A new family of bacterial ribosome hibernation factors – Nature

Production of P. urativorans biomass

As a model organism, we used the bacterium P. urativorans. Freeze-dried cells of P. urativorans were obtained from the American Type Culture Collection (ATCC 15174). The cell pellet was rehydrated in 15 ml of pre-chilled marine broth 2216 medium (Sigma-Aldrich) and incubated in a shaker (SciQuip Incu-Shake Mini) at 150 r.p.m. at 10 °C for 7 days, according to the American Type Culture Collection protocol. To isolate ribosomes corresponding to structures 1–3 (Extended Data Table 1), this culture was then used to inoculate 1 l of pre-chilled marine broth 2216 medium and incubated at 150 r.p.m. for 4 days at 10 °C until the culture reached an optical density at 600 nm (OD₆₀₀) of 0.272. The cells were then placed on ice for 10 min and centrifuged for 5 min at 4 °C and 5,000g, yielding approximately 1 g of cell pellet. To isolate ribosomes from stationary cells, P. urativorans cultures were allowed to reach the stationary phase (OD₆₀₀ of 1.5) and remain in this phase for 4 days before pelleting these cells for 5 min at 4 °C and 5,000g and using this pellet for ribosome isolation.

Ribosome isolation

To lyse the cells, the pellets were rapidly resuspended in 1 ml of buffer A (50 mM Tris-HCl pH 7.5, 20 mM magnesium acetate and 50 mM KCl), transferred to 2-ml microcentrifuge tubes containing approximately 0.1 ml of 0.5 mm zirconium beads (Sigma-Aldrich BeadBug), and disrupted by shaking for 30 s at 6.5 m s⁻¹ speed in a bead beater (Thermo FastPrep FP120 Cell Disrupter). The sample was then centrifuged for 5 min at 4 °C and 16,000g to remove cell debris, and the resulting supernatant was collected and centrifuged for 1 min at 16,000 r.p.m. and 4 °C to remove the remaining debris. To analyse polysome profiles, we analysed 0.1 ml of crude P. urativorans lysates per time point, using 10–40% sucrose gradients in buffer A after 3 h of centrifugation at 35,000 r.p.m. and 4 °C in a SW41 rotor (Beckman Coulter). To isolate ribosomes for structural analysis, the cell lysate corresponding to 30 min of ice treatment was then mixed with PEG 20,000 (25% w/v) to a final concentration of 0.5% (w/v) and centrifuged for 5 min at 4 °C and 16,000g to precipitate insoluble aggregates. Then, the supernatant was mixed with PEG 20000 (powder) to the final concentration of about 12.5% (w/v) and centrifuged for 5 min at 4 °C and 16,000g to precipitate ribosomes. To monitor precipitation of ribosomes, we analysed lysates and their fractions using size-exclusion chromatography with Superdex Increase 200 10/300 in buffer A (Extended Data Fig. 1). The obtained ribosome-containing pellet was dissolved in 50 μl of buffer A, and the solution was passed twice through PD Spin Trap G-25 microspin columns (GE Healthcare) to clear crude ribosomes from small molecules. The obtained solution had an OD_260nm of 34.89 and an OD_260nm/280nm of 1.71, corresponding to a ribosome concentration of 512 nM. This solution was split into 10-μl aliquots and frozen at −20 °C for subsequent cryo-EM and mass spectrometry analyses.

Mass spectrometry analysis of crude samples of P. urativorans ribosomes

For each measurement shown in supplementary datasets 1 and 2, a 10-μl aliquot of crude P. urativorans ribosome solution was reduced with 4.5 mM dithiothreitol and heated at 55 °C. The sample was alkylated with the addition of 10 mM iodoacetamide before proteolytic digestion with 0.2 μg Promega sequencing-grade trypsin and incubation at 37 °C for 16 h. The resulting peptides were desalted by Millipore C18 ZipTip, following the manufacturer’s protocol, with final elution into aqueous 50% (v/v) acetonitrile. Desalted peptides were dried under vacuum before being resuspended in aqueous 0.1% trifluoroacetic acid (v/v) for LC–MS/MS.

