Cryo-EM structure of Mycobacterium smegmatis ribosome reveals two unidentified ribosomal proteins close to the functional centers

Cryo-EM structure of Mycobacterium smegmatis ribosome reveals two unidentified ribosomal proteins close to the functional centers.

the IC and EM was mixed at 37°C for 10 s and quenched with 17% formic acid. The dipeptides were isolated and analysed in HPLC as described earlier (Degiacomi et al., 2016;Holm et al., 2016). The activity of the ribosome was determined by fitting the data with hyperbolic function and estimating the reciprocal of the minimal ribosome concentration needed to reach the saturation ( Figure S2A). For tri-peptide formation, an IC with 1 µM 70S ribosome, and an EM with 5 µM EF-G were prepared with all M. smegmatis as well as E.
coli components. After 15-min incubation at 37 °C, equal volumes of IC and EM were rapidly mixed and quenched at different time points using a quench-flow instrument (BioLogic QFM-400). The peptides were analysed as mentioned in (Degiacomi et al., 2016;Holm et al., 2016) . The time curves were fitted with single exponential and the rates were determined.

Sample preparation and data collection
3.5-µL aliquots of purified ribosomes at a concentration of ~100 nM were applied to carbon-coated grids (Quantifoil, R2/2). Grids were blotted with filter paper at both sides for 2.5 s using FEI vitrobot Mark IV (100% humidity, 4°C) and vitrified in liquid ethane. The cryo-grids were transferred to a transmission electron microscope (FEI Titan Krios) operating at 300 kV for data collection. Images were recorded, at a nominal magnification of 22500 X, on Gatan K2 Summit detector in counting mode. Under these conditions, the pixel size at the object scale is 1.32 Å. The nominal defocus used to collect data ranged from -1.5 µm to -2.2 µm. All image stacks were collected with UCSF Image4 (Li et al., 2015) at low-dose conditions with each stack containing 32 dose-fractionated frames. The total exposure time for each stack was 8 s and the dose rate for each pixel was ~8 counts.

Image processing
MOTIONCORR (Li et al., 2013) was used to align the 32-frame stacks and to generate motion-corrected micrographs. The contrast transfer function (CTF) parameters of each micrographs was estimated using CTFFIND3 (Mindell and Grigorieff, 2003). The micrographs were first screened using SPIDER (Shaikh et al., 2008). Particles were picked automatically using SPIDER, and 2D classification was performed with RELION (Scheres, 2012). Particles in good 2D classes ( Figure S3A) were chosen for further 3D classification using RELION. To improve the resolution of the density map to facilitate the modeling, several rounds of 3D classification were performed ( Figure S2B). In the first round, the differences among the 8 classes were dominated by conformational variations of the 30S subunit (largely inter-subunit rotation) and the presence/absence of tRNA. Six similar classes were combined and used for the second round 3D classification (into three classes). Two of the final three classes were combined for high-resolution 3D refinement. After refinement the density of the 30S subunit was not as good as the 50S subunit, indicating that the 30S subunit was still heterogeneous in conformation. Then, another round of 3D classification was performed, with a 30S soft-mask applied during classification ( Figure S3B). Only one class were selected for the high-resolution refinement for the 30S subunit (47,338 particles). Final rounds of refinement for the 50S and 30S subunits were done with respective soft-masks applied, and with a final set of particles which were re-extracted from the dose-reduced micrographs (frame 2 to 15 using MOTIONCORR). The final resolution estimations for the MS50S and MS30S are 3.08 Å and 3.45 Å, respectively (based on the gold-standard FSC = 0.143 criterion). Local resolution variations were estimated using ResMap (Kucukelbir et al., 2014).

Model building and refinement
The modelling was similarly performed as previously described (Wu et al., 2016). The crystal structure of E.
coli 50S and 30S subunits (PDB accession number 4KIX, 4KIY) (Pulk and Cate, 2013) were fitted in the MS50S and MS30S maps, respectively. All ribosomal proteins and rRNAs were aligned with their E. coli counterparts using Clustal Omega (Sievers et al., 2011) to analyze the conserved regions. For the conserved domains of ribosomal proteins, we mutated the E. coli residues to the corresponding M. smegmatis residues manually during the model building. The mycobacterium-specific extensions/domains of ribosomal proteins were built de novo using COOT (Emsley et al., 2010). For rRNA models, base-pair information was reviewed manually.
The atomic models of the MS50S and MS30S were refined against corresponding maps using real-space refinement (phenix.real_space_refine) (Afonine et al., 2012) in PHENIX (Adams et al., 2010). After the refinement, manual adjustments were conducted using COOT. These procedures were done in several rounds.
Further refinement was carried out using REFMAC (Murshudov et al., 1997) in Fourier space according to the previously published protocols (Amunts et al., 2014). Cross validations were also performed following the same procedure described elsewhere (Wu et al., 2016). The final models were evaluated using MolProbity .           Branch b1 represents eukaryotes, and branch b2 represents bacteria. Branch b2.2 is actinobacteria and they have proteins uL41 and bL37. The evolution tree was generated using the software MEGA7 (Kumar et