A conserved protein inhibitor brings under check the activity of RNase E in cyanobacteria

Abstract The bacterial ribonuclease RNase E plays a key role in RNA metabolism. Yet, with a large substrate spectrum and poor substrate specificity, its activity must be well controlled under different conditions. Only a few regulators of RNase E are known, limiting our understanding on posttranscriptional regulatory mechanisms in bacteria. Here we show that, RebA, a protein universally present in cyanobacteria, interacts with RNase E in the cyanobacterium Anabaena PCC 7120. Distinct from those known regulators of RNase E, RebA interacts with the catalytic region of RNase E, and suppresses the cleavage activities of RNase E for all tested substrates. Consistent with the inhibitory function of RebA on RNase E, depletion of RNase E and overproduction of RebA caused formation of elongated cells, whereas the absence of RebA and overproduction of RNase E resulted in a shorter-cell phenotype. We further showed that the morphological changes caused by altered levels of RNase E or RebA are dependent on their physical interaction. The action of RebA represents a new mechanism, potentially conserved in cyanobacteria, for RNase E regulation. Our findings provide insights into the regulation and the function of RNase E, and demonstrate the importance of balanced RNA metabolism in bacteria.


Introduction
RNA degradation is one of the essential processes for RNA metabolism in all organisms.In bacteria, mRNAs usually have an average lifetime of only a few minutes (1)(2)(3)(4) .The rapid degradation of mRNA molecules serves as an important mechanism for rapid adjustment of protein levels in response to the changing environmental conditions.In Esc heric hia coli and many other bacteria, the endoribonuclease RNase E plays a central role in mRNA degradation; it also participates in the maturation of ribosomal RNAs, tRNAs, as well as many small regulatory RNAs (5)(6)(7) .RNase E homologs from most bacteria share a very similar overall architecture, with a conserved catalytic N-terminal half and a divergent noncatalytic C-terminal half ( 8 ) .RNase E from E. coli functions as a tetramer.Its N-terminal half ( 1-529 aa ) includes several subdomains: an RNase H-like domain, an S1 domain that contributes to substrate binding, a 5 monophosphate ( 5 p ) sensing domain ( 5 sensor domain ) that binds to the 5 p of monophosphorylated RNA substrates, a DNase I domain that harbors the RNA hydrolyzing center, a Zn-link, and a small folded domain.Both the Zn-link and the small folded domain are involved in tetramerization.The C-terminal half ( 530-1061 aa ) , though largely disordered, also contains several functional motifs, including two RNA binding sites, and several protein recognition sites that can recruit the exoribonuclease PNPase, the RNA helicase RhlB, and the glycolytic enzyme enolase to form the RNA degradation complex called RNA degradosome (9)(10)(11)(12) .
RNase E cuts single-stranded RNA substrates preferentially at AT-rich sites with only modest sequence specificity (13)(14)(15)(16)(17) .At least two potential pathways for substrate recognition by RNase E have been identified: a 5 end-dependent mode that relies on 5 monophosphate recognition (18)(19)(20) , and an internal entry mode by which substrates are cleaved regardless of their 5 end phosphorylation status ( 21 ,22 ) .In the 5 enddependent mode, the 5 p of the substrate binds to a pocket residing within the 5 sensor domain of RNase E. Such a binding mode could facilitate orientation of the substrates to the catalytic center and allow RNase E to scan linearly from the 5 end for sites of cleavage ( 18-20 ,23 ) .Inactivating the 5 p binding pocket was shown to affect the processing of many cellular transcripts without leading to severe growth defects ( 19 , 22 , 24 ) .The mechanism for the direct entry mode is less clear, and may involve the recognition of a duplex region in the substrate by several residues from the RNase H-like domain and the small domain ( 20 ,25 ) .
The RNase E encoding gene, rne , is essential in all bacteria investigated.Interestingly, chloroplast RNase E is dispensable for viability in Arabidopsis, though its absence will lead to plastid ribosome deficiency ( 26 ) .RNase E depletion in E. coli increased the average half-life of bulk mRNA from about 2.5 min to over 10 min ( 27 ,28 ) .RNase E, together with RNase III, accounted for the initiation of decay of ∼72% of the E. coli transcriptome ( 29 ) .Since RNase E has such a global impact on RNA metabolism with poor substrate selectivity, its cellular activity needs to be properly controlled.However, only a few mechanisms of RNase E regulation were identified in a limited number of bacterial species.For instance, E. coli RNase E maintains the abundance of its own transcripts at a proper level by cleaving at its own 5 -UTR region ( 30 ) , and so does Synechocystis RNase E ( 31 ) .Additionally, several proteins regulate the activity of RNase E by physical interactions.For example, the regulator of ribonuclease activity A ( RraA ) binds to the two RNA binding sites of the noncatalytic region of RNase E, and interferes with the substrate binding activity of RNase E ( 32 ,33 ) .The E. coli ribosomal protein L4 could interact with different sites of the C-terminal region of RNase E and inhibit RNase E activity on the artificial substrate LU13 in vitro , but it seems to only affect the degradation of certain mRNAs involved in stress-response in vivo ( 34 ) .Dip, a protein from the giant phage f KZ, could inhibit the activity of RNase E of its host Pseudomonas aeruginosa by binding to the RNA binding sites in the noncatalytic region ( 35 ) .Recently, a bacterial cell wall peptidoglycan hydrolase, AmiC, was shown to stimulate the activity of RNase E, potentially by enhancing RNase E multimerization ( 36 ) .It is yet unknown if any of these regulation mechanisms exist in the other RNase E-containing organisms.
Cyanobacteria, capable of oxygen-evolving photosynthesis and evolutionarily related to chloroplasts of higher plants, are a unique phylum of gram-negative prokaryotes with great morphological diversity ( 37 ) .An rne gene is present in each sequenced cyanobacterial genome, and attempts to inactivate the rne gene in several cyanobacterial species all failed (38)(39)(40) , implying that this endoribonuclease has an essential role in cyanobacteria.Cyanobacterial RNase E proteins also consist of a conserved catalytic domain at the N-terminal and an intrinsically disordered non-catalytic region at the C-terminal ( 41 ) .The catalytic region is similar to that of E. coli but lacks the small folded domain, while the noncatalytic region shows no detectable similarity.The non-catalytic region shows no detectable similarity to that of E. coli RNase E, but contains several subregions well conserved in cyanobacteria.In the filamentous cyanobacterium Anabaena PCC 7120 ( Anabaena hereafter ) , RNase E can interact with the exoribonuclease PN-Pase, the exoribonuclease RNase II, and the RNA helicase CrhB, suggesting the presence of an RNase E-based RNA degradosome in this organism (41)(42)(43) .
Cyanobacterial RNase E has catalytic properties and substrate preference similar to its counterparts from other bacteria in vitro , and acts on a large panel of RNA species ( 17 , 38 , 41 , 44 , 45 ) .The transcriptome-wide cleavage sites of RNase E were mapped in the unicellular cyanobacterium Synechocystis sp.PCC 6803, revealing a consensus site featured with adenine residues at positions -4 and -3 upstream and uridine residues immediately downstream of the cleavage site, especially at position + 2 ( 17 ) .Synechocystis RNase E has been shown to participate in regulating various cellular activities, such as photosynthesis, plasmid replication, and the maturation of crRNAs and tRNAs ( 17 , 24 , 38 , 46 , 47 ) .Nevertheless, how the activity of RNA degradation is controlled in cyanobacteria remains little understood.Here, we report the identification of RebA, previously annotated as a protein of unknown function, by co-immunoprecipitation with RNase E from the cell lysate of Anabaena .We show that this protein binds to the catalytic domain of RNase E and inhibits its binding and cleavage activity on various RNA substrates, and its action on RNase E plays a role in cell morphology control.

