Translational GTPase BipA Is Involved in the Maturation of a Large Subunit of Bacterial Ribosome at Suboptimal Temperature

BPI-inducible protein A (BipA), a highly conserved paralog of the well-known translational GTPases LepA and EF-G, has been implicated in bacterial motility, cold shock, stress response, biofilm formation, and virulence. BipA binds to the aminoacyl-(A) site of the bacterial ribosome and establishes contacts with the functionally important regions of both subunits, implying a specific role relevant to the ribosome, such as functioning in ribosome biogenesis and/or conditional protein translation. When cultured at suboptimal temperatures, the Escherichia coli bipA genomic deletion strain (ΔbipA) exhibits defects in growth, swimming motility, and ribosome assembly, which can be complemented by a plasmid-borne bipA supplementation or suppressed by the genomic rluC deletion. Based on the growth curve, soft agar swimming assay, and sucrose gradient sedimentation analysis, mutation of the catalytic residue His78 rendered plasmid-borne bipA unable to complement its deletion phenotypes. Interestingly, truncation of the C-terminal loop of BipA exacerbates the aforementioned phenotypes, demonstrating the involvement of BipA in ribosome assembly or its function. Furthermore, tandem mass tag-mass spectrometry analysis of the ΔbipA strain proteome revealed upregulations of a number of proteins (e.g., DeaD, RNase R, CspA, RpoS, and ObgE) implicated in ribosome biogenesis and RNA metabolism, and these proteins were restored to wild-type levels by plasmid-borne bipA supplementation or the genomic rluC deletion, implying BipA involvement in RNA metabolism and ribosome biogenesis. We have also determined that BipA interacts with ribosome 50S precursor (pre-50S), suggesting its role in 50S maturation and ribosome biogenesis. Taken together, BipA demonstrates the characteristics of a bona fide 50S assembly factor in ribosome biogenesis.


INTRODUCTION
The ability of bacteria to respond, adapt, and grow at suboptimal temperature is known as cold shock response, which is activated in Escherichia coli under a condition of a sudden drop in culturing temperature (usually from 37 to 15 • C) (Barria et al., 2013). Suboptimal temperature can cause both physiological and morphological changes in the bacteria and even cell death if the temperature shift is beyond bacterial tolerance. The majority of cold-inducible proteins involves RNA metabolism, indicating that regulation of RNA metabolism is crucial for suboptimal temperature adaptation (Barria et al., 2013). The most notable examples include the following: (i) cold shock protein (Csp) family proteins, which are RNA chaperones and able to prevent formation of RNA secondary structures (Phadtare et al., 2003;Barria et al., 2013;Uppal and Jawali, 2015); (ii) DEAD-box RNA helicase (DeaD) that can unwind RNA secondary structure to promote degradation (Charollais et al., 2004;Prud'homme-Généreux et al., 2004;Resch et al., 2010); and (iii) RNase R, the only 3 -5 exonuclease in E. coli that degrades double-stranded RNA without the help of a helicase (Cairrão et al., 2003;Awano et al., 2010;Barria et al., 2013).
The translational stress response involves a key component of protein-synthesizing machinery, the ribosome. Bacteria utilize a series of protein factors that bind ribosome to modulate translation in order to cope with stress. For example, the RelA/SpoT homolog (RSH) proteins bind ribosome when there is a surge of uncharged tRNAs caused by the shortage of amino acids. Consequently, binding of RSH to ribosome triggers the alarmone synthesis and is followed by the stringent response (Hauryliuk et al., 2015). Notably, there are two proteins that bind to the ribosome to exert their stress response function: ObgE and BPI-inducible protein A (BipA). ObgE is an essential GTPase, and its homologs have been ubiquitously found across all kingdoms of life (Sato et al., 2005;Jiang et al., 2006). An analysis of those immature 50S particles, accumulated in cells with the ObgE depleted, showed that late assembling r-proteins (L33, L34, and L16) were under-represented, indicating that ObgE is an important factor during the late step of 50S biogenesis (Jiang et al., 2006).
BipA has been implicated in various functions of pathogenic bacteria, such as Salmonella typhimurium (S. typhimurium), Pseudomonas aeruginosa (P. aeruginosa), and enteropathogenic E. coli (EPEC). BipA is upregulated by ∼7-fold when S. typhimurium is "attacked" by bactericidal/permeability increasing (BPI) antimicrobial peptide (Qi et al., 1995). BipA is also known as TypA, referring to tyrosine phosphorylated protein A, and this is the case in P. aeruginosa where TypA was found to be involved in virulence, antimicrobial resistance, and biofilm formation (Neidig et al., 2013). In EPEC, BipA has been reported to upregulate the virulence and reduce flagella-mediated motility (Grant et al., 2003). The bipA deletion would result in growth delay of E. coli K12 strain while cultured at low temperature (e.g., 20 • C) (Pfennig and Flower, 2001). Furthermore, Choudhury and Flower (2015) reported that the ribosomal particle distribution of the bipA deletion strain changes dramatically, with the accumulation of the 30S and the presence of 50S precursor (pre-50S), indicating a possible role of BipA in ribosome biogenesis (Choudhury and Flower, 2015). Interestingly, a transposon-mediated random insertion mutation into E. coli K12 strain genome revealed that the disruption of the rluC gene can suppress the phenotype of the bipA deletion (Krishnan and Flower, 2008). RluC is a pseudouridine synthase, which converts three uridines (U955, U2504, and U2580) in 23S rRNA to pseudouridines ( 955, 2504, and 2580). Pseudouridine is a uridine isomer where the uracil base is linked to the pentose sugar by a carbon-to-carbon bond, resulting in the ability to form an additional hydrogen bond and thereby increasing the stability compared to uridine (Kierzek et al., 2013). In addition, Choudhury and Flower (2015) discovered that the deletion of DeaD exacerbates both the growth and ribosomal particle distribution defects (Choudhury and Flower, 2015). Since DeaD is involved in 50S biogenesis, it is not surprising that bipA and deaD double mutations could cause an evident defect in ribosome biogenesis. Crystal structures of the apo-, as well as the nucleotides [GDP, (p)ppGpp, and GDPCP] bound BipA, showed no significant difference in their overall conformations (Fan et al., 2015;Kumar et al., 2015). However, a drastic conformational change was observed in domains III, V, and the C-terminal domain (CTD), upon GTP form BipA binding to 70S ribosome (Kumar et al., 2015). The conformational change of the CTD was particularly interesting because the CTD loop extends into the ribosomal A site to establish extensive contacts with the tRNA acceptor stem; however, the precise role of such interactions is largely unknown (Kumar et al., 2015). Nevertheless, the structure of BipA with ribosome demonstrated an activated form of BipA in the ribosome, with the catalytic residue, histidine 78 (H78), situated close to the 23S rRNA sarcin-ricin loop (SRL) in a ratcheting ribosome as well as interacting with the bound nucleotide (GDPCP). Collectively, BipA is an authentic ribosome-dependent trGTPase. However, how BipA in association with 70S ribosome is correlated with bacterial stress response as well as how BipA functions in conditional translation for bacterial cells at suboptimal temperature require further studies.
Notably, recent quantitative mass spectrometry data nicely showed the accumulated pre-50S intermediates in bipA mutant cells at suboptimal temperature, with several r-proteins absent, further demonstrating the role of BipA in 50S subunit assembly (Gibbs et al., 2020). Furthermore, a paper published during our manuscript preparation revealed that bacteria can remodel their protein expression relevant to biofilm formation in a temperature-dependent manner by modulating BipA abundance (Del Peso Santos et al., 2021). With the desire to support the ongoing effort in BipA-related studies and to expand our efforts on structural study of ribosome-associated proteins and biofilm formation as well as its relevant pathogenesis and resistance (Tanaka et al., 2008;Selmer et al., 2012;Kumar et al., 2015;Yu et al., 2015;Ero et al., 2016;Kumar et al., 2016;Yang et al., 2017), here we report additional evidence to demonstrate the role of BipA in ribosome biogenesis, specifically in large subunit 50S maturation, and conditional protein translation at suboptimal temperature through combinative approaches. In particular, our findings by tandem mass tag (TMT)-based quantitative proteomic analysis identified proteins relevant to RNA metabolism (e.g., DeaD) with increased expression levels upon BipA deletion, such an effect can be suppressed by a further mutation of RluC or complementation by BipA. The upregulation of these protein possibly indicates that bacterial cells can compensate for the loss of BipA during 50S biogenesis.