Peptides were loaded onto a mClass nanoflow UPLC system (Waters) equipped with a nanoEaze M/Z Symmetry 100-Å C18, 5-µm trap column (180 µm × 20 mm, Waters) and a PepMap, 2 µm, 100 Å, C18 EasyNano nanocapillary column (75 μm × 500 mm, Thermo). The trap wash solvent was aqueous 0.05% (v/v) trifluoroacetic acid and the trapping flow rate was 15 µl min⁻¹. The trap was washed for 5 min before switching flow to the capillary column. Separation used gradient elution of two solvents: solvent A—aqueous 0.1% (v/v) formic acid; solvent B—acetonitrile containing 0.1% (v/v) formic acid. The flow rate for the capillary column was 330 nl min⁻¹ and the column temperature was 40 °C. The linear multi-step gradient profile was: 3–10% B over 7 min, 10–35% B over 30 min, 35–99% B over 5 min and then proceeded to wash with 99% solvent B for 4 min. The column was returned to initial conditions and re-equilibrated for 15 min before subsequent injections.

The nanoLC system was interfaced with an Orbitrap Fusion Tribrid mass spectrometer (Thermo) with an EasyNano ionization source (Thermo). Positive ESI-MS and MS2 spectra were acquired using Xcalibur software (v4.0, Thermo). Instrument source settings were: ion spray voltage—1,900 V; sweep gas—0 a.u.; ion transfer tube temperature—275 °C. MS1 spectra were acquired in the Orbitrap with 120,000 resolution, the scan range of m/z 375–1,500, the AGC target of 4 × 10⁵, and the maximum fill time of 100 ms. Data-dependent acquisition was carried out in top speed mode using a 1-s cycle, selecting the most intense precursors with charge states >1. Easy-IC was used for internal calibration. Dynamic exclusion was carried out for 50-s post precursor selection and a minimum threshold for fragmentation was set at 5 × 10³. MS2 spectra were acquired in the linear ion trap with: scan rate—turbo; quadrupole isolation—1.6 m/z; activation type—HCD; activation energy—32%; AGC target—5 × 10³; first mass—110 m/z; maximum fill time—100 ms. Acquisitions were arranged by Xcalibur to inject ions for all available parallelizable time.

Peak lists in Thermo.raw format were converted to.mgf using MSConvert (v3.0, ProteoWizard) before submitting to database searching against the P. urativorans subset of the UniProt database (3 August 2022, 2,349 sequences; 769,448 residues)⁵² appended with 118 common proteomic contaminants. Mascot Daemon (v2.6.0, Matrix Science) was used to submit the search to a locally running copy of the Mascot program (Matrix Science, v2.7.0). Search criteria specified: enzyme—trypsin; maximum missed cleavages—2; fixed modifications—carbamidomethylation of protein C termini; variable modifications—acetylation of protein N-termini, deamidation of Asn and Gln residues, N-terminal conversion of Gln and Glu to pyro-Glu, oxidation of Met and phosphorylation of Ser, Thr and Tyr residues; peptide tolerance—3 ppm; MS/MS tolerance—0.5 Da; instrument—ESI-TRAP. Peptide identifications were passed through the percolator algorithm to achieve a 1% false discovery rate assessed against a reverse database. The search data are summarized in supplementary datasets 1 and 2, for which molar percentages of each identified protein were calculated from Mascot emPAI values by expressing individual values as a percentage of the sum of all emPAI values in the sample, as previously described⁵³. To calculate the relative abundance of each cellular protein before and after 30 min of ice treatment (as shown in supplementary dataset 1), their total spectrum counts in the ice-treated sample were divided by the corresponding total spectrum counts of the control (non-ice-treated) sample. An infinite value for a few proteins means that in the control sample we have not been able to detect evidence for a protein by spectral counting.

