RNA localization in bacteria

Abstract

One of the most important discoveries in the field of microbiology in the last two decades is that bacterial cells have intricate subcellular organization. This understanding has emerged mainly from the depiction of spatial and temporal organization of proteins in specific domains within bacterial cells, e.g., midcell, cell poles, membrane and periplasm. Because translation of bacterial RNA molecules was considered to be strictly coupled to their synthesis, they were not thought to specifically localize to regions outside the nucleoid. However, the increasing interest in RNAs, including non-coding RNAs, encouraged researchers to explore the spatial and temporal localization of RNAs in bacteria. The recent technological improvements in the field of fluorescence microscopy allowed subcellular imaging of RNAs even in the tiny bacterial cells. It has been reported by several groups, including ours that transcripts may specifically localize in such cells. Here we review what is known about localization of RNA and of the pathways that determine RNA fate in bacteria, and discuss the possible cues and mechanisms underlying these distribution patterns.

Abbreviations

RNA

ribonucleic acid

sRNA

small RNA

ncRNAs

non-coding RNAs

RNAP

RNA polymerase

FISH

fluorescence in situ hybridization

RNA: New Insights on an Ancient Molecule

In light of the central dogma in molecular biology– DNA codes for RNA, which in turn encodes a protein^1,2 - ribonucleic acid (RNA) has been discussed for many years mainly in the context of mediating information flow from the latent genetic information, which is embedded in the genome, to the proteome. Still, additional roles for RNA in cell physiology have been uncovered over the years. In the 1980s, it was established that RNA molecules may have catalytic activities, both in prokaryotes, e.g., cleavage of phosphodiester bonds by RNase P during maturation of tRNA in E. coli,^3,4 and in eukaryotes, e.g., self-splicing of an rRNA exon in Tetrahymena.^5-7 The novel observation that RNA molecules can splice themselves and, hence, recombine to produce new combinations of the genetic material, in addition to the earlier knowledge on reverse transcription discovered in 1970,⁸ led Walter Gilbert to coin a new term in molecular biology - “the RNA world.”⁹ Gilbert proposed that at the beginning of evolution, there were RNA molecules that catalyzed their own synthesis, without the need for protein enzymes. In this world, RNAs served both as holders of the genetic information and as molecules with enzymatic activities. Only during the next stage in evolution did RNA molecules begin to synthesize proteins, which were simply better enzymes than their cognate RNA molecules. Finally, DNA evolved by reverse transcription, followed by the evolution of double-stranded repair-prone DNA, which took over the role of the genetic material storage. Keeping in mind that eukaryotic organisms evolved after or from prokaryotes,¹⁰ when one comes to contemplate the hypothesis that the current form of life evolved from an RNA-only form of life, it is reasonable to unravel prokaryotic RNA-related mechanisms.

Today, it became obvious that in addition to being an intermediate product between DNA and proteins and the sole form of genetic information in some viruses, RNAs that are not translated into proteins regulate gene expression by a variety of mechanisms, both in prokaryotic and eukaryotic cells.^11-13 Hence, studies of different RNA-related aspects in different organisms have been initiated in laboratories all over the world. To understand a biological phenomenon, molecular biologists ask a rather common set of questions: “how” - what is the underlying molecular mechanism; “when” - what is the temporal order of events; and “where”- where do these events take place spatially. The first two questions had been raised and discussed with regard to RNA-related processes both in prokaryotes and eukaryotes. However, the third question was considered to be relevant only for eukaryotes until recently. This situation is currently undergoing a significant change.

The remarkable improvements in the microscopy field, both technologically (optical improvements and novel probes) and computationally (algorithms), enabled scientists to look at RNA molecules in live cells and describe their spatial distribution. Naturally, imaging of RNAs and of their related counterparts was initiated in the larger eukaryotic cells. However, the relatively new understanding that bacteria are well-organized organisms, which regulate the localization of their macromolecules,¹⁴ combined with the new perception that basic RNA-related processes in prokaryotes and eukaryotes are rather similar, led to a growing interest in understanding the spatial and temporal subcellular organization of prokaryotic RNA. This awareness came in an era of important discoveries regarding the distribution of other bacterial macromolecules and the intricate arrangement of the bacterial cell, which we briefly review below.

