Phase-Separated Bacterial Ribonucleoprotein Bodies (BR-bodies) Organize mRNA Decay

Abstract

In bacteria, mRNA decay is controlled by megadalton scale macromolecular assemblies called, “RNA Degradosomes,” composed of nucleases and other RNA decay associated proteins. Recent advances in bacterial cell biology have shown that RNA degradosomes can assemble into phase-separated structures, termed bacterial ribonucleoprotein bodies (BR-bodies), with many analogous properties to eukaryotic P-bodies and stress granules. This review will highlight the functional role that BR-bodies play in the mRNA decay process through its organization into a membraneless organelle in the bacterial cytoplasm. This review will also highlight the phylogenetic distribution of BR-bodies across bacterial species, which suggests that these phase-separated structures are broadly distributed across bacteria, and in evolutionarily related mitochondria and chloroplasts.

Graphical/Visual Abstract

Graphical abstract

Bacterial RNA degradosomes and RNA have been found to phase-separate into membraneless assemblies called bacterial-RNP bodies (BR-bodies). BR-bodies have similar properties to eukaryotic P-bodies and stress granules to organize the mRNA decay/processing machinery. Since bacteria typically lack membrane bound organelles, BR-bodies, and more generally biomolecular condensates, are likely to play a critical role in organizing their biochemical pathways.

1. INTRODUCTION

1. Biomolecular Condensates – Organization without a lipid membrane

1.1. Biomolecular Condensate organizational principles

Many biomolecules are concentrated into dynamic self-assembled structures called biomolecular condensates (Fig 1). These structures act to organize biochemical reactions in cells by increasing the local concentration of their molecular components thereby increasing the reaction kinetics (Banani, Lee, Hyman, & Rosen, 2017; Courchaine, Lu, & Neugebauer, 2016; Nakamura, DeRose, & Inoue, 2019). In addition, biomolecular condensates can selectively recruit enzymes in a manner that segregate on-pathway from off-pathway reactions to prevent unfavourable reactions (Banani et al., 2017; Ditlev, Case, & Rosen, 2018). The ability of biomolecular condensates to organize specific biomolecules together in space provides a similar organizational role as membrane bound organelles. In contrast to their membrane bound counterparts, biomolecular condensates provide permeable barriers that can permit rapid exchange of constituents with the more dilute environment. Biomolecular condensates are commonly composed of protein and RNA (Banani et al., 2017; Weber & Brangwynne, 2012) although protein-DNA or protein-only assemblies have also been identified (Bergeron-Sandoval, Safaee, & Michnick, 2016; S. Boeynaems et al., 2018; Jiang et al., 2015; Sear, 2008). These structures range in size from tens of nanometers to tens of micrometers and have been found in the nucleus, cytoplasm, mitochondria, and chloroplasts of eukaryotic cells or on the surface of membranes (Banani et al., 2017; Clifford P. Brangwynne, 2013; Cioce & Lamond, 2005; Lamond & Spector, 2003). Interestingly, these have now been identified in the bacterial cytoplasm and the bacterial cell surface (N. Al-Husini, Tomares, Bitar, Childers, & Schrader, 2018; Heinkel et al., 2019).

Figure 1.

Overview of biomolecular condensates. A.) Cartoon highlighting the organization of biomolecules via phase separation. On the left, a well-mixed solution of macromolecules A and B. On the right, demixing of molecules A and B by phase separation into a biomolecular condensate. B.) Cartoon of material states of biomolecular condensates. Left, Liquid-liquid phase separation can generate biomolecular condensates with rather dynamic internal diffusion as represented as the particle trajectory by the red arrow. Middle, phase separation to form a gel or hydrogel can lead to biomolecular condensates with slower internal diffusion. Right, liquid-solid phase separation leading to very slow internal motion.

Biomolecular condensates are dynamic macro molecular assemblies which are thought to assemble through the process of liquid-liquid phase separation (LLPS) (Fig 1A). These structures show selective permeability and constituent molecules can be rapidly exchanged between the condensates and the surrounding liquid phase (Fig 1). Depending on the extent of molecular dynamics, these can be liquid droplets, hydro-gels, or insoluble aggregates (Banani et al., 2017; Nakamura et al., 2019) (Fig 1B). The cellular functions of condensates can also be correlated with their physical state. More liquid-like assemblies are able to facilitate biochemical reactions, whereas more gel or solid-like assemblies can sequester certain biomolecules for storage (Banani et al., 2017; S. Boeynaems et al., 2018). Liquid-to-solid transitions of biomolecular condensate have also been associated with neurological diseases in humans (Aguzzi & Altmeyer, 2016; Li, King, Shorter, & Gitler, 2013; Ramaswami, Taylor, & Parker, 2013).

The molecular principles governing LLPS are paramount to understanding the function of biomolecular condensates. Since the physiochemical aspects of LLPS has been reviewed in-depth recently in these articles (Alberti, Gladfelter, & Mittag, 2019; Banani et al., 2017; Choi, Holehouse, & Pappu, 2020; Courchaine et al., 2016; Mitrea & Kriwacki, 2016) it will only be briefly described here. Multivalent protein-protein and protein-RNA intra and inter-molecular interactions are the key molecular driving forces underlying the formation of condensates. Multivalency is required in the form of “stickers” (protein-protein or protein-nucleic acid binding sites), and the material properties of biomolecular condensates are also impacted by the length and flexibility of “spacer” regions (Choi et al., 2020; Harmon, Holehouse, Rosen, & Pappu, 2017). Proteins with large intrinsically disordered regions (IDRs) are often effective spacers (Clifford P. Brangwynne, Tompa, & Pappu, 2015) since IDRs lack ordered three dimensional structure. However, IDRs may also form β-sheet structures that may promote assembly into amyloid-like irreversible aggregates (Peran & Mittag, 2019). Many types of stickers are known to exist, including ionic bonds, cation-pi stacking, pi-pi stacking, and many specific protein-protein or protein RNA-interaction domains (Choi et al., 2020). The large diversity of macromolecular interactions comprising the “stickers” makes bioinformatics prediction of a protein’s capacity to phase-separate a major challenge. Despite the complexity, some rules about protein sequence features have been identified that promote LLPS and which affect material properties of biomolecular condensates. Repeated sequence elements can provide the basis for the multivalent intermolecular interactions, such as the electrostatic interactions between blocks of oppositely charged protein residues (Banani et al., 2017; Nott et al., 2015). Pi-stacking is another important feature for LLPS (Vernon et al., 2018). The presence of Arg residues helps to promote cation-pi stacking with purine bases, while Lys residues promote cation-pi stacking with pyrimidine bases (Steven Boeynaems et al., 2019). Furthermore, Arg and Tyr tend to be negatively correlated with the saturation concentration of proteins, suggesting that higher Arg and Tyr content leads to an enhancement in LLPS potentially through cation-pi interactions (J. Wang et al., 2018). High Gly content is associated with fluidity, whereas high Glu and Ser content tends to promote hardening of biomolecular condensates (J. Wang et al., 2018). In addition to repeated sequence elements, protein IDRs often contain specific RNA binding domains which are crucial for RNA targeting (Drino & Schaefer, 2018). Despite the broad presence of proteins in biomolecular condensates, LLPS may also be promoted by other macromolecules. RNA based biomolecular condensates have been found to form largely through intermolecular base-pairing interactions (Van Treeck & Parker, 2018). Other negatively charged biopolymers, such as polyphosphate, have been long known to form protein independent condensates in the presence of divalent cations (Strauss, Smith, & Wineman, 1953). Even lipids can form phase-separated domains within the lipid bilayer (Crowe & Keating, 2018; Heberle & Feigenson, 2011), suggesting that condensates may be more broadly utilized across biopolymers.

