BR-bodies provide selectively permeable condensates that stimulate mRNA decay and prevent release of decay intermediates

Summary

Biomolecular condensates play a key role in organizing RNAs and proteins into membraneless organelles. Bacterial RNP-bodies (BR-bodies) are a biomolecular condensate containing the RNA degradosome mRNA decay machinery, but the biochemical function of such organization remains poorly defined. Here we define the RNA substrates of BR-bodies through enrichment of the bodies followed by RNA-seq. We find that long, poorly translated mRNAs, small RNAs, and antisense RNAs are the main substrates, while rRNA, tRNA, and other conserved ncRNAs are excluded from these bodies. BR-bodies stimulate the mRNA decay rate of enriched mRNAs, helping to reshape the cellular mRNA pool. We also observe that BR-body formation promotes complete mRNA decay, avoiding the build-up of toxic endo-cleaved mRNA decay intermediates. The combined selective permeability of BR-bodies for both enzymes and substrates together with the stimulation of the sub-steps of mRNA decay provide an effective organization strategy for bacterial mRNA decay.

Graphical Abstract

eTOC blurb

Al-Husini et al. show that BR-bodies are enriched with mRNAs, sRNAs, and antisenseRNAs, while they exclude tRNAs, rRNAs, and other highly structured ncRNAs. BR-bodies assemble preferentially with long, poorly translated mRNAs, where RNase E and degradosome-associated exoribonucleases help to stimulate the multi-step mRNA decay pathway.

Introduction

The bacterial cytoplasm has been thought to be poorly organized due to the broad lack of membrane bound organelles which compartmentalize biochemical pathways by selective membrane permeability (Abbondanzieri and Meyer, 2019; Kerfeld et al., 2018; Surovtsev and Jacobs-Wagner, 2018). Rapid advances have recently demonstrated biomolecular condensation promotes the formation of membraneless compartments that selectively concentrate and organize proteins and nucleic acids (Banani et al., 2017; Courchaine et al., 2016; Shin and Brangwynne, 2017). Biomolecular condensation occurs most commonly through liquid-liquid phase separation of multivalent intrinsically disordered scaffolding proteins together with RNA resulting in cytosolic liquid-like droplets. The formation of these condensates is driven by networks of weak multivalent interactions between scaffolding proteins and their recruited client proteins (Banani et al., 2017). BR-bodies are the first biomolecular condensates in bacteria that were shown to directly assemble through liquid-liquid phase separation (Al-Husini et al., 2018). Phase separation occurs through the RNase E scaffold protein’s intrinsically disordered region (IDR) together with RNA degradosome client proteins and RNA (Al-Husini et al., 2018). These BR-body forming capabilities were previously shown to increase cell survival during stress exposure (Al-Husini et al., 2018), however, the biochemical function of these BR-bodies in organizing mRNA decay is poorly understood.

Many eukaryotic biomolecular condensates have been shown to concentrate a subset of cellular proteins and RNAs, such as nucleoli with rRNA transcription and processing machinery, Cajal bodies with snRNP assembly, and germ granules with germ cell-related mRNPs (Banani et al., 2017; Courchaine et al., 2016; Shin and Brangwynne, 2017; Weber, 2017). Most similar to BR-bodies, eukaryotic P-bodies organize the mRNA decay machinery by selective permeability to decay related proteins including decapping enzymes, deadenylation factors, translation repressors, and the major cytoplasmic nuclease Xrn1 (Hubstenberger et al., 2017). Stress granules also share similarity with BR-bodies: these granules are assembled with Xrn1 and some overlapping translation repressors, but they lack the decapping factors and deadenylases and instead contain translation initiation factors (Jain et al., 2016). Recently it was revealed in purified stress granule cores that many long, poorly translated mRNAs are enriched in stress granules, showing that stress granules also exhibit selective permeability for substrate RNAs (Khong et al., 2017). Interestingly, this selectivity may come from the RNA itself and its abilities to form intermolecular base pairs, yielding toxic RNA assemblies (Van Treeck et al., 2018). While both P-bodies and stress granules share poorly translated mRNAs as substrates, P-bodies tend to be devoid of all ribosomes (Hubstenberger et al., 2017), whereas stress granules contain stalled initiation complexes but lack the large ribosome subunit (Reineke et al., 2012). Additionally, translational repressors such as microRNAs (miRNAs) associate in P-bodies (Eulalio et al., 2008), yet the interplay with both ribosomes or inhibitory bacterial small RNAs (sRNAs) which also silence mRNAs through imperfect base-pairing is unknown for BR-bodies.

The biochemical function of P-bodies and stress granules in the context of mRNA decay is less well understood. P-bodies have been found to stimulate mRNA decay (Sheth and Parker, 2003) and recent mathematical models have found that smaller P-bodies might act as more efficient sites of mRNA decay (Pitchiaya et al., 2019), yet P-bodies are not necessary for mRNA decay (Eulalio et al., 2007). P-bodies can also store untranslated mRNAs (Hubstenberger et al., 2017; Sheth and Parker, 2003), suggesting that the mRNA fate may depend on the specific mRNA identity and context of cellular conditions (Wang et al., 2018). How stress granules affect mRNA decay and or mRNA function remains to be established, but their lack of deadenylases and decapping factors suggests that they are more likely utilized for mRNA storage than for decay (Ivanov et al., 2019; Protter and Parker, 2016).

In most bacterial mRNAs, decay initiates by endonuclease cleavage by RNase E, which provides the rate-limiting step of the process (Bandyra and Luisi, 2018; Hui et al., 2014; Mohanty and Kushner, 2016). Upon endonuclease cleavage, 3’−5’ exoribonucleases degrade the mRNA decay intermediates generated by RNase E cleavage (Bandyra and Luisi, 2018; Hui et al., 2014; Mohanty and Kushner, 2016). RNase E’s scaffolding activity of the RNA degradosome proteins, including conserved 3’−5’ exoribonuclease PNPase, help to stimulate the decay process (Ait-Bara and Carpousis, 2015; Bandyra and Luisi, 2018; Lopez et al., 1999). RNase E and 3’−5’ exonucleases work cooperatively with DEAD-box RNA helicases to ensure mRNA decay intermediates do not accumulate, which are only occasionally detected in wild type cells and accumulate when the IDR of RNase E or exoribonucleases are mutated (Coburn et al., 1999; Khemici and Carpousis, 2004; Morita et al., 2004). Caulobacter crescentus RNase E is known to recruit PNPase, RhlB, and RNase D into its RNA degradosome and BR-bodies (Al-Husini et al., 2018; Hardwick et al., 2011; Voss et al., 2014). The assembly of BR-bodies independent of RNA degradosome protein association was found to stimulate mRNA decay for the RNase E mRNA (Al-Husini et al., 2018), however, the exact mRNA decay steps stimulated by biomolecular condensation are not understood. Additionally, RNase E is known to control global mRNA decay in E. coli (Clarke et al., 2014; Hammarlof et al., 2015; Ono and Kuwano, 1979), with mutations blocking focus-formation leading to a slowdown in global decay rates (Hadjeras et al., 2019; Lopez et al., 1999), yet the entire collection of RNA substrates that BR-bodies act on in C. crescentus has not yet been defined.

In this study, we combine BR-body enrichment with RNA sequencing (RNA-seq) to define the RNA substrates of BR-bodies, and we use global mRNA half-life profiling of BR-body mutant strains to define their biochemical roles in facilitating mRNA decay. BR-body enrichment defines substrates predominantly as mRNAs and small RNAs (sRNAs) that act like miRNAs to silence mRNAs by base pairing. BR-body mRNA substrates are predominantly longer more poorly translated mRNAs, and we find that the RNA length is an intrinsic property that stimulates BR-body assembly in vitro. We find rRNA and the nucleoid are physically excluded from BR-bodies, thereby providing selective permeability of target substrates and rejecting molecules that compete for mRNA substrates. Additionally, global mRNA half-life profiling reveals that BR-bodies stimulate decay of BR-body enriched mRNAs by stimulating both the initial endo-cleavage by RNase E and the subsequent exonucleolytic decay step by degradosome associated exonucleases. This study therefore provides new functional insights into how the organization of the mRNA decay machinery into biomolecular condensates can facilitate robust biochemical pathways.

Results

RNase E cleavage stimulates rapid mRNA decay in C. crescentus

As E. coli RNase E provides the rate limiting cleavage in mRNA decay (Hammarlof et al., 2015; Ono and Kuwano, 1979) (Fig 1A), we sought to determine if the C. crescentus enzyme also provides this critical function. We measured and compared the mRNA half-lives using a modified RNA-seq assay for the wild type RNase E and a strain with an RNase E active site mutation (ASM) during exponential growth (Callaghan et al., 2005) (Fig 1B). The ASM contains D->N point mutations of the two residues which coordinate the catalytic Mg²⁺ ion in the active site (Al-Husini et al., 2018; Callaghan et al., 2005). The resulting bulk mRNA half-lives were 3.6 min for the wild type, and 7.6 min for the ASM, suggesting that endonuclease activity stimulates bulk mRNA decay (Fig S1A). The approximately 2-fold slowdown in mRNA decay rate observed with the ASM is slightly lower than the 5-fold decrease observed in an E. coli RNase E TS strain (Ono and Kuwano, 1979) perhaps due to the ASM mutant’s ability to maintain degradosome formation. While bulk estimates of sRNAs and tRNAs half-life are too stable to calculate, we observed a 3.0 min bulk half-life for cis-encoded antisense RNAs (asRNAs), whose lifetime increased slightly to 5.0 minutes in the ASM mutant. By examining the individual mRNA half-lives, we found the median mRNA half-life is 1.8 minutes with the active RNase E, while the median mRNA half-life rose to 3.9 minutes in the ASM mutant (Fig 1C). The median half-life is likely overestimated in the wild-type because we measure 553 more mRNA half-lives in the ASM strain and these mRNAs were predominantly undetectable by the wild type’s 3 min time-point. Across individual mRNAs we find that 96% of mRNAs have a longer half-life in the ASM strain as compared to wild type with an average slowdown of 2.4 fold (Fig S1), showing that initiation of mRNA decay by RNase E also stimulates global mRNA decay in C. crescentus (Fig 1A, S1B). ASM mutation did not affect median RNA half-lives across sRNAs or asRNAs, suggesting that these RNAs may be degraded by other RNases (Fig 1C).