RNA Biogenesis and Turnover

Gene expression is one of the fundamental processes in all cells and, thus, the pathway by which the information from the DNA is copied to RNA and translated into proteins is termed the ‘central dogma of molecular biology’,^1,2 despite of the known cases of different paths of information flow mentioned above. The first step in gene expression, i.e., the synthesis of RNA from DNA by RNA polymerase (RNAP) is termed transcription. To initiate transcription, bacterial RNAP binds to specific regions in the DNA, which contain common consensus sequences that define the promoter region, and, typically, a particular thymine that marks the transcription start site. Once the polymerase is bound, it melts the DNA to form a transcription bubble; single nucleoside triphosphates enter the secondary channel and pair with the template strand, and, after several rounds of abortive initiation, the RNAP moves along the DNA strand in the 3′-5′ direction, polymerizing the RNA in the 5′-3′ direction; once the RNAP reaches a terminator sequence, it pauses and, subsequently, falls off.^15-18 The principles underlying transcription of DNA are universal and have evolved in ancient organisms, such as bacteria. The forms of transcription products are also similar in different cell types and consist of mRNAs (mRNAs), which are translated to proteins, and non-coding RNAs (ncRNAs).¹¹ An important group of bacterial ncRNAs are the small RNAs (sRNAs), which function in controlling post-transcriptional events.¹⁹

Obviously, transcription takes place near the genome in all cell types. However, whereas in eukaryotes this event is confined to the nucleus, in prokaryotes, there is no membrane to spatially delimit the process. Still, with the improvements in wide-field microscopy and the advent of super-resolution fluorescence microscopy, it became obvious that the distribution of RNAP in the bacterial cell significantly overlaps with the nucleoid, contributing to the robustness of transcription.²⁰

Following transcription, the eukaryotic transcripts leave the nucleus to be translated, stored or degraded.²¹ Where are bacterial RNAs translated or degraded? As described below, the translation machinery of many bacteria, among them E. coli and B. subtilis, has been shown to be largely segregated from the DNA and from the transcription machinery^20,22 although in some organisms, exemplified by Caulobacter crescentus, the ribosomes and chromosomal DNA share the same space.²³ As for the RNA degradation machinery, various components of the E. coli and B. subtilis RNA degradosome, including RNase E, RNase III, RNase P, and poly(A) polymerase and RNase Y, respectively, were shown to be associated with the membrane.^24-27 Taken together, these findings imply that the machineries that determine RNA fate in many types of bacteria localize outside the nucleoid/transcription machinery zone, and suggest that RNA transcripts are re-positioned to execute their function and to be ultimately degraded.

The Bacterial Cellular Architecture

The nucleoid

Compaction and organization of the bacterial chromosome is facilitated by nucleoid-associated proteins (NAPs), which are also involved in global transcription regulation. Using super-resolution fluorescence microscopy to observe the major NAPs in live E. coli cells, it has been shown that their distribution varies, with some scattered throughout the nucleoid and others forming compact clusters.²⁸ Recent studies shed light on how meticulously the nucleoid constituents are packaged, both spatially and temporally. By combining biochemical, imaging and computational methodologies, Church and coworkers derived the three-dimensional architecture of the Caulobacter crescentus chromosome and have shown it to be an ellipsoid with specific regions residing near the poles.²⁹ Kleckner and coworkers explored the three-dimensional packaging of the E. coli nucleoid over time and its positioning relative to the cell at high spatial and temporal resolutions. They revealed that the E. coli nucleoid is a dynamic helical ellipsoid. This ellipsoid is discrete and, although it occupies the majority of the cell space, it is separated from the poles, mainly from the old pole, except for during segregation of the replicated chromosomes (Fig. 1A).³⁰

Figure 1.