Bacterial Strains and Culturing
The E. coli strains used in this project are listed in Table 1. Strains were grown in Lysogeny broth (LB) at 37 • C for optimal growth and at 25 • C for suboptimal growth. The antibiotics chloramphenicol (34 µg/ml), gentamicin (15 µg/ml) and kanamycin (30 µg/ml) were added when required for selection.
Deletion of rluC gene was carried out using the pRed/ET system (Heermann et al., 2008), replacing rluC with gentamicin resistance cassette through sequence-specific homologous recombination. Linear DNA fragments consisting of the homologous arms (50-bp sequences upstream and downstream of rluC), sandwiching the gentamicin antibiotic cassette, were transformed into pRed/ET-harboring E. coli K12 strain (K12WT) using electroporation.
Deletion of genomic bipA and spoT genes was carried out using the pDS132 (Philippe et al., 2004) suicide plasmid. Approximately 500 bp upstream and downstream of bipA and spoT were amplified as the homologous arms. The upstream and downstream arms were combined through restriction site ligation and cloned into pDS132 to obtain pDS132-bipA and pDS132-spoT plasmids. The plasmids were transformed into target cells using electroporation and two rounds of selections. Kanamycin-resistant cells were selected from the first round of selection, and sucrose-sensitive cells were selected from the second round of selection.
Knock-in of lacZ downstream of deaD and obgE was done using pVIK111 (Zheng et al., 2007) suicide vector. In-frame insertion was created by amplifying approximately 500 bp of 3 -end of the target gene, while excluding the stop codon, and cloned into pVIK111 yielding pVIK111-deaD and pVIK111-obgE. The plasmids were then transformed into E. coli K12 strain (K12WT), bipA, bipA (pCA24N-BipA), and rluC/ bipA using electroporation. Kanamycin-resistant clones were selected.

Site-Directed Mutagenesis
Modifications of pCA24N-BipA were done using the QuikChange method to produce pCA24N-BipA H78A , pCA24N-BipA H78Q , and pCA24N-BipA T544_D552del . Primers were designed to have equal lengths of nucleotides extending toward 5 -and 3 -ends from the point of modification. Briefly, the first step was PCR amplification of pCA24N-BipA to produce two single-stranded circular DNAs by carrying out the forward and reverse amplification in separate reactions. Secondly, the samples were mixed in the presence of 1 µl DpnI enzyme (NEB R ), which digested the template DNA. Next was activation at 37 • C for 2.5 h, followed by deactivation at 80 • C for 20 min, and then denaturation at 98 • C for 15 min. Lastly, samples were incubated at room temperature to allow the denatured products to re-anneal. Chloramphenicol-resistant clones were selected for sequence verification.

Growth Assay
Growth curves were determined using a 96-well microplate. The overnight cultures were diluted in LB to OD 600 ≈ 0.02 in a final volume of 100 µl/well. Plates were shaken in a shaker incubator at 200 RPM and temperature of 37 • C for optimal growth and 25 • C for suboptimal growth. The cell density was measured by a TECAN Spark TM 10M multimode microplate reader without microplate lid every half an hour (37 • C) or every hour (25 • C) under the settings of 600-nm light absorbance, 25 flashes, and 120-ms wait time between wells. Growth curves were performed in biological triplicates.

Swimming Motility Assay
Overnight cultures were diluted to OD 600 ≈ 1.0 in 0.9% (w/v) NaCl. Soft LB agar (0.3% w/v) was inoculated with overnight cultures by using a drawing needle to dip into the diluted overnight cultures and then pierced through the soft agar to the middle. The plates were incubated at room temperature for 24 and 48 h. Results were recorded in the form of images and measurements of swimming diameter.

Tandem Mass Tag-Mass Spectrometry-Based Quantitative Proteomics
Tandem mass tag-mass spectrometry (TMT-MS) was carried out as described in a previous work (Park et al., 2019), with slight modifications. K12WT, bipA, bipA (pCA24N-BipA), rluC, and rluC/ bipA were cultured in 200 ml LB at 25 • C in a shaking incubator at 200 RPM to OD 600 ≈ 1.0. The cultures were immediately cooled on ice before centrifugation at 4,500 RPM for 8 min at 4 • C. The resulting cell pellets were resuspended in 1 ml TMT lysis buffer [100 mM tetraethylammonium bromide (TEAB) pH 8.5, 1 mM PMSF, 1% Triton X-100, and EDTA-free protease inhibitor cocktail]. Lysis was done using sonication for 5 min at 40% amplitude with pulses set at 5 s on followed by 5 s off. Lysates were incubated at 4 • C on a rotator for 1 h and then clarified by centrifugation at 15,000 RPM for 45 min at 4 • C. The clarified lysates were filtered through a 0.22-µm spin filter (Corning R Costar R Spin-X R ) and sent to the Proteomic Core Facility of the Biological Research Center of Nanyang Technological University (Singapore) for TMT-MS services. The TMT-MS was done with technical triplicates.