Despite the rapid advancement in knowledge into the biophysical mechanisms controlling biomolecular condensate formation, less is known about their cellular, biochemical, and organismal functions. In eukaryotes, condensates have been observed in different biological contexts ranging from basic and essential cellular processes such RNA transcription (Hnisz, Shrinivas, Young, Chakraborty, & Sharp, 2017), to neurodegenerative disease progression in Amyotrophic Lateral Sclerosis (ALS), Alzheimer disease and Huntington disease (Patel et al., 2015; Ramaswami et al., 2013; Ross & Poirier, 2004). At the organismal level, Ataxin-2 IDRs contribute to learning and memory processes, and when the IDR is mutated, this leads to learning defects (Bakthavachalu et al., 2018). Additionally, germ granules help sequester germ-cell specific mRNAs in the embryo and deliver them to what will ultimately become the germ line cells (C. P. Brangwynne et al., 2009; Seydoux, 2018). At the biochemical level, cGAS proteins sense cytoplasmic DNA as part of the innate immune response, with phase-separation stimulating the catalysis of cyclic-G-A dinucleotides in the STING pathway (Du & Chen, 2018). At the cellular level, nucleoli act as organizational centers for ribosome synthesis in the nucleus where rRNA transcription, processing and RNA modification take place (D. Chen & Huang, 2001; Mao, Zhang, & Spector, 2011). DVl clusters in Wnt signaling and signaling clusters in T cell activation are examples of condensates involved in cell signaling pathways (Schaefer & Peifer, 2019). Most importantly for the focus of this review, are stress granules and processing bodies (P-bodies) which play critical roles in mRNA metabolism and homeostasis (Decker & Parker, 2012; Khong & Parker, 2020; Panas, Ivanov, & Anderson, 2016; Protter & Parker, 2016). P-bodies and stress granules sequester poorly translated mRNAs for decay or storage upon cellular stresses, and thus play critical roles in organizing post-transcriptional mRNA metabolism in eukaryotes (Decker & Parker, 2012; Standart & Weil, 2018).

1.2. The emergence of biomolecular condensates in organizing the bacterial cytoplasm

While dramatic gains have been made in the last decade to discover roles for biomolecular condensates in eukaryotic cells, much less progress has been made in bacterial cells. Since bacteria generally lack membrane-bound organelles, it is possible that biomolecular condensates may provide a general mechanism to organize biochemical pathways in their cytoplasm. So far, very few examples of biomolecular condensates have been identified in bacteria, (BR-bodies, carboxysomes, DEAD Box RNA helicases, and ABC transporters) (N. Al-Husini et al., 2018; Heinkel et al., 2019; Hondele et al., 2019; J. S. MacCready, J. L. Basalla, & A. G. Vecchiarelli, 2020; H. Wang et al., 2019). C. crescentus (Ccr) Ribonuclease E (Ccr-RNase E) was the first bacterial protein identified that forms LLPS condensates both in vivo and in vitro termed BR-bodies (N. Al-Husini et al., 2018). BR-bodies are composed of Ccr-RNase E (the major bacterial mRNA decay nuclease), RNA degradosome components, and RNAs (N. Al-Husini et al., 2018). The intrinsically disordered C-terminal domain (CTD) of Ccr-RNase E, which is composed of alternating blocks of positive and negative charges, is both necessary and sufficient for the assembly of BR-bodies (N. Al-Husini et al., 2018). Ccr-RNase E was the first bacterial protein found to undergo LLPS at physiological concentrations and without the use of crowding reagents in vitro (Fig 2B). In vitro Ccr-RNase E condensates have a strict salt dependence, and deletion of positively charged residues blocks phase-separation in vivo, suggesting electrostatic forces play a significant role in BR-body condensation (N. Al-Husini et al., 2018). RNA also plays a key role in BR-body assembly, as depletion of cellular mRNA by rifampicin treatment leads to a loss in BR-bodies, and BR-bodies were found to highly enriched for long poorly-translated mRNAs (N. Al-Husini et al., 2018; Nadra Al-Husini et al., 2020).

Figure 2.

C. crescentus BR-bodies organize mRNA decay through biomolecular condensation. A.) Domain organization of the Ccr-RNase E. Structured N-terminal domain (NTD) containing the catalytic E/G domain and “small domain” for multimerization are shown in neon green. In red, the intrinsically disordered CTD with patches of negative (red) and positive (blue) amino acids that facilitate self-assembly. Below, cartoons of the RNase E monomer which assembles into a tetramer. The RNA degradosome which assembles with a suite of RNA decay related proteins which associated predominantly with the CTD is also shown with red ovals as degradosome proteins. B.) Ccr-RNase E phase separates in vitro (N. Al-Husini et al., 2018), forming a biomolecular condensate. C.) Subcellular localization of BR-bodies in C. crescentus and E. coli. RNase E subcellular localization is highlighted in green and the cell envelope is colored black. D.) Evidence of liquid like properties for C. crescentus and E. coli BR-bodies. Left, RNase E-YFP foci were observed to fuse in vivo, suggestive of a liquid-like state. Images were taken every 10s for C. crescentus (N. Al-Husini et al., 2018) and 200ms for E. coli (Strahl et al., 2015). Right, single molecule particle trajectories showed confined diffusion in C. crescentus cells in the presence of mRNA. E.) BR-bodies compete with ribosomes for mRNA substrates. Free mRNA (red) can either be directed down a decay path by the RNA degradosome or be translated by ribosomes. Both Eco- and Ccr-RNase E foci are rifampicin sensitive, suggesting they require mRNA. Translation inhibitor treatments and translation initiation factor depletion experiments in C. crescentus suggest that the untranslated mRNA pool directs BR-body assembly. In addition, C. crescentus BR-bodies were shown to exclude ribosomes. While eukaryotic P-bodies and stress granules tend to lead to mRNA storage upon stress, its unclear whether BR-bodies can provide a similar storage function. In conditions of logarithmic growth, BR-bodies accelerate the mRNA decay rate.

Subcellular localization patterns of RNase E in the α-proteobacteria like C. crescentus and γ-proteobacteria like E. coli (Eco) differ substantially (Fig 2C). In the α-proteobacterium C. crescentus, it was discovered that RNase E is nucleoid-associated (Montero Llopis et al., 2010) and can be observed throughout the nucleoid-filled cytoplasm (Bayas et al., 2018)(Fig 2C). Additionally, RNase E from A. tumefaciens and S. meliloti also appear to be cytoplasmic (N. Al-Husini et al., 2018), although nucleoid-association has not been verified. Interestingly, Eco-RNase E is attached to the inner membrane via an amphipathic helix between comprising residues 568-582 in the IDR (Khemici, Poljak, Luisi, & Carpousis, 2008). Importantly, Eco-RNase E forms membrane anchored foci which require the membrane attachment sequence (Strahl et al., 2015). It’s likely that Eco-RNase E foci are BR-bodies, as replacement of the Ccr-RNase E intrinsically disordered CTD with Eco-RNase E’s CTD restored RNase E foci but moved them to the inner membrane (N. Al-Husini et al., 2018; Strahl et al., 2015). Ccr-RNase E recruits the components of the RNA degradosome into BR-bodies, acting as the primary scaffold for this multi-protein complex (N. Al-Husini et al., 2018). Eco-RNase E similarly recruits its degradosome proteins, as Eco-RhlB colocalization depends on its RNase E scaffolding site (Vanzo et al., 1998).

While Eco-RNase E hasn’t been shown to undergo phase-separation directly through in vitro assays, time-lapse imaging has shown similar dynamic properties to Ccr-RNase E. RNase E-YFP fusions in both organisms appear to show dynamic foci that depend on the presence of RNA (Fig 2D) (N. Al-Husini et al., 2018; Strahl et al., 2015). These RNase E-YFP foci have been observed to fuse in both organisms (N. Al-Husini et al., 2018; Strahl et al., 2015), suggesting they both have liquid-like properties (Fig 2D). In opposition to solid protein aggregates which would maintain their irregular shapes, RNase E foci fusion is followed by rapid relaxation to spherical shapes suggesting liquid-like behaviour (N. Al-Husini et al., 2018). Additional single molecule particle tracking experiments of Ccr-RNase E have shown that these molecules undergo confined diffusion in the presence of mRNA, suggesting dynamic movement within BR-bodies (Bayas et al., 2018) (Fig 2D). When cells were depleted of mRNAs, RNase E-YFP maintained a similar diffusion rate, however, fewer confined RNase E molecules were observed (Bayas et al., 2018). The dynamic assembly and disassembly of BR-bodies in logarithmic growth, is regulated by the presence of RNA, followed by its dissociation upon cleavage (N. Al-Husini et al., 2018). Taken altogether, these data suggest that BR-bodies show biomolecular condensate behaviour both in vitro and in vivo.