Strains and culture conditions
All strains used in this study are listed in Supplementary Table S1.
The cyanobacterial strains were constructed by transferring the relevant plasmids into the wild type strain of Anabaena or Synec hococcus elong atus PCC 7942 by conjugation as previously described ( 48 ,49 ) .Construction of markerless strains using the CRISPR / Cpf1-based genome editing system was according to our previously developed method ( 50 ) .The genotypes of all obtained strains were verified by PCR.
The wild type strain of Anabaena and its derivatives were grown in the mineral medium BG11 under continuous illumination ( 30 μmol m −2 s −1 ) at 30 • C with the shaking speed of 180 rpm.When needed, neomycin ( 50 μg ml −1 ) , spectinomycin ( 5 μg ml −1 ) , or streptomycin ( 2.5 μg ml −1 ) was added into the media.The rne conditional mutant strain CTrne , in which the expression of RNase E was controlled by the artificial CT promoter ( 42 ) , was maintained in BG11 supplemented with 0.3 μM CuSO 4 and 1 mM theophylline.To perturb cellular RNase E levels, the strain CTrne was pre-cultured in BG11 medium supplemented with 0.3 μM CuSO 4 and 1 mM theophylline until reaching an OD 750 of 0.5.Subsequently, the cells underwent three washes using BG11 medium without copper.After the washes, the cells were then transferred into two different culture conditions, both with an initial OD 750 of 0.3, BG11 medium without copper for RNase E depletion and BG11 medium supplemented with 3 μM CuSO4 and 2 mM theophylline for RNase E overexpression.The RNase HII overexpression strain of Anabaena ( OE-RnhB ) and the RebA overexpression strain of Synec hococcus elong atus PCC 7942 ( OE-SynRebA ) were cultured BG11 medium without copper.Induction of Anabaena RNase HII or Synechococcus RebA was achieved by supplementing the cultures with 3 μM CuSO 4 and 2 mM theophylline.

Construction of plasmids
All plasmids and related oligonucleotides are listed and additionally described in Supplementary Tables S2 and S3, respectively.

Construction of recombinant protein expression plasmids
The vectors pET28a ( Invitrogen ) , pHStag ( GenBank accession number: MK948096; 42 ) and pNStrep ( GenBank accession number: OP902607 ) were used to construct recombinant protein expression plasmids.The plasmid for expressing the full-length RebA protein bearing an N-terminal Strep tag ( pNStrep-RebA ) , which has a codon-optimized ORF of rebA cloned in the vector pNStrep, was synthesized by GenScript Biotech Corporation.The plasmids for expressing the RebA variants with single residue mutation were constructed by site-directed mutagenesis using pNStrep-RebA as the parent plasmid.The plasmids expressing the RebA_C and RebA_DC were constructed by amplifying the rebA regions from Anabaena genomic DNA with the primer pairs Pall1338F901 / Pall1338R1269 and Pall1338F1 / Pall1338R960 and cloning them into pHSTag via the NdeI-XhoI site, respectively.The plasmid for expressing the catalytic region of E. coli RNase E ( 1-529 aa ) was constructed by assembling the insert fragment amplified from the E. coli genomic DNA with the primer pair Pb1084F1 / Pb1084R1587 and the vector fragment amplified from pET28a with PV_1 / PV_2 via seamless cloning.To construct the plasmid for expressing RneN396 ( the catalytic region containing the first 396 residues of RNase E ) , the gene fragment was amplified from Anabaena genomic DNA with the primer pair Palr4331F1 / Palr4331R1188 was cloned into pET28a via the NdeI and XhoI sites.The plasmids pHT Alr4331, pHT Alr4331N412, pHT Alr4331C, which were used to express the full-length of RNase E, RneN412 ( the catalytic region containing the first 412 residues of RNase E ) and Rne_C ( the non-catalytic region of RNase E ) , respectively, were described previously ( 41 ) .The plasmids for expressing the RneN412 variants with single residue mutation were constructed by site-directed mutagenesis using pHTAlr4331N412 as the parent plasmid.

Construction of cyanobacterial genome editing plasmids
The CRISPR / Cpf1-based genome editing vectors pCpf1b and pCpf1b-Sp were used to create the plasmids for making genome modified Anabaena strains, following the method-ology described previously ( 50 ) .The plasmid pICTrne was used to construct a conditional mutant of rne in Anabaena , in which the native promoter of the rne gene was replaced with the artificial and inducible CT promoter.To construct this plasmid, we first made an intermediate plasmid by inserting a fragment of spacer sequence ( prepared by annealing the oligonucleotide pair cr_alr4331F26mF / cr_alr4331F26mR ) into pCpf1b-sp through the two AarI sites.This plasmid was then linearized with BamHI and BglII.At the same time, two rne regions were amplified, respectively, from Anabaena genomic DNA with Palr4331F1596m / Palr4331R556m and Palr4331F1b / Palr4331R990, and the CT promoter was amplified from the plasmid pCT ( 42 ) using the primer pair PcoquwF / PV_19.The four fragments were assembled into pICTrne via seamless cloning.The plasmids pCpf1b-rebA and pCpf1b-rnhB were used to create the markerless deletion strains for rebA ( gene id: all1338 ) and rnhB ( gene id: alr4332 ) , respectively.To construct pCpf1b-rebA , the BamHI-and BglII-linearized vector pCpf1b-Sp and two rebA regions amplified from Anabaena genomic DNA with the primer pairs Pall1338F1005m / Pall1338R11 and Pall1338F1247a / Pall1338R2225 were assembled into an intermediate plasmid via seamless cloning.Then this intermediate plasmid was digested with AarI, and ligated with the spacer fragment generated by annealing the oligonucleotide pair cr_all1338R1130F / cr_all1338R1130R, resulting in pCpf1b-rebA .The plasmid pCpf1b-rnhB was created similarly but with different oligoribonucleotides: rnhB regions were amplified with Palr4332F797m / Palr4332R1m and Palr4332F679 / Palr4332R1464, and the spacer fragment was prepared with cr_alr4332R379F / cr_alr4332R379R.