Figure 1. RNase E cleavage is needed for rapid mRNA decay in C. crescentus.

A.) Cartoon of RNA degradosome mediated mRNA decay. The C. crescentus RNase E (black) degradosome components are shown in colored circles. The mRNA containing a 5’ terminal phosphate (red star) is shown in red. The first “entry” step of mRNA decay pathway is the rate limiting step performed by RNase E endonuclease cleavage. After initial cleavage, further decay occurs by 3’ to 5’ exoribonucleases. After exo decay into short RNA oligos, oligoRNases convert the small oligos into nucleotide mono-phosphates to complete decay. B.) RNA-seq measurement of wild type (WT, JS38) or active site mutant RNase E variant (ASM, JS299) after treatment with 200 μg/mL rifampicin for the indicated amount of time. Each row represents a different transcript whose RNA level is normalized to the level in untreated cells. Grey color represents a value too low to be determined. C.) Box-plots of RNA half-lives based on the data in panel A. Two-tailed T-test between FL and ASM mRNA decay rates (n=465 mRNAs) with uneven variance yield a p-value of 1.3 × 10⁻²³.

BR-bodies engage on longer poorly translated mRNAs

As RNase E’s endonuclease activity stimulates bulk mRNA-decay, we sought to determine the cellular RNA substrates of BR-bodies. Importantly, as native BR-bodies are highly labile and not detectible in a cell lysate, we utilized the ASM mutant of RNase E which blocks dissolution of BR-bodies and allows them to be stabilized during enrichment (Al-Husini et al., 2018). Initial affinity purification attempts of BR-bodies by an N-terminal HA-tagged ASM yielded highly purified RNase E, however, the bodies dissolved during the hours-long incubation times of purification and RNA was not detectable (Fig S2). Therefore we performed an enrichment of BR-bodies by rapidly separating BR-bodies away from cellular contents by differential centrifugation, similar to the procedure used to isolate stress granule cores (Khong et al., 2018; Wheeler et al., 2016) (Fig S2). Importantly, high yield RNA isolation was dependent on the presence of BR-bodies, as mock-treated lysates of a strain containing the NTD mutant (JS221 – a variant that is unable to form BR-bodies) yielded 3.5x lower RNA levels (Fig S2D). Enriched BR-body RNA levels were then compared to RNA levels in the cell lysate using RNA-seq (Fig 2A,B). By examining the fraction of reads among mRNAs and non-coding RNAs we noticed that the lysate contained 69.7% ncRNAs (63.4% rRNA, 5.8% tRNA, 0.005% sRNA, and 0.002% asRNA) and 25.7% mRNAs, while the BR-body enriched samples contained 13.3% ncRNAs (12.7% rRNA, 0.005% tRNA, 0.007% sRNA, and 0.006% asRNA) and 63.4% mRNA (Fig 2C).

Figure 2. BR-bodies are enriched for mRNAs.

A.) Differential centrifugation-based enrichment of BR-bodies. An aliquot of the cell lysate and enriched BR-bodies were RNA-extracted and libraries for RNA-seq were generated and sequenced. Red lines represent enriched mRNAs, black lines represent excluded RNAs. In the magnified view of the BR-body, RNase E tetramers are shown in green, with degradosome components as red circles, and Arg-rich regions in blue. B.) Normalized read density for the parB operon (top) and for a tRNA gene (bottom). C.) Fraction of RNA-seq reads mapping to each RNA category from whole-cell lysate (WCL, blue) or BR-body enriched samples (orange). Values are averages ± σ. D.) Log₂ ratio of the BR-body enriched sample RPKM compared to the cell lysate RPKM values of two biological replicates. RNAs with p-values >0.05 are colored in red Fraction of total reads mapping to non-coding RNA (blue) and mRNA (orange) in the lysate and in the enriched BR-body samples. p-values were determined using a negative binomial model in edgeR. E.) Box plots of BR-body enrichment for RNAs of each RNA category.

To explore which RNA species were enriched or depleted, we compared the log₂ ratio of enrichment (BR-body read density / total-RNA read density) for each RNA (Fig 2D, Table S1). Two biological replicates of this assay gave relatively similar agreement in the level of BR-body enrichment observed with R² = 0.95 (Fig 2D). Many conserved and highly structured ncRNAs, including tRNAs (−3.5 median enrichment), rRNA (−2.3), RNase P (−2.4), tmRNA (−2.8), 6S RNA (−1.3), and SRP RNA (−1.9) were highly depleted in BR-bodies (Fig 2E). As contaminating polysomes have been observed in centrifugation based RNA granule preps (Matheny et al., 2019), the strong depletion of rRNA suggest that enriched BR-bodies are not contaminated with polysomes. Conversely, mRNAs (1.6 median enrichment), sRNAs (1.1), and asRNAs (1.9) were predominantly enriched in BR-bodies (Fig 2E). We noticed that the level of RNA enrichment positively correlated with the length of the RNAs (Fig 3A, 3B) similar to stress granules and P-bodies (Khong et al., 2017; Matheny et al., 2019; Padron et al., 2019). While hundreds of sRNAs are known to exist in C. crescentus (Schrader et al., 2014; Zhou et al., 2015), only three have been functionally explored. Two trans-encoded sRNAs who can regulate target mRNAs through base pairing were found to be slightly enriched in BR-bodies (crfA 0.67 and chvR 0.27) (Frohlich et al., 2018; Landt et al., 2010), while another was found to be slightly depleted (gsrN −0.24) (Tien et al., 2018). Pairing of sRNAs with mRNAs typically occurs through the Lsm protein Hfq (Santiago-Frangos and Woodson, 2018; Vogel and Luisi, 2011), and we observe that 252/257 of these RNAs that associate with Hfq (Assis et al., 2019) are enriched in BR-bodies (Table S1).

Figure 3. Long poorly structured RNAs have enhanced association with BR-bodies.

A.) Log₂ ratio of the BR-body enriched RPKM vs cell lysate RPKM. RNAs were divided into bins based on their length and the medians of enrichment are highlighted (black bars). Conserved ncRNAs are indicated. Right, plot of RNA size distributions for BR-body enriched, depleted, or neither enriched nor depleted. Two-tailed T-tests with unequal variances resulting in p-values less than 0.05 are highlighted with asterisks. n=number of RNAs for depleted (n=51), neither (n=152), and enriched (n=2046). P-values are 2.6 × 10⁻⁶ (D to N), 1.5 × 10⁻⁵³ (E to N), and 1.0 × 10⁻³⁹ (D to E). B.) In vitro RNase E CTD-YFP biomolecular condensate recruitment assay with the indicated labeled RNAs. CTD-YFP was incubated with the indicated Cy-5 RNAs grouped by length before droplet formation was induced . Condensate assays were either performed in the presence of folded RNAs with Mg²⁺ (left) or with heat denatured RNAs lacking Mg²⁺ in the presence of PEG(8000) crowding agent (right). Quantification of droplet area is shown for each RNA on the right in the folded and denatured conditions. On the bottom, each sample is quantified compared to the protein only condition. Asterisks signify samples with p-values < 0.001 from one-way ANOVA. n displayed is the number of droplets analyzed.

As RNase E cleavage stimulates mRNA decay, we explored which mRNA features correlate with BR-body enrichment. The largest correlation coefficients with BR-body enrichment were observed for mRNA abundance (R=−0.51), GC% (R=0.47), mRNA length (R=0.41), translation level (R=−0.40), and codon optimality (nTEI) (R=0.30) (Fig S3). The negative correlation between mRNA abundance and BR-body enrichment suggests that BR-body mediated decay likely plays a role in shaping the cellular mRNA pool. The finding that longer poorly translated mRNAs are highly correlated with BR-body enrichment suggesting a similar mRNA preference as eukaryotic stress granules (Khong et al., 2017; Padron et al., 2019; Van Treeck et al., 2018). As the mRNA GC% is also correlated with nTEI (R=0.57) and sRNAs contain no GC bias in enriched RNAs (Fig S3B), this GC% correlation may relate to the underlying codon usage. Surprisingly, the mRNAs with lower nTEI are poorly enriched in BR-bodies, while those with more highly optimal codons are more highly enriched. As C. crescentus RNase E is arg-rich and arg is known to form pi-stacking interactions with purines (Boeynaems et al., 2019), we further explore the purine content of enriched mRNAs. We found that G% is positively correlated with BR-body enrichment of mRNAs (R=0.49), while A% is negatively correlated (R=−0.52), and the combined GA% led to a slightly negative correlation (−0.20) (Table S1). This suggests that RNase E’s in vivo RNA selectivity is not driven primarily by cation-pi stacking with purines. Interestingly, poly-G tends to form aggregated RNA structures in vitro (Boeynaems et al., 2019; Van Treeck et al., 2018) and its possible G-rich RNA aggregates may promote BR-body formation in vivo. Lower correlations were observed for translation efficiency (ribosome profiling footprint level normalized to the total RNA level (Schrader et al., 2014)) (−0.16), Shine-Dalgarno strength (R=−0.04), 5’ UTR length (R=0.02), and average tRNA adaptation index (TAI) (R=−0.01) (Fig S3C).