β-Galactosidase Assay
The β-galactosidase assay was performed using 96-well microplates by referring to a reported work (Schaefer et al., 2016). The obgE:: pVIK111 and deaD:: pVIK111 strains were cultured in 10 ml LB at 25 • C with shaking at 200 RPM. When the OD 600 reached the value of 0.2, 0.5, and 1.0, 20 µl of the culture was collected and mixed with 80 µl permeabilization solution (100 mM Na 2 HPO 4 , 20 mM KCl, 2 mM MgSO 4 , 0.04% sodium deoxycholate, 5 mM β-mercaptoethanol, and 1 mg/ml lysozyme) in a 96-well microplate (Nunc TM MicroWell TM , Thermo Scientific, Waltham, MA, United States). Permeabilizing samples were stored at 4 • C until all the samples were prepared, and then, 25 µl of permeabilized samples were mixed with 150 µl of substrate solution (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 1 mg/ml ONPG, and 5 mM β-mercaptoethanol) in a 96-well microplate (Nunc TM MicroWell TM , Thermo Scientific, Waltham, MA, United States) and mixed well before loading the plate without its lid into the TECAN Spark TM 10M multimode microplate reader. The OD 420 absorbance was measured every 5 min for a total of 80 min incubation at 37 • C. The settings for the OD 420 absorbance measurement were 25 flashes and 120-ms wait time between wells. Biological triplicates were analyzed for all strains.
Purification of Pre-50S Particles E. coli bipA strain was cultured in large scale (six flasks of 800 ml LB) at 25 • C with shaking at 200 RPM, until the OD 600 ≈ 0.4-0.5. The cultures were immediately cooled down on ice before centrifugation at 4,000 RPM for 15 min using JLA-8.1000 rotor (Beckman Coulter, Brea, CA, United States). The resulting pellets were pooled and resuspended in 30 ml lysis buffer (20 mM HEPES pH 7.5, 10.5 mM MgOAc, 100 mM NH 4 Cl, 0.5 mM EDTA pH 8.0, 6 mM β-ME, 20 U/ml DNase I, and 1 mg/ml lysozyme) and aliquoted into Eppendorf tubes to go through three rounds of freeze and thaw for cell lysis. The lysates were centrifuged at 14,500 RPM for 30 min at 4 • C, and the supernatants were pooled. The supernatant was diluted using lysis buffer to a concentration of 80 A 260 units in a final volume of 800 µl. Then, 10-25% (w/w) sucrose gradients were formed by mixing sucrose solution with overlay buffer (

Purification of BipA
BipA was overexpressed from the plasmid pNIC28-Bsa-bipA in E. coli BL21 (DE3) and purified through three consecutive chromatography steps. Cells transformed with the plasmid were cultured large scale at 37 • C until the OD 600 ≈ 0.8 and then overexpression was induced by adding isopropyl β-Dthiogalactoside (IPTG) to a final concentration of 125 µM. Culture temperature was reduced to 16 • C upon IPTG induction for overnight incubation. Cells were harvested by centrifugation at 4,000 RPM for 15 min at 4 • C. The resulting cell pellet was resuspended in 100 ml of lysis buffer [50 mM NaH 2 PO 4 pH 8.0, 300 mM NaCl, 10% (v/v) glycerol, and 5 mM β-ME] and lysed using LM20 Microfluidizer R (Microfluidics TM , Newton, MA, United States) with 20,000 PSI pressure. The lysate was clarified by centrifugation at 20,000 RPM for 1 h at 4 • C using a JA-25.50 rotor, filtered through a 0.45-µm filter, and kept on ice before loading onto HisTrap R HP 5 ml Ni 2+ column for affinity chromatography using ÄKTA Purifier (GE Healthcare Life Sciences, Marlborough, MA, United States). The column was then washed with two column volumes (CV) of lysis buffer. The elution was performed by gradually increasing the concentration of His elution buffer [50 mM NaH 2 PO 4 pH 8.0, 300 mM NaCl, 10% (v/v) glycerol, 5 mM β-ME, and 500 mM imidazole). The fractions containing BipA as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) were pooled and diluted with 25 mM Tris pH 8.0 buffer to a final concentration of 50 mM NaCl before loading onto HiTrap R Q Fastflow 5-ml column for anion exchange chromatography using ÄKTA Purifier. The column was then washed with two CVs of Q buffer A (25 mM Tris pH 8.0, 50 mM NaCl, and 5 mM β-ME). The elution was carried out by gradually increasing the concentration of Q buffer B (25 mM Tris pH 8.0, 1 M NaCl, and 5 mM β-ME). The fractions containing BipA as determined by SDS-PAGE were pooled and concentrated to a volume of 5 ml using Amicon R Ultra-15 Centrifugal Filter Units with 30-kDa cut-off membrane (MERCK, Darmstadt, Germany). A concentrated sample was loaded onto HiLoad R 16/60 Superdex R 200-pg column equilibrated with GF buffer (25 mM Tris pH 8.0, 100 mM NaCl, and 5 mM β-ME) for size exclusion chromatography using ÄKTA Explorer. The fractions containing BipA were pooled and concentrated to 13 mg/ml using Amicon R Ultra-15 Centrifugal Filter Units with 10-kDa cut-off membrane (MERCK, Darmstadt, Germany). The concentrated proteins were snap frozen using liquid nitrogen and stored at −80 • C.

Ribosome Binding Assay
For at least 30 min on ice, 50 µl of BipA (13 mg/ml) in GF buffer was incubated with 100 µM GDPCP. For western blotting of polysome profiling fractions, 10 A 260 units of bipA cell lysate was prepared, and BipA pre-incubated with GDPCP was added five times in excess (10 A 260 = 23.9 nM; BipA concentration = 1.195 µM) and incubated on ice for 1.5 h. The sample was layered onto 5%-45% sucrose gradient in overlay buffer (10 mM HEPES pH 7.5, 50 mM KCl, 10 mM NH 4 Cl, 10.25 mM MgOAc, and 0.25 mM EDTA pH 8) in 13.2-ml thin-wall polypropylene tubes (Beckman Coulter, Brea, CA, United States) and centrifuged at 36,000 RPM for 1.5 h (ω 2 t = 7.6746 × 10 10 ) at 4 • C using an SW 41 Ti rotor (Beckman Coulter, Brea, CA, United States). The gradients were fractionated by 10 drops per 1.5-ml Eppendorf tubes. Fractions were precipitated by adding 2.5 times sample volume of icecold 100% ethanol and one-tenth sample volume of 3 M NaOAc and incubation at −20 • C overnight. The precipitated ribosomal samples were recovered by centrifuging the samples at 14,500 RPM for 30 min at 4 • C, then air-dried before resuspending in 10 µl of RNase-free water. Recovered ribosomal samples were then loaded onto 10% polyacrylamide gel followed by semi-dry transfer onto nitrocellulose membrane. Recombinant BipA was detected by western blotting using 1:2,000 of HRPconjugated anti-His 6 antibody (Santa Cruz Biotechnology, Dallas, TX, United States), and then, Clarity ECL western blotting substrates (Bio-Rad, Hercules, CA, United States) was applied before visualization using ChemiDoc TM (Bio-Rad, Hercules, CA, United States). BipA binding was also analyzed by co-pelleting through sucrose cushion. Two separate samples were prepared on ice. Then, 50 µl of BipA (13 mg/ml) in GF buffer was incubated with 100 µM GDPCP for at least 30 min on ice. The sample with BipA pre-incubated with GDPCP only was prepared by mixing 1X Buffer G (5 mM HEPES pH 7.5, 50 mM KCl, 10 mM NH 4 Cl, 10 mM MgOAc, and 6 mM β-ME) and 144 µM of BipA pre-incubated with GDPCP and topped up to 50 µl with RNase-free water. The BipA pre-incubated with GDPCP complex with pre-50S particles was prepared by mixing 1X Buffer G with 144 µM of BipA pre-incubated with GDPCP and 24 OD 260 units (0.576 µM) of pre-50S particles and then topped up to 50 µl with RNase-free water. Then, samples were layered onto 1.1 M sucrose in 1X Buffer G and centrifuged at 45,000 RPM in a TLA-100 rotor for 16 h at 4 • C. Post-centrifugation, 1 µl of supernatant was aliquoted from each sample. The remaining supernatant was discarded, and the pellet was washed with 1X Buffer G three times to remove the sucrose. The pellets were then resuspended in 20 µl of 1X Buffer G. The RNA concentration of sample with BipA pre-incubated with GDPCP complex with pre-50S particles was measured using a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States) and adjusted to 24 OD 260 units using 1X Buffer G if dilution was required. Samples for PAGE were prepared by mixing 1 µl of the sample, 2.5 µl of 4X loading buffer, 1 µl of β-mercaptoethanol, and 5.5 µl of RNase-free water and incubated at 70 • C for 10 min. Then, samples were loaded onto 4-12% NuPAGE Bis-Tris Gel (Invitrogen, Waltham, MA, United States) and run in 1X MES buffer at 200 V.