Synechococcus elongatus PCC 7942
The plasmid pPrbcL-RebA, which was used to complement the rebA mutant and to overexpress RebA protein in Anabaena , was constructed by assembling the vector fragment amplified from pCT with the primer pair P25TNotI-F / P25TBamHI-R, the promoter region of the rbcL gene amplified with PPrbcL-F / PPrbcL-R and the rebA ORF amplified from the plasmid pNStrep-RebA with Pall1338bF1c / Pall1338bR1272c via seamless cloning.The plasmids for overexpressing the RebA variants in Anabaena , which were based on pPrbcL-RebA, were constructed by site-directed mutagenesis.The plasmid pCT-RnhB, which was utilized for the overexpression of RNase HII ( encoded by rnhB ) in Anabaena , was constructed by assembling the vector fragment, amplified from pCT using the primer pair PV_19 / PV_26, with the ORF region of rnhB , amplified from Anabaena genomic DNA using Palr4332F1m / Palr4332R675.Similarly, the plasmid pCT-SynRebA, employed for the overexpression of the RebA homolog ( encoded by Synpcc7942_0395 ) in Synechococcus elongatus PCC 7942, was constructed as pCT-RnhB, except that the ORF for Synechococcus RebA was amplified using Psyn7942-rebAF1 / Psyn7942-rebAR1122.

Construction of plasmids for bacterial tw o-h ybrid assays
The bacterial adenylate cyclase two-hybrid plasmids were constructed using the vectors pKT25a ( GenBank accession number: OQ032554 ) and pUT18Ca ( GenBank accession number: OP902608 ) , which were modified from pKT25 and pUT18C ( 51 ) , respectively.pKT25a and pUT18Ca have identical cloning site, thus allowing the same fragment to be cloned into both vectors.All the two hybrid plasmids were constructed by seamlessly cloning as previously described ( 42 ) .
All the constructed plasmids were verified by DNA sequencing.

Bacterial two-hybrid assays
The two-hybrid system of Bacterial Adenylate Cyclase Two-Hybrid ( BATCH ) was used to test for protein-protein interactions ( 51 ) .Pairs of two-hybrid plasmids were cotransformed into the c y a − strain BTH101.The transformants were grown on LB plates containing 50 μg / ml ampicillin, 25 μg / ml kanamycin, 0.5 mM IPTG ( isopropyl β-d -1-thiogalactopyranoside ) , and 40 μg / ml X-gal ( 5-bromo-4chloro-3-indolylα-d galactopyranoside ) in the dark at 30 • C for 1-3 days for interaction screening.Transformants that contain plasmids encoding interacting proteins are expected to have colonies with blue color.The strain co-transformed with pKT25-zip and pUT18C-zip was used as a positive control ( 51 ) , and that co-transformed with the empty vectors of pKT25a and pUT18Ca was used as a negative control.

Purification of recombinant proteins
E. coli BL21 ( DE3 ) was used as the host strain for recombinant protein expression.Proteins with a His tag was purified with the Ni-NTA resin ( Genscript ) according to the product manual.Proteins with a Strep tag were purified using the Strep-Tactin XT resin ( IBA-Lifesciences ) according to the product manual.When necessary, the proteins eluted from Ni-NTA resin or Strep-Tactin XT resin were further passed through a size-exclusion column to improve the purity.Purified proteins were finally dialyzed into the storage buffer ( 50 mM Tris-HCl, 500 mM NaCl, 2 mM EDTA and 25% glycerol, pH 8.0 ) , and stored at -80 • C.

Synthesis of RNA substrates
The short labelled RNAs 5 p-LU13-FAM ( 5 p-GA GA CA GUA UUUG-F AM ) and 5 OH-LU13-F AM ( 5 OH-GA GA CA GUA UUUG-FAM ) were chemically synthesized by Genscript Biotech Corporation.The 5 triphosphorylated RNAs of 9S RNA, the psbO mRNA and psbAI mRNA were produced by in vitro transcription using the T7 High Yield RNA Transcription Kit ( Vazyme ) .The 5 monophosphorylated RNAs were prepared by providing five-fold excess of GMP over GTP in the in vitro transcription reactions.The DNA templates for the transcription of 9S RNA was prepared as previously described ( 41 ) .The templates for psbO mRNA and psbA mRNA transcription were amplified from the genomic DNA of Anabaena using the primer pairs of Pall3854F219ma / Pall3854R1095 and Palr4866F65maa / Palr4866R1144a.The transcribed RNAs were purified using RNA purification columns.

RNase E cleavage activity assays
The cleavage activity of RNase E was measured in the reaction buffer containing 20 mM Tris-HCl ( pH 8.0 ) , 100 mM NaCl, 5 mM MgCl 2 , 0.1 mM DTT, 5% glycerol, 0.1% Triton X-100, 1 mg / ml BSA and 2 U / μl Murine RNase inhibitor.To investigate the effect of RebA on RNase E activity, RneN412 was first incubated with indicated amount of RebA in the re-action buffer for 30 min on ice.Subsequently, RNA substrate was added into the mixture at 30 • C for indicated time periods.The reaction was stopped by adding an equal volume of 2X RNA loading buffer ( 95% formamide, 5 mM EDTA, 0.025% SDS, 0.025% bromophenol blue, 0.025% xylene-cyanol ) .After treatment at 95 • C for 5 min, cleavage products in the reaction mixture were separated on 5% ( for mRNA substrates ) or 15% ( for oligoribonucleotide substrates ) polyacrylamide gels that contained 7 M urea.For FAM-labeled RNA substrates, the gels were detected with GE Typhoon Trio Imager, while for non-labeled substrates, the gels were stained with GelRed ( Biotium ) and imaged with a common gel imaging system.The intensity of RNA bands was quantified using ImageJ.
The inhibitory effect of RebA on RNase E activity was evaluated with the RNAs of 5 p-LU13-FAM, 5 OH-LU13-FAM, 9S RNA, psbO mRNA and psbAI mRNA.The cleavage products were separated on PAGE gels, and the intensities of the gel bands were used to calculate the inhibition efficiency ( I ) of RebA on the RNase activity using the formula: where S ( total ) is the intensity of the band in the blank control reaction that contains RNA only, S ( RneN412 + RebA ) is the intensity of the uncut substrate band when RebA and RneN412 are both added in the reaction, and S ( RneN412 ) is the intensity of the uncut substrate band when only RneN412 is included in the reaction.

RNA-binding assays
RNA-binding assays were carried out in the buffer containing 20 mM TrisHCl ( pH 8.0 ) , 200 mM NaCl, 0.1 mM DTT, 2 mM EDTA, 5% glycerol and 1 mg / ml BSA ) .To test the effect of RebA on the binding activity of RNase E, RneN396 and RebA was first incubated with indicated amounts of RebA in the binding buffer for 30 min on ice, then RNA substrate was added into the mixture for another 30 min of incubation at 30 • C. The incubation mixtures were then analyzed on 5% native polyacrylamide gels at 4 • C.