Cryo-EM grid preparation and data collection for P. urativorans ribosomes

To prepare ribosome samples for cryo-EM analyses, 8–10-μl aliquots of crude ribosomes were thawed on ice and loaded onto glow-discharged (20 mA, 60 s or 90 s, PELCO easiGlow) Quantifoil grids (R1.2/1.3, 200 mesh, copper), using 2 μl of the sample per grid. The grids were then blotted for 1 or 2 s at 100% humidity (using blotting force −5) and vitrified using liquid nitrogen-cooled ethane in a Vitrobot Mark IV (Thermo Scientific). The grids were screened using Smart EPU (Thermo Scientific) with a 200-kV Glacios electron cryo-microscope (Thermo Scientific) with a Falcon 4 detector located at the York Structural Biology Laboratory, University of York, UK. The dataset corresponding to structures 1–3 (Extended Data Table 1) was collected on a 300-kV Krios cryogenic electron microscope (Thermo Scientific) located at the electron Bio-Imaging Centre, Diamond Light Source, UK using the parameters detailed in Extended Data Table 1. A total of 9,637 micrograph videos were recorded in aberration-free image shift mode using defocus targets of −2.4, −2.1, −1.8, −1.5, −1.2 and −0.9 μm. The dataset corresponding to the stationary phase sample of P. urativorans ribosomes was collected using a 200-kV Glacios cryogenic electron microscope (Thermo Scientific) with a Falcon 4 detector located at the York Structural Biology Laboratory, University of York, UK. For each video, the grids were exposed to a total dose of 50 electrons Å⁻² across 5.65 s. A nominal magnification of ×150,000 was applied, resulting in a final calibrated object sampling of 0.94 Å pixel size. A total of 4,997 micrograph videos were recorded in aberration-free image shift mode using defocus targets of −1.4, −1.2, −1.0, −0.8 and 0.6 μm.

Cryo-EM data processing for P. urativorans ribosomes

The cryo-EM dataset corresponding to structures 1–3 (Extended Data Table 1) was processed using RELION 3.1⁵⁴ as summarized in Extended Data Table 1 and Extended Data Fig. 1). In brief, to determine structures 1–3, a total of 180,467 particles were picked from 8,903 micrographs using the Laplacian of Gaussian picker (220–330 Å particle diameter; 0.6 s.d. threshold). Particle images were initially downscaled threefold and extracted in a 180-pixel box (2.169 Å effective pixel size). Four rounds of two-dimensional (2D) classification were carried out to clean the dataset, with 83,841 ‘good’ particles selected for 3D refinement. These particles were rescaled to full size and extracted in a 540-pixel box. The initial 3D refinement generated a map at 3.5 Å resolution, using an initial 3D reference imported from a previous Glacios dataset that had been low-pass filtered to 60 Å. Heterogeneity was apparent at the Balon-binding site and decoding centre. At this point, contrast transfer function (CTF) refinement was carried out to account for beam tilt, trefoil fourth-order aberrations and magnification anisotropy. The CTF was estimated per particle, and the astigmatism was estimated per micrograph. Subsequent particle polishing and 3D refinement generated a map at 2.51 Å resolution, thereby providing the most accurate angular assignments for subsequent focused classification. Initial attempts at focused classification using masked classification without alignment were not successful, so signal subtraction was first carried out using masks to define Balon density and P-site density. This effectively separated particles into three groups corresponding to differential factor occupancy: ribosome (empty); structure 1 (P. urativorans ribosome–Balon–RaiA, with partial EF-Tu occupancy); and structure 2 (P. urativorans ribosome–Balon–tRNA–mRNA, with partial EF-Tu occupancy). The overall density for EF-Tu in the above structures 1 and 2 was weak, indicating substoichiometric amounts. Focused classification with signal subtraction was therefore carried out, first with a loose mask around the entire EF-Tu molecule. This revealed that about 44% of Balon-associated ribosomes were bound by EF-Tu. However, within this subset, the density for domain I of EF-Tu was poor, indicating conformational flexibility. Further signal subtraction was therefore carried out using a mask to isolate the density for EF-Tu domain I. Subtracted particles were recentred in a 200-pixel box and classification with local angular searches was carried out, revealing that domain I was present in two slightly rotated orientations with respect to domains II and III of EF-Tu. Particles corresponding to these two orientations of domain I were re-extracted and refined separately, leading to interpretable density at a local resolution of about 4 Å for EF-Tu domain I, at which point the details of the nucleotide-binding site became visible. Structure 3 represents the map with the best density for EF-Tu but is heterogeneous with respect to tRNA or RaiA in the P site. In this structure, EF-Tu is also associated with the globular C-terminal domain of the L7/L12 stalk (protein bL12). A map filtered to about 6 Å shows this the most clearly (Fig. 5a and Supplementary Fig. 1). Finally, sharpened maps weighted by estimated local resolution were calculated. All reported estimates of resolution are based on the gold-standard Fourier shell correlation at 0.143, and the calculated Fourier shell correlation is derived from comparisons between reconstructions from two independently refined half-sets.