Organization of the nucleoid and the ribosomes in E .coli cells. Shown are schematic presentation of non-dividing (top) and dividing (bottom) E. coli cells. (A) The nucleoid (gray) is organized as dynamic helical ellipsoid that occupies the majority of the cell space³⁰. (B) The ribosomes (red) are distributed mainly near the membrane and the poles, well-segregated from the nucleoid lobes.²⁰

RNA polymerase and ribosomes organization

Where are the transcription and translation machineries located relative to the nucleoid? Electron microscopy images, showing actively transcribed and concomitantly translated chromosome portions, extracted from lysed E. coli cells, led Miller and coworkers to suggest in 1970 that transcription is completely coupled to translation in bacteria.³¹ This suggestion has been accepted as a dogma that prevailed for many years.³² The short half-life of bacterial RNA^33,34 supported this notion. One implication of this coupling is that the machineries that catalyze transcription and translation co-localize. However, re-visiting the transcription-translation coupling idea with modern microscopic technologies led to reassessment of the percentage of transcripts that are translated while tethered to the DNA. Using fluorescence, as well as super-resolution microscopy, most of the RNA polymerase molecules, but only a small percentage of the ribosomes, were shown to co-localize with the nucleoid lobes in E. coli cells (Fig. 1B). In fact, only 4% of the RNAPs, or less, overlap with the ribosome-rich regions, suggesting that the two machineries are spatially well-separated from each other.^20,35 These new findings imply that most of the translation in E. coli occurs on free mRNAs that have been relocated from the nucleoid zone to the ribosome-rich regions. The question then arises: how do RNA localize in bacteria?

The cytoplasm

Nowadays, it is clear that the bacterial cytoplasm, although an aqueous environment, is more crowded than expected.^36,37 Assuming that bacteria rely on diffusion for transport of molecules that are required for the different intracellular processes, including cytokinesis, it is important to understand the nature of this environment. Still, the diffusion mode within the cytoplasm has been described in some studies as normal and in others as anomalous.^38-40 Recently, Jacobs-Wagner and coworkers reported that the bacterial cytoplasm exhibits glass-like properties.⁴¹ This characteristic affects the motion of cytoplasmic components in a size-dependent manner, i.e., the larger the component, the slower its exhibited motion. This is very relevant for an environment whose constituents size span several orders of magnitude, from ions and metabolites to mega-complexes and chromosome. Since RNAs are bulky molecules, particularly when associated with RNA-binding proteins, this new understanding of the cytoplasm has obvious implications on how RNAs are distributed intra-cellularly, a process suggested by the strong segregation of the transcription and translation machineries described above.

Spatial distribution of proteins

The vast amount of information that accumulated in the last two decades demonstrates that proteins are spatially and temporally localized to particular sites within the bacterial cell. In addition to the membrane, periplasm and midcell, the poles of rod shaped bacteria are emerging as a significant subcellular region to which various macromolecules localize. The intricate organization of bacterial proteins and the mechanisms underlying their subcellular distribution can be exemplified by reviewing polar proteins localization.⁴²

The idea that proteins are specifically localized in bacteria was relatively easy to comprehend, even when bacteria were considered non-compartmentalized, mainly because it was well known that proteins containing hydrophobic regions localize specifically to the membrane.⁴³ The poles of rod-shaped bacteria contain hydrophobic proteins, which localize to the membrane portion of the poles, but also soluble proteins, which constitute the cytoplasmic portion of the poles. What brings these proteins to the poles and maintains them there? The membrane in the poles is more negatively curved than in the rest of the cell. Several proteins have been shown to be attracted to this geometric cue, membrane-bound and soluble proteins alike, e.g., DivIVA in B. subtilis⁴⁴ and enzyme I in E. coli, respectively.⁴⁵ Another polar cue that can be recognized by proteins is cardiolipin, an anionic lipoprotein that is enriched in polar membrane regions, e.g., ProP in E. coli.⁴⁶ Nucleoid occlusion, i.e., the process that prevents cell division at nucleoid-occupied regions, is also involved in localizing proteins to the cell poles, which are DNA-free regions, e.g., PopZ in C. crescentus.⁴⁷

Due to their unique features, the poles of rod-shaped bacteria emerge as hubs for central regulatory systems.¹⁴ For example, the chemotaxis complex in E. coli, which controls the movement of bacterial cells along gradients, clusters at the poles. Interestingly, this complex has been suggested to localize by stochastic self-assembly of the clusters, and not by an active transport mechanism.⁴⁸

In eukaryotes, another mechanism that underlies specific protein localization and establishment of cell polarity is RNA trafficking and localized translation.⁴⁹ New findings described below suggest that this mechanism also operates in bacteria.