1.3. Similarities between BR-bodies and eukaryotic P-bodies and stress granules

BR-bodies have many similar properties similar to eukaryotic P-bodies and stress granules. Bacterial mRNA decay is most often rate-limited by RNase E, (Hammarlöf, Bergman, Garmendia, & Hughes, 2015; Ono & Kuwano, 1979) which forms the core of BR-bodies, while eukaryotic Xrn1 also rate limits cytoplasmic mRNA decay (Kedersha et al., 2005), and is a core component of P-bodies and stress granules (Hubstenberger et al., 2017; Youn et al., 2019). P-bodies and stress granules contain translationally repressed mRNAs that often accumulate after acute or chronic cell stress exposure and typically are trigged by a reduction in cellular mRNA translation (Brengues, Teixeira, & Parker, 2005; Hubstenberger et al., 2017; Khong et al., 2017; Shah, Zhang, Ramachandran, & Herman, 2013). BR-bodies can also be strongly induced upon a variety of cell stresses and growth conditions, which tend to negatively correlate with cellular translation as judged by polysome content (N. Al-Husini et al., 2018). Translation inhibitor treatments showed RNase E competes with ribosomes for untranslated mRNAs, forming BR-bodies readily when puromycin was added while dissolving BR-bodies in the presence of chloramphenicol or tetracycline (N. Al-Husini et al., 2018). Additionally, directly inhibiting translation initiation by depleting IF3 from cells also triggered robust BR-body formation (N. Al-Husini et al., 2018). Interestingly, P-bodies, stress granules, and BR-bodies all appear to occlude translating ribosomes (Nadra Al-Husini et al., 2020; Hubstenberger et al., 2017), perhaps ensuring that actively translating mRNAs are not aberrantly degraded or stored. Additionally, P-bodies and stress granules contain miRNAs (Liu, Valencia-Sanchez, Hannon, & Parker, 2005), and BR-bodies contain small noncoding RNAs (sRNAs) (Nadra Al-Husini et al., 2020) which can both inhibit translation, suggesting the mRNAs are indeed translationally repressed. Therefore, P-bodies, stress granules and BR-bodies all compete for a common pool of mRNAs with the translational machinery suggesting they may play a common role in shaping global translation.

Early work in P-bodies provided evidence that these structures could act as sites of mRNA decay (Parker & Sheth, 2007; Sheth & Parker, 2003), and smaller P-bodies appear to be suited to stimulating mRNA decay whereas larger P-bodies may be better for storage (Pitchiaya et al., 2019). More recent work in P-bodies and stress granules suggests that these structures can also act as storage sites for mRNAs upon cellular stress (Hubstenberger et al., 2017; Ivanov, Kedersha, & Anderson, 2019; Padron, Iwasaki, & Ingolia, 2019; Protter & Parker, 2016). BR-bodies have been found to stimulate mRNA decay (Nadra Al-Husini et al., 2020), however their role in mRNA storage has not yet been explored (Fig 2E). Truncation of the RNase E CTD in C. crescentus or E. coli, which blocks degradosome formation and prevents condensation, slows mRNA decay (N. Al-Husini et al., 2018; Nadra Al-Husini et al., 2020; Lopez, Marchand, Joyce, & Dreyfus, 1999). The rate could be restored in C. crescentus by introducing the CTD without the degradosome protein binding sites, suggesting that formation of a condensate stimulates mRNA decay (Fig 2E) (N. Al-Husini et al., 2018). In E. coli, deletion of the RNase E membrane targeting sequence (MTS) led to a slowdown in mRNA decay (Hadjeras et al., 2019). Importantly, the MTS is needed for foci formation in vivo, and the ΔMTS variant still forms a degradosome (Hadjeras et al., 2019; Strahl et al., 2015). Since the ΔMTS variant still forms a degradosome but cannot assemble membrane associated foci, this suggests that the observed slower mRNA decay rates with this mutant result from the failure to phase separate and not from a defect in RNA degradosome assembly. While the stimulation of mRNA decay by phase separation has been observed, it’s currently unclear whether BR-bodies can also act as mRNA storage sites.

While stress-granules and P-bodies are conserved from yeast to mammals, phylogenetic range of BR-bodies across bacteria has not yet been fully defined. Bioinformatics sequence analysis of bacterial mRNA decay nucleases can be used to predict the likelihood of BR-body formation across bacteria. While RNase E is the most broadly distributed mRNA decay nuclease (Ait-Bara & Carpousis, 2015; Hui, Foley, & Belasco, 2014; Mohanty & Kushner, 2016), some bacteria instead use RNase Y or RNase J to perform the rate-limiting step of their mRNA decay pathways (Laalami, Zig, & Putzer, 2014). Some bacteria contain RNase G which is homologous to RNase E’s catalytic region, but lacks the small domain needed for tetramerization and lacks the IDR so it is highly unlikely to phase-separate. Despite the similarity of the nuclease domains, rng cannot perform the same functional role as rne in E. coli unless it is dramatically overexpressed (Masaru Tamura, Moore, & Cohen, 2013). To examine the potential for BR-bodies across bacteria, a representative of each major bacterial clade and plant chloroplasts were selected and the disorder content and propensity of phase-separation based on pi-stacking are provided in Table I. Plant chloroplasts were included because RNase E from ancestral cyanobacteria has been maintained in the chloroplast. The disorder content is from IUPred long (Meszaros, Erdos, & Dosztanyi, 2018) and propensity score (Pscore) from phase separation predictor (Vernon et al., 2018). As expected, the single structured domain of RNase G correlates to low disorder (0-9%) and low PScores (from −0.8-0.7). Conversely, RNase E sequences show all RNase E genes in bacteria contain a large IDR (ranging from 43-57%) with a high Pscore, (1-7) while the A. thaliana (Ath) chloroplast RNase E that has 13% disorder and a Pscore of 1. Across bacteria and organelles, RNase Es all have a rather large extension at the C-terminus as in C. crescentus or E. coli, or at the N-terminus as in the A. thaliana chloroplast, or both N- and C-terminus as in Actinobacteria (Ait-Bara & Carpousis, 2015; Ait-Bara, Carpousis, & Quentin, 2015). Overall, due to high disorder and Pscores, RNase E variants are highly likely to form BR-bodies across bacteria, and potentially even chloroplasts.

Table I.

List of major mRNA decay nucleases across bacteria and chloroplasts.

				Ribonuclease
				RNG		RNE		RNJ		RNY
	Group		Species	%Disorder	PScore	%Disorder	PScore	%Disorder	PScore	%Disorder	PScore
Bacteria	Actinobacteria		Mycobacterium smegmatis	N/A		57	4	6	1	N/A	N/A
	Bacteroidetes		Bacteroides thetaiotaomicron	3	−0.4	N/A		N/A	N/A	3	1
	Chlamydiae		Chlamydia trachomatis	5	−0.2	N/A		N/A	N/A	16	0.8
	Cyanobacteria		Synechococcus elongatus	N/A		43	2	7	0.3	N/A	N/A
	Firmicutes	Bacillales	Bacillus subtilus	N/A		N/A		4	0.4	20	0.6
		Clostridia	Clostridia difficile	0	−0.8	N/A		0	0.2	7	0.7
		Lacto-bacillales	Lactobacillus acidophilus	N/A		N/A		5	0.5	22	0.2
	Mollicutes		Mycoplasma pneumonia	N/A		N/A		0	−0.2	0.4	−0.4
	Proteo-bacteria	α	Caulobacter crescentus	9	0.3	47	4	10	1	N/A	N/A
		β	Neisseria gonorrhoeae	1	0.7	46	1	N/A	N/A	N/A	N/A
		γ	Eschericia coli	0	−0.7	51	3	N/A	N/A	N/A	N/A
		δ	Myxococcus xanthus	N/A		57	7	8	0.7	N/A	N/A
		ε	Campylobacter jejuni	0	−0.3	N/A		14	0.4	0.6	0
	Spirochaetales		Borrelia burgdorferi	N/A		N/A		N/A	N/A	2	−0.8
Plants	Chloroplast (nucleus)		Arabidopsis thaliana	N/A		13	1	26	1	N/A	N/A

