Cryo-EM structures of the Mycobacterium 50S subunit reveal an intrinsic conformational dynamics of 23S rRNA helices

Pathogenic organisms encounter a broad range of stress conditions within host micro-environment and adopt variety of mechanisms to stall protein translation and protect translational machinery. Structural investigations of the ribosomes isolated from pathogenic and non-pathogenic Mycobacterium species have identified several mycobacteria-specific structural features of ribosomal RNA and proteins. Here, we report a growth phase-dependent conformational switch of domain III and IV helices (H54a and H67-H71) of the mycobacterium 23S rRNA. Cryo-electron microscopy (cryo-EM) structures (∼3-4 Å) of the M. smegmatis (Msm) 50S ribosomal subunit of log-phase manifested that, while H68 possesses the usual stretched conformation in one of the maps, another one exhibits an unprecedented conformation of H68 curling onto a differently oriented H69, indicating an intrinsic dynamic nature of H68. Remarkably, a 2.8Å cryo-EM map of the Msm stationary-state 50S subunit unveiled that H68 preferably acquires folded conformation in this state (closely mimicking dormant state). Formation of a bulge-out structure by H68 at the inter-subunit surface of the stationary-state 50S subunit due to the rRNA conformational changes prevents association with 30S subunit and keeps an inactive pool of the 50S subunit representing a ribosome-protection mechanism during dormancy. Evidently, this dynamic nature of H68 is an integral part of the cellular functions of mycobacterium ribosome, and irreversibly arresting H68 flexible motion would stall ribosome function. Thus, this conformational change may be exploited to develop anti-mycobacterium drug molecules. Significant statement Bacteria utilize several mechanisms to reprogram the protein synthesis machinery so that their metabolism is reduced in the dormant state. Mycobacteria are capable of hiding themselves in a dormant state during physiological stresses. Our study identified a hitherto-unknown folded conformation of the helix 68 (H68) of domain IV of mycobacterial 23S rRNA, which is predominantly present in the stationary state (closely mimicking latency). Our results suggest that this conformational transition is instrumental in keeping an inactive pool of the 50S subunit in the stationary state. Irreversibly arresting such conformational dynamics would lead to protein synthesis shutdown in mycobacteria during dormancy. Thus, this folded conformation of H68 offers an excellent therapeutic intervention site to treat mycobacterial latent infection. Highlights Identification of a hitherto-unknown folded conformation of the helix 68 of mycobacterial 23S rRNA H68 conformation transition represents a new ribosome protection mechanism in dormant mycobacteria The conformational switch of mycobacterial H68 offers an excellent therapeutic intervention site


3
Significant statement 51 Bacteria utilize several mechanisms to reprogram the protein synthesis machinery so that 52 their metabolism is reduced in the dormant state. Mycobacteria are capable of hiding 53 themselves in a dormant state during physiological stresses. Our study identified a hitherto-54 unknown folded conformation of the helix 68 (H68) of domain IV of mycobacterial 23S 55 rRNA, which is predominantly present in the stationary state (closely mimicking latency). 56 Our results suggest that this conformational transition is instrumental in keeping an inactive The ribosome, one of the most intriguing biomolecular machines (Spirin, 2002), is a 80 ribonucleoprotein complex having an intricate molecular architecture (Schmeing and 81 Ramakrishnan, 2009). Several actinobacterium-specific unique structural features have been 82 identified in mycobacterium 70S ribosome (Kushwaha and Bhushan, 2020). The most 83 striking one is a long (more than 100 nucleotides) helical expansion segment of 23S rRNA, 84 H54a (referred to as 'Steeple' (Shasmal and Sengupta, 2012)), that has been identified 85 emerging from the bottom of the L1 stalk side of the 50S (Fig. S1A, B). The tip of H54a   In this study, we set out to search for growth phases-specific structural alteration and  Tris-HCl pH 7.6, 10mM Mg-acetate, 100mM NH 4 Cl, 5mM β-ME). Cells were lysed in 143 Buffer 'A' by sonicating 10-12 times (amplitude -90% for 30s with 45s interval), then 144 50mg/ml lysozyme was added, and cell suspension was incubated at RT for 30 mins, on ice 145 for 20 mins, followed by French Press at 5psi. Cell debris was removed by centrifugation 146 (13,000 rpm, 4 o C, 2 hrs.   angle) at 20 o C. 50nM of purified 50S from both phases were placed in the cuvette to get the 173 scattering of free 50S subunit. For the re-association reaction, 50nM of purified 50S and 30S 174 subunits from both phases were mixed in ribosome re-association buffer (20 mM Tris-HCl 175 pH 7.6, 10 mM NH 4 Cl, 50 mM KCl, 10 mM Mg-acetate, 5 mM β-ME), and quickly added to 176 the instrument cuvette to get the scattering pattern. The intensity of scattered light was 177 9 recorded for 5 minutes and 10 sec. The data was plotted in the Origin software and values of 178 starting 10 seconds were omitted while plotting.