Deletion of Genomic bipA Gene Affects E. coli Growth at Suboptimal Temperature
While the growth defects of E. coli bipA knock-out strains at suboptimal culturing temperature (mostly at 20 • C) have been reported by several groups (Pfennig and Flower, 2001;Krishnan and Flower, 2008;Choudhury and Flower, 2015;Choi and Hwang, 2018), the phenotype is not well consistent and understood. For a better understanding of the importance of BipA in cold stress conditions, we first determined the growth curves of E. coli K12 strain (K12WT) and its corresponding bipA knock-out strain ( bipA), both harboring the empty pCA24N vector, under optimal (37 • C, Figure 1A) and suboptimal (25 • C, Figure 1B) temperatures. While the two strains demonstrated a similar growth rate at 37 • C (Figure 1A), the bipA strain revealed a notable growth retardation at 25 • C, resulting from a significantly longer lag phase ( Figure 1B). Further corroborating the role of BipA in cold stress is the finding that BipA expression from plasmid pCA24N-BipA could restore the growth of bipA strain, albeit not entirely ( Figure 1B). In line with a previous report (Choudhury and Flower, 2015), RluC deficiency ( rluC) can complement the growth defect of bipA at 25 • C ( Figure 1B).
Next, we sought to investigate the effect of the loss of genes linked to bacterial stress response. Namely, we examined whether the growth defect is exacerbated if alarmone synthetase genes relA and spoT were deleted in the bipA background. Our results showed that the double mutant strains ( relA/ spoT) demonstrated a slight growth defect compared to the wild type, whereas the growth curve was similar to relA/spoT knock-out in the bipA background (triple mutations) (Figure 1C), implying that alarmone level perhaps has little influence on the growth of E. coli at suboptimal temperature.
GTP hydrolysis activity is important for trGTPase turnover on ribosome and in turn its physiological function through conformational change (Kumar et al., 2015;Ero et al., 2016). To study the significance of GTP hydrolysis for BipA functioning under sub-optimal growth temperature, mutations were introduced into the plasmid-borne bipA gene and subsequently transferred to the bipA strain. We mutated the proposed catalytic residue histidine 78 to alanine (H78A) or glutamine (H78Q) to abolish the GTP hydrolysis, respectively, based on structure and sequence comparison of BipA with other trGTPases such as EF-G and EF-Tu (Scarano I, Krab et al., 1995;Gao et al., 2009;Schmeing et al., 2009;Koripella et al., 2015). In addition, the C-terminal loop (CTL) of BipA was truncated (T544_D552del) given that the CTL is believed to be essential for BipA binding with 70S ribosome (deLivron et al., 2009;Kumar et al., 2015). Growth complementation results showed that under an optimal growth condition (37 • C), leaky expression of BipA mutants BipA H78A , BipA H78Q , and BipA T544_D552del would have a negative effect on the growth of bipA strain whereas the native BipA expression had no effect on cell growth ( Figure 1A). In line with the role of GTP hydrolysis essential for BipA turnover, these findings likely suggest that the aforementioned mutations cause BipA turnover defects, resulting in the "trapped" ribosomes leading to decrease in translation and, ultimately, affecting cell growth. By being "trapped, " the ribosomes would be prevented from carrying out its task due to the bound translational factors being unable to dissociate itself from the ribosome, similar to that in which EF-G was trapped by fusidic acid .
Interestingly, the growth defects that vary in magnitude were observed when the bipA strains transformed with diverse bipA mutations were grown at suboptimal temperature of 25 • C ( Figure 1D). Namely, the growth complementation of bipA strain by plasmid-borne BipA was not achieved for BipA H78A , BipA H78Q , and BipA T544_D552del mutants, whereas BipA could completely restore the growth of bipA strain ( Figure 1D). In particular, the expression of BipA H78A and BipA T544_D552del mutants resulted in further notable defects in bipA strain at suboptimal conditions, suggesting that ribosome binding and GTP hydrolysis are crucial for the role of BipA in bacterial growth under cold shock stress.
Taken together, our results demonstrated that the elimination of alarmone synthesis by removing RSH proteins has a minute effect on cell growth of at low temperature, whereas the loss of bipA would cause a significant growth defect, which could be complemented or suppressed by pCA24N-BipA supplementation and genomic rluC deletion, respectively.

Loss of bipA Gene Causes Swimming Motility Defect in E. coli at Suboptimal Temperature
It was recently reported that E. coli with bipA deletion demonstrated motility defects while incubated at 20 • C (Choi and Hwang, 2018). Hence, we would like to examine The cells with BipA H78Q demonstrated the same growth rate as the strain bipA (pCA24N), while preceding with BipA H78A that preceded with BipA T544_D552d el . Note that biological triplicates were analyzed, and pCA24N without gene of interest was transformed into cells not harboring any plasmid so that all strains have the similar cellular burden of holding a plasmid. The "p" in the bracket alone or preceding "BipA" represents plasmid pCA24N.
the swimming motility of our E. coli strains using agar plate assay and incubation at room temperature, with the agar plate images after 24-and 48-h incubation shown (Figures 2A,B). The results of 24-h incubation demonstrated that the swimming motility was severely diminished for bipA strain, and it could be complemented by plasmid harboring native bipA gene, but not the bipA H78A , bipA H78Q , and bipA T544_D552del mutants (Figure 2A). Interestingly, all bipA strains produced chemotactic rings after 48-h incubation ( Figure 2B). Despite that the rings for the BipA mutants are notably smaller than those for the strains expressing BipA, the results clearly showed that swimming motility was not diminished, but rather reduced, for bipA and the bipA mutants. Furthermore, bipA strain expressing BipA H78Q produced significantly larger rings than bipA strain expressing BipA H78A or BipA T544_D552del after 48-h incubations (Figure 2B), which suggests that H78Q substitution retains BipA function to a certain extent, likely the ribosome binding and transition state stabilizing. In line with this hypothesis, bipA (pCA24N-BipA H78A ) produced a significantly larger ring than bipA (pCA24N-BipA T544_D552del ) after 48-h incubation, indicating that the truncation of BipA CTL causes more severe motility defects for E. coli (Figure 2B), and note that BipA CTL is required for BipA binding to ribosome (deLivron et al., 2009;Kumar et al., 2015).
On the other hand, while RluC-deficient ( rluC) strain behaved similarly to E. coli wild-type strain (K12WT), the rluC/ bipA strain demonstrated significant suppression on the swimming defect of bipA (Figures 2A,B); this revealed a functional link between BipA and RluC as previously reported (deLivron et al., 2009).
Agar plate assay alone does not provide sufficient evidence to conclude whether the chemotactic ring represents the swimming motility of the bacteria, given that the possible influence of cell growth on the ring formation cannot be completely ruled out. The rings observed on semi-solid agar might represent bacterial growth instead of swimming motility because all the strains except K12WT and bipA (pCA24N-BipA) did not present the typical chemotactic rings where highly motile populations form an outer layer of the ring (Koster et al., 2012;Cremer et al., 2019;Liu et al., 2019). Furthermore, the cells might be defective in swimming and were tumbling, or the observations were restricted to semi-solid environment (Kinosita et al., 2020). Thus, inverted microscopy was employed to detect the diluted overnight cultures of K12WT, bipA, bipA (pCA24N-BipA), and rluC/ bipA cells (Supplementary Movies 1-4), and these animated movies show that all four strains demonstrated swimming motility in liquid media. Taken together, strains with bipA yielded larger The bipA strain expressing BipA mutants presented significant increase in chemotactic ring size, where the strain bipA (pCA24N-BipA H78Q ) yielded chemotactic ring larger than the strains bipA (pCA24N-BipA H78A and pCA24N-BipA T544_D552d el ). The "p" in the bracket alone or preceding "BipA" represents plasmid pCA24N. chemotactic rings at room temperature, and deletion of bipA caused a significant delay in the appearance of chemotactic rings, which could be complemented by introducing functional BipA or suppressed by genomic deletion of rluC.