While mRNA length correlated rather strongly with BR-body enrichment (R=0.41), we also noted that sRNA and asRNA length correlated positively with BR-enrichment, although to a weaker extent due in part to the limited RNA size ranges of sRNAs and asRNAs (R=0.20 and R=0.22). Since RNase E’s CTD forms electrostatic interactions with the RNA, longer RNAs would have a higher degree of multivalency, suggesting they may stimulate phase-separation. Conversely, shorter RNAs such as generated by RNase E cleavage would have a lower degree of multivalency favoring dissolution. To explore the role of RNA length and RNA structure content on BR-body assembly, we added Cy5 dye-labeled RNAs of different sizes to purified CTD of RNase E which is both necessary and sufficient for BR-body formation (Al-Husini et al., 2018) (Fig 3B). A sample containing longer Cy5-poly-A RNA (average size 500 nt) was recruited into the CTD droplets. We also generated short Cy5-RNAs containing significant structure content (pre-5S rRNA and tRNA) and a short RNA with low structure content (poly-A <100 nt). We observed that all tested RNAs displayed a 1.1- to 2.5-fold increase in fluorescence intensity within CTD droplets versus the diffuse background. Comparing RNA-free conditions (RNase E only) to the indicated RNAs, long unstructured poly-A stimulated droplet size, while short unstructured poly-A had a smaller stimulation of droplet size (increase of 136.1 ± 8.2% [poly-A 500 nt], 55.0 ± 8.0% [poly-A <100 nt]). Interestingly, the pre-5S rRNA was recruited efficiently into the droplets and led to a mild increase in CTD droplet size (53.5 ± 7.9%), while the tRNA was not efficiently recruited and led to a reduction in CTD droplet size (decrease of 35.4 ± 7.8%). To explore whether structure content in the pre-5S and tRNA samples limited their recruitment into CTD droplets, we heat denatured the RNAs and co-assembled each with under conditions that lack Mg²⁺ and include PEG(8000) crowding agent to induce droplet formation. Both pre-5S rRNA and tRNA showed an increase in BR-body recruitment under denaturing conditions, suggesting suggests that the high level of RNA structure content limits pre-5S rRNA and tRNA co-assembly with BR-bodies. Under the denatured RNA conditions, long poly-A RNA maintained similar degree of recruitment and did not effect CTD droplet size, while all three shorter RNAs led to reduction in droplet sizes (decreases of 61.4 ± 5.1% [pre-5S rRNA], 68.5 ± 6.2% [tRNA], 45.5 ± 8.6% [poly-A <100 nt]). Interestingly, the pre-5S rRNA and tRNA assemble into smaller distinct zones within the protein-rich CTD phase, that display a 2.5- to 9.2-fold increase in fluorescence intensity versus the dilute background. The reduction in CTD droplet size by short and structured RNAs suggests that the in vivo preference of BR-bodies for long unstructured RNA substrates is an intrinsic property. As the size of condensates is a complex parameter, follow up studies will be needed to determine whether a difference in density, a shift in the binodal, an alteration of surface tension, or an alteration of coarsening kinetics is the source of the RNA-mediated change in droplet size. This also suggests that cleavage of the internal RNA, which decreases RNA length and the degree of multivalency, results in a BR-body size reduction.

BR-bodies are selectively permeable

As BR-bodies compete with ribosomes for free mRNA (Al-Husini et al., 2018), and rRNA was highly depleted from BR-bodies (Fig 2), we explored whether BR-bodies might exclude ribosomes in vivo. In C. crescentus cells expressing a ribosomal protein L1-eYFP fusion and an RNase E-eCFP fusion as the sole copies, we observe low ribosome density at the sites of BR-bodies (Fig 4A, S4). The exclusion of ribosomes from the BR bodies is further revealed by super-resolution microscopy in fixed C. crescentus cells expressing L1-PAmCherry and RNase E-eYFP as the sole copies; these images show that the ribosomes are distributed throughout the cells but are excluded from the RNase E foci. In measurements of 11 cells, we find that only 5% of the L1 localizations occur within 100 nm of an RNase focus (Fig 4B, S5A). This anti-localization is contrasted with the control case of cells expressing RNase E-eYFP and the degradosome component aconitase-PAmCherry; as expected, RNase E and aconitase strongly co-localize (Fig S5A). As the cytoplasm is filled with the nucleoid in C. crescentus, and RNase E was found to associate with the nucleoid (Montero Llopis et al., 2010), we also sought to explore whether the nucleoid was excluded from BR-bodies. We indeed observe exclusion of the nucleoid by DAPI staining (Fig 4C), suggesting BR-bodies create distinct subcellular compartments for mRNA decay (Fig S4,S5). Additionally, by heterologously expressing both the structured catalytic NTD and intrinsically disordered CTD of RNase E in E. coli cells, we find that the NTD colocalizes with the E. coli nucleoid, while the CTD forms nucleoid excluded condensates throughout the body of the cell (Fig 4C,S4).

Figure 4. Ribosomes and the nucleoid are excluded from BR-bodies.

A.) Dual labeled strain expressing ribosomal protein L1-eYFP (yellow) and BR-body scaffold RNase E (RNE)-eCFP (blue) as the sole copies (JS350). Red line is the source of the signal intensity plot for both fluorophores on right. B.) Super-resolution images of RNE-eYFP (green) and L1-PAmCherry (magenta) (JS545). White cell outlines from phase image. Red arrow indicates position of BR-body. In merged image, overlapping eYFP/PAmCherry signal is displayed in white. Scale bar is 500nm. C.) ASM-eYFP (JS299) (top), CTD-eYFP (JS230) middle, and NTD-eYFP (JS231) bottom colocalized with the DAPI stained nucleoid. Signal intensity plot for both fluorophores on right was generated from the red line. For E. coli cells, the black bar denotes the ribosomes rich cell pole which is known to exclude the nucleoid (Bakshi et al., 2012). Scale bars are 2 μm.

We used mRNA FISH and the Ms2 RNA labeling system to colocalize the highly-translated rsaA mRNA, which is known to have an unusually long mRNA half-life in C. crescentus (Lau et al., 2010), together with BR-bodies (Fig 5, S6). The rsaA mRNA is within the bottom quartile of mRNAs for BR-body enrichment (Log₂BR-enrichment=0.96). Using mRNA FISH or the Ms2 RNA labeling system, we observe that most cells containing the rsaA mRNA have a single RNA focus, in line with previous observations that C. crescentus mRNAs are limited in their movement from their genetic loci due to their high nucleoid/cytoplasm ratio (Gray et al., 2019; Montero Llopis et al., 2010). By mRNA FISH, we observed that 28% of mRNA foci show some extent of overlap with BR-body foci in diffraction-limited images, however, the rsaA mRNA is predominantly observed outside BR-bodies. Visualization of the rsaA mRNA by the Ms2-tagged RNA visualization system (Golding and Cox, 2004) using a Ms2 coat protein capsid assembly mutant (LeCuyer et al., 1995) in live cells also showed similar fractions of colocalization with BR-bodies (27%) (Fig 5A). By mRNA FISH and Ms2-tagging we predominantly observed cells with a single rsaA mRNA focus, in line with a previous report which found mRNAs colocalized to their DNA loci in Caulobacter (Montero Llopis et al., 2010). As a negative control, we also colocalized the rsaA mRNA with mCherry-PopZ, a polar protein known to form a matrix that excludes ribosomes with no known role in mRNA turnover (Bowman et al., 2008; Ebersbach et al., 2008). Here 12% of rsaA mRNA foci colocalized with a mCherry-PopZ foci, suggesting that a significant amount of the rsaA mRNA overlap with BR-bodies may be due to the poor resolution obtained by diffraction-limited images relative to the small cell size (Fig 5A). To explain the poor colocalization of BR-bodies with the rsaA mRNA (28%) we hypothesized that if mRNA decay occurs within BR-bodies we would expect a low fraction of colocalization due to the rapid internal mRNA decay. Previous work showed that BR-bodies are known to dynamically assemble and disassemble on the sub-minutes scale and that RNA cleavage is needed to disassemble BR-bodies (Al-Husini et al., 2018). Therefore, inhibiting RNase E endonuclease activity may lead to an accumulation of the rsaA mRNA in catalytically inactive BR-bodies. In line with this hypothesis, when RNase E’s endonuclease cleavage is blocked, colocalization of the rsaA mRNA and BR-bodies increased significantly with a majority of rsaA mRNAs (74%) becoming colocalized with BR-bodies (Fig 5B). Taken together, these results suggest that mRNA association with BR-bodies results in a short-lived assembly, wherein mRNA decay is stimulated followed by a rapid disassembly of the BR-body.

Figure 5. RNase E endonuclease activity limits rsaA mRNA colocalization in BR-bodies.

A.) rsaA mRNA weakly colocalizes with BR-bodies. Left, rsaA mRNA visualization in fixed cells by mRNA FISH or by the Ms2 tagged system in living cells. Fluorescein mRNA FISH probes were probed with either RNE-mCherry (JS403) or mCherry-PopZ (JP369) as a negative control. In live cells the Ms2-coat protein double mutant (Ms2DM)-mCherry fusion with an array of 96 tandem Ms2 RNA hairpins fused to the 3’ end of the rsaA gene was imaged with RNE-msfGFP (JS287). Right, quantitation of the fraction of rsaA mRNA foci colocalized with RNase E or PopZ foci. B.) Left, mRNA FISH (Quasar 670 fluorophore) in either wild type RNE-eYFP (JS38) or with ASM-eYFP (JS299). Right, quantitation of the fraction of rsaA mRNA foci colocalized with RNase E or ASM foci.

BR-bodies accelerate RNase E endo cleavage and degradosome exonucleolytic steps

To explore the functional organization of the RNA degradosome into BR-bodies we explored how mutants affecting BR-body assembly modulate mRNA decay using global mRNA half-life profiling (Fig 6). The NTD mutant lacks the IDR and therefore the ability to form a condensate or to assemble the RNA degradosome. In contrast, the degradosome binding site mutant (ΔDBS) lacks only the scaffolding activity of the IDR for degradosome exoribonucleases, while it retains condensate formation properties allowing the separation of functions (Al-Husini et al., 2018). The levels of RNase E variants were within a 2-fold range, allowing for a variant to variant comparison in activity (Fig S1D). In the strain expressing wild-type RNase E, the bulk mRNA half-life was the shortest (3.6 min) and mRNAs with lower BR-body enrichment tend to have longer half-lives, suggesting BR-body enrichment stimulates mRNA decay at a global level (Fig 6B). The NTD mutant had a modest bulk slowdown in decay (3.6 to 4.8 min), similar to an E. coli RNase E CTD truncation (Lopez et al., 1999). Importantly, the NTD mutant also showed a strong increase in half-lives occurring predominantly in mRNAs that are enriched in BR-bodies (Fig 6A,B). The ΔDBS mutant also showed a similar slowdown of bulk mRNA decay (3.6 to 4.5 min) with longer half-lives of mRNAs that are enriched in BR-bodies (Fig 6A,B).