(B) Sucrose Gradient Density Centrifugation:
180 Two sets of re-association reactions were prepared, one for log phase (L) and another for 181 stationary phase (S). For both sets of reactions, 0.5 μM of 50S was mixed with 0.5 μM of 30S 182 in the ribosome re-association buffer to a final reaction volume of 300 μl and incubated at 183 37 o C for 25 mins, then at RT for 10 mins and finally on ice for 10 mins. After incubation, the 184 reaction mixture was layered on top of 33 ml of sucrose gradient (10%-40% w/v sucrose in 185 ribosome re-association buffer). Ribosomal subunits were separated by sucrose density 186 gradient ultracentrifugation (28,000 rpm, 4 o C, 5 hr. 25 mins) in Sorvall AH-629 rotor.

187
Fractions of 500 μl were collected from bottom to top using peristaltic pump. Absorbance at 188 260 was measured for all the fractions to generate the ribosome profile. Two additional sets 189 of re-association assay were prepared in the same way as mentioned above, with minute    Good 2D class averages were selected to be used as a template for auto-picking of particles 224 from the entire dataset. A total of 202808 particles were picked by auto-picking, extracted 225 and subjected to multiple rounds of 2D classification to remove bad particles. 3D Initial 226 11 models were generated from selected good 2D class averages. Initial model of 50S was used 227 as a reference for the first round of 3D classification. The 3D classification resulted in 4 228 major classes (two 70S, one 50S and one 30S), bad classes were discarded. The class 229 representing 50S was given for multiple rounds of reference-based 3D classification (shown

252
Two ribosome datasets were used to isolate the stationary state 50S subunits (S-50S).

253
Movies were collected from microscope by automated data acquisition using EPU software 254 (dataset-1; 1719 movies, dataset-2; 1321 movies). The initial steps of data processing were 255 done in Relion 3. Alignment of movie frames and CTF determination were done as described 256 previously. For dataset-1, a few thousand particles were manually picked and extracted with a 257 box size of 350 pixels. Extracted particles were used for reference free 2D class averaging. complex dataset. The second S-50S dataset was processed the same way as described before.

267
The structures of the ligand (EF-G2) unbound S-50S 3D classes showed identical features as 268 seen in the unbound 3D classes of S-50S particles extracted from first Msm S-ribosome-EF-269 G2 dataset. Therefore, particles of unbound S-50S classes of these two datasets were merged.

270
Combined 50S particles were subjected to a series of 2D classification to remove some bad 271 particles. Finally, good 2D class averages were subjected to reference-based 3D classification 272 (shown in Fig. S8). The final refined map consisted of 172451 particles, which was processed 273 further for particles polishing, CTF refinement and 3D auto refinement as described above

316
The overall architecture of 23S rRNA in the L1-50S is very similar to the available 317 structures of mycobacterial ribosome (Fig. 1C, E). In contrast, an unusual conformational 318 change in the H67-H71 region was observed in L3-50S (Fig. 1B), where H68 attains a 319 compact mushroom head-like bulge-out feature on the H67 stalk (Fig. 1D, F). The bulge-out 320 structure at the intersubunit surface of L3-50S (Fig. 1D, F) likely prevent its association with 321 the 30S subunit, and thus L3-50S needs a conformational switch to L1-50S, which is   Interestingly, analysis of sucrose density gradient ribosome profile of subunit re-343 association showed significant reduction in 70S ribosome formation by the 30S and 50S 344 subunits purified from stationary phase (S-30S, S50S) as compared to that of the log phase 345 (L-30S, L50S) ( Fig. 2A, B). Rayleigh light scattering analysis also showed that stationary-346 state 30S and 50S subunits do not associate as efficiently as the log-phase 30S and 50S 347 subunits do (Fig. 2C, D). Further, re-association assay by sucrose density gradient ribosome 348 profiling showed that association of S-50S and L-30S gets hindered, while L-50S and S-30S 349 subunit efficiently associates to form 70S ribosome (Fig. 2E, F), suggested that 50S subunits, 350 not the 30S subunits, of the stationary phase are responsible for inefficient re-association.