^*Disorder data from IUPred long (Meszaros et al., 2018), PScore from Phase separation Predictor (Vernon et al., 2018)

1.4. Compositional variation of the RNA degradosome

The RNA degradosome and degradosome-like protein complexes characterized so far show extreme compositional variation across bacteria, chloroplasts, and mitochondria (Table II and III). Therefore, unlike the core proteins of P-bodies which contain homologs of Xrn1, Edc3, Dcp1/2, Pat1, Ccr4-Nt, Dhh1, and LSm1-7 from yeast to humans (Luo, Na, & Slavoff, 2018), the BR-body core of RNA degradosome proteins are expected to vary significantly across bacteria. Since the compositional variation of RNA degradosomes across bacteria has been recently reviewed in depth (Ait-Bara & Carpousis, 2015; Ait-Bara et al., 2015; Tejada-Arranz, de Crecy-Lagard, & de Reuse, 2019) it will be briefly discussed here. RNA degradosome complexes are composed of two core components: RNases and RNA helicases, while a majority of RNA degradosomes also contain a third core-component, metabolic enzymes (Tables II and III). While most bacteria contain RNase E scaffolded RNA degradosomes, RNases such as RNase J, RNase Y, or PNPase can also form the scaffold of the RNA degradosome in other organisms (Tables II and III) (Durand & Condon, 2018; Szczesny et al., 2012; Tejada-Arranz et al., 2019). Many RNA degradosomes contain both endo- and exo-ribonucleases, while some have only been identified to contain a single nuclease (Table II). The phosphorolytic 3′-5′ exoribonuclease PNPase is found in the RNA degradosome across many different bacterial species, while it is the core component of the mitochondrial degradosome (Table II and III). RNA helicases are a near universal feature in bacterial RNA degradosomes, as 14/17 bacterial RNA degradosomes contain DEAD-box RNA helicases, although the identity of the helicase is highly variable across species (Table II and III). The physical association of the RNases and RNA helicases in the RNA degradosome is thought to facilitate the coordination of their individual activities to promote RNA decay (Marcaida, DePristo, Chandran, Carpousis, & Luisi, 2006). Most bacterial RNA degradosomes (9/17) contain a metabolic enzyme such as enolase, aconitase or phosphofructokinase as a core component (Table II and III) (Tejada-Arranz et al., 2019). The biochemical and physiological role of metabolic enzymes in the RNA degradosome has sparked a lot of interest where they have been hypothesized to transduce the metabolic status of the cell to RNA degradation activity (Kuhnel & Luisi, 2001; Marcaida et al., 2006; Morita, Kawamoto, Mizota, Inada, & Aiba, 2004; Morita, Maki, & Aiba, 2005). In addition to their primary enzymatic roles in metabolism, these enzymes are likely to perform moonlighting functions in the RNA degradosome complex (Jeffery, 1999; Pal-Bhowmick, Vora, & Jarori, 2007). Indeed, aconitase and enolase from other organisms is known to have moonlighting activities involved in RNA metabolism (Chandran & Luisi, 2006; Constable, Quick, Gray, & Hentze, 1992; Kuhnel & Luisi, 2001; Morita et al., 2004; Pal-Bhowmick et al., 2007).

Table II.

RNase E degradosome components identified to associate with RNase E in diverse organisms.

Organism	Exoribonuclease	RNA Chaperone	Metabolic Enzyme	Other Function
Escherichia coli	PNPase (Major) (Carpousis, Van Houwe, Ehretsmann, & Krisch, 1994), RNase II (Lu & Taghbalout, 2014), RNase R (Carabetta, Silhavy, & Cristea, 2010)	RhlB (Major) (Py, Higgins, Krisch, & Carpousis, 1996) RhlE (Khemici, Toesca, Poljak, Vanzo, & Carpousis, 2004), CsdA (Prud’homme-Genereux et al., 2004), SrmB (Khemici et al., 2004), Hfq (Butland et al., 2005)	Enolase (Major) (Chandran & Luisi, 2006), FabZ (Butland et al., 2005), YfiQ (Butland et al., 2005)	RraA (Gao et al., 2006), RraB (Gao et al., 2006), DnaK (Miczak et al., 1996), GroEL (Miczak et al., 1996), Ppk (Blum et al., 1997), PAPI (Raynal & Carpousis, 1999), 70S ribosomes (Tsai et al., 2012), Srd (Qi, Alawneh, Yonesaki, & Otsuka, 2015), Dip (Van den Bossche et al., 2016), FruR (Butland et al., 2005), RluB (Rajagopala et al., 2014), RlmB (Butland et al., 2005), RlmN (Butland et al., 2005), RmlL (Butland et al., 2005), TufA (Butland et al., 2005), TufB (Butland et al., 2005), MinD (Taghbalout & Rothfield, 2007)
Caulobacter crescentus	PNPase (Major)(Hardwick et al., 2011) RNase D (Voss, Luisi, & Hardwick, 2014)	RhlB (Major)(Voss et al., 2014), RhlE (Aguirre et al., 2017), DEAD/DbpA (Aguirre et al., 2017)	Aconitase (Major) (Hardwick et al., 2011) OdhA (Aguirre et al., 2017), MetK (Aguirre et al., 2017), NudC (Aguirre et al., 2017), Acetoacetyl-CoA reductase (Aguirre et al., 2017)	Rho (Aguirre et al., 2017), S1 ribosomal protein (Aguirre et al., 2017)
Mycobacterium tuberculosis	PNPase (Major) (Kovacs, Csanadi, Megyeri, Kaberdin, & Miczak, 2005) RNase J (Major) (Kovacs et al., 2005)	RhlE (Major) (Kovacs et al., 2005), CspA (Kovacs et al., 2005), CspB (Kovacs et al., 2005)		DnaK (Plocinski et al., 2019), GroEL (Plocinski et al., 2019), Ppk (Plocinski et al., 2019), Rv2908c (Plocinski et al., 2019), Rv3920c (Plocinski et al., 2019), Acetyltransferase (Kovacs et al., 2005)
Pseudomonas haloplanktis	PNPase (Ait-Bara & Carpousis, 2010)	RhlB (Ait-Bara & Carpousis, 2010)
Pseudomonas aeruginosa	PNPase (Van den Bossche et al., 2016)	DeaD (Van den Bossche et al., 2016)	NuoC (Van den Bossche et al., 2016), AtpA/AtpD (Van den Bossche et al., 2016)	ClpA (Van den Bossche et al., 2016), DnaK (Van den Bossche et al., 2016), DnaA (Van den Bossche et al., 2016), FliC (Van den Bossche et al., 2016) S3 ribosomal protein (Van den Bossche et al., 2016), Dip (Van den Bossche et al., 2016), PA3309 (Van den Bossche et al., 2016)
Pseudomonas syringae	RNase R (Purusharth et al., 2005)	RhlE (Purusharth et al., 2005)
Anabaena nostoc	PNPase (J. Y. Zhang et al., 2014)
Synechocystis	PNPase (J. Y. Zhang et al., 2014)	CrhR (Rosana, Whitford, Fahlman, & Owttrim, 2016)
Streptomyces coelicolor	PNPase (Lee & Cohen, 2003)
Vibrio cholera	PNPase (Major) (Kimura & Waldor, 2019)	RhlB (Kimura & Waldor, 2019)	Enolase (Kimura & Waldor, 2019)
Rhodobacter capsulatus		DpbA (Jager et al., 2001), SrmB (Jager et al., 2001)		Rho (Major) (Jager et al., 2001)
Rhizobium leguminosarum		DbRhl (Y. Zhang & Hong, 2009), Hfq (Y. Zhang & Hong, 2009)
Arabidopsis thaliana (Chloroplast)				RHON 1 (Stoppel et al., 2012)

Table III.

RNase Y/J or PNPase degradosome components identified to interact with the scaffolding nuclease in diverse bacteria.