The Effect of BipA Mutations on the Defect of Ribosome Assembly
The loss of bipA had been found to cause ribosome assembly defect with the accumulation of pre-50S, which can be alleviated by expressing BipA from a plasmid or genomic deletion of rluC (Krishnan and Flower, 2008). Very recently, BipA has been implicated in ribosome (specifically large subunit) assembly at low temperature growth (Choi and Hwang, 2018;Gibbs et al., 2020). Here, we examine whether the bipA H78A , bipA H78Q , and bipA T544_D552del mutants affect the complementation of ribosome assembly defect in the bipA cells through sucrose gradient sedimentation analysis. As shown in Figure 3A, compared to the wild-type strain K12WT, bipA deficiency resulted in significantly reduced 70S ribosome and 50S subunit populations (Figure 3A), and simultaneously, a minor peak between the 30S and 50S peaks was observed, which is likely representing a population of pre-50S particles and is consistent with the previous results (Choi and Hwang, 2018;Gibbs et al., 2020). Similar to that presented by Gibbs et al. (2020), the pre-50S peak was not observed in the ribosomal particle distribution of bipA expressing exogenous BipA (pCA24N-BipA), suggesting that BipA is involved in ribosome assembly, particularly in the maturation of the 50S subunit ( Figure 3A). On the other hand, bipA strain with rluC genomic deletion ( rluC/ bipA) yielded ribosomal particle distribution without a notable pre-50S peak, which differs from the previous study (Choudhury and Flower, 2015), and concurrent rluC deletion failed to fully compensate the ribosome assembly defect of bipA deficiency ( Figure 3A). Note that all the strains demonstrated a similar profile of polysome level.
Comparisons among the bipA strains expressing pCA24N-BipA mutants revealed very interesting results on the ribosomal particle distribution (Figure 3B). The expression of BipA H78A in the bipA strain resulted in a very similar profile to that of the control ( bipA strain), suggesting that the H78A mutation might have rendered BipA non-functional ( Figure 3B). In contrast, the ribosomal particle distribution of bipA strain with the expression of BipA T544_D552del showed the lowest 70S and 50S peaks of all strains as well as an abnormal 30S peak, demonstrating the importance of intact CTL (Figure 3B). The relatively skewed 30S peak might be an accumulation of heterogeneous population of ribosomal subunits and likely includes the pre-50S particles; therefore, the lower 50S peak perhaps resulted from fewer 50S subunits matured in the strain bipA (pCA24N-BipA T544_D552del ). The expression of BipA H78Q in the bipA strain generated a medium level of compensation for BipA, demonstrated by the absence of pre-50S peak. In addition, the ribosomal particle distribution was significantly different from that of the bipA (pCA24N-BipA H78A ) strain, but similar to that of the rluC/ bipA strain, implying that H78Q substitution of BipA might have retained the function of BipA to a certain degree (Figures 3B,C). As rluC deletion is known to compensate the loss of BipA, the similarity in complementation would therefore support the notion that glutamine is able to partially substitute histidine as the catalytic residue of BipA GTPase activity (Koripella et al., 2015). FIGURE 3 | Ribosomal particle distribution showing ribosome assembly defects caused by bipA deletion and BipA mutants. Peaks corresponding to polysomes, 70S ribosome, and free subunits are indicated. (A) Using K12WT (black) as a reference, bipA (red) presented accumulation of ribosomal subunits and reduction of 50S and 70S ribosomal particles, deduced from the higher 30S peak and lower 50S and 70S peaks, respectively. The pre-50S peak appeared as a minor peak between 30S and 50S peaks in bipA ribosomal particle distribution. Although with a lower 70S peak, the profile of bipA (pCA24N-BipA) (green) was similar to K12WT. The rluC/ bipA (blue) yielded similar 50S and 70S peaks as K12WT, but similar 30S peak as bipA. Notably, the region between the peak of 30S and 50S subunits was slightly elevated indicating maturing pre-50S. (B) The bipA (pCA24N-BipA H78A ) (yellow) yielded almost identical ribosomal particle distribution as bipA. While bipA (pCA24N-BipA H78Q ) (brown) had lesser 70S ribosome, the pre-50S peak was not visible. The bipA (pCA24N-BipA T544_D552d el ) (purple) produced the least 70S and 50S particles with a skewed 30S peak, which may include a large population of pre-50S particles. (C) A comparison between sucrose gradient profiles of the strains rluC/ bipA and bipA (pCA24N-BipA H78Q ). The data showed similarity in terms of reduced pre-50S with elevated area under the peak between 30S and 50S peaks, likely representing a population of pre-50S that had mature further than what was seen in bipA. The "p" in the bracket alone or preceding "BipA" represents plasmid pCA24N. Peaks corresponding to subunits (30S, pre-50S, and 50S), monosomes (70S), and polysomes are indicated. Top and bottom of each gradient are marked with arrows.
Taken together, ribosome assembly defect caused by the loss of endogenous BipA could be partially complemented by introducing BipA and the mutant BipA H78Q or suppressed by genomic rluC deletion, but not the BipA H78A and BipA T544_D552del , demonstrating varied and complicated effects of GTP hydrolysis and ribosome binding (e.g., CTL) of BipA in ribosome assembly at low temperature.