Co-immunoprecipitation
To prepare the magnetic beads covalently coated by RNase E polyclonal antibodies ( Anti-Rne ) , the purified polyclonal antibodies against RNase E was bound to Protein A / G MagBeads ( Genscript ) , followed by cross-linking using dimethyl pimelidate dihydrochloride ( Sigma-Aldrich ) according to the manufacture's instruction.For co-immunoprecipitation of cellular proteins associated with RNase E, about 200 ml fresh Anabaena cells ( OD 75 0 ≈ 0.5 ) grown in BG11 was collected by centrifugation at 5000 g for 10 min at room temperature.After washing once with 5 ml buffer W ( 50 Tris-HCl, 150 mM NaCl, pH 8.0 ) , cells were resuspended in 1 ml lysis buffer ( 50 mM Tris-HCl, pH8.0, 150 mM NaCl, 0.1% Triton X-100 ) supplemented with cOmplete™ Protease Inhibitor Cocktail ( Roche ) and 50 μg / ml of RNase A. Subsequently, the cells were lyzed by FastPrep-24 ( MP Biomedicals ) .Following centrifugation at 12 000 g for 20 min, the supernatant of cell lysate was transferred into a 2 ml tube containing 0.2 ml of Anti-Rne beads.After 5 hours of incubation at 4

SPR assays
The binding constants between RneN396 and RebA were determined by surface plasmon resonance ( SPR ) at 25

Western blotting
Conventional Western blotting was performed as previously described ( 41 ) .The polyclonal antibodies for detecting RNase E and RebA were produced by immunizing rabbits with the purified proteins ( FrdBio ) .The production of the antibodies against FtsZ was described previously ( 52 ) .Far-Western blotting assay was performed as previously described ( 41 ,42 ) .Briefly, to test the interaction between RebA and RNase E regions, RneN396 ( 1 μg ) , RneC ( 1 μg ) , and RebA ( 10 ng, used as positive control ) were separated in duplicate on two 10% SDS-PAGE gels and electrotransferred onto two PVDF membranes.Both membranes were blocked in PBS-T ( PBS supplemented with 0.1% Tween 20 ) containing 5% skimmed milk for 1 h.Then, one membrane was incubated with 10 μg of RebA in PBS-T containing 1% skimmed milk, while the other was incubated in the same solution but without RebA.Subsequently, both membranes were immunodetected with the antibodies against RebA.The interactions between subregions of RebA ( i.e.RebA_C and RebA_DC ) and RNase E regions were detected in the same way.

Microscopy
Microscopic images were taken by using a bright-field SDP-TOP EX30 microscope ( Sunny Optical Technology ) with an oil immersion lens objective ( 100 / 1.28 ) .Autoexposure was used for all images.The cells for microscopy were from cultures at exponential growth phase with an OD 750 between 0.3 and 0.5 unless otherwise indicated.

Quantification and statistical analysis
For cell size measurement, at least two hundred cells from several randomly selected microscopic images of each culture condition were used.The length and width of each cell were measured using ImageJ.The plotting and statistical analysis of data were performed using GraphPad Prism.P values were calculated using the nonparametric Mann-Whitney test.Statistical details for individual experiments, such as number of samples and experimental replicates, have been described in the figure legends and method details.

Identification of RebA, a novel RNase E-binding protein, uni ver sally present in cyanobacteria
Aiming to identify new components of the RNA degradosome or RNase E regulators in Anabaena , we performed a co-immunoprecipitation ( CoIP ) assay using the polyclonal antibodies against RNase E. The proteins precipitated were separated on SDS-PAGE and each gel band was subjected to mass spectrometry analysis ( Figure 1 A).Most of the bands corresponded to degraded fragments of RNase E, likely due to that the intrinsically disordered nature of the noncatalytic region of RNase E makes the protein very susceptible to proteolytic degradation.Among the known partners of RNase E (i.e.RNase II, PNPase and CrhB), only RNase II was detected in one band; the missing of the others could be attributed to factors such as lower expression, lower affinity for RNase E, or the partial degradation of their recognition regions in RNase E. Interestingly, there existed one band corresponding to a protein of unknown function (protein id: All1338) together with a degraded product of RNase E. Using bacterial two-hybrid assays, we further showed that All1338 interacted directly with RNase E, but not with the three known RNase E partners, RNase II, PNPase and CrhB (Figure 1 B).These results demonstrated that All1338 is a new interacting partner of RNase E. We thus named All1338 as RebA ( R Nase E b inding protein A ) in this study.
A RebA homolog could be found in each sequenced cyanobacterial genome by BLAST analysis (Supplementary Table S4).The length of RebA homologs varies greatly, ranging from 218 aa in the marine picocyanobacterium Prochlorococcus CCMP 1986 to 423 aa in Anabaena PCC 7120.The phylogenetic tree of RebA homologs from representative strains has a topology similar to the phylogenetic tree of cyanobacterial species (Supplementary Figure S1) ( 53 ), indicating that rebA is an ancient gene that has already existed in the common ancestor of cyanobacteria.Outside cyanobacteria, RebA homologs are only confidently identified in several species of photosynthetic Paulinella , and these Paulinella sequences are grouped together with those of marine Synechococcus and Prochlorococcus strains (Supplementary Table S4; Supplementary Figure S1), suggesting their cyanobacterial origin.Indeed, photosynthetic Paulinella is a lineage of amoebae that had obtained the capacity of photosynthesis 90-140 million years ago via endosymbiosis of a cyanobacterium belonging to the Proc hlorococcus -Synec hococcus lineage ( 54 ).
A Pfam search showed that the major part of RebA did not match to any known protein domains, except that its first 42 residues displayed a low similarity (e-value = 0.00022) to a helix-turn-helix domain.We also performed a disorder analysis on RebA using PONDR-FIT.It showed that the N-terminal and C-terminal regions of RebA are ordered with high probability, while the middle part is disordered (Figure 1 C).Sequence alignment with RebA and its homologs showed that the protein can be divided into four distinct domains, which we named as N, V, E and C, respectively (Figure 1 C; Supplementary Figure S2).The domains N and C are two conserved regions located at the N-terminus ( ∼60 aa) and C-terminus ( ∼90 aa), respectively, corresponding to the two ordered regions identified by PONDR-FIT analysis.The highly variable V domain ( ∼200 aa) and the glutamic acid-rich, negatively charged E domain ( ∼50 aa), are located between N and C. The different lengths of RebA homologs are mainly due to the length variation of their V and E regions.

The C domain of RebA interacts with the 5 sensor domain of RNase E
We next identified the regions in RebA and RNase E responsible for interaction.As the C domain of RebA is the most conserved, we first tested its role in interaction.We expressed and purified, respectively, the recombinant proteins of the fulllength RebA, the C region of RebA (RebA_C), the truncated form of RebA without the C region (RebA_DC), the catalytic core of RNase E (RneN396, corresponding to 1-396 aa) and the noncatalytic domain of RNase E (RneC, corresponding to 401-687 aa).RebA, RebA_C and RebA_DC were all purified in monomeric form.The interactions between these proteins were examined by far-Western blot assays.The result showed that RebA and RebA_C, but not RebA_DC and RneC, was able to interact with RneN396 (Figure 1 D).Since RneN396 does not contain the Zn-link region required for RNase E tetramerization, its interaction with RebA indicates that the tetrameric state of RNase E is not a prerequisite for RebA binding.
Similar to the E. coli RNase E, the catalytic region of Anabaena RNase E can be further divided into several structurally distinct subregions (i.e.S1, 5 sensor, RNase H, and DNase I) that are all required for the enzyme activities (Figure 1 E).We then examined the interaction between each of these subregions and RebA using bacterial two-hybrid assays.The result showed that interaction occurred only between the 5 sensor subregion of RNase E and RebA (Figure 1 E).To further confirm the role of the 5 -sensor subregion in the interaction, we introduced several point mutations into this region within the catalytic domain of RNase E. Most of the resulting variants, which correctly folded and retained enzyme activity similar to that of the wild-type protein, lost their ability to interact with RebA (Supplementary Figure S3).Thus, it appears that, apart from the 5 sensor subregion, the other subregions make only minimal, if any, contributions to the binding of RebA.
Based on these interaction assays, we concluded that the interaction between RebA and RNase E was mostly mediated by the C domain of RebA and the 5 sensor domain of RNase E. We also measured the binding affinity between RebA and RneN396 by surface plasmon resonance (SPR) assay (Figure 1 F).Their dissociation constant was determined to be 10.3 nM, indicating that the two proteins form a stable complex.The high affinity between RebA and RneN396 allowed us to visualize a clear band of RebA-RneN396 complex in a native gel (Figure 1 G).By further analyzing the complex for the molar ratio between them with SDS-PAGE (Figure 1 G), we estimated that RneN396 and RebA form a complex with a stoichiometry of 1:1.