Organism (Degradosome Scaffold)	Exoribonuclease	RNA Chaperone	Metabolic Enzyme	Other Function
Bacillus subtilis RNase Y (Lehnik-Habrink et al., 2011)	PNPase (Major) (Lehnik-Habrink et al., 2011) RNase J1/J2 (Major) (Lehnik-Habrink et al., 2011)	CshA (Major) (Lehnik-Habrink et al., 2011)	Enolase (Major) (Lehnik-Habrink et al., 2011) PfkA (Major) (Lehnik-Habrink et al., 2011) GapA (Commichau et al., 2009)	YlbF (A. DeLoughery, Dengler, Chai, & Losick, 2016), YmcA (A. DeLoughery et al., 2016), YaaT (A. DeLoughery et al., 2016)
Staphylococcus aureus RNase Y (Roux, DeMuth, & Dunman, 2011)	PNPase (Major) (Roux et al., 2011) RNase J1/J2 (Major) (Roux et al., 2011)) RNase P (Major) (Roux et al., 2011)	CshA (Major) (Giraud et al., 2015)	Enolase (Major) (Roux et al., 2011) PfkA (Major) (Roux et al., 2011)	Flotillin (Koch et al., 2017)
Streptococcus mutans RNase Y (Mu, Shinde, Zou, Kreth, & Merritt, 2019)	RNaseJ1/J2 (Major) (Mu et al., 2019)	CshA (Major) (Mu et al., 2019)	Ldh (Major) (Mu et al., 2019) Enolase (Mu et al., 2019)	FtsZ (Mu et al., 2019), DnaK (Mu et al., 2019), DnaJ (Mu et al., 2019)
Streptococcus pyogenes RNase Y (Kang, Caparon, & Cho, 2010)			Enolase (Major) (Kang et al., 2010)
Helicobacter pylori RNase J (Redko et al., 2013)		RhpA (Major) (Redko et al., 2013)		70S ribosomes (Redko et al., 2013)
Homo sapiens (Mitochondria) PNPase (Silva, Camino, & Aguilera, 2018)		SUV3 (Silva et al., 2018)		GRSF1 (Pietras et al., 2018)

In addition to core components, a plethora of additional species-specific RNA degradosome associated proteins have been identified with functions known in various cellular processes (Tables II and III). These additional interacting partners are often present at lower stoichiometry and suggest novel functions of the RNA degradosome in concert with other cellular processes. Cell cycle-regulators including the DNA replication initiator DnaA and cell division proteins FtsZ and MinD have been identified as minor degradosome components (Table II), suggesting a potential link between RNA degradosome activity and bacterial cell cycle processes. Interestingly, in a temperature-sensitive mutant of Eco-RNase E, E. coli cells fail to divide, suggesting that cell division is linked to the RNA degradosome (Cam, Rome, Krisch, & Bouché, 1996; Murashko & Lin-Chao, 2017; M. Tamura et al., 2006). A majority of these degradosome proteins are known to associate with microdomains in the RNA degradosome scaffold’s IDR (Ait-Bara et al., 2015). IDR sequence analysis of γ-proteobacterial RNase E yielded 30 microdomains, identified as short linear motifs (SLiMs) (Ait-Bara et al., 2015). Interestingly, the binding partners of only a subset of these SLiMs have been identified, suggesting that the diversity of degradosome components is actually greater than currently known (Ait-Bara et al., 2015). The dramatic diversity of SLiMs and client proteins likely allows RNA degradosomes to evolve very quickly to better adapt bacteria to their diverse niches.

1.5. Conservation of IDR sequences in major bacterial mRNA decay nucleases

While the C-terminal IDRs across α-proteobacteria shared very low sequence homology, these proteins all held blocks of “charge patterning” across the C-terminal IDR (N. Al-Husini et al., 2018) (Fig 3). Charge patterning is calculated for all IDRs in each RNase E protein using a 11 amino acid sliding window (Fig 3). Patches of positively- and negatively-charged amino acids alternate throughout the IDRs, potentially setting up electrostatic protein-protein interactions that facilitate phase-separation. Indeed, deletion of the positively charged patches in the CTD abolished phase separation in C. crescentus, and the in vitro salt dependence suggests electrostatic interactions are critical for phase-separation (N. Al-Husini et al., 2018). While length of IDRs and amino acid identity vary dramatically, charge patterning is a shared feature in α-proteobacteria irrespective of IDR size (Fig 3). While most α-proteobacteria lack genetic tools, A. tumefaciens and S. meliloti have well-established genetic tools allowing them to be tested for BR-body capacity. Both A. tumefaciens and S. meliloti contain charge patterning in their RNase E CTD, and both were found to have RNase E foci (N. Al-Husini et al., 2018), suggesting there’s a strong likelihood for BR-bodies across this clade. Interestingly, even γ-proteobacteria contain charge patterning in their IDRs (Fig 3), suggesting that charge patterning may be critical for phase-separation across clades. Indeed, when comparing bacteria representing each major clade of organisms that contain RNase E, blocks of charge patterning is present across RNase E IDRs (Fig 4) with the exception of M. xanthus (Mxa) RNase E. In Mxa-RNase E, charge patterning occurs more broadly, as the S1 domain insert IDR is highly negative, and the entire CTD is highly positive. There is also a micro-patterning of charged residues in the CTD, as positive and negative amino acids are present in alternating orders (Fig 3). Interestingly, the charged residues are separated by Gly spacers and were found to promote liquid-like droplets in other IDR scaffolds (J. Wang et al., 2018). Mxa-RNase E has the highest Pscore based on potential pi-interactions (Table I), suggesting that perhaps it has evolved to utilize pi-pi interactions.

Figure 3.

α-proteobacterial RNase E IDR sequences contain charge-patterning. Phylogenetic tree containing several α-proteobacterial RNase E sequences (black) and two γ proteobacterial RNase E sequences (orange). Sequences are aligned to the catalytic NTD which is represented as a light-grey box. To the right, the IDR is colored corresponding to a 10 amino-acid sliding window of the electrostatic charge (scale below).

Figure 4.

RNase E IDRs show charge patterning across bacterial phyla. A representative RNase E from each major bacterial class or from plant chloroplasts is shown to highlight the charge-patterning of RNase E IDRs. Sequences are aligned to the catalytic domain which is represented as a light-grey box. Below, the IDR is colored corresponding to a 11 amino-acid sliding window of the electrostatic charge (scale below). For Mxa-RNase E, the individual amino acids are displayed based on their charge to reveal micro-charge patterning (right).

Other major mRNA decay nucleases RNase J and RNase Y are broadly distributed in bacteria lacking RNase E although it is less clear as to whether they make BR-bodies. RNase Y ranges from 0-22% disorder across bacteria, and Pscores range from −0.8 to 1. RNase Y in B. subtilis (Bsu-RNase Y) was shown to have an IDR, and assemble into dimers and tetramers in vitro (Lehnik-Habrink et al., 2011). Like Eco-RNase E, Bsu-RNase Y is a membrane anchored nuclease, and is thought to be the core of the B. subtilis RNA degradosome (Lehnik-Habrink et al., 2011). Interestingly, Bsu-RNase Y shows some charge patterning within the IDR which is also predicted to have a coiled-coil structure (Fig 5A). RNase Y may have a propensity to assemble BR-bodies, as localization pattern shows patchiness when imaged in log-phase in LB-media (Cascante-Estepa, Gunka, & Stulke, 2016). Bsu-RNase Y was found to assemble into foci, and time-lapse TIRF microscopy showed that these foci can fuse suggesting they have liquid-like properties (Fig 5B) (Hamouche et al., 2020). Interestingly, Bsu-RNase Y foci are not dissociated by rifampicin treatment, suggesting assembly may not depend on RNA (Hamouche et al., 2020). Bsu- RNase J is a known degradosome component of the RNase Y degradosome (Lehnik-Habrink et al., 2011). Bsu-RNase J can predominantly form a dimer in vitro (Mathy et al., 2010), but can form modest amounts of tetramers, suggesting a potential for higher-order assemblies. Bacterial RNase Js have relatively low disorder content (0-14%) and Pscores that range from −0.2 to 1, whereas A. thaliana chloroplast RNase J has a 26% disorder content and a Pscore of 1 (Table I). RNase J scores higher than RNase G, but considerably lower than RNase E suggests that RNase J has an intermediate capacity to assemble BR-bodies. Interestingly, RNase J in B. subtilis localized non-uniformly in the cytoplasm concentrating near the cell poles (Cascante-Estepa et al., 2016), but it’s unknown if this localization depends on RNase Y or if it occurs when unbound. While it is currently unclear whether RNase J phase separates in B. subtilis, there is a low predicted disorder and Pscore, suggesting if it phase-separates it is likely not a primary scaffold (Table I). Across RNase Y and J proteins, the disorder content and Pscores are moderate, suggesting that some of these proteins may have the capacity to phase separate (Table I). It’s important to note that computational prediction of phase-separation is in its infancy, and no algorithm has yet been developed that captures all proteins known to phase separate. Therefore, experimental determination of bacterial RNase phase separation across species will be an important goal for the field to assess which organisms and nucleases assemble BR-bodies.