The cryo-EM dataset corresponding to ribosomes from stationary P. urativorans was processed using cryoSPARC v4.3.0⁵⁵ as summarized in Supplementary Fig. 2. In brief, a total of 909,391 particles were picked from 4,997 micrographs using the Blob picker (190–260 Å particle diameter). Particle images were extracted in a 400-pixel box (without downscaling). Five rounds of 2D classification were carried out to clean the dataset, with 80,882 ‘good’ particles selected for ab initio reconstruction and subsequent homogeneous refinement. This generated a final map at 5.1 Å resolution, which was used for rigid-body docking of structure 1 to assess the presence of Balon in the ribosomal A site (Supplementary Fig. 2).

Model building, refinement and deposition

The atomic models of P. urativorans ribosomes and the ribosome-binding proteins were produced using Coot v0.8.9.2⁵⁶ and AlphaFold⁵⁷. As a starting model, we used the atomic model of ribosomal proteins generated by AlphaFold2 and the atomic model of rRNA from the coordinates of T. thermophilus ribosomes (PDB code 4y4o). These rRNA and protein models were morph-fitted into the cryo-EM maps using ChimeraX 1.4⁵⁸ and Phenix 1.20.1⁵⁹ and then rebuilt using Coot on the basis of the information about the genomic sequence of P. urativorans (RefSeq GCF_001298525.1). In the ribosome complex with Balon, mRNA and tRNA, the mRNA molecule was modelled as poly-U, and the tRNA molecule was modelled as U_1–72A₇₃C₇₄C₇₅A₇₆.

The density corresponding to Balon was initially identified as a non-ribosomal protein, which was initially modelled as a poly-alanine chain to determine its backbone structure. This poly-alanine backbone model was then used as an input file for a search of proteins with similar fold in the PDB using the National Center for Biotechnology Information tool for tracking structural similarities of macromolecules, Vast⁶⁰. This search identified the archaeal protein aeRF1 from Aeropyrum pernix as the most similar known structure to Balon, suggesting that Balon is a bacterial homologue of aeRF1 (Supplementary Table 1). We therefore searched for P. urativorans proteins that have a similar sequence to that of A. pernix aeRF1. Using three iterations of Markov model-based search with HHMER⁶¹ in the UniProt database, we found that P. urativorans encodes a hypothetical protein (UniProt ID A0A0M3V8U3) with sequence similarity to aeRF1 and Pelota. This protein, which we termed Balon, had a sequence that matched the cryo-EM map and was used to create its atomic model. The resulting atomic structures of Balon in complex with the ribosome, RaiA and EF-Tu or Balon in complex with the ribosome, tRNA and mRNA were then refined using Phenix real-space refinement, and the refined coordinates were validated using MolProbity⁶².

Purification and cryo-EM analysis of M. smegmatis ribosomes and their complexes with Msmeg1130, Rv2629 and EF-Tu

M. smegmatis 70S ribosomes, isolated from strain mc²155, were prepared as previously described⁶³. The full-length Rv2629 sequence was PCR amplified from M. tuberculosis H37Rv genomic DNA and the full-length M. smegmatis EF-Tu and Msmeg1130 were amplified from M. smegmatis mc²155 gDNA and cloned into the pET28a plasmid with a 6×His-SUMO tag. E. coli BL21 (DE3) Star cells were transformed with the constructs and grown at 37 °C in the LB medium with kanamycin. The expression of the proteins of interest was induced with isopropyl-β-d-thiogalactoside once the culture reached an OD_600nm of about 0.6. The purification protocol for mycobacterial proteins included a two-step Ni-affinity (HisTrap HP column), ion exchange (HiTrap Q HP or Source 15Q) and size-exclusion chromatography (Superdex 200 16/600). Rv2629, Msmeg1130 and M. smegmatis EF-Tu were flash frozen in liquid nitrogen and stored in 20 mM HEPES-KOH pH 7.5, 10 mM MgCl₂, 200 mM KCl, 300 mM l-arginine and 1 mM dithiothreitol.