RebA inhibits the binding and the cleavage activities of RNase E
RebA does not contain any domains or motifs that may confer the activities for RNA cleavage or binding, suggesting that it does not act on RNA directly.We here assessed RebA's potential cleavage and binding activities on synthetic RNAs in vitro , using the catalytic region of RNase E as a positive control.Our initial focus was on a synthetic 13-nucleotide oligoribonucleotide, 5 p-LU13-FAM, which features a 5 -monophosphate and a 3 FAM modification and has previously been employed as a substrate for cyanobacterial RNase E previously ( 42 ).The RNA remained intact even when RebA was added at a 20-fold molar excess in enzymatic assays (Figure 2 A).Additionally, it exhibited no binding to RebA, even when RebA was at up to 50-fold molar excess in binding assays (Figure 2 B).Similarly, RebA displayed no discernible activity on several other substrates (details provided below) with different 5 ends and lengths (Supplementary Figure S4A).However, we found that RebA when present at more than 2-fold molar excess could strongly interfere with the substrate binding activity of RNase E (Figure 2 C), suggesting that it may function as a regulator of the RNase E activity.We thus further tested the RNase E activity in the presence of RebA.RneN412 (1-412 aa) that contains the Zn-link domain for tetramerization was used in most enzymatic assays in this study, as it is easier to purify than the full-length RNase E, while having a similar enzymatic activity ( 41 ).
It is well known that RNase E cleaves many RNA substrates preferentially when they have a 5 -monophosphate end, and such preference involves the recognition of 5 monophosphate by the 5 senor domain ( 18 ,55 ).We thus tested first with the substrate 5 p-LU13-FAM.As expected, RneN412 alone efficiently cleaved most of 5 p-LU13-FAM within 30 minutes (Figure 2 D).In the presence of RebA, however, the enzyme activity was significantly inhibited, and this inhibition became stronger with increasing amount of RebA (Figure 2 D).According to the inhibition curve, half-inhibition requires a RebA to RneN412 ratio of 1:1.When the ratio reached 4:1, the activity dropped by about 90%.
We further examined the activity of RneN412 in the presence of RebA on two substrates that do not have a 5monophosphate for recognition by the 5 senor domain: 5 OH-LU13-FAM that is similar to 5 p-LU13-FAM, but with a 5 -hydroxyl end, and E. coli 9S RNA (246 nt) that has a 5triphosphate end.These two substrates normally do not have a 5 -monophosphate for recognition by the 5 senor domain, and are cleaved by RNase E with lower efficiency ( 21 ,56 ).Surprisingly, the activity of RneN412 on them was also inhibited by RebA, and the inhibition curves were similar to that for 5 p-LU13-FAM (Figure 2 D).
Furthermore, we tested the inhibition of RebA on RNase E with two in vitro transcribed Anabaena mRNAs, psbAI mRNA (1209 nt) and psbO mRNA (1314 nt), which encode the photosystem II components D1 and PsbO, respectively.We chose these two mRNAs because their 5 boundaries had been clearly defined ( 57 ) and their 3 boundaries could be predicted with high confidence due to the presence of typical Rhoindependent terminators (Supplementary Figure S4B).These mRNAs, longer than the substrates used above, are expected to have more complex structures.Both mRNAs, either with a 5 -ppp end or a 5 -p end, could be processed by RneN412, and this activity was again inhibited by the presence of RebA, although the maximal inhibition efficiency achieved for these long substrates was lower than those with shorter ones (Figure 2 D).
Since the catalytic activity of RNase E is conserved in bacteria, we wondered whether RebA had a cross-species activity.Thus, we tested the cleavage activity of the catalytic domain of E. coli RNase E (EcRneN529, 1-529 aa) using 5 p-LU13-FAM and 9S RNA as substrates in the presence of RebA (Supplementary Figure S5A).It showed that RebA had no effect on the activity of EcRneN529, consistent with the lack of used to analyze the cleavage products.For the reactions containing 9S RNA, psbA1 mRNA or psbO mRNA, 50 ng / μl substrate and 6% urea-PAGE were used, and gels were stained with GelRed after electrophoresis.The intensities the substrate bands from all three replicative assays were quantified using Image J and used to make the plots.
Taken together, we conclude that RebA is an inhibitor specifically for cyanobacterial RNase E, and it by interacting with the 5 sensor domain inhibits the activity of RNase E on a large spectrum of RNA substrates, irrespective of the nature of their 5 ends.

RebA plays a role in cell morphology control
To explore the physiological function of RebA, we created a rebA deletion mutant ( rebA ) by markerless deletion using a CRISPR / Cpf1-based gene editing system, and an overexpression strain of rebA (OE-RebA) by expressing an extra copy of rebA controlled by the strong promoter of rbcL (P rbcL ) on a replicative plasmid (Supplementary Figure S6A-D).The growth of rebA and OE-RebA in liquid medium and on agar plate were compared with that of the WT.While rebA grew only slightly slower than the WT, the overexpression strain OE-RebA grew poorly, especially on agar plate (Figure 3 A,  B).Microscopic images showed that cells of rebA appeared significantly shorter than WT cells (Figure 3 C, D).Note that the extent of morphological changes is dependent on growth conditions: cells were shortened more dramatically in fastgrowing cultures; however, cell morphology became closer to that of WT when cells were growing slowly (for example when the culture reached high cell densities or was illuminated under low light intensity) (Supplementary Figure S6E, F).By contrast, cells of OE-RebA were significantly longer and wider than those of WT (Figure 3 C, D), and this phenotype was not influenced by growth conditions.These results suggested that overexpression of RebA affected negatively cell division or favored lateral peptidoglycan synthesis, leading to elongated cells, while the deletion of rebA resulted in an opposite phenotype.
It is known that a decrease in the activity of the RNase E in E. coli leads to cell elongation ( 58-60 ), a phenotype similar to that reported here.Since both RebA and RNase E are highly conserved in cyanobacteria, we sought to determine whether the effect of RebA overexpression could be a conserved function in some other cyanobacteria.For this purpose, we chose Synechococcus elongatus PCC 7942, also a rod-shaped cyanobacterium like the shape of E. coli .When the RebA homolog from Synechococcus elongatus PCC 7942 was overexpressed form a replicative plasmid, a strong cell division defect was observed, with cell length increased by about 5 folds as compared with the wild type (Supplementary Figure S7).