Figure 5.

The B. subtilis RNA degradosome scaffold RNase Y likely forms BR-bodies. A.) Charge-patterning of the mRNA decay nuclease Bsu-RNase Y. The sequence is colored corresponding to a 11 amino-acid window of electrostatic charge (scale below). RNase Y domain architecture is shown below, with its IDR shown in a black box (Lehnik-Habrink et al., 2011). TM=transmembrane, CC=coiled-coil, KH=RNA binding domain, HD=nuclease active site. B.) TIRF microscopy time-lapse of and Bsu-RNase Y-msfGFP fusion (Hamouche et al., 2020). Red arrows mark two foci that ultimately fuse together, suggesting liquid-like properties. Images were taken every 100ms (Hamouche et al., 2020).

2. Compositional control and evolution of BR-bodies

2.1. BR-body organization stimulates mRNA decay

BR-body organization helps to stimulate the rapid mRNA decay process in bacteria by organizing the RNA decay machinery together with its mRNA substrates (Fig 6). BR-bodies dynamically assemble and disassemble during log growth with RNA, with mRNA decay triggering their disassembly (N. Al-Husini et al., 2018). RNA degradosome components are recruited into BR-bodies, requiring their specific protein-protein interaction sites for entry (N. Al-Husini et al., 2018). The high concentration of mRNA substrates and RNA degradosomes in BR-bodies stimulates mRNA decay in two distinct ways: 1) RNase E endonuclease cleavage of mRNA is stimulated within BR-bodies and 2) RNA degradosome exoribonuclease decay is stimulated on RNase E cleaved mRNA fragments (Nadra Al-Husini et al., 2020). C-terminal deletions of RNase E that block BR-body formation and degradosome scaffolding lead to longer mRNA half-lives in both C. crescentus and E. coli (N. Al-Husini et al., 2018; Nadra Al-Husini et al., 2020; Lopez et al., 1999). Mutations in Ccr-RNase E that block C-terminal degradosome formation without impacting foci formation led to enhanced mRNA decay (N. Al-Husini et al., 2018), suggesting that phase-separation stimulates RNase E cleavage. Additionally, deletion of the MTS in Eco-RNase E, which disrupts foci formation (Strahl et al., 2015), and likely ablates phase-separation without disrupting degradosome formation, led to decreased rates of mRNA decay (Hadjeras et al., 2019). Additionally, C. crescentus mRNAs that were found to be more highly enriched in BR-bodies, tended to have shorter mRNA half-lives (Nadra Al-Husini et al., 2020). How then does enhanced RNase E catalysis occur within the BR-body? It’s possible that cross subunit catalysis plays a critical role in enhancing the mRNA decay rate. In Eco-RNase E, a 5′-Phosphate (5′-P) sensor mutation or CTD truncation mutant are viable alone, but when combined in the same gene they prove lethal (Ali & Gowrishankar, 2020). Interestingly, when an active site mutant that contained a CTD, and which cannot cleave RNA and fails to support E. coli growth is co-expressed in cells containing a 5′-P sensor CTD truncation mutant, the cells have restored viability (Ali & Gowrishankar, 2020). This suggests that multiple RNase E molecules coordinate their RNase activity, either across monomers in a tetrameric RNase E complex, or across RNase E tetramers through formation of BR-bodies.

Figure 6.

BR-bodies facilitate the mRNA life cycle. BR-bodies are stimulated by untranslated mRNAs causing condensation of RNA degradosomes with mRNA. mRNA acts as a scaffold to which RNA degradosomes can self-assemble. The mRNA decay pathway is shown on the right, with both the monophosphate stimulated decay pathway and the direct entry pathways highlighted. The localization of RppH, the enzyme which catalyzes the conversion to a 5′-monophosphate containing RNA which enhances the RNase E cleavage rate. Within the condensate, mRNA decay is stimulated by high concentrations of the RNA degradosome and poorly translated mRNA (top middle). The initial step of mRNA decay is controlled by the endoribonuclase RNase E and is stimulated within the BR-body (right). The subsequent steps of mRNA performed by RNA degradosomes associated exoribonucleases that are also stimulated within the BR-body (right). Upon cleavage of the mRNA down into small oligonucleotides by both the endoribonuclase RNase E and RNA degradosome-associated exoribonucleases, the multivalent bridging functionality of the mRNA is lost, causing a dissolution of the BR-body, releasing short oligo RNA products and RNA degradosomes that can reassemble on new mRNAs.

RNA degradosome component localization within BR-bodies likely also helps to stimulate the multi-step mRNA decay pathway in bacteria. As RNase E provides the rate limiting endonuclease cleavage, degradosome associated exoribonucleases such as PNPase then trigger rapid 3′-to-5′ exonucleolytic decay. In cells where PNPase was deleted, the CTD of RNase E was truncated, or degradosome binding sites in the CTD were deleted, mRNA decay intermediates began accumulating (Morita et al., 2004). Additional in vitro experiments with mini E. coli RNA degradosomes showed omission of either RhlB or PNPase led to the build-up of mRNA decay intermediates (G. A. Coburn, Miao, Briant, & Mackie, 1999). This suggests that the coordinated activity of the degradosome components helps to degrade the endo-cleaved fragments generated by RNase E. Therefore, degradosome component localization within BR-bodies ensures mRNA decay intermediates are degraded within the body and not released. The ability to phase-separate into BR-bodies may facilitate a sensitive switch-like change in RNA decay activity that ensures rapid complete mRNA decay. A similar mechanism has been proposed for super enhancers to induce transcription, where large switch-like Hill coefficients are generated by highly multivalent interactions between TFs and super-enhancers to stimulate transcription activity (Hnisz et al., 2017). After decay of the mRNA into small oligonucleotides, the bridging function of the mRNA is lost, reducing the multi-valency and allowing the BR-body to dissolve. This is supported by the fact that BR-body enrichment found that these assemblies contain long, poorly translated mRNAs and are depleted of short RNAs. In addition, in vitro BR-body assemblies are stimulated by long, poorly translated RNAs, while short or highly structured RNAs do not stimulate assembly (Nadra Al-Husini et al., 2020). As DEAD Box RNA helicases have been shown to disassemble RNP condensate in vitro (Hondele et al., 2019), it is also possible that the degradosome-associated DEAD Box RNA helicase RhlB provides a similar function. Interestingly, an RhlB binding site deletion of Ccr-RNase E yielded a larger fraction of the protein in BR-bodies than wild type (N. Al-Husini et al., 2018), suggesting this may be due to defects in disassembly. Once dissolved, the RNA degradosomes can be released so they can be reused on new RNA substrates to catalyze another round of mRNA decay. The overall biophysical mechanism provided by BR-bodies may also help to ensure that mRNA decay does not aberrantly initiate on actively translating mRNAs, as ribosomes are productively excluded from BR-bodies (Nadra Al-Husini et al., 2020).

