues involved in bridge contacts were used for distance measurements [41]. For intersubunit bridge B1a the distance between the center of mass of Arg91 and Arg92 of S13 and the center of mass of nucleotides A887 and C888 of H38 of the 23S rRNA were measured. For intersubunit bridge B1b the distance between the center of mass of Ile8, Asp67, Arg69, and Arg70 of S13 and the center of mass of Asp112, Arg114, and Tyr142 of L5 were used.