RNase E and RebA control cell morphology in an opposite manner
Given that RebA interacted with, and inhibited the activity of RNase E, the phenotypes of rebA and OE-RebA could be attributed to altered cellular activities of RNase E. We tested this assumption by modulating the RNase E levels in cells.We initially attempted to create a rne deletion strain but failed, implying that the gene could be essential in Anabaena , as suggested by similar studies in other cyanobacteria ( 17 ,38-40 ).We therefore made a conditional mutant of rne (CTrne ), in which the upstream region (555 bp) of the rne ORF was replaced with the CT promoter, inducible by copper and theophylline (Figure 4 A).The strain CTrne was maintained in the presence of 0.3 μM Cu 2+ and 1 mM theophylline.Under such conditions, CTrne showed optimal growth (Supplementary Figure S8A).When cells of CTrne were transferred to BG11 medium without inducers, the cellular level of RNase E gradually decreased and became nearly undetectable after 5 days (Figure 4 B).At the same time point, cells lost viability (Supplementary Figure S8B), providing experimental evidence for the essential functions of RNase E. By observation under a microscope of CTrne cells, we found that the level of RNase E also had an effect on cell morphology.When RNase E was overexpressed in the presence of high concentrations of inducers (3 μM Cu 2+ and 2 mM theophylline), cells became shorter and wider, a phenotype similar to that of the rebA mutant.In contrast, when RNase E levels decreased following the removal of the inducers, cells became longer, similar to that of the rebA overexpression strain OE-RebA (Figure 4 C to E).Thus, RNase E and RebA both regulate cell morphology, but with opposing effects.
In the genomes of Anabaena and many other cyanobacteria, the rne gene is immediately upstream of rnhB that encodes RNase HII, and the two genes are likely cotranscribed.The phenotypes of CTrne could thus be the result of altered expression of the downstream rnhB gene.To rule out this possibility, we constructed a deletion mutant ( rnhB ) and an overexpression strain (OE-RnhB) of rnhB .Both strains could grow well in BG11 without apparent morphological changes of the cells (Supplementary Figure S9), indicating that the observed phenotypes of CTrne were indeed due to the altered expression of rne .

Interaction-dependent regulation of the RNase E activity by RebA is required for maintenance of cell morphology
Since RebA could interact with and inhibit the activity of RNase E, we sought to determine the importance of their interaction in vivo .For this purpose, we initially screened for the mutations in RebA that could disrupt RebA-RNase E interaction (Supplementary Figure S10).Most residues in RebA_C were individually mutated into a charged residue (glutamate or arginine) to maximize the effects of the mutation, followed by bacterial two hybrid assays to check the effects of their mutations on interaction.Those residues whose mutation led to a significantly decreased interaction were then subjected to a second round of mutagenesis, with each being replaced into a small, neutral residue (alanine or serine), to confirm their roles on interaction.Ultimately, 11 residues whose mutations significantly disrupted the interaction in both rounds of mutagenesis were identified (Figure 5 A), providing another evidence for the interaction between RebA_C and RNase E. For some of the residues, we expressed and purified the recombinant RebA mutants with alanine substitution, and examined their inhibitory activities on RNase E. As expected, these RebA derivatives had a greatly reduced capacity of inhibition on the RNase E activity (Figure 5 B).Next, the genetic effect of these mutations in RebA was examined following ectopic expression in WT.While overexpression of the WT RebA inhibited growth and led to the formation of elongated cells similar to that caused by RNase E depletion, that of its interactiondefective mutants had much weaker effects on the growth and showed wild-type-like cell morphology (Figure 5 C to E).These results all together indicate that the inhibition of RNase E by RebA is required for a balanced RNase E activity in vivo , and this regulation is dependent on their physical interaction.To further confirm the relationship between RebA and RNase E, we examined the effects of rebA inactivation while varying the levels of RNase E in the cells.The rebA gene was thus deleted from the chromosome in the CTrne strain, leading to strain CTrne :: rebA .The CTrne :: rebA strain displayed constantly a shorter-cell phenotype in the presence 0.3 μM copper and 1 mM theophylline (Supplementary Figure S8C), inducer concentrations that allowed RNase E to be produced at a level without causing cell shape changes in CTrne (Figure 4 C).Under such conditions, the absence of RebA in CTrne:: rebA mimicked therefore RNase E overexpression.Following RNase E depletion by removal of the inducers from the medium, the cells of CTrne :: rebA changed their shape and maintained viability for significantly longer time than the cells of CTrne (Supplementary Figure S8C), demonstrating that rebA inactivation could slow down the effects of RNase E depletion.These results suggest that a proper level of RNase E activity in the cells is necessary for cell growth and morphology maintenance, since the absence of RebA inhibitor leads to an enhanced activity of RNase E in vivo , and partly suppresses the lethal effect caused by decreased levels of RNase E.