2.2. Diversity of degradosome components and their potential roles in BR-bodies

In addition to the coordination of major degradosome proteins in the mRNA decay cycle, minor degradosome components share some functionalities that are present in eukaryotic P-bodies and stress granules. In P-bodies and stress granules, decapped and deadenylated mRNAs are found together with miRNAs which repress target mRNA translation and lead to faster rates of decay (Decker & Parker, 2012; Parker & Sheth, 2007). While bacteria do not contain a canonical m7G 5′-cap, 5′-PPP, 5′-nucleoside tetraphosphates, and 5′-NAD groups found on RNAs stabilize them as compared to a 5′-P (Bird et al., 2016; Garrey & Mackie, 2011; Luciano, Levenson-Palmer, & Belasco, 2019). Interestingly, in C. crescentus cells grown in the cold, Ccr-NudC, an enzyme catalyzing the removal of the 5′-NAD cap, was found to associate with the RNA degradosome (Aguirre et al., 2017) suggesting BR-bodies may also organize decapped RNA. Conversely to eukaryotes however, in bacteria poly-A tails tend to lead to a destabilization of the mRNAs through coordinated activity of poly-A polymerase and PNPase (Glen A. Coburn & Mackie, 1998; Dreyfus & Regnier, 2002; Marujo et al., 2003). Eco-poly-A polymerase was found to associate with RNase E (Raynal & Carpousis, 1999). While bacteria do not have miRNAs, small noncoding RNAs (sRNAs) can imperfectly base pair and silence target mRNAs in a similar manner through the protein Hfq (De Lay, Schu, & Gottesman, 2013; Storz, Vogel, & Wassarman, 2011). Interestingly, many RNA degradosomes contain Hfq, and Ccr-Hfq pulldown experiments found overlapping RNAs with BR-bodies (Nadra Al-Husini et al., 2020; Assis et al., 2019). Interestingly in Eco-RNase E, CTD truncation leads to defects in sRNA silencing, suggesting that BR-bodies may play a role in facilitating sRNA induced mRNA silencing (De Lay et al., 2013; Ikeda, Yagi, Morita, & Aiba, 2011). Finally, many protein chaperones have been identified to associate with RNase E, which are known to help to disaggregate stress granules (Mateju et al., 2017). In Ccr-RNase E, PNPase, and RhlB were found to partially enter the insoluble protein fraction upon heat shock suggesting they may form aggregates under these conditions (Schramm, Schroeder, Alvelid, Testa, & Jonas, 2019). In E. coli, Eco-GroEL has been found to associate with Eco-RNase E and its activity is linked to RNase E (Miczak, Kaberdin, Wei, & Lin-Chao, 1996; Sohlberg, Lundberg, Hartl, & von Gabain, 1993). Since RNase E’s large disordered region may help to facilitate phase-separation, it may also lead to susceptibility of RNase E to aggregate which might require protein chaperones to prevent.

While many RNA degradosomes contain a metabolic enzyme, the functional role of metabolic enzymes in BR-bodies is still a mystery. For a subset of these enzymes, some functional link to RNase E has been established. For Eco-Ppk, which catalyzes polyphosphate formation, enzyme activity may help to module PNPase activity (Blum, Py, Carpousis, & Higgins, 1997). Since PNPase is a phospholytic enzyme, changes in the local phosphate concentration could directly alter the activity of PNPase within a BR-body. Interestingly, poly-P has been found to phase separate on its own in the presence of Mg²⁺, and can readily coalesce with eukaryotic disease associated Arg-rich repeat proteins (Steven Boeynaems et al., 2019), suggesting that poly-P itself may help facilitate BR-bodies, but so far has not been tested. Enolase, a major component of the Eco-RNA degradosome, has been found to trigger a detachment of RNase E from the inner membrane when cells are shifted to anaerobic conditions (Murashko & Lin-Chao, 2017). While it is currently unknown how enolase triggers the membrane detachment, E. coli mainly utilizes glycolysis during anaerobic growth, and it’s possible that this shift is somehow sensed through the increased activity of this glycolytic enzyme. Interestingly, this change in localization of RNase E alters the stability of a small RNA, dicF, which leads to a drop in FtsZ levels causing the cells to filament which may reduce phagocytosis (Murashko & Lin-Chao, 2017). Surprisingly, many metabolic enzymes have recently been found to have a “moonlighting” RNA binding activity, including most proteins involved in glycolysis (Queiroz et al., 2019; Shchepachev et al., 2019). This suggests that the metabolic degradosome components may be much more directly linked to RNA metabolism than previously anticipated. Interestingly, in human cells the TCA enzyme aconitase has a well-established role as an iron-responsive RNA binding protein which regulates the expression levels of many iron import related genes (Constable et al., 1992). Ccr-aconitase is part of the Ccr-RNA degradosome, although its currently unknown whether it may play a similar role in C. crescentus. The moonlighting roles of these metabolic components in BR-bodies will be an important area for future studies.

Because a rather large array of minor degradosome proteins have been discovered an exciting area of future studies will be to define how their biochemical activities will impact BR-bodies. For instance, Rho the transcription termination factor has been identified as a degradosome component in E. coli, C. crescentus, and R. capsulatus (Aguirre et al., 2017; Jager et al., 2001). In E. coli, it was found that the Eco-RNase E CTD deletion was synthetically lethal in combination with rppH deletion (Anupama, Leela, & Gowrishankar, 2011; Bouvier & Carpousis, 2011). Surprisingly, point mutation in the rho gene could restore viability, suggesting that Rho may play a functional role, and that this required RNase H1 which acts on RNA-DNA hybrids (Anupama et al., 2011). This suggests that Rho may help to link BR-bodies with decay of nascent transcripts. Ribosomes were also identified as minor degradosome components of the Eco-RNA degradosome, and some ribosomal proteins were identified in Ccr- and Pae-RNA degradosomes (Table II) (Aguirre et al., 2017; Tsai et al., 2012; Van den Bossche et al., 2016). In vitro biochemical experiments showed 70S association with RNase E did not affect mRNA decay activity, however, it did inhibit 5S rRNA processing (Tsai et al., 2012), suggesting that it may help to regulate the balance of mRNA decay and rRNA processing.

Many ribosome modification enzymes and DEAD box RNA helicases associated with rRNA assembly are found to associate with RNase E (Butland et al., 2005; Rajagopala et al., 2014). The observation that most rRNA genes in E. coli are colocalized to the same subcellular location (Gaal et al., 2016) suggests that rRNA processing may occur similarly in bacteria as to the eukaryotic nucleolus. Interestingly, in C. crescentus, RNase E foci are commonly observed to be colocalized to the rRNA genes in rapidly growing cells (Bayas et al., 2018). These observations suggest that BR-bodies may act similarly to a nucleolus to organize rRNA processing machinery into a biomolecular condensate. Functionally however, CTD truncation of Ccr-RNase E led to no changes in 9S rRNA processing rate or steady state levels of 5S rRNA (N. Al-Husini et al., 2018), and in vitro the catalytic NTD of Ccr-RNase E is able to perform this processing step (Hardwick, Chan, Broadhurst, & Luisi, 2011), suggesting BR-bodies are not required for rRNA processing. Interestingly, the ΔDBS mutant of Ccr-RNase E which ablates all degradosome protein binding in the RNase E CTD, led to a reduction of 9S rRNA processing rate, however, steady state levels of 5S appeared to be unchanged (N. Al-Husini et al., 2018), suggesting BR-bodies likely have complex influences on rRNA processing. As rRNA processing is thought to occur cotranscriptionally, and Eco-RNAP foci colocalize with pre-rRNA (Weng et al., 2019), suggesting that the coordination of BR-bodies with RNAP clusters for rRNA processing will be an important area of future studies (Jin, Mata Martin, Sun, Cagliero, & Zhou, 2017).