M. smegmatis ribosome complexes (20 μl) with Msmeg1130 or Rv2629 were prepared by incubating 2 µM M. smegmatis mc²155 70 S ribosomes with 30 µM Msmeg1130 or Rv2629 in 1× buffer B (20 mM HEPES-KOH pH 7.5, 60 mM KCl, 10 mM MgCl₂, 1 mM DTT) for 15 min at room temperature. Ribosome complexes with Msmeg1130 and EF-Tu were prepared by pre-incubating 60 µM EF-Tu with 1 mM GDP or 1 mM GDPCP for 5 min at 37 °C and mixed with M. smegmatis mc²155 70S ribosomes to final concentrations of 2 µM ribosomes, 30 µM Msmeg1130 and EF-Tu followed by a 15-min incubation at room temperature.

Cryo-EM grid preparation and data collection for M. smegmatis ribosomes

Plasma-cleaned Quantifoil (R2/1, 200 mesh, gold) grids (Electron Microscopy Sciences) were used for sample application. Grids were blotted in 85% humidity at room temperature for 22 s and plunge frozen in liquid nitrogen-cooled ethane using the Leica EM GP2 cryo-plunger. Then, cryo-EM micrographs were recorded with a Titan Krios G3i electron microscope (300 kV) equipped with FalconIII (ThermoFisher) and K3 (Gatan) direct electron detectors. A total of 11,031 (dataset 4, Extended Data Table 1 and Supplementary Fig. 3), 9,546 (dataset 5, Supplementary Fig. 4) and 10,161 (dataset 6, Extended Data Fig. 5) micrograph videos were acquired in the counting mode with a pixel size of 0.839 Å per pixel (K3) or fast integrating mode with 0.85 Å per pixel (FalconIII). On the basis of the relative ice thickness, patch CTF fit, length and curvature of motion trajectories, 9,440, 9,438 and 9,465 micrographs were selected for further processing. For all datasets with M. smegmatis 70S ribosomes, the particles were picked using the circular ‘blob’ picker in cryoSPARC and filtered on the basis of defocus-adjusted power and pick scores. Particles were then subjected to one (dataset 5) or two (datasets 4 and 5) rounds of reference-free 2D classification. The selected particles were used to generate ab initio volumes that were sorted using ‘heterogeneous refinement’. Selection of the classes with Msmeg1130 or Rv2629 or Msmeg1130–EF-Tu bound to M. smegmatis 70S was carried out using 3D classification analysis, further classified and polished by focused 3D variability analysis (Msmeg1130EF-Tu datasets) with a spherical mask around EF-Tu. This approach allowed us to remove ‘bad’ or noisy particles, re-extract to full-size (512-pixel box) particles with solid density for factors and carry out non-uniform with CTF refinements, which yielded the final reconstructions for M. smegmatis 70S complexes. The model was assembled from individual parts. The non-rotated M. smegmatis 70S ribosome model (PDB code 5o61) structure with P-site tRNA in the active site⁶³ was rigid-body fitted into the 3.0-Å charge density maps using UCSF Chimera 1.14⁶⁴. The models predicted by AlphaFold2^65,66 for Rv2629 and Msmeg1130 were docked into the density maps and adjusted in Coot v0.8.9.2⁵⁶. mRNA was modelled as poly-U and tRNA^Phe was modelled in the P site. The complete model was refined using five cycles of real-space refinement in Phenix 1.19.2⁵⁹.

A total of 11,846 micrographs (GDPCP dataset, Extended Data Fig. 6) were selected for further processing. The model predicted by AlphaFold2^65,66 for EF-Tu was used for the fitting into the density maps. The Thermus aquaticus EF-Tu(GDP) model (PDB code 1tui)⁴² was used for fitting domain I and switches I and II into the EF-Tu density map. Local refinement combined with particle subtraction produced higher-quality maps for EF-Tu in complexes formed in the presence of GDP (Extended Data Fig. 5) and GDPCP (Extended Data Fig. 6).