Discussion
RNase E is widely present in bacteria and it plays a central role in RNA metabolism.Due to its broad substrate specificity, its activity must be under check according to changes of internal or external conditions.Yet, only a few regulators of RNase E have been described ( 32-34 ,36 ), and most of them are present only in a limited number of bacteria, suggesting that distinct regulation mechanisms may have been evolved in different organisms.Here, we report that RebA, whose homologs are universally present in cyanobacteria, could inhibit the activity of RNase E. RebA was co-isolated with RNase E from Anabaena cell lysate by co-immunoprecipitation.This interaction was further confirmed by bacterial two hybrid assays, and the identification of mutations that disrupted their interaction associated with the expected phenotypes.High affinity interaction occurred between the conserved C-terminal region of RebA and the 5 -sensor domain within the catalytic region of RNase E. RebA showed no RNA-binding or cleavage / degradation activities; it however strongly inhibits the substrate binding and cleavage activities of RNase E. Furthermore, the cellular levels of RebA and RNase E were shown to have opposite effects on cell morphology, demonstrating the inhibitory role of RebA on the RNase E activity in vivo .Taken together, we conclude that RebA is an inhibitor of RNase E in Anabaena .By varying the relative amounts of RebA and RNase E, we were able to provide evidence on the importance of a balanced activity of RNase E in Anabaena .
The activity of RNase E is regulated through multiple mechanisms, including constrained subcellular localization, transcriptional autoregulation, and the formation of RNA degradosome ( 30 , 42 , 61 , 62 ).Additionally, RNase E activity can also be regulated by transacting components, such as the E. coli proteins of RraA, RraB, ribosomal L4 protein, AmiC ( 32 , 34 , 36 , 63 ).These regulators mostly associate with the noncatalytic region of RNase E, and either enhancing or disrupting their interactions with RNase E did not severely affect cell growth and RNA metabolism of the corresponding bacteria, suggesting their limited regulatory effects in vivo .We found here that RebA acts on RNase E in a distinct mechanism: it binds to the catalytic region of RNase E, and inhibits RNase E activity by interfering with substrate binding (Figure 6 A).Thus, the effect of RebA on RNase E activity represents a new regulatory mechanism for RNase E functions.Furthermore, RebA could inhibit the cleavage of various substrates by RNase E in vitro , and its overexpression led to severe growth defect, similarly as that caused by RNase E depletion, indicating that the control of RNase E by RebA has a global effect on cellular RNA metabolism.
Several substrate binding motifs have been identified within both the catalytic and noncatalytic domains of E. coli RNase E ( 18 , 20 , 64 , 65 ).Among them, the 5 -sensor domain within the catalytic region, which has a binding pocket that specifically accommodates the 5 monophosphate end of RNA, is vital for the binding and the cleavage of 5 -monophosphorylated substrates ( 18 ,66 ).How much this domain contributes to the recognition of other types of substrates has been unclear.We showed here that when the 5 -sensor domain was bound to RebA, the binding and the cleavage of the RNA substrates with 5 -OH, 5 -p or 5 -ppp by RNase E were all inhibited.This result implies that the 5 -sensor domain plays a critical role in the binding and cleavage of all RNA species, regardless of their 5 -end states.The substrate-binding surface within the 5 -sensor domain, or the RNA binding channel near it, may be blocked or become inaccessible upon RebA binding (Figure 6 A).
Based on previous studies, an RNase E-based RNA degradosome, which consists of RNase E, PNPase, RNase II and CrhB, has been suggested to exist in Anabaena ( 7 ).Here, the low dissociation constant between RebA and RNase E indicates that the two proteins form a stable complex in vivo .In this regard, we propose RebA as a new component of the RNA degradosome in Anabaena (Figure 6 B).Of particular interest is the recent discovery of an autoinhibitory pocket residing within the RNase H-like domain of E. coli RNase E ( 20 ).This pocket was suggested to destabilize the activated conformational state of RNase E, thereby facilitating the rapid release of cleaved product from the enzyme to prevent potential product inhibition ( 20 ).However, it appears that this autoinhibition mechanism is absent in cyanobacterial RNase Es due to the lack in the cyanobacterial counterparts of key pocket residues present in the RNase E of E. coli .Given that RebA effectively inhibits the activity of Anabaena RNase E across a range of substrates, it is plausible that RebA serves a function analogous to the autoinhibitory pocket of E. coli RNase E by forming a constant association with RNase E. Nevertheless, it is worth noting that RebA's expression levels may fluctuate under different growth conditions, which could enable it to modulate the metabolism of bulk RNAs in response to environmental fluctuations.
RebA and RNase E both influence cell morphology of Anabaena , and the effect of RebA on cell morphology was  mediated by RNase E. A decrease of RNase E or an increase of RebA led to cell elongation, likely a consequence of cell division defect.To determine if such a regulation is common in cyanobacteria, we overexpressed the RebA homolog in the unicellular cyanobacterium Synechococcus elongatus PCC 7942, which is evolutionarily distant from Anabaena.The result demonstrates that RebA overexpression in Synechococcus led to a strong cell elongation phenotype (Supplementary Figure S7).Interestingly, a decrease in the RNase E amount in E. coli was shown to have a very similar effect on cell morphology (58)(59)(60).In E. coli , processing the polycistronic ftsA-ftsZ transcripts by RNase E results in a higher FtsZ to FtsA ratio ( 67 ).E. coli RNase E can also enhance FtsZ translation by degrading the small RNA DicF, which binds to the ribosome binding site of ftsZ mRNA and thereby inhibits its translation ( 60 ).The mechanisms at play in cyanobacteria remain to be investigated.Nevertheless, regulation of cell morphology by RNase E in different bacteria, at least for those with a rod shape such as E. coli and Synechococcus elongatus PCC 7942 or an ovoid shape such as Anabaena , suggests a convergent mechanism for the coordination between global RNA metabolism and cell growth.
RebA, like RNase E, is encoded by each sequenced cyanobacterial genome.Although RebA homologs from dif-ferent cyanobacteria have variable lengths; their C-terminal regions, the potential RNase E-interacting region, are highly conserved (Supplementary Figure S2; Supplementary Table S4).Such an observation suggests that the regulation of RNase E by RebA is an ancient and highly conserved mechanism of RNA metabolism control in cyanobacteria.This is further supported by the similar effects of RebA overexpression on cell morphology in both Anabaena PCC 7120 and Synec hococcus elong atus PCC 7942, as well as by a recent finding showing that a RebA homolog and RNase E are present in the same component during the Grad-Seq analysis in Synechocystis PCC 6803 ( 68 ), another unicellular cyanobacterium phylogenetically distant from Anabaena PCC 7120.Our study provides a new control mechanism on gene expression and open the door for our understanding on how RNA metabolism is regulated in different cyanobacterial species in response to changes of physiological or environmental conditions.

Figure 1 .
Figure 1.RebA interacts with RNase E both in vivo and in vitro .(A) Identification of the proteins in complex with RNase E by co-immunoprecipitation from Anabaena cell lysate, using polyclonal antibodies against Anabaena RNase E. The co-immunoprecipitated proteins were separated by a 10% SDS-PAGE, and the proteins in the visible bands after Coomassie-blue staining were identified by mass spectrometry.Those bands corresponding to the degraded products of RNase E were marked as 'RNase E*'.(B) Bacterial two hybrid assay for the interaction between RebA and RNase E. No interaction was found between RebA and other known RNase E partners.(C) Disorder (upper) and domain (lower) analyses for RebA.Prediction of intrinsically disordered residues in RebA was performed with PONDR-FIT ( http:// original.disprot.org/pondr-fit.php).The line at 0.5 of Y-axis is the threshold abo v e which it is predicted to be disordered.T he domain architecture of R ebA w as defined based on the alignment of R ebA homologs: N, the N-terminal conserved region; V, a highly variable region; E, a region rich in glutamic acid and aspartic acid; C, the C-terminal conserved region.(D) Determination of the main regions within RebA and RNase E responsible for interaction by Far-Western blotting assays.The N-terminal region of RNase E (RneN396) and the C-terminal region of RNase E (RneC), together with RebA, RebA_C (C-terminal region of RebA) or RebA_DC (RebA without the C-terminal region), were separated on two sets of SDS-PAGE gels, followed by transfer onto PVDF membranes.One set of the membranes (upper) were incubated respectively with purified RebA, RebA_C or RebA_DC, and detected with the polyclonal antibodies against RebA.The other set of the membranes (bottom) were incubated with BSA prior to the detection with RebA antibodies.The upper weaker band in the lanes of RebA_C likely corresponds to a protein contaminant that reacts nonspecifically with RebA antibodies.(E) Interaction of different domains of RNase E (RNase H, S1, 5 sensor and DNase I, upper panel) with RebA, determined using bacterial two-hybrid assays, with the full-length catalytic region of RNase E as a positive control (lo w er panel).(F) Quantification of R ebA-RNase E interaction using surf ace plasmon resonance (SPR) assa y.K on , K dis and K d are the estimated association rate, dissociation rate, and dissociation constant, respectively.(G) Stoichiometry of RebA-RNase E complex.RneN396, RebA and a mixture of RneN396 and RebA were incubated for 30 min, then separated on a native gel (left gel).The band of the protein complex (indicated by an '*') was e x cised and analyzed with SDS-PAGE (right gel), with purified RneN396 and RebA at defined molar ratios loaded for comparison.The intensity of the protein bands in the SDS-PAGE was quantified with Image J.