Loss of bipA Resulted in Upregulation of Proteins Involved in RNA Metabolism
TMT is an isobaric mass tag-based multiplexed quantitative proteomics method by mass spectrometry (Thompson et al., 2003). Tryptic peptides from different samples are labeled with different isobaric tags for accurate relative quantitation of protein expression across the samples. Using tandem mass spectrometry, proteins can be identified by the fragment ions of peptides, and their expression levels quantitated with reported ion intensities. Next, we sought to further investigate the effect of BipA on protein expression and the rationale behind the suppression of bipA phenotypes by genomic deletion of rluC for cells under suboptimal temperature with TMT approach. Volcano plots of the TMT proteomic datasets were used to determine significant changes in protein expression between different conditions. The cut-off for fold change (FC) with statistical significance (p < 0.05) was determined to be 1.5 (log 2 abundance ratio = 0.585). Differentially expressed proteins were shortlisted for further data analysis and interpretation (Supplementary Tables 2-7). The data was reliable as validated by the protein expression level of the deleted genes.
As compared to wild-type strain K12WT, several proteins demonstrated higher expression level in the bipA strain in response to cold stress ( Figure 4A). Particular attention was drawn to two proteins DeaD and ObgE given that both have been implicated in 50S subunit biogenesis (Charollais et al., 2004;Sato et al., 2005). Other proteins, with significantly higher expression level in bipA but not implicated in ribosome assembly, include CspA, RNase R, and RpoS. In contrast, by introducing BipA (pCA24N-BipA) to the bipA strain, the expression levels of DeaD, ObgE, CspA, RNase R, and RpoS become similar to those in the strain K12WT ( Figure 4B). Furthermore, a comparison of expression levels of these five proteins in the bipA (pCA24N-BipA) and bipA strains showed lower levels for the former (Figure 4C). In addition, the proteins relevant to cell motility were found significantly upregulated while the bipA strains express exogenous BipA (pCA24N-BipA) (Figures 4B,C and  Supplementary Tables 3, 4), suggesting that BipA has a direct or indirect influence on bacterial motility. Notably, upregulation of these proteins also rationalized our motility assay for the role of BipA, as observed in Figure 2.
Furthermore, the expression levels of these proteins (except for RpoS) in the rluC/ bipA double mutant strain were similar to those in the strain K12WT, within cut-off value for different expression level (Figure 4D). In the case of RpoS, it was significantly upregulated in the rluC strain as compared to the strain K12WT (Figure 4E), implying that rluC genomic deletion would affect RpoS expression and thereby leading to an additional effect in the rluC/ bipA strain. A comparison of protein expression level of the rluC/ bipA strain with that of the bipA strain also revealed significantly lower levels of DeaD, ObgE, and CspA ( Figure 4F). Notably, the expression level of RNase R in the rluC/ bipA strain was about 0.7 times lower than that in the bipA strain with high statistical significance (log 2 abundance ratio = 0.432), indicating that RNase R was indeed downregulated in rluC/ bipA, but did not meet the FC cut-off. These observations together demonstrated that the upregulation of the aforementioned proteins likely was ascribed to compensation of the bipA loss in E. coli when cultured at suboptimal temperature.
Interestingly, the upregulation of GrcA, a stress-induced alternate pyruvate formate-lyase subunit, was observed in rluC strains. The mRNA of GrcA can be cleaved by MazF, leading to leaderless mRNA with anti-Shine-Dalgarno sequence removed; therefore, the resultant mRNA is favorably translated by a ribosome (Vesper et al., 2011). The significance of GrcA in bacterial stress and cold shock response remains poorly understood, but it is of interest for further study.

Ribosome Maturation Factor DeaD Is Upregulated in bipA Strain at Suboptimal Growth Temperature
To further validate the upregulation of 50S biogenesis factors detected by TMT-MS, β-galactosidase reporter assay was employed. The lacZ gene was inserted downstream of the target gene in the genome, and its activity (β-galactosidase activity) could be easily measured. Therefore, a fusion design (target gene-lacZ) with the stop codon of the target gene excluded can be used to assess the expression level of this target protein based on the output of the β-galactosidase (LacZ) activity. In our case, the β-galactosidase activity was used to reflect the expression level of the fused DeaD (DeaD-LacZ) protein, which would be consequently used to validate our observation by TMT-MS.
Upon the bipA deletion ( bipA), the β-galactosidase activity was dramatically increased for all tests at OD 600 ∼0.2, ∼0.5, and ∼1.0, demonstrating DeaD was highly expressed (Figures 5A-C). However, with further deletion of the gene rluC ( rluC/ bipA), the elevated β-galactosidase activity was reduced, almost to the level of the wild-type strain K12WT. Similarly, exogenous expression of BipA in its deletion strain ( bipA + pCA24N-BipA) was also able to complement the effect of bipA deletion on DeaD expression. Collectively, DeaD-LacZ activities for the three strains (K12WT, bipA + pCA24N-BipA, and rluC/ bipA) show similar time course, but lower than that for the bipA deletion strain ( bipA), suggesting that upregulation of DeaD was triggered by bipA deletion, which can be complemented or suppressed by overexpressing BipA or rluC genomic deletion, respectively. These data are in line with the expression changes in DeaD detected by TMT-MS and are in support of upregulated RNA helicase DeaD, which likely plays an important role in ribosome assembly at low temperature upon the loss of bipA. However, the precise and detailed mechanism remains unknown.

BipA Binds Pre-50S Ribosomal Subunit and 70S Ribosome in vitro
We next tested whether BipA can bind pre-50S particle for its proposed function as 50S maturation factor through in vitro reconstitution. First, we mixed the purified BipA, which was pre-incubated with GDPCP with clarified lysate of the bipA bipA. We put DeaD, ObgE, RpoS, CspA, and RNase R in the negative log 2 abundance ratio side of the plot, meaning peptide reads of the proteins were lesser in the presence of pCA24N-BipA. The blue dot labeled as BipA was an indication that BipA was indeed expressed. (D) The rluC deletion in bipA background produced a volcano plot similar to (A), but the DeaD, ObgE, CspA, and RNase R were close to K12WT as differential expressions against K12WT were not detected. As in (A,D), higher readout of RpoS was detected in rluC/ bipA. Red dots labeled with BipA and RluC showed that these bipA and rluC were indeed deleted. (E) Genomic deletion of rluC only yielded six differentially expressed proteins relative to K12WT, and out of which, two interesting changes were RpoS and GrcA upregulations. The significant negative log 2 abundance ratio of RluC indicates that the gene was indeed removed. (F) Comparisons between rluC/ bipA and bipA showed a reduced expression level of the DeaD, ObgE, and CspA to wild-type level. The RNase R readout was found to be reduced by the loss of rluC, close to the cut-off for FC. The pBipA refers to pCA24N-BipA. cells followed by sucrose gradient centrifugation and ribosome fractionation, and subsequently, western blot was employed to detect BipA. Our data clearly showed that BipA is able to co-sediment with both large and small ribosomal units, as well as the pre-50S ribosomal unit ( Figure 6A). BipA could bind to even small ribosomal unit 30S, perhaps through its β-barrel domain FIGURE 5 | β-galactosidase activity assays demonstrated the upregulation of DeaD in the strain bipA. The cells were harvested at OD 600 ≈ 0.2 (A), 0.5 (B), and 1.0 (C), respectively, and the activity was assessed by measuring the breakdown of ONPG through recording OD 420 . bipA cells (red) harvested at all three OD 600 presented the highest β-galactosidase activity, indicated by the significantly stronger OD 420 absorbance over time as compared to K12WT (black), bipA (pCA24N-BipA) (green), and rluC/ bipA (blue). The pBipA refers to pCA24N-BipA.
II, implying a yet unknown function in bacterial translational machinery ribosome ( Figure 6A). Second, we combined BipA pre-incubated with GDPCP and purified pre-50S population of bipA strain, layered on top of 1.1 M sucrose cushion and performed high-speed centrifuge to find out if BipA would cosediment with pre-50S. As shown in Figure 6B, the results clearly demonstrated that BipA was indeed co-sedimented with pre-50S, corroborating that BipA is capable of binding to the pre-50S ribosomal particle.