Evolutionary analysis of Balon

To assess phylogenetic distribution of Balon in bacterial species, we carried out three iterations of homology search using the sequence of Balon from P. urativorans (UniProt ID A0A0M3V8U3) as an input for a profile hidden Markov model-based analysis with HMMER. For each search iteration, we used the following search options: -E 1 –domE 1 –incE 0.01 –incdomE 0.03 –seqdb uniprotrefprot. The resulting dataset was reduced first by removing protein sequences that lacked information about their Phylum (21 sequences), then by removing sequences that were shorter than 300 amino acids as they typically lacked one or two of the three domains of Balon/aeRF1 (which included 806 sequences), then by removing sequences that were annotated as a protein fragment (34 sequences), and finally by removing duplicated sequences (31 sequences). This resulted in the dataset including 1,896 sequences of Balon homologues from 1,565 bacterial species (supplementary dataset 3).

To gain insight into a possible evolutionary origin of Balon from the archaeo-eukaryotic family of aeRF1 proteins, we carried out a complementary analysis in which we searched for bacterial homologues of the archaeal aeRF1 using three iterations of HMMER. As an input for the first iteration, we used the sequence of aeRF1 from the archaeon A. pernix (UniProt ID Q9YAF1), which we identified as being one of the closest structural homologues of Balon. For each iteration, we used the database of reference proteomes restricted to the bacterial domain of life, using these search options: -E 1 –domE 1 –incE 0.01 –incdomE 0.03 –seqdb uniprotrefprot. The resulting dataset was reduced first by removing sequences lacking information about their phylum (21 sequences), then by removing sequences that were lacking at least one of the three domains of aeRF1 proteins (sequences shorter than 300 amino acids, which included 1,422 sequences), then by removing sequences annotated as a protein fragment (5 sequences), and finally by removing duplicated sequences (104 sequences). This resulted in the dataset of 1,617 sequences of bacterial aeRF1 homologues from 1,353 bacterial species (supplementary dataset 4).

To map the identified Balon homologues on the tree of life, we combined the results of the previous searches in supplementary datasets 3 and 4 and removed repetitive entries, which resulted in a dataset of 1,898 protein sequences from 1,572 bacterial species (supplementary dataset 5). We then aligned the combined sequences using Clustal Omega⁶⁷ with default parameters, which resulted in a multiple sequence alignment (supplementary dataset 6) and a phylogenetic tree (supplementary dataset 7). To compare phylogenetic distribution of Balon, RMF and RaiA-type hibernation factors, we repeated the homology search using HMMER (with the same parameters as for our Balon homologues searches) for RaiA (using the E. coli sequences of RaiA as an input; supplementary dataset 8) and RMF (using the E. coli sequence of RMF as an input; supplementary dataset 9).