Figure 6. BR-bodies accelerate initial cleavage and exonucleolytic steps of mRNA decay.

A.) RNA-seq measurement of wild type (JS38), NTD truncation (JS221), or DBS mutant (JS233) strains after treatment with 200 μg/mL rifampicin for the indicated amount of time. Each row represents a different transcript whose RNA level is normalized to the level in untreated cells. B.) mRNA half-lives for each RNase E mutant across four bins of BR-body enrichment. Only simple mRNAs with a single ORF and TSS are shown. Asterisks indicate samples with p-values ≤0.05 based on a T-test with unequal variance. p-values can be found in Table S1. C.) qRT-PCR (>100nt fragments) and RNA-seq (<50nt fragments) RNA half-life measurements of the rne mRNA for the indicated strains. Each half-life measurement was performed on the same RNA samples split between the two assays.

To explore whether different steps in mRNA decay were affected in the RNase E mutants, we used quantitative reverse transcription PCR (qRT-PCR), which examines the integrity of longer pieces of RNA >100nt in length, compared to the shorter 15–50nt fragments measured by RNA-seq. The RNA degradosome facilitates the multi-step mRNA decay process, in which the endonuclease RNase E makes the first initial cleavage followed by exonuclease activity of degradosome associated nucleases (Fig 6C). Therefore, half-lives measured by qRT-PCR will likely be most sensitive to the initial cleavage step, while half-lives measured by RNA-seq will likely be sensitive to both endo-cleavage and the partially cleaved mRNA decay intermediates generated by exoribonucleases (Fig 6C). We then tested the three RNase E variants by RNA-seq and qRT-PCR measurements on the same total RNA samples, yielding dual half-life measurements for 4 substrate mRNAs (rne, dnaA, ctrA, and gcrA) and the 9S rRNA which is known to be processed by RNase E (Hardwick et al., 2011). For the wild type RNase E, qRT-PCR half-lives ranged from 0.51–0.81 minutes, while RNA-seq measures ranged between 0.71–1.1 minutes. Out of the 4 mRNAs tested, only one showed a half-life that was significantly shorter by qRT-PCR than by RNA-seq with a difference of 0.3 minutes (Fig 6C, Table S2). qRT-PCR mRNA half-lives were all longer than wild type for the NTD mutant (ranging from 1.5–2.7 minutes), which were partially restored in the ΔDBS mutant (qRT-PCR mRNA half-lives 0.80–1.9 minutes) suggesting condensation stimulates RNase E endonuclease activity. When comparing to the RNA-seq derived half-lives, 3 out of 4 mRNA half-lives measured for the NTD mutant were found to be significantly shorter by qRT-PCR than by RNA-seq with differences on the order of 0.2–0.7 minutes (Fig 6C, Table S2). Finally, for the ΔDBS mutant, 4/4 mRNA half-lives were found to be significantly shorter by qRT-PCR than by RNA-seq with larger differences on the order of 1.2–1.8 minutes (Fig 6C, Table S2). Indeed, RT-PCR products of the full length ctrA mRNA show that initiation of mRNA decay is impaired in the NTD strain, but is not affected in the ΔDBS mutant (Fig S1E). Rates of 9S rRNA processing into the 5S rRNA, an essential function of RNase E that does not require the C-terminal IDR (Hardwick et al., 2011), were virtually identical for the wild type and NTD mutants (Table S2), with a significant slowdown in both qRT-PCR and RNA-seq measured in the ΔDBS mutant. The longer mRNA half-lives measured by RNA-seq relative to qRT-PCR for the NTD and ΔDBS mutant suggests that recruitment of degradosome associated exoribonucleases (PNPase and RNase D) into BR-bodies stimulates the secondary steps of mRNA decay (Fig 7).

Figure 7. Model of BR-body mediated mRNA decay.

As translation levels drop, RNA degradosomes (RNase E catalytic region (green), Arg-rich CTD patches (blue), and degradosome proteins (red) are highlighted) can engage on the mRNA to initiate decay. On the left, is the pathway of decay for soluble mRNA decay where initiation of decay by RNase E (green spheres) and exoribonuclease (red spheres) steps are indicated and can occur slowly. On the right, is the pathway that can occur inside BR-bodies. First, the RNA degradosome and mRNAs phase-separate into a biomolecular condensate via the multivalent interactions with the mRNA and the RNase E CTD. Inside the condensate RNase E endonuclease and degradosome exoribonuclease activity are both accelerated from the high-local concentration. Once the mRNA fragments are cut to a small size to lower valence of the interaction with RNase E’s CTD, the BR-body can dissolve releasing both RNA degradosomes and oligonucleotides that can be converted into nucleotides by oligoribonuclease.

Discussion

BR-bodies exhibit selective permeability

Selective permeability is a hallmark for membrane bound organelles. Though bacteria generally lack membrane bound organelles, bacterial microcompartments (BMCs) such as carboxysomes have been shown to exhibit selective permeability to facilitate the carbon fixation pathway (Kerfeld et al., 2018), however, BMCs exhibit a rather narrow species distribution. Biomolecular condensates are widespread across eukaryotic cells (Banani et al., 2017), yet their identification across bacteria has only recently been explored (Abbondanzieri and Meyer, 2019; Al-Husini et al., 2018; Monterroso et al., 2019; Wang et al., 2019). Overall, BR-body condensates exhibit selective permeability by assembling with or allowing in degradosome proteins, mRNAs, sRNAs, and asRNAs, and by rejecting ribosomes and the nucleoid. Such ability to organize enzymes (degradosomes) and their substrates (mRNAs) into biomolecular condensates may provide a more broadly utilized mechanism for bacterial subcellular organization. Indeed, rubisco itself was recently shown to phase separate with the cyanobacterial protein CcmM which facilitates the assembly of the carboxysome shell (Wang et al., 2019). While BR-bodies and CcmM-Rubisco condensates are currently the only bacterial biomolecular condensates whose components have been directly shown to form liquid-like droplets in physiological conditions (Al-Husini et al., 2018; Wang et al., 2019), the cell division protein FtsZ and its inhibitor SlmA were shown to form condensates upon addition of crowding reagents (Monterroso et al., 2019). In addition, other bacterial proteins such as the polar protein scaffold PopZ (Lasker et al., 2020; Zhao et al., 2018) and the nucleoid associated protein DPS (Janissen et al., 2018) have been shown to act as selectively permeable scaffolds, yet phase separation has not been directly observed.

Across domains of life, there is a well-known antagonistic relationship between the processes of mRNA translation and mRNA decay (Dreyfus, 2009; He and Jacobson, 2015). Recent data have suggested that translation initiation and elongation can both impact global mRNA decay (Chan et al., 2018; Presnyak et al., 2015), and translating ribosomes were found to be physically excluded from P-bodies and stress granules (Hubstenberger et al., 2017; Reineke et al., 2012). Here we showed that BR-bodies physically exclude ribosomes, providing a physical separation of mRNA decay and translation in bacteria (Al-Husini et al., 2018). The exclusion of ribosomes from BR-bodies may also explain why long poorly translated mRNAs are enriched BR-body substrates. While eukaryotic RNA silencing machinery is found in P-bodies and stress granules (Hubstenberger et al., 2017; Jain et al., 2016), sRNAs were found to be enriched in BR-bodies (Fig 3A) and are known to base pair with mRNAs often through the RNA chaperone Hfq (Fei et al., 2015; Santiago-Frangos and Woodson, 2018; Vogel and Luisi, 2011). Indeed, Hfq bound RNAs identified by HITS-CLIP (Assis et al., 2019) including both mRNAs and sRNAs are enriched in BR-bodies (Fig S6, Table S1), and an E. coli CTD truncation mutant led to reduced sRNA induced mRNA decay (Fei et al., 2015), suggesting mRNA silencing may trigger BR-body localization. Cis-encoded asRNAs were also found to be enriched in BR-bodies (Fig 3) similar to stress granules where many short asRNAs associate through pairing to their complimentary mRNAs (Van Treeck et al., 2018). In bacteria asRNAs can be generated by internal promoters or by errors in rho-dependent transcription termination, and interestingly Rho protein was found to associate with RNase E when cells are grown at cold temperature (Aguirre et al., 2017), suggesting a link between these two processes. Importantly, mRNAs are the only class of substrate whose decay is globally stimulated by RNase E, while sRNAs and asRNAs have similar median half-lives in the ASM (Fig 1B, Table S1).

Most highly structured ncRNAs involved in translation-related processes (RNase P, tRNAs, SRP RNA, tmRNA, and rRNA) are highly depleted from BR-bodies. It is possible that the high level of secondary structure may exclude these ncRNAs from BR-bodies as observed for other condensates (Boeynaems et al., 2019; Langdon et al., 2018), and/or strong association with specific RNA binding proteins (such as ribosomal proteins) limits these RNAs from entering BR-bodies. Interestingly, tmRNA, which functions to rescue non-stop mRNAs, was found to form diffraction limited foci in C. crescentus (Russell and Keiler, 2009), and we find tmRNA to be highly depleted from BR-bodies (Fig 3A), suggesting that ribosome rescue occurs in distinct cytoplasmic bodies.

BR-bodies stimulate mRNA decay entry and prevent intermediate release

Despite broad identification of biomolecular condensates across organisms and cell types, the functional mechanisms of biomolecular condensates on their internal biochemical processes remains poorly understood. The high concentration of reactants and enzyme within the condensate has been proposed to have a stimulatory effect. This is true for the innate immune signaling protein cGAS, where condensation with DNA has a stimulatory effect on the rate of cGAMP production (Du and Chen, 2018). Similarly, condensate stimulation of actin polymerization has been shown for Arp2/3 N-WASP condensates (Banjade and Rosen, 2014; Li et al., 2012). In other cases, condensates have been proposed to allow substrate selectivity by partitioning only certain substrates together with the enzyme within the condensate (Banani et al., 2017).