351
A comparative analysis of ribosome profiles of purified 70S ribosome from Msm log and 352 stationary phases clearly showed that occurrence of the 50S subunit peak along with 70S 353 ribosome is higher in stationary phase as compared to log phase (Fig. S7Ai, Bi). Moreover, 354 when stored Msm70S ribosome purified from log and stationary phases were again loaded 355 onto sucrose density gradient, only a single 70S ribosome peak was observed in log phase 356 ribosome profile. In contrast a clearly detectable peak of the 50S subunit was also observed 357 along with the 70S ribosome peak (Fig. S7Aii, Bii), indicating that the equilibrium between 358 70S ribosome and its subunits tends to shift towards ribosomal subunits in stationary phase.

359
A plausible explanation for this observation would be that dissociated subunits are unable to 360 efficiently re-associate to form active 70S ribosome due to some structural changes, creating  Visualization of the H67-H71 region in the refined (prior to b-factor sharpening) and 397 sharpened maps of S-50S (Fig. S11) revealed the emergence of a mushroom head-like bulge 398 out density of H68-H69 in the refined map while moving from high to low contour levels 399 18 (density appears fragmented in the sharpened map even at lower contour level (Fig. S11)).

400
The densities corresponding to H67 and H70-H71 on the other hand were clearly visible at 401 high contour level even in the sharpened map (Fig. S11). Although it is clear that H68 prefers into the bulge-out density (Fig 3D).

479
Upon H68 conformational transition from elongated to rotated structure in L3-50S, the H68- (represents a state closely mimicking dormancy) (Fig. 7). Thus, our study has unveiled a 518 novel mechanism of ribosome protection during dormancy in Mycobacterium. 519 Alternative mycobacterial ribosomes, having distinct translational features, contribute to regulation, but it also identifies a novel target for developing antimycobacterial agents. As 543 observed in our study, H68 prefers to adopt a folded conformation in the stationary state 544 (closely mimicking dormancy). Irreversibly arresting H68 in the folded conformation during 545 dormant state would lead to translation shutdown in mycobacteria due to unavailability of 546 active 70S ribosome for protein synthesis, even when the growth condition becomes 547 favorable which ultimately results in cell death (Fig. 7).        . The presence of 70S ribosome (red box), 50S (green box) and 30S subunits (blue box) is clearly visible (multiple rounds of 2D classification were done). Selected good 2D class averages were subjected to reference-based 3D classification (multiple rounds). 3D classification of the cryo-EM data processing resulted in 70S ribosome, 50S and 30S subunits density maps (lower panel). The same approach was taken to process both the log-and stationary phase datasets. Multiple rounds of 2D and 3D classifications were performed to extract the 50S subunits of log-and stationary phase ribosome datasets. Class 2 and 3 were merged and subjected to multiple rounds of reference-based 3D classification to get the homogenous classes. After multiple rounds of 3D classification, 3 different types of sub-classes were generated (Sub-class 1 (SC1) with unrotated H68, subclass 2 (SC2; 2264 particles) with rotated H68 and steeple out and sub-class 3 (SC3) with rotated H68). SC1 and C1 were merged to get a total of 29,350 particles of unrotated H68, whereas SC3 and C4 were merged to get a total of 21,806 particles of rotated H68. SC2 was refined to get map L2-50S. Each particle sets of unrotated/elongated H68 and rotated H68 were subjected to particle polishing, CTF refinement and 3D refinement to generate high resolution map referred as L1-50S and L3-50S, respectively.     The 50S subunit class of dataset one (S_50S_1) consisted of 115236 particles, subjected to a series of 2D averaging and 3D classification. Finally, five classes of 50S were generated, out of which one class was EF-G bound (C1) having ~ 5,700 particles, whereas all other classes showed the same 3D conformation with rotated H68 and steeple out (C2-C5; total consisting of 1,04,076 particles). The four classes showing the same conformation were combined and subjected to 3D refinement to generate map RF1. 50S refined map (RF2) from another Msm S-ribosome-EF-G2 complex dataset having the same conformation were combined with 50S refined map (RF1) of Msm S-ribosome-EF-G2-GDPNP complex dataset. Combined 50S particles were subjected to a series of 2D classification to remove some bad particles. Finally, good 2D class averages were subjected to reference-based 3D classification, resulting in formation of 11 major 3D Classes (SC1-SC11), out of which 9 classes were found to be identical (rotated H68 and steeple out), and further merged as set of 3-3 classes to generate 3 Super classes (SuC1-SuC3). All three super classes were combined to generate a final refined map with 1,72,541 particles. 3D refined map was subjected to particle polishing, CTF refinement and 3D refinement to generate high resolution map named as S-50S.   The density corresponding to H70-71 is seen at high threshold in both the maps, whereas, the fused density attributable to the folded H68-H69 is weak. The mushroom head-like density is visualized in the refined map at lower threshold and allowed us to build the backbone trace model of this region. This density, however, appears fragmented in the sharpened map suggesting conformational variability of this region.