DISCUSSION
GTP Hydrolysis and CTL of BipA Are Crucial for 50S Biogenesis at Low Temperature The highly conserved bipA gene has been shown to be significant for bacterial growth and ribosome assembly at suboptimal temperature (Choudhury and Flower, 2015;Ero et al., 2016;Choi and Hwang, 2018;Gibbs et al., 2020). Here, we report that bipA deletion and mutagenesis significantly affect E. coli swimming motility (Figure 2). The experiment was carried out at room temperature (suboptimal for E. coli growth) and revealed that the swimming defect of the bipA strain can be complemented or suppressed by the expression of plasmidborne BipA (pCA24N-BipA) or the concurrent deletion of rluC gene ( rluC/ bipA), respectively (Figure 2). The possible reason for this phenotype might be the decelerated global translation due to retardation of ribosome biogenesis, which subsequently shuts down energy-costly pathways, one of which involves the cell motility (Ottemann and Miller, 1997;Martínez-García et al., 2014). The observations of agar plate assay for these mutations correlate with the change of growth trend that can be seen in growth curves (Figure 1) as well as with ribosomal particle distribution observed in Figure 3, suggesting that the loss of bipA leads to a condition affecting growth, motility, ribosome assembly, and protein translation. Indeed, the ribosome assembly defect likely slows down protein translation as the concentrations of 70S ribosomes and polysomes are decreased (Peil et al., 2008;Gibbs et al., 2020). This retardation in protein translation then negatively affects the downstream processes including bacterial growth and motility (Rudra and Warner, 2004).
In this work, we showed that the phenotypes presented by the bipA deletion ( bipA) could be complemented by pCA24N-BipA or suppressed by rluC genomic deletion, suggesting that BipA is important for E. coli to thrive at suboptimal temperature, and pseudouridylation on the correct nucleotides (such as RluC involving U955, U2504, and U2580 of 23S rRNA) can be helpful for ribosome biogenesis in E. coli and therefore growth as well as motility. While there is no reported relation to bacterial motility, Choudhury and Flower (2015) proposed that suboptimal temperature caused the ribosome to be dependent on BipA for efficient assembly, and loss of 23S rRNA modification by RluC actually led to an alternative folding pathway that is BipA independent. This further suggests that the improvement of bipA swimming is perhaps not motility regulation related. The unexpected appearance of the chemotactic rings for the strains bipA and bipA with plasmid-expressing BipA mutants after 48-h incubation suggests that the loss of bipA did not diminish swimming motility. Instead, the swimming was delayed probably due to the decreased rate of protein translation and cell growth caused by ribosome assembly defect. This is supported by the microscopy visualization of the bacteria cells grown overnight in liquid culture demonstrating their swimming motility (Supplementary Movies 1-4).
In this study, the plasmid-borne BipA, with H78A or H78Q mutation (catalytic residue H78) or CTL truncation (T544_D552del), was transformed into the bipA strain to examine its effect on complementation for BipA, respectively. The rationale behind the mutations was based on previous mutagenesis studies on catalytic histidine of EF-Tu and EF-G (H84 and H91, respectively) and structural study where the CTL interacts directly with the A-loop of 50S subunit (Supplementary  Figure 11). The H84Q substitution in EF-Tu showed a reduction FIGURE 6 | In vitro binding assays showed that BipA pre-incubated with 5X excess GDPCP bound to various ribosomal particles including pre-50S particles. (A) Ribosomal particle distribution of in vitro reconstitution of BipA-GDPCP with ribosomal particles. By analyzing the fractions from ribosomal particle distribution using western blot, most of the BipA pre-incubated with GDPCP were detected in the junk fractions before 30S peak, and the band intensity decreased from 30S toward polysomes. Peaks corresponding to subunits (30S, pre-50S, and 50S), monosomes (70S), and polysomes are indicated. Top and bottom of each gradient are marked with arrows. (B) SDS-PAGE check of ribosome fractions from (A). BipA interaction with pre-50S was observed based on the presence of co-pelleting through sucrose cushion using ultracentrifugation. The band representing co-pelleted BipA is indicated by the black arrow in the "Complex pellet" lane, which consists of pelleted BipA pre-incubated with GDPCP-bound ribosome complex, suggesting that binding occurs. of GTP hydrolysis by 35%, and H84A abolished the GTPase activity (Scarano I, Krab et al., 1995), while the activity of ribosome-associated GTP hydrolysis of EF-G with H91Q substitution was comparable to native protein, and it was found to be defective in organic phosphate release (Koripella et al., 2015). The finding that the expression of BipA H78A did not restore the ribosomal particle distribution as did the wild-type BipA was also observed in a recently published study where the expression of BipA H78A led to slow growth for both WT and bipA strains (Gibbs et al., 2020). On the other hand, Gibbs et al. (2020) expressed suspicion that BipA H78A is able to bind 70S ribosome but is unable to facilitate GTP hydrolysis as well as the subsequent factor release, which ultimately led to halt in translation. The BipA H78Q presented interesting outcomes that could be summarized in three points: (1) similar growth curve as bipA at suboptimal temperature despite longer lag phase at optimum condition (Figures 1C,D); (2) shorter lag phase than bipA (pCA24N-BipA H78A ) and bipA (pCA24N-BipA T544_D552del ) at suboptimal temperature ( Figure 1D); (3) similar ribosomal particle distribution as rluC/ bipA, albeit with significantly fewer 70S ribosomal particles ( Figure 3C). Based on the aforementioned examples of H84Q and H91Q, glutamine substitution might have rendered BipA less efficient in GTP hydrolysis and/or organic phosphate release (Scarano I, Krab et al., 1995;Koripella et al., 2015), leading to the retention on 70S ribosome for longer period of time and negatively impacting the assembly efficiency. Interestingly, BipA T544_D552del seemed to have an inhibitory effect on E. coli growth and ribosome assembly at suboptimal temperature as deduced from the exacerbated growth defect as compared with the strain bipA ( Figure 1D) and lesser 50S and 70S observed in ribosomal particle distribution ( Figure 3B). Kumar et al. (2015) reported that the region L543-E553 of CTL projects deep into the peptidyl transferase center (PTC), while the region N536-K542 is critical for ribosome binding. Therefore, the truncation of T544-D552 could potentially play a role in BipA association and function by establishing direct contact with the A-loop of 23S rRNA (Supplementary Figure 11). This corroborates with the study by deLivron et al. (2009) revealing that alanine substitutions of amino acids within the CTL hinder the binding of BipA to the 70S ribosome.
Collectively, growth curves, ribosomal particle distribution, and swimming motility assays demonstrated a correlation whereby the defect in ribosome assembly at suboptimal temperature leads to a delay in growth and swimming motility (Pfennig and Flower, 2001;Choudhury and Flower, 2015;Choi and Hwang, 2018). Perhaps, BipA is a ribosome biogenesis factor that is crucial for ribosome biogenesis at suboptimal temperature, as a previously reported role for BipA is incorporating the ribosomal protein L6 into the 50S ribosome (Choi and Hwang, 2018). In the recent study by Gibbs et al. (2020), structural block 3 of 50S ribosomal subunit demonstrated a growth conditiondependent assembly, including suboptimal temperature. While it was not evident that BipA has a direct role in delaying block 3 folding, the loss of bipA led to the accumulation of pre-50S without ribosomal protein L17, an r-protein associated with block 3 (Gibbs et al., 2020).