In BR-bodies, the organization of the mRNA decay factors into a biomolecular condensate appears to perform multiple biochemical roles. BR-body condensation is stimulated upon the presence of long untranslated mRNA due to the multi-valent interactions, which are then reduced when the RNA is cleaved leading to the dissolution of the BR-body (Fig 7) (Al-Husini et al., 2018). The NTD mutant of RNase E which cannot form condensates is slow in the initial mRNA cleavage step by RNase E (Fig 6) (Al-Husini et al., 2018), suggesting condensation stimulates initial cleavage. The stimulation of the initial cleavage rate is likely due to the condensate and not disruption of the RNA degradosome as a mutant lacking all degradosome binding sites (ΔDBS) in the IDR was still able to accelerate initial RNase E cleavage (Table 1) (Al-Husini et al., 2018). BR-body accelerated initial mRNA cleavage is likely conserved across bacteria as E. coli RNase E mutants that disrupt focus formation through IDR deletion or through deletion of a critical inner-membrane attachment helix required for focus formation both lead to slower global rates of mRNA decay (Hadjeras et al., 2019; Lopez et al., 1999; Strahl et al., 2015). Interestingly, the RNase E ΔDBS mutant lacking binding sites for exoribonucleases PNPase and RNase D was also found to have a lag between initial cleavage and decay of smaller RNA fragments suggesting a build-up of mRNA decay intermediates (Fig 6, Table 1). A similar buildup of mRNA decay intermediates is observed in E. coli when the conserved degradosome component PNPase is mutated (Coburn et al., 1999; Khemici and Carpousis, 2004; Morita et al., 2004). The accumulation of decay intermediates may explain why the ΔDBS strain grows more slowly than wild type or the NTD strain (Al-Husini et al., 2018). This suggests that another biochemical function of BR-bodies is to prevent the buildup of toxic mRNA decay intermediates (Fig 7). By spatially coordinating the assembly of poorly translated mRNAs and RNA decay enzymes into BR-bodies, bacteria can both accelerate the endo and exonuclease reaction rates and simultaneously prevent premature release of toxic reaction intermediates.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
α-RNase E antibody	Ben Luisi (U. Cambridge)	N/A
Goat α-Rabbit IgG (H+L) Secondary Antibody, HRP	Thermo scientific	31460; RRID: AB_228341
Bacterial and Virus Strains
Caulobacter crescentus NA1000	Lucy Shapiro, Stanford University School of Medicine	N/A
E. coli TOP10	Thermofisher scientific	C404010
E. coli BL21(DE3)	NEB	C2527I
Chemicals, Peptides, and Recombinant Proteins
Vanillic acid	Fluka	94770–50G
Xylose	Sigma Aldrich	X1500–500G
Gentamycin	Sigma Aldrich	G1264–5G
Kanamycin	Sigma Aldrich	K1377–5G
Chloramphenicol	Sigma Aldrich	C0378–25G
Spectinomycin	Sigma Aldrich	S4014–5G
Streptomycin	Sigma Aldrich	S6501–25G
Rifampicin	Sigma Aldrich	R7382–1G
Lysozyme	Lucigen	R1804M
RNase free DNase I	Roche	04 716 728 001
0.1% w/v polylysine	Sigma Aldrich	P8920–100ML
EDTA-free protease	Rouch	11873580001
T4 Polynucleotide Kinase	NEB	M0210L
T4 RNALigase2 truncated KQ (T4Rn12tr KQ)	NEB	M0373L
SuperScript III Reverse Transcriptase	invetrogen	56575
Phusion DNA Polymease	Thermo Scientific	F-530L
Triton™ X-100	Sigma Aldrich	T8787–100 ML
β-mercaptoethanol (βME)	Amresco	M131–100ML
Tergitol solution	Sigma Aldrich	NP-40s
para-formaldehyde (37%)	Sigma Aldrich	F8775
Superase-In Rnase inhibitor	Invetrogen	AM2694
RNAprotect Bacterial Reagent	QIAGEN	76506
SYBR-Gold nucleic acid gel stain	Invitrogen	S11494
TRizol Reagent	Ambion	15596018
EDTA-free protease inhibitor tablet	Roche	11873580001
Stellaris RNA FISH hybridization buffer	Biosearch technologies	SMF-HB1–10
Stellaris RNA FISH wash buffer A	Biosearch technologies	SMF-WA1–60
Stellaris RNA FISH wash buffer B	Biosearch technologies	SMF-WB1–20
Antifade Mounting Medium	VECTASHIELD	H-1000
Universal miRNA cloning linker	NEB	S1315S
T4 RNA ligase 1	NEB	MO 473M
Amino allyl UTP	Jena Bioscience	NU-821–776
pCp- Cy5	Jena Bioscience	NU-1706-CY5
cy5-NHS ester	Lumiprobe	13020
Critical Commercial Assays
Luna® Universal One-Step RT-qPCR Ki	NEB	E3005L
Qubit™ RNA HS Assay Kit	Thermo Fisher Scientific	Q32852
Qubit™ dsRNA HS Assay Kit	Thermo Fisher Scientific	Q32851
Core Kit- rRNA remocval beads	illumina	MRZ116C
Ribo Zero rRNA remocval reagents	illumina	RZNB56
CircLigase ssDNA Ligase	epicentre	CL4115K
Pierce Anti-HA Agarose	Thermo Fisher Scientific	26182
Pierce HA peptide	Thermo Fisher Scientific	26184
rsaA- Flourescin Dye FISH probes	Stellaries	SS508160-01-48
rsaA- Cy5 (Quasar 670 Dye) FISH probes	Stellaries	SS531045-01-45
Deposited Data
BR body enrichment RNA-seq datasets	NCBI GEO	GSE133522
BR-body mutant mRNA half-life profiling datasets	NCBI GEO	GSE133532
Experimental Models: Organisms/Strains
See table S3 (strains and plasmids table)
Oligonucleotides
See table S3 (oligonucleotides table)
Recombinant DNA
See table S3 (strains and plasmids table)
Software and Algorithms
microbeJ	Ducret et al. 2016	http://www.microbej.com/
imageJ	Rueden et al. 2017	https://imagej.nih.gov
Other

STAR Methods

RESOURCE AVAILABILITY

All unique/stable reagents generated in this study are available from the Lead Contact without restriction.

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Jared M. Schrader (Schrader@wayne.edu).

Materials Availability

All strains and plasmids will be provided by request to the lead contact.

Data and Code Availability

High-throughput RNA-sequencing data is deposited in the NCBI GEO omnibus (accession numbers GSE133522 & GSE133532).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Caulobacter crescentus cell growth

All Caulobacter crescentus strains used in this study (Table S3) were derived from the wild-type strain NA1000 (Evinger and Agabian, 1977), and were grown at 28°C in peptone-yeast extract (PYE) medium or M2 minimal medium supplemented with 0.2% D-glucose (M2G) (Schrader and Shapiro, 2015). When appropriate, the indicated concentration of vanillate (500μM), xylose (0.2%), gentamycin (0.5 μg/mL), kanamycin (5 μg/mL), chloramphenicol (2 μg/mL), spectinomycin (25 μg/mL), and/or streptomycin (5 μg/mL) was added. Strains were analyzed at mid-exponential phase of growth (OD 0.3–0.6). Optical density was measured at 600 nm in a cuvette using a Nanodrop 2000C spectrophotometer. Replacements strains containing a xylose inducible copy of RNase E and a vanillate inducible RNase E variant were first grown in media containing xylose overnight, then washed 3 times with 1mL growth media, and resuspended in growth media containing vanillate, diluted, and grown overnight. Log-phase cultures were then used for the downstream experiment. Ribosome exclusion: JS348 and JS350 cells were grown overnight in PYE-Gent-Spec in 3 dilutions. Log-phase cells were split into two tubes and either treated with 10% Ethanol for 10 min. or left untreated as indicated. 1μL of the cells under each condition was spotted on a M2G 1.5% agarose pad and imaged using YFP, CFP, and TX-Red filter cubes as indicated. Nucleoid exclusion: JS299 cells were grown in PYE-Gent-Kan media containing 0.2% xylose overnight. The next day, the log-phase cells were washed 3 times with PYE media, and used to inoculate PYE-Gent-Kan media containing 0.5mM vanillate, diluted, and grown overnight. Log-phase cells were treated with 5 ng/μL DAPI for 15 minutes (cultures were covered with foil while shaking) and spotted on a M2G 1.5% agarose pad and imaged using both YFP and DAPI filter cubes.

E. coli cell growth

E. coli strains were grown at 37°C and cultured in LB medium (L3522, Sigma), supplemented with the indicated concentration of kanamycin (30 μg/mL) or ampicillin (50 μg/mL). For induction, BL21 DE3 cells were induced with isopropyl-3-Dthiogalactopyranoside (IPTG) (1μM) for two hours and TOP10 cells were induced with 0.0004% arabinose for one hour. Strains were analyzed at mid-exponential phase of growth (OD₆₀₀ 0.3–0.6). Optical density was measured at 600 nm in a cuvette using a NanoDrop 2000C spectrophotometer. Nucleoid exclusion (E. coli): JS230 and JS231 cells (TOP10) were grown at 37°C in LB medium supplemented with 50 μg/mL ampicillin. For induction, log-phase cells were induced with 0.0004% arabinose 1 hour. After induction, cells were treated with 5ng/μl DAPI for 15 minutes before spotting on 1.5% agarose pad and imaging.

Strain and Plasmid Construction

All DNA oligos used for strain generation are listed in table S3.