Figure 2 .
Figure 2. RebA inhibits RNase E activities.(A) No clea v age or degradation activities of RebA on the substrate 5 p-LU13-FAM, a 1 3-bp RNA (LU1 3) with a monophosphate at the 5 end and a 3 FAM label at the 3 end.RNA was used at 50 nM.BSA and RneN412 were used as a negative control and a positive control, respectively.The leftmost lane ( −) contained RNA only.After 30 minutes of incubation, the reactions were stopped and analyzed by urea-PAGE.RebA also showed no cleavage / degradation activities on other substrates tested (Supplementary Figure S4).(B) No binding activities of RebA using 5 p-LU13-FAM RNA.The RNA was incubated in the binding buffer with RneN412 (positive control) or with increasing amounts of RebA, respectively.One picomole of RNA was used in the binding mixtures.After 30 minutes of incubation, the reactions were analyzed by native-PAGE.(C) Effect of RebA on the binding activity of RneN396 on 5 p-LU13-FAM.The assay condition was similar to that presented for panel B, except that RebA and RneN396 were incubated on ice for 30 minutes prior to the addition of RNA.(D) The cleavage of various RNA substrates by RneN412 in the presence of RebA. 5 OH-LU13-FAM: similar to 5 p-LU13-FAM but with a h y dro xyl group at the 5 end; 5 ppp-9S RNA: E. coli 9S RNA with a triphosphorylated 5 end; psbAI mRNA and psbO mRNA: mRNAs encoding the photosystem II components D1 and PsbO, respectively.RneN412 were used at 0.4 nM for 5 p-LU13-FAM, 10 nM for 5 OH-LU13-FAM, and 200 nM for the other substrates, respectively.All the reactions were performed at 30 • C for indicate time periods.For the reactions containing 5 p-LU13-FAM or 5 OH-LU13-FAM, 50 nM substrate was included, and 15% urea-PAGE was

Figure 3 .
Figure 3.Effect of RebA level on the growth and morphology of Anabaena cells.(A) Growth of WT (wild type), rebA , OE-RebA (RebA overexpression stain using the promoter P rbcL ), C-rebA ( rebA -complemented strain) in BG11 liquid medium.The optical densities at 750 nm (OD 750 ) at different time points are shown as means with standard deviation (n = 3).(B) Growth of the same strains as in panel A on solid media.The cultures at OD 7 50 = 0.5 were serially diluted with the dilution factor of 2, and 5-μl aliquots of each dilution were spotted on a BG11 agar plates.The plate was then placed under normal growth conditions for 8 days before photographed.(C) Microscopic images of the filaments of different strains.Samples were taken from fresh liquid cultures at OD 750 ≈ 0.3.Scale bars:10 μm.(D) Distribution of the cell lengths and the cell widths of the samples shown in panel C. A total of 256 cells for each sample was measured using Image J. The violin plotting and statistical analysis of data were performed with the GraphPad Prim software.P values were calculated using the nonparametric Mann-Whitney test in comparison to WT. Significant P-values are marked by * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001; n.s: non-significant ( P > 0.05).

Figure 4 .
Figure 4. Characterization of a conditional mutant of rne (CT-rne ) and the effects of different le v els of RNase E on cell morphology.(A) Schematic representation of the genotype of Anabaena wild type strain (WT) and that of CT-rne .In CT-rne , the promoter of rne was replaced with the CT promoter, inducible by Cu 2+ and theophylline (Tp) ( 37 ).(B) Levels of RNase E in CT-rne cells following removal of the inducers or addition of high amounts of inducers in the media.CT-rne cells pre-cultured in the presence of 0.3 μM Cu 2+ and 1 mM theophylline, conditions allowing cells to display a morphology similar to that of WT, were transferred into a medium without Cu 2+ and Tp for RNase E depletion, or into a medium supplied with 3 μM Cu 2+ and 2 mM Tp for RNase E overproduction.Cells were collected at the indicated time points for Western blot analysis using antibodies against RNase E. WT cells grown in BG11 was used as a control.Each lane contained 50 μg of cellular proteins.(C) Microscopic images of the same samples as those used in panel B. 'DP' and 'OE' are abbreviations for 'depletion' and 'overexpression' of RNase E, respectively.Scale bars: 10 μm.(D) Distribution of cell lengths and cell widths of the samples shown in panel C. Cell measurement, data plotting and statistical analysis were performed as described in Figure 3 C. (E) Summary of the relationships between cell morphological changes and the depletion or overexpression of RebA and RNase E. L: cell length; W: cell width; L / W: the ratio of cell length to width.

Figure 5 .
Figure 5. Significance of RebA-RNase E interaction.(A) Identification of key residues within RebA_C involved in the interaction with RNase E. Each residue was replaced by alanine, and the interaction was assessed by using bacterial two hybrid assays.Only those mutations that led to significantly decreased interaction were shown here after an initial screening.The result of all mutations within RebA_C was shown in Supplementary Figure S10.WT RebA_C was used as a positive control.(B) The activity of RNase E on 5 p-LU13-FAM in the presence of mutant RebA RebA L368A , RebA I370A , R ebA V387A or R ebA I388A .WT R ebA w as used as a control.T he reaction conditions and data analy sis w ere the same as those f or Figure 2 D. (C) Gro wth of strains with o v erproduction of RebA interaction-defective mutants on BG11 agar plates.WT strain and a WT RebA overproduction strain were included as controls.The experiment was performed as described in Figure 3 .(D) Microscopic images of filaments of different strains grown in BG11 liquid media.Samples were taken from fresh cultures at OD 750 ≈ 0.3.Scale bar: 10 μm.(E) Distribution of cell lengths and cell widths of the samples shown in the panel D. Cell measurement, data plotting and statistical analysis were performed as described in Figure 3 C.

Figure 6 .
Figure 6.A proposed mechanism of RebA-RNase E interaction, and a proposed RNase E-dependent RNA degradosome in Anabaena based on current data.(A) Proposed model of RNase E activity regulated by RebA.In the absence of RebA, the subunits in the RNase E tetramer can efficiently bind to RNA substrates, mostly in a 5 -sensor-domain-dependent w a y.Such a binding ensures a high clea v age activity of RNase E. In contrast, when most or all 5 sensor domains are bound by RebA, RNase E tetramer binds to substrate less efficiently, resulting in a low cleavage activity.The non-catalytic region of RNase E subunits is not shown in the diagrams for simplification.(B) Schematic representation of Anabaena degradosome.RebA and PNPase binds to the 5 sensor domain and the conserved motif C4, respectively.RNase II and CrhB both binds to the catalytic domain of RNase E, but their precise recognition sites on RNase E ha v e not been determined.Only one subunit of the RNase E tetramer is shown for simplicity.
• C with gentle rotation, the beads were collected and washed twice with the lysis buffer.Proteins captured by the Anti-Rne beads were then eluted