Loss of bipA Leads to Upregulation of RNA Metabolism
TMT-MS revealed a significant upregulation of a number of proteins involved in RNA metabolism and chaperoning in the bipA strain, but with expression comparable to the wild-type level in the strains bipA (pCA24N-BipA) and rluC/ bipA. Contrary to pCA24N-BipA complementation, the rluC/ bipA strain presented an expression of majority of the proteins at above wild-type level while it suppressed bipA phenotypes. Furthermore, the findings from TMT-MS seem to correlate with growth curve, ribosomal particle distribution, and swimming motility assays in a sense that significant changes observed in bipA were found to be reversed in complementation strains. Furthermore, the proteins with significant changes in expression in the bipA strains were involved in ribosome assembly, stress response, and growth. In particular, the expression of DeaD, ObgE, CspA, and RNase R was increased in the bipA strain, but decreased in the complementation strains, indicating their involvement in the phenotypes of the bipA strain. On the other hand, the expression of RpoS was higher in the absence of BipA expression as compared to K12WT, including the rluC/ bipA strain. This observation can be explained as the influence of rluC deletion given that the RpoS in the rluC strain was presented BipA GTPase 50S Ten out of 24 proteins have previously been reported to play a role in cold sensitivity when deleted or mutated (bold).
higher than the wild-type level, and thus, an additive effect of RpoS reads would be observed in the strain rluC/ bipA. To date, no relationship between RluC and RpoS has been reported according to our knowledge, and it is an intriguing topic for further studies. Note that an increased sensitivity to several antibiotics caused by the inactivation of rluC has been reported (Murakami et al., 2005;Toh and Mankin, 2008). This is not surprising as the pseudouridines synthesized by RluC are situated in PTC (U955, U2504, and U2580) where a number of antibiotics bind; hence, the rluC deletion increased the antibiotic susceptibility, and rpoS could be upregulated as a stress response in the rluC strain (Conrad et al., 1998). Out of the five aforementioned proteins, DeaD and ObgE have been implicated in the biogenesis of ribosome large subunit, suggesting their upregulations may have a functional relationship with BipA in vivo. Similar to BipA, DeaD has been found to be associated with pre-50S particles (Charollais et al., 2004). DeaD has a role in rRNA structural rearrangement using its helicase activity, which aids in 50S biogenesis at low temperature (Charollais et al., 2004). Charollais et al. (2004) also demonstrated that the overexpression of DeaD was able to complement the growth defect of srmB at low temperature. The finding that the DEAD-box RNA helicase SrmB facilitates 50S biogenesis during early maturation phase (Charollais et al., 2003) suggests an overlap of functions between DeaD and SrmB (Charollais et al., 2004). Such overlapping functions could be possible with BipA and DeaD as well, since both strains deaD and bipA showed accumulation of pre-50S, and deaD/ bipA double knock-out produced an additive effect (Choudhury and Flower, 2015). From the study by Kitahara and Suzuki (2009), DeaD was shown to be likely to contribute to efficient assembly of circularly permuted rRNAs, implying that the cell would lose the ability to assemble rRNAs with scrambled domains in the absence of DeaD. This finding could be linked to a recent hypothesis on "limited parallel processing" of rRNA and serve as an explanation for the upregulation of DeaD upon the bipA deletion ( bipA), which is likely to be a strategy for E. coli to cope with the loss of bipA at suboptimal temperature (Davis et al., 2016). Davis et al. (2016) found that ribosome assembly is dynamic after observing pre-50S from L17-deficient cells mature into the 50S albeit at a slower rate, which the author considered a "limited parallel processing" (Davis et al., 2016). Such behavior confers the bacteria the flexibility in ribosome assembly should there be any factors that are undesirable for ribosome assembly. In addition, Davis et al. (2016) also found that 23S rRNA matures in the form of cooperative folding blocks, whereby different regions of 23S rRNA mature independently in parallel and come into contact with each other to form the tertiary structure. Collectively, their results indicate that ribosome maturation and assembly can occur in multiple pathways, demonstrating the flexibility of bacterial ribosome assembly when there is a bottleneck caused by the shortage of assembly factors or r-proteins.
The importance of RNA secondary structure destabilization at low temperature was demonstrated by Awano et al. (2007), where they presented the growth defect of deaD at 15 • C being complemented by the overexpression of CspA. The CspA is an RNA chaperone that is able to destabilize RNA secondary structure at low temperature with low substrate specificity (Phadtare, 2011). In addition, RNase R, the only 3 -5 exonuclease in E. coli, is also able to complement the cold sensitivity of deaD, indicating their tight relationship during cold shock (Awano et al., 2010). The RNase R consists of a helicase and an RNase domain that function independently, and its mutant with only helicase activity was able to complement the cold sensitivity of deaD, revealing that the helicase activity was the key during cold shock (Awano et al., 2007). Notably, upregulation of RNase R by at least sevenfold is observed during cold shock in E. coli (Cairrão et al., 2003;Guan et al., 2005). Hence, the upregulation of DeaD, CspA, and RNase R in bipA is probably an attempt by the cell to compensate the loss of BipA during 50S biogenesis in which BipA plays a yet unknown role.
In addition to the evidence presented in our work for BipA, the loss of various ribosome assembly factors results in the development of the common phenotype of cold sensitivity coupled with ribosome assembly defect. As summarized in Table 2, 10 out of 24 proteins have previously been reported to play a role in cold sensitivity when deleted or mutated (bold), and several others demonstrated involvement in cold shock. As shown in this study, the deletion of RluC suppresses the cold sensitivity of bipA; the overexpression and deletion of RhlE complement the cold sensitivity of deaD and srmB, respectively (Jain, 2008); and LepA is released from the membrane during stress, including cold shock (Pech et al., 2011). Other than these, a recent report revealed that r-protein L6, RplF, an essential protein involved in ribosome biogenesis at low temperature, was missing from the pre-50S fraction of E. coli in the absence of bipA (Bosl and Böck, 1981;Shigeno et al., 2016;Choi and Hwang, 2018). These observations strengthen the connections among cold sensitivity, RNA metabolic process, ribosome biogenesis, and ribosome assembly factors and together support BipA as a bona fide ribosome assembly factor. In addition to the structure of BipA bound to 70S ribosome, it would be of interest to characterize the structure of BipA in complex with the intermediate state of ribosome (like pre-50S in Figure 3) during its biogenesis, which could offer atomic insights into how BipA facilitates ribosome assembly.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
Y-GG directed the project. KJG performed most of the experiments. RE, J-EP, and BK participated in some of the experiments. KJG, X-FY, JZ, SKS, and Y-GG analyzed the data. KJG, RE, X-FY, and Y-GG wrote the manuscript with some input from all other co-authors.