JS403: NA1000 rne::rne-mCherry Gent^R

The RNase E insert was digested by NdeI/KpnI from pRNE-YFPC-1 and ligated to NdeI/KpnI digested pChyC-4. Resulting plasmids were transformed into E. coli and selected on LB gent plates, miniprepped, and sequence verified. Resulting pRNE-ChyC-4 was transformed into NA1000 cells, selected on PYE-gent plates, and verified by fluorescence microscopy.

JS25: NA1000 vanA::Ms2(V75E/A81G)-mCherry Spec^R Strp^R

pVMs2(V75E/A81G)-mCherryC-1 plasmid was generated by IDT as a gblock fragment. The gblock fragment was assembled into pVChyC-1 (cut with NdeI and KpnI) via Gibson assembly. The insert was sequence verified then transformed into NA1000 cells via electroporation and selected on PYE spec/str. Resulting colonies were screened for vanillate inducible mCherry expression by fluorescence microscopy.

MS2 double mutant gblock GCGAGGAAACGCATATGATGGCTTCTAACTTTACTCAGTTCGTTCTCGTCGACAATGGCGGAACTGGCGACGTGACTGTCGCCCCAAGCAACTTCGCTAACGGGGTCGCTGAATGGATCAGCTCTAACTCGCGTTCACAGGCTTACAAAGTAACCTGTAGCGTTCGTCAGAGCTCTGCGCAGAATCGCAAATACACCATCAAAGTCGAGGTGCCTAAAGTGGCAACCCAGACTGTTGGTGGAGAGGAGCTTCCTGTAGCCGGCTGGCGTTCGTACTTAAATATGGAACTAACCATTCCAATTTTCGCTACGAATTCCGACTGCGAGCTTATTGTTAAGGCAATGCAAGGTCTCCTAAAAGATGGAAACCCGATTCCCTCAGCAATCGCAGCAAACTCCGGCATCTACGGTACCTTAAGATCTCG

JS 430: NA1000 vanA::Ms2(V75E/A81G)-mCherry rsaA::rsaA-96 array Kan^R Spec^R Strp^R

This strain was constructed at multiple steps. First, The Ms2 96 array was moved into pFlagC-2 by Gibson reaction. This was done by PCR amplifying the array from BAC2(P(lac,ara)- mRFP1 – 96bs) plasmid (Golding and Cox, 2004) with the following Gibson primers:

1. j5_00268_(Ms2array)_forward

   CTCCGGAGAATTCCGATTAGCTGCGCATCCCCC

2. j5_00269_(Ms2array)_reverse

   GGACAAAAACAGAGAAAGGAAACGACAGAGGCACCGGTCCGAC

The amplified 96 array fragment was gel purified and Gibson Ligated into pFlagC-2 plasmid (Thanbichler et al., 2007) amplified with the following Gibson primers:

1. j5_00270_(pflgc2EcoRIAgeI)_forward

   GGAAACGACAGAGGCACCGGTCCGACTACAAGGATGACG

2. j5_00271_(pflgc2EcoRIAgeI)_reverse

   CCTTAAGATCTCGAGCTCCGGAGAATTCCGATTAGCTGCGC

The Gibson ligation reaction was transformed into E. coli top10 cells and selected on LB-kan plates. Resulting kan^R colonies were then screened by PCR for the insert, and then verified by sanger sequencing(genewiz) pFlagC-2–96array. Second, he last 621 bp of rsaA gene was cloned into pFlagC-2–96array. This was done by PCR amplification of the rsaA fragment using the following primers:

1. rsaA F KpnI

   ATAAGGTACCCTGAACCTGACCAACACCGG

2. rsaA R EcoRI

   ATATTAGAATTCTTAGGCGAGCGTCAGGACTTCG

The PCR amplified fragment was gel purified, cut with Kpn1/EcoR1 and ligated to Kpn1/EcoR1 cut pFlagC-2–96array. The ligation was transformed into E. coli top10 cells and selected on LB-kan plates. Resulting kan^R colonies were then screened by PCR for the insert, and then verified by sanger sequencing(genewiz) making pFlagC-2-rsaA-96array. The purified plasmid was then transformed into NA1000 cells via electroporation and plated on PYE-kan plates.

Third, the pVMS2(V75E/A81G)-mCherryC-1 plasmid was transformed and cells were selected on PYE kan spec strep plates. Resulting colonies were screened for integration at the vanA locus (Thanbichler et al., 2007).

JS287: NA1000 vanA::Ms2(V75E/A81G)-mCherry rsaA::rsaA-MS2 96 array rne::rne-msfGFP Kan^R Spec^R Strp^R Gent^R

JS430 cells were transduced with phage lysate from JS87 and selected on PYE gent plates. Resulting colonies were verified to have GFP expression.

JS348: NA1000 L1::L1-eYFP acnA::acnA-mCherry Gent^R Spec^R

This strain was made by transducing JS290 cells(plasmid from Jared) with aconitase mcherry phage lysate from JS134 strain (Al-Husini et al., 2018) and the cells were plated on PYE-gent-spec-strp plates. The growing colonies were restreaked three times, sequentially, on PYE-gent-spec-strp plates and validated by imaging

JS350: NA1000 L1::L1-eYFP rne::rne-eCFP Gent^R Spec^R Strp^R

This strain was generated by Transducing JS255 cells with L1-eYFP phage lysate from JS 290 strain. The cells were plated on PYE-gent-spec-strp plates. The growing colonies were restreaked three times, sequentially, on PYE-gent-spec-strp plates and validated by imaging.

JS302: NA1000 vanA::rne(HA-ASM)_mut2)-eYFP rne::pXrnessrAC Gent^RKan^R

This strain was constructed by generating the pVRNE-HA-ASM _mut2)-eYFPC-4 from pVRNE(ASM _mut2)eYFPC-4 plasmid (Al-Husini et al., 2018) by site directed mutagenesis using the following primers:

1. HA add for

   TACCCGTACGACGTCCCGGACTACGCCCTGATCGACGCAGCACACG

2. HA add rev

   CATCTTCTTCGACATCATATGGTCGTC

First, the fragment was PCR amplified using the T4 PNK kinased oligos and using pVRNE(ASM mut2)YFPC-4 plasmid as a template. The resulting PCR reaction was gel purified and self-ligated. The ligation was transformed into E. coli TOP10 cells and selected on LB-gent plates. Resulting gent^R colonies were then verified by sanger sequencing (genewiz). Second the purified plasmid was transformed into NA1000 cells by electroporation and the transformants were selected by plating on PYE-Gent plates. Resulting colonies were grown in the presence and absence of vanillate and verified by fluorescence microscopy. Third, the NA1000 cells were transduced with phase lysate from JS8 cells (Al-Husini et al., 2018) and the cells were plated on PYE-Gent-Kan-xylose plates. The resultant colonies were patched on PYE-Gent-Kan plates with and without xylose to confirm the depletion phenotype.

JS545: NA1000 L1::L1-PAmCherry rne::rne-eYFP gent^R spec^R str^R

Ribosomal protein L1 PAM fusion was generated by cloning the PAmCherry gene into pL1-YFPC-4 (Bayas et al., 2018). First NheI/AgeI digested pL1-YFPC-4 and PAmCherry insert was ligated to create pL1::PAMC-4. The resulting pL1::PAMC-4 plasmid was selected on LB-Kan plates and sequence verified. The plasmid was then transformed into NA1000 cells by electroporation, and the resulting gentR colonies were screened using fluorescence microscopy. Phage lysate from strain JS51 was then transduced into strain cells harboring pL1::PAMC-4 and selected on PYE-spec-gent plates. Resulting colonies were verified to express both fusions by fluorescence microscopy.

JS546: NA1000 rne::rne-PAmCherry acnA::acnA-eYFP gent^R spec^R str^R

pRNE-YFPC-1 was first digested with NdeI/KpnI and ligated with pPAmCherryC-4 (NdeI/KpnI digested) and transformed into E. coli cells. Resulting gent^R colonies were screened for the insert and sequence verified yielding pRNE-PAmCherryC-4. pRNE-PAmCherryC-4 was transformed into NA1000 cells and selected on PYE gent plates. Gent resistant colonies were verified to have RNE-PAmCherry expression by fluorescence imaging. Next cells were transduced with phage lysate from JS251 (Al-Husini et al., 2018) and selected on PYE gent-spec-str plates and expression of both fusions was verified by fluorescence microscopy.

METHOD DETAILS

In vitro droplet assembly assay

The RNase E CTD-YFP was purified as described in (Al-Husini et al., 2018). Cy5 dye-labeled RNAs were generated by in vitro transcription using a 9:1 ratio of UTP:aminoallyl-UTP (9S and tRNA samples (t Hoen et al., 2003)) followed by conjugation of a Cy5 NHS ester or by 3’ ligation of Cy5 pCp using T4 RNA ligase 1 (yeast total RNA, Poly-A (England et al., 1980)). Shorter poly-A samples were generated by base hydrolysis followed by 3’ repair by T4 PNK before ligation of 3’ Cy5 pGp. Denatured Cy5-labeled RNAs were prepared by heating to 85 °C for 2 minutes immediately before addition. Indicated RNAs were incubated with DTT, RNase E CTD, and NaCl. Either MgSO₄ (folded RNA) or PEG(8000) (denatured RNA) was then added to the solution to induce droplet formation. The final concentrations of each component are 20 μM RNase E CTD, 100 mM NaCl, 500 μM DTT, 10 ng/μL RNA, and either 10 mM MgSO₄ or 10% PEG(8000).

10 μL of the protein-RNA mixture placed in a well formed by a SecureSeal™ Spacer (EMS) and sealed with a glass coverslip. The droplets were incubated for 1 hour coverslip-side down before imaging. Images were taken with with an Eclipse Ti-E inverted microscope (Nikon) with a Plan Apo-λ 100x/1.45 oil objective with 518F immersion oil (Zeiss). Images were taken in both phase-contrast and fluorescence imaging with a halogen lamp as the white light source. A Spectra X light engine was used for fluorescence excitation, and Chroma CFP/YFP/mCh (77074157) and Cy5 (77074160) excitation filter cubes and CFP/YFP/mCh (77074158) and Cy5 (770741161) emission filter sets were used for fluorescence imaging. An Andor Ixon Ultra 897 EMCCD camera was used for all imaging. Nikon Element AR software was used to control the setup.