Generation of Balon knockout strains

The gene knockout in P. arcticus 273-4 (DSMZ 17307, ref. seq. NC_007204.1) was generated on the basis of a suicide-vector-based approach originally developed previously⁶⁸. In brief, the suicide vector pBC19 was constructed on the basis of the design of pJK100. Regions of 500–600 base pairs serving as homologies for the targeted deletion were PCR amplified from genomic DNA isolated from the respective strains. A third fragment containing a kanamycin resistance cassette flanked by two loxP sites was amplified from the pCM184 (Addgene no. 46012) vector. The three fragments were each amplified using primers carrying overlaps necessary for Gibson Assembly (NEBuilder HiFi DNA Assembly Master Mix, NEB). These three fragments were then assembled together and fused with the pKNOCK-Tc vector (Addgene no. 46259) and digested with EcoRI and KpnI using Gibson Assembly. The pKNOCK-Tc backbone carries a R6Kγ origin of replication requiring the pir gene for replication; therefore, the Gibson reactions were first transformed into the E. coli MDS42pir strain⁶⁹. A desired clone was then identified and purified and transformed into the conjugative strain E. coli BW29427 that is auxotrophic for diaminopimelic acid (DAP). This donor strain was then grown at 37 °C in LB broth medium supplemented with 50 mg ml⁻¹ kanamycin, 20 mg ml⁻¹ tetracycline and 100 mg ml⁻¹ DAP overnight and mixed with cultures of P. arcticus recipients, both grown for 48 h at 20 °C in LB broth. Aliquots of 1 ml of the cultures were centrifuged (2 min, 4,000 r.p.m.) and washed twice in PBS (8 g l⁻¹ NaCl, 0.2 g l⁻¹ KCl, 1.44 g l⁻¹ Na₂HPO₄, 0.24 g l⁻¹ KH₂PO₄, pH 7.4) and finally resuspended in 1 ml PBS. The cultures were then mixed gently at ratios of 100 ml donor to 100 ml recipient and 100 ml donor to 400 ml recipient and spot plated onto LB agar plates containing 100 mg ml⁻¹ DAP and grown for 24 h at 20 °C. The grown conjugation mixtures were then scraped and suspended in 1 ml LB broth, after which 100 ml of the suspension and 10× and 100× dilutions were spot plated onto LB agar plates containing 50 mg ml⁻¹ kanamycin and grown at 20 °C. Putative conjugant colonies became visible after 72–96 h. These were picked and checked for their sensitivity against 20 mg ml⁻¹ tetracycline. Tetracycline-sensitive colonies (with presumably the kanamycin resistance cassette inserted into the targeted genomic site) were re-streaked onto LB agar plates supplemented with 50 mg ml⁻¹ kanamycin. Perturbation of the target gene was then validated using PCR, with one primer annealing outside of the genomic homology region, and the other to the kanamycin resistance gene. In this manner, the expected fragment was produced only from colonies where the resistance cassette was genomically inserted into the desired locus. These PCR fragments were then sequence-verified using Sanger sequencing. Additionally, plasmid pBC16 was also constructed (on the basis of pJK100) carrying just the loxP-kanamycin resistance marker-loxP cassette cloned into the pKNOCK-Tc backbone carrying multi-cloning sites flanking the knock-in cassette. This plasmid can be used to clone homology cassettes using traditional restriction endonuclease-based cloning. To assess the growth of the produced strains, they were incubated in the BioTek TS800 microplate reader while measuring their growth using Agilent BioTek Gen 5 software.

Analysis of P. arcticus rRNA

Cell pellets of log phase and stationary phase cells of wild-type P. arcticus and P. arcticus with the Balon-coding gene deleted were obtained by centrifuging 120 ml of an actively growing P. arcticus culture and 25 ml of an 11-week-old culture of P. arcticus at 5,000g for 5 min. The biomass of each of the four resulting cell samples was taken in the same quantity of 170 mg and resuspended in 750 µl of TRIzol reagent (Invitrogen) with 20 µl of zirconia beads and vortexed for 5 min to lyse the cells. Then, 150 µl of chloroform was added to the lysed cells and vortexed for 2 min. After 15 min, the lysates were centrifuged for 15 min at 13,000 r.p.m, resulting in phase separation of the samples. The upper aqueous phase, containing the RNA, was transferred to a clean Eppendorf tube, and mixed with 350 µl of isopropanol. After inverting the samples and allowing them to stand for 10 min, the samples were centrifuged for 10 min at 13,000 r.p.m. The resulting supernatant was decanted, and the RNA pellets were washed with 1 ml of 70% ethanol and centrifuged for 15 min at 13,000 r.p.m. After decanting the ethanol, the pellets were dried in a speed vacuum and resuspended in 18 µl of RNase-free water. The resuspended RNA was incubated at 55 °C for 5 min to allow for complete resuspension. Then, 2 µl of each sample or 1 µl RiboRuler Low Range RNA ladder (Thermo Scientific) was mixed with 5 volumes of 5× glyoxal loading dye (61.2% dimethylsulfoxide (v/v), 20.4% glyoxal (v/v), 12.2% 10× BPTE buffer (300 mM Bis-Tris, 100 mM PIPES (piperazine-N,N′-bis(2-ethanesulfonic acid)), 10 mM EDTA) (v/v), 4.8% glycerol (v/v)) with 0.2 mg ml⁻¹ ethidium bromide and incubated at 55 °C for 1 h. Glyoxalated RNA samples were separated on a 1.2% agarose 1× BPTE gel before being visualized on a PhosphorImager (Typhoon FLA9000; GE Healthcare).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.