2.3. How broadly distributed are BR-bodies?

While RNase E appears to be broadly capable of forming BR-bodies across bacteria, it’s unclear to what extent other RNases may form similar roles. RNase J and Y based RNA degradosomes are predicted to have less disorder in the scaffold proteins (Fig 5, Table I), but limited localization experiments have shown evidence of BR-body formation. Bsu-RNase Y, the scaffold of the Bsu-RNA degradosome, was observed to form membrane anchored patchy foci (Hamouche et al., 2020; Lehnik-Habrink et al., 2011). Additionally, the Y-complex component Bsu-YaaT, which acts as a specificity factor for Bsu-RNase Y, was found to form similar membrane anchored patchy foci (Aaron DeLoughery, Lalanne, Losick, & Li, 2018). Upon deletion of Y-complex component ylbF or rny, localization shifted to the cytoplasm where larger foci appeared (Aaron DeLoughery et al., 2018). Conversely, Bsu-RNase Y degradosome components Bsu-enolase, Bsu-CshA, and Bsu-RNase J showed polar foci, while Bsu-PFK was diffuse throughout the cytoplasm (Cascante-Estepa et al., 2016). Quantitative ribosome profiling found that all Bsu-RNase Y degradosome proteins are between a 2-23 fold excess of Bsu-RNase Y while yaaT was translated at 61% the level of Bsu-RNase Y (Lalanne et al., 2018) potentially explaining the difference in localization patterns. Conversely to Eco- and Ccr-RNase E scaffolded RNA degradosomes, addition of rifampicin led to no changes in patchiness of Bsu-RNase Y (Hamouche et al., 2020), suggesting that RNase Y foci do not require RNA to assemble. Conversely, Bsu-RNase J polar foci become diffuse throughout the cytoplasm upon rifampicin treatment (Cascante-Estepa et al., 2016), suggesting a relationship between the membrane anchored Bsu-RNase Y scaffold and its clients localized in the cytoplasm. Membrane localization of Bsu-RNase Y degradosomes is likely critical, as the deletion of the Bsu-RNase Y membrane tether leads to a decrease in fitness (Khemici, Prados, Linder, & Redder, 2015). The evidence suggests Bsu-RNase Y forms BR-bodies with distinct properties from RNase E BR-bodies. Fusion of Bsu-RNase Y foci suggests that this protein may be able to phase-separate (Fig 5B), however direct observation of phase separation has not yet been observed, and no subcellular localization experiments have been published for RNase J-scaffolded degradosomes.

The RNA decay machinery within the mitochondria and chloroplasts were derived from α-proteobacterial and cyanobacterial ancestors and experimental observations suggest they inherited the ability to form BR-bodies. Phylogenetic analysis indicates that RNase E appears to have been inherited vertically and the ancestral RNase E is predicted to have contained the IDR (Ait-Bara & Carpousis, 2015; Ait-Bara et al., 2015). During the endosymbiotic event leading to mitochondria, an α-proteobacteria was engulphed by a eukaryotic cell. The best characterized α-proteobacterial degradosome is C. crescentus, which contains RNase E as the scaffold, with PNPase, DEAD Box RNA helicase RhlB, and aconitase as major components (Hardwick et al., 2011). Interestingly, mitochondria have lost RNase E, but do form RNA degradosomes with PNPase and a DEAD box RNA helicase SUV3 (Table III) that localize into degradation foci (D-foci) (Borowski, Dziembowski, Hejnowicz, Stepien, & Szczesny, 2013; Szczesny et al., 2010). D-foci were most likely inherited from α-proteobacterial BR-bodies upon endosymbiosis. Additionally, both C. crescentus and mitochondrial RNA degradosomes have been identified to localize to the nucleoid (Borowski et al., 2013; Montero Llopis et al., 2010), while a fraction of mitochondrial PNPase is also used in the intermembrane space to import some RNAs from the cytoplasm (H. W. Chen et al., 2006; G. Wang et al., 2010). Disruption of human PNPase led to accumulation of mitochondrial mRNA decay intermediates (Borowski et al., 2013) similar to proteobacterial relatives C. crescentus and E. coli, suggesting it’s the major mRNA decay nuclease in this organelle. Similarly, purified PNPase and SUV3 coordinately degrade dsRNA in vitro (D. D. Wang, Shu, Lieser, Chen, & Lee, 2009) and associate with another protein, GRSF1, which helps stimulate the decay of mitochondrial RNAs containing G-quadruplexes (Pietras et al., 2018). Understanding the role of D-foci on mitochondrial mRNA processing and decay is important for human health, as genetically inherited diseases are linked to mitochondrial mRNA processing defects (Van Haute et al., 2015; Vedrenne et al., 2012; von Ameln et al., 2012). In contrast to mitochondria, chloroplasts maintained RNase E after cyanobacteria became endosymbionts with plant cells (Table II), although chloroplast degradosomes acquired a new RNA binding protein RHON1 (Stoppel et al., 2012) and lost ability to associate with PNPase (Baginsky et al., 2001). While chloroplast RNase E localization has not been tested, mRNA FISH showed chloroplast mRNAs localize into foci when under stress, including oxidation damaged RNA (Uniacke & Zerges, 2008). The observations of mitochondrial and chloroplast RNP-bodies suggest that these structures were acquired from ancestral bacterial BR-bodies.

2.4. Can BR-bodies become sites of RNA storage instead of decay?

The internal dynamics of RNP-bodies tend to govern their activities, and eukaryotic RNP-condensates tend to have a rather large dynamic range from liquid-like to solid-like. Some RNP condensates, such as the nucleolus, appear to be more liquid-like, which facilitates internal biochemical processes of rRNA transcription, processing, and rRNA modification (D. Chen & Huang, 2001). Others, such as p-granules or stress granules, are more gel-like which facilitates a function in mRNA storage (Decker & Parker, 2012; Sheth & Parker, 2003). BR-body foci were found to be rather dynamic when C. crescentus or E. coli cells are growing exponentially (N. Al-Husini et al., 2018; Strahl et al., 2015). BR-bodies assembled and disassembled on the minutes timescale, individual RNase E molecules have dynamic internal motion, and fusion of BR-bodies is readily observed (N. Al-Husini et al., 2018; Prud’homme-Genereux et al., 2004) (Fig 2D). Upon the addition of cell stress, however, BR-body intensity increases, as most of the RNase E in the cell transitions into BR-bodies (N. Al-Husini et al., 2018). At this stage, increased occurrence of BR-bodies could stem from an increase in the formation rate, a decrease in dissolution rate, or a combination of both. Interestingly, the increase in BR-body intensity correlates with a decrease in mRNA decay upon entry into stationary phase, where cells stop growing exponentially (H. Chen, Shiroguchi, Ge, & Xie, 2015; Dressaire et al., 2018). The large slowdown in metabolic activity and low ATP levels resulting from the transition into stationary phase has been found to lead to physical changes in the nucleoid filled bacterial cytoplasm leading to a glass-like state that dramatically slows diffusion of large cytoplasmic objects on the size scale of BR-bodies (Parry et al., 2014; Weber, Spakowitz, & Theriot, 2012). Since DEAD Box RNA helicases use ATP hydrolysis activity to remodel or dissolve RNP-condensates (Hondele et al., 2019), it’s possible that the drop in ATP levels could lead to a solidification of BR-bodies under such conditions, switching BR-body function from stimulation of mRNA decay to mRNA storage. An important goal for the field will be to determine whether BR-bodies can act as mRNA storage granules under conditions where their internal dynamics may allow jellification or solidification.

Conclusion

Biomolecular condensates likely play a key role in the subcellular organization of bacterial RNA degradosomes. Like eukaryotic P-bodies and stress granules, BR-bodies provide a broadly conserved mechanisms of gene-regulation across bacteria. An important future goal for the field will be to determine which bacteria have the capacity to assemble BR-bodies. RNA degradosomes have been shaped by evolution to have a wide array of different proteins that associate with the major scaffolding nuclease (Ait-Bara & Carpousis, 2015; Ait-Bara et al., 2015; Burger, Whiteley, & Boshoff, 2011). Another important goal for the field will be to define the functional roles of these diverse degradosome proteins on mRNA decay. Since bacteria generally lack membrane bound compartments to organize their biochemical pathways, another important goal will be to define the biochemical diversity of their biomolecular condensates. While the mRNA decay machinery can assemble into BR-bodies, proteins associated with carboxysome assembly and trafficking (Joshua S MacCready, Joseph L Basalla, & Anthony G Vecchiarelli, 2020), ABC transporters (Heinkel et al., 2019), and DEAD-box RNA helicases (Hondele et al., 2019) have all been found to phase-separate in vitro without the use of crowding reagents. Our perspective is that we’re about to see an explosion of new bacterial biomolecular condensates that will organize a diverse array of biochemical pathways across all domains of life.