All images were processed using ImageJ. Droplets were outlined with the microbeJ plugin using a Default selection method and a dark background, adjusting Z-scores manually per image set. Only droplets with a YFP fluorescence intensity greater than 300 F.U. were included. Particles were manually separated using the particle slicer tool as needed. Histograms were generated using Prism 8 with a bin number of 5. A one-way ANOVA was used to analyze significance between the groups compared to the no-RNA (RNase E only) population.

Super-resolution imaging of BR-bodies and exclusion analysis

C. crescentus JS545 (L1-PAmCherry/RNaseE-eYFP) and JS546 (acnA-eYFP/RNase E-PAmCherry) cells were grown in liquid M2G media with 1.0 μg/mL gentamycin and 25 μg/mL spectinomycin to OD ~0.6 and fixed at log phase using formaldehyde cross-linking with 1% fixation buffer (11% HCOH in 1x PBS at pH 7.3). The fixed cells were then spotted onto a pad of 1.5% agarose in M2G media and fiduciary marker 0.35 μm Fluoresbrite carboxylate YG beads were added at a concentration of 5.0 × 10⁷ spheres/mL. The cells were sandwiched between two coverslips and imaged on an Olympus IX71 inverted epifluorescence microscope with a 100× objective (NA 1.40). Each camera pixel corresponds to 49 nm × 49 nm in the sample.

Super-resolution microscopy was performed as described previously (Tuson and Biteen, 2015). Briefly, the eYFP fusions were imaged under 488-nm excitation (Coherent Sapphire 488–50) at a power density of ~2.5 × 10⁶ μW/mm². The PAmCherry fusions were photo-activated at 406 nm (Coherent Cube 406 laser) for 30 ms at 5.0 × 10⁵ μW/mm². The PAmCherry was imaged under 561-nm excitation (Coherent Sapphire 561–50) at ~2.5 × 10⁶ μW/mm². The fluorescence emission was filtered with appropriate filters and imaged on a 512 × 512 pixel Photometrics Evolve electron-multiplying charge-coupled device (EMCCD) camera. Each channel was imaged sequentially. The PAmCherry was imaged first to avoid leakage of the eYFP signal (which extends to ~600 nm) into the 561-nm channel.

Super-Resolution Image Reconstruction

Single molecules were detected and fit as previously described using the SMALL-LABS algorithm (Isaacoff et al., 2019). In these fixed cells, individual molecules were fluorescent for multiple frames without moving. We therefore avoided over-counting single molecules with a spatio-temporal filter (Bayas et al., 2018): for each localized molecule, we fit the probability of finding an additional localization within 1.5 times the average 95% confidence interval of the fit over time to a biexponential decay, and used the short decay time as a temporal threshold. All single-molecule detections within 1.5 times the average 95% confidence interval and within the calculated temporal threshold were therefore combined as a single localization. We additionally avoided false positives by including only molecule detections that were tracked for at least five frames. Images were reconstructed as a histogram of number of fits in each 25 nm × 25 nm pixel on a grid, and then a Gaussian filter of 25 nm width was applied to the image to account for localization precision. To determine colocalization, RNase E foci were identified within these super-resolution histograms as collections of 25 nm × 25 nm pixels with counts greater than the average plus two standard deviations. A 1-pixel padding was added around each focus.

Fluorescence In Situ Hybridization (FISH)

5 mL of cells were grown in M2G medium in a 28 °C shaker incubator in the presence of the appropriate antibiotics. For JS38 and JS299 the growth medium was supplied with 0.2% xylose, 0.5μg/mL gentamycin, and 5μg/mL kanamycin. Next day, the 5 mL overnight cultures were washed 3 times with M2G media and used to inoculate 25 mL M2G medium (in 3 dilutions) containing 500μM vanillate, 0.5μg/mL gentamycin, and 5μg/mL kanamycin and grown for 8 hours. JS403 and JP469 cells were grown in M2G medium with 0.5μg/mL gentamycin. For all the strains, log-phase cultures were then fixed with 7.5% para-formaldehyde (Sigma) for 15 min at 28 °C followed by incubation at 4 °C for 30 minutes. The fixed cells were harvested by centrifugation (11,000xg /3 min.). The cell pellets were washed 2 times in ice-cold 1% PBS (140mM NaCl, 3mM KCl, 8mM sodium phosphate, and 1.5mM potassium phosphate [PH 7.5]). The cell pellets were resuspended in ice-cold 100% ethanol and stored at −20°C for less than 1 week. When ready to proceed, the cells were pelleted at 4 °C (2655 x g for 5 minutes) and washed 3 times with GTE buffer (50 mM glucose, 20mM tris-HCl PH7, and 10mM EDTA). The cells were lysed with 4000U of RNase-free lysozyme (Lucigen) at 37 °C for 2 hours. Permeabilized cells were pelleted again and resuspended in 100 μL of hybridization buffer (Stellaris RNA FISH) containing 1μL of the rsaA FISH custom Assay DNA probes (Stellaris) or were RNase A treated prior to probing as a control. FISH probes with fluorescein Dye were used for probing rsaA of JS 403, and JP369 mCherry-popZ cells. FISH probes with Quasar 670 Dye were used for probing rsaA of JS38 and JS299 cells. The samples were incubated with the probes in the dark at 42 °C overnight. Next day, the cells were pelted and washed in wash buffer A (Stellaris RNA FISH) at 42 °C for 30 minutes. The cells were pelleted and washed with 1 mL of wash buffer B (Stellaris RNA FISH) for 5 minutes at 42 °C. the cells were pelleted and resuspended in GTE buffer supplied with 0.1% TX100. The suspended cells were spotted on a polylysine coated microscope slide, dried, washed 3 times with 1% PBS and air-dried again. A 5 μL drop of mounting medium (VECTASHIELD) was added, and the slides were covered with coverslips and the samples were imaged with a Nikon Eclipse NI-E with CoolSNAP MYO-CCD camera and 100x Oil CFI Plan Fluor (Nikon) objective, driven by Nikon elements software using appropriate filter cubes (Cy5, GFP, Texas Red). Using microbeJ (Ducret et al., 2016) the fluorescent foci were identified using the “maxima” function in microbeJ with “foci” selected as shape, with tolerance and Z-score parameters tuned for each image. Aberrant foci with area < 0.01 μm² and length > 1 μm were removed, and the segmentation option was used to split adjoined foci.

Visualization of the rsaA mRNA with the MS2 system

Strain JS287 containing a non-dimerizable mutant of the Ms2 coat protein fused to mCherry at the vanillate locus and an array of 96 Ms2 RNA hairpins (gift of Ido Golding) fused to the 3’ end of the rsaA gene were grown into log phase and induced with 0.5 mM vanillate for four hours and imaged on an M2G agarose pad. As a control, (JS25) lacking the integrated hairpins was imaged where no fluorescent foci were detected (Fig S6). Fluorescent foci were identified using microbeJ (Ducret et al., 2016) using the “maxima” function in microbeJ with “foci” selected as shape, with tolerance and Z-score parameters tuned for each image. Aberrant foci with area < 0.01 μm² and length > 1 μM were removed, and the segmentation option was used to split adjoined foci. To measure colocalization of the rsaA mRNA with BR-bodies, we manually scored each mRNA focus for overlapping an RNase E focus or not. A minimum of 100 foci per strain were used for this analysis.

Exclusion analysis of fluorescent microscopy images

Multi-channel images of test molecules with a BR-body protein fusion (either RNase E or aconitase) were aligned using the “align image by line ROI” plugin in FIJI. Next, fluorescent foci were identified using the “maxima” function in microbeJ with foci selected as shape, with tolerance and Z-score parameters tuned for each image. Aberrant foci with area < 0.01 μm² and length > 1 μM were removed, and the segmentation option was used to split adjoined foci. For each focus, we manually drew a line-slice across the focus and recorded the fluorescence intensity of each channel, then ran a pearson correlation function on the intensities of each channel. In the case of the RNase E NTD expressed in E. coli where no foci form, the density of the nucleoid by DAPI intensity was used for the line slice instead. Resulting correlation coefficients were then reported for each focus, with a correlation coefficient of 1 representing perfect correlation, 0 representing no correlation, and −1 representing perfect anti-correlation (Fig S4). As a positive control we imaged (acnA-mCherry/RNE-eYFP) which both go into BR-bodies (Al-Husini et al., 2018), as an uncorrelated control we examined (AcnA-mCherry/RNEAconBS-YFP) where RNEAconBS-YFP still forms foci but AcnA-mCherry is diffuse across the cytoplasm (Al-Husini et al., 2018), and as a negative control we imaged (mCherry-PopZ/RNE-msfGFP) which the polar protein PopZ matrix is known to exclude ribosomes from the cell pole (Bowman et al., 2008) which also excludes the RNE-msfGFP protein. A minimum of 30 foci from 30 different cells were selected for each correlation distribution reported. Two-tailed T-tests with uneven variance were used to analyze statistical significance.

BR-bodies Enrichment by Differential Centrifugation

5mL of JS299 cells were grown overnight in PYE medium supplied with 0.2% xylose, 0.5μg/mL gentamycin, and 5μg/mL kanamycin. The overnight cultures were then washed 3 times with PYE growth media and used to inoculate 30 mL of PYE medium supplied with 500μM vanillate, 0.5μg/mL gentamycin, and 5μg/mL kanamycin. The cells were grown overnight and then pelleted at 11,000xg for 5 min. The cell pellet was resuspended in 2.5 mL of lysis buffer (35mM NaCl, 20mM Tris-HCl-pH 7.4, 1mM β-mercaptoethanol, one tablet EDTA-free protease inhibitor (roche) per 25 mL of buffer, 1U/mL Superase IN, and 10U/ml RNase-free DNase I). The cell suspension was flash frozen dropwise in liquid nitrogen before lysis in a mixer-mill. After collecting a small scoop of the frozen lysate for the whole cell lysate sample, the cell lysate was spun at 2000xg for 5 min. to clear membranes. The pellet was resuspended in 200 μL of lysis buffer and the samples were spun again at 10,000xg for 10 min. The resulting pellet was resuspended again in 200 μL of lysis buffer and subjected to another spin at 20,000xg for 10 minutes. The BR-bodies enriched pellet was resuspended in 200 μL of lysis buffer and RNA was extracted from both whole cell lysate and BR-bodies enriched fractions. RNA extraction was performed by adding 1mL of 65°C Trizol to the samples and incubating at 65°C for 10 min, then 200μL of chloroform was added and incubated for 5 min at room temperature. The samples were then spun at max speed in a microcentrifuge for 10 min at room temperature and the aqueous layer was removed and incubated with 700μL of isopropanol. The samples were then precipitated at −20°C for 1 hour, spun at 20,000 g for 1 hour at 4°C, and washed three times with 80% ethanol. The pellet was air dried and resuspended in 10mM Tris pH 7.0. RNA-seq library construction was performed as described in (Aretakis et al., 2018) using 1.0 μg of total RNA. Raw sequencing data is available in the NCBI GEO database with accession number GSE133522.

BR-body purification using HA-ASM pulldown

5 mL cultures of HA-ASM (JS302) and the untagged ASM (JS299) were inoculated and grown overnight in PYE-Gent-kan-xylose. The next day, these cultures were used to inoculate a 40 mL culture containing PYE/Gent/kan/xylose media and were diluted to be grown overnight. The log phase cultures were pelleted by centrifugation and washed 3x with 15 mL each of PYE. The washed cells were used to inoculate 50 mL of PYE-gent-kan-van and the cells were grown for 8 hours. The cells were harvested by centrifugation 11000xg for 10 min, resuspended in 0.5 mL of lysis buffer (35mM NaCl, 20 mM Tris-HCl (7.4), 1mM BME, 1U/ml Superase In, 10U/5mL RNase-free Dnase I). The pellets were flash frozen dropwise in liquid nitrogen and stored at −80°C. The cells were lysed using mixer-mill and the cell lysates were thawed and transferred into Eppendorf tubes. The samples were spun at 2000xg for 5 min to remove membranes. 500 μL of the supernatant was used for HA-affinity purification. The purification was carried out as following:

100 μL of anti-HA-beads were washed three times (in 1.7 mL Eppendorf tubes) with 1mL of HA bead wash buffer each (20 mM Tris-HCl (7.4), 35mM NaCl, 1mM BME, 1U/ml Superase In). The cell lysate samples were added to the washed beads and incubated for 1 hour at 4°C on a nutator. The samples were spun at 12,000xg for 10 seconds and the flow through samples were collected. The beads were washed 3x with 1mL HA bead wash buffer for 20 min each. 100 μL of the HA-peptide solution was then added to the beads and were incubated at 30°C for 20 min and the first eluates were collected (E1). E2 samples were collected by adding another 100 μL of the HA peptide solution to the beads and incubating at 30°C for another 20 min. The third elution was done by adding 200 μL of the HA peptide to the samples and transferring them to Eppendorf tubes, incubating at 30°C for 5 min and collect the samples by spinning at 12,000xg for 10 seconds. The collected fractions were analyzed by imaging and western blot using anti-RNase E antibody (Gift from Luisi lab).

Analysis of BR-body enrichment

To calculate log₂BR-body enrichment, we took RNA-seq libraries from total RNA and BR-body enriched samples and calculated the RPKM values. This normalizes the RNA both the expected number of fragments based on size and to the overall number of reads in the sample. The log₂BR-body enrichment was calculated as the log₂(BR-body enriched RPKM/lysate RPKM). We used the edgeR package (Robinson et al., 2009) which uses a negative binomial distribution based model to calculate statistical significance of differential RNA abundance data comparing the lysate samples to the BR-enriched samples. mRNA length was estimated from transcriptional units mapped from previous RNA-seq datasets (Schrader et al., 2014; Zhou et al., 2015). To exclude complexities relating to multi-gene operons, simple mRNAs were used for the analysis. Simple mRNAs were defined as those containing a single TSS and a single CDS. mRNA levels, translation levels, and translation efficiency data from cells grown in M2G were from (Schrader et al., 2014). 5’ UTR length was calculated from (Schrader et al., 2014; Zhou et al., 2015). Shine-Dalgarno affinity was taken from (Schrader et al., 2014). TAI was calculated for C. crescentus by utilizing the online tool version of the TAI calculator (Sabi et al., 2017). C. crescentus was selected as the organism, and a FASTA file was uploaded with sequences of all of C. crescentus protein coding genes. stAIcalc (Sabi et al., 2017) calculated a TAI value for each gene as the geometrical mean of the TAI value of the codons making up that gene. nTE was calculated as described in as in (Pechmann and Frydman, 2013). RNA-seq data used in the calculation was from C. crescentus collected in M2G minimal media (Schrader et al., 2014). Hfq HITS-CLIP RNAs used were from (Assis et al., 2019).

RNA decay rates by RNA-seq and qRT-PCR

JS38, JS221, JS233, and JS299 cells were grown in M2G-kan-gent media containing xylose overnight. The next day, the log-phase cultures were washed 3 times with M2G media and resuspended in 20 mL M2G-kan-gent media with 500 μM vanillate and grown for 8 hours. 1 mL of log-phase (OD600 0.3–0.6) Caulobacter crescentus cells untreated (0 min) or treated with (200 μg/mL) rifampicin for the indicated time points (1, 3, and 9 min). At the indicated time point the samples were added immediately to 2 mL of RNAprotect Bacterial Reagent (QIAGEN), immediately vortexed, and incubated at room temperature for 5 min. The cells were pelleted at (20,000xg) for 1 min and resuspended in 1mL of 65 °C hot TRizol Reagent (Ambion) and incubated 65 °C for 10 min in a heat block. 200μL of chloroform were added to the samples and the tubes were incubated at room temperature for 5 min before spinning at (20,000xg) for 10 min. RNA samples were chloroform extracted once and precipitated using isopropanol (1x volume isopropanol, 0.1X volume 5M NaOAc pH 5.2) overnight at −80 °C. The RNA samples were spun at 20,000 x g at 4 °C for 1 hour, pellets were washed with 80% ethanol for 10 min, air dried, and resuspended in 10 mM Tris-HCl (pH 7.0). The RNA-Seq libraries were made using 5μg of total RNA samples, rRNA was removed by ribozero gram negative kit, and library construction was performed according to protocol (Aretakis et al., 2018). Raw sequencing data is available in the NCBI GEO database with accession number GSE133532. The qRT-PCR, reactions were performed using 200 ng of total RNA samples according to the NEB Luna universal one-step-qRT-PCR kit on a stratagene MX3000P qRT-PCR machine. qRT-PCR were performed with primers for 5SrRNA, 9SrRNA, rne, ctrA, gcrA, and dnaA genes (Table S3).

To measure RNA-decay rates we performed linear curve fitting of the ln (fraction RNA remaining) at each time point of RNA extraction. For RNA-seq, the fraction remaining was calculated as the RPKM of each time point divided by the RPKM measured in the untreated 0’ sample. For qRT-PCR, the Ct was converted into amount of RNA using a standard curve, and the amount of RNA at each time points was divided by the amount of RNA measured in the untreated 0’ sample. Slopes of the linear curve fits were then converted into mRNA decay rates using the following equation (mRNA half-life=−ln(2)/slope). To reduce biases in mRNA half-life induced by rifampicin lag (Chen et al., 2015), we excluded multi-gene operons from half-life measurement analysis and focused only on simple mRNAs. Simple mRNAs were defined as those containing a single TSS and a single ORF. For bulk mRNA half-life calculations of RNA-seq data the fraction of all mRNA reads was compared to the total fraction of reads which includes a majority of stable tRNA reads.

To examine levels of full-length mRNA, Isolated total RNA was used as the template for reverse transcription reactions according to Invitrogen first-strand cDNA synthesis protocol using Superscript III. 1 μg of total RNA was mixed with 2 pmoles of ctrA RT FL R primer and 1μL of dNTP mix (10 mM each dATP, dGTP, dCTP and dTTP). The mixture was heated to 65 °C for 5 min and incubated on ice for at least 1 min. After incubation the contents of the tube was collected by brief centrifugation and added 4 μL of 5X First-strand buffer, 1 μL of DTT (0.1M), 1 μL of SUPERase-In RNase inhibitor (20 Units/ μL) and 1 μL Superscript III (200 Units/ μL) and incubated at 55 °C for 60 min. The reaction was inactivated by heating at 70 °C for 15 min. The resultant cDNA was used as a template for amplification in PCR. The 50 μL PCR reaction was setup by adding 3 μL of ctrA RT FL R primer (10 μM) and ctrA RT FL F primer (10 μM), 2 μL of cDNA from reverse transcriptase reaction, 1 μL dNTP mix (10 mM), 25 μL of Betaine (5 M), 10 μL of 5X HF-buffer, 5.5 μL of ddH₂O and 0.5 μL of Phusion enzyme (2 Units/ Units/ μL). Reactions were run for 15 PCR cycles and 25 μL from each sample was run on a 1% agarose gel and imaged after ethidium bromide staining.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical analyses were performed in Microsoft excel or edgeR. Statistical details can be found in the figure legends or methods section including the information on the statistical tests used.

Supplementary Material

3

Table S1. Processed RNA-seq data for BR-body enrichment and mRNA half-life profiling experiments. Related to STAR Methods.

4

Table S3. List of oligonucleotides, strains, and plasmids used. Related to STAR Methods.

Highlights.

• Differential centrifugation enrichment identified RNAs in BR-bodies.

• BR-bodies are enriched in mRNAs, sRNAs, and antisenseRNAs.

• BR-bodies exclude tRNAs, ribosomes, and the nucleoid.

• BR-bodies globally stimulate the multi-step bacterial mRNA decay pathway.