Subcellular localization of RNA

In eukaryote, mRNA is synthesized in the nucleus and devoted pathways are involved in exporting the transcripts outside the nucleus and in targeting them to the different sub-domains where they are translated. In prokaryotic cells, in which the nucleoid and the cytoplasm are not separated by a membrane, mRNA synthesis and translation were assumed to be strictly coupled. The extent of this coupling is currently being reevaluated.

Already in 1993, Hahn et al. used whole-cell hybridization with digoxigenin-labeled probes to detect transcripts of the thiostrepton resistance (tsr) gene in fixed Streptomyces violacelatus cells. Distinct signals for the tsr RNA were observed throughout the hyphae.⁵⁰ A couple of years later, the same technique was applied to study temporal expression of nprM, which encodes an extra-cellular protease in Bacillus megaterium, during different growth phases. A strong nprM mRNA signal was observed during exponential and early stationary phases, but only a weak signal was detected in late stationary phase (Hönerlage, Hahn and Zeyer, 1995). Of note, the nprM mRNA signal was detected all through these relatively big bacterial cells. In 1998, Molin and coworkers employed in situ RT-PCR to study population heterogeneity by means of detecting lac mRNA expression in single cells.⁵¹ These pioneering studies laid the foundation for further exploration of RNA localization in bacteria.

The improvements in imaging methodologies and the development of new fluorescent probes enabled more detailed studies of mRNA localization in bacterial cells in recent years.⁵² The first to track RNA molecules in live bacterial cells were Golding and Cox, who used the RNA-binding coat protein of MS2 phage fused to GFP (MS2-GFP) to detect specific transcripts, which were tagged with MS2 binding sites.⁵³ By introducing repeats of MS2 binding sites upstream or downstream to the coding sequence of candidate RNAs, they were able to visualize three distinct characteristic dynamics of RNA molecules: localized motion, motion spanning the entire cell, and dynamic chain-like structures. In the case of localized motion, the random but restricted movement was interpreted as RNA tethered to the DNA template. The transcripts that traversed the entire cell spent more time at the poles, most likely because of hydrodynamic coupling between the RNA and the cell wall. The third pattern represents stochastic gene transcription.⁵³

A different approach to view RNA molecules in bacterial cells has been used by Broude and coworkers, who exploited a fluorescence protein complementation methodology that relies on the two fragments of spilt eIF4A protein, each fused to a different RNA-binding protein, which are brought together and fluoresce due to binding of both RNA-binding proteins to their target RNA sites. Using this approach, lacZ was observed as distributed uniformly in the cytoplasm, 5S RNA was detected mainly at sites outside the nucleoid, and short artificial untranslated transcripts were viewed as concentrated at cell poles.⁵⁴ Using an improved fluorescence protein complementation methodology, the same group reported the detection of a short non-coding RNA that follows a helical path along the cell axis; only a small fraction of the transcripts were detected at the poles, the middle and quarter points of the cells, in line with the localization of the mRNA-encoding plasmid.⁵⁵ Russell and Keiler (2009) used fluorescence in situ hybridization (FISH) to study the localization of endogenous RNA molecules in fixed bacterial cells. They detected the tmRNA ssrA in Caulobacter crescentus cells in a helix-like distribution, similar to the distribution of the tmRNA-binding protein SmpB, which is required for ssrA localization.⁵⁶ Pilhofer et al., also used FISH to visualize nifH transcripts in Klebsiella oxytoca and Azotobacter vinelandii. The nifH transcripts displayed uneven distribution in A.vinelandii cells and specific foci at one or both poles in K. oxytoca cells.⁵⁷

Two models for localization of endogenous RNA in bacteria have been recently presented. The first model, suggested by the Jacobs-Wagner group, states that transcripts stay near their transcription site in the nucleoid (Fig. 2A). Using primarily FISH, but also the MS2 system, this group followed several mRNAs in C. crescentus and the lacZ mRNA in E. coli. They observed limited dispersion of the transcripts from their site of transcription and suggested that the chromosome serves as a template for the localization of these transcripts.⁵⁸ The same study also demonstrated that unlike in E. coli and B. subtilis, the ribosomes in C. crescentus co-localize with the nucleoid, suggesting a tight coupling between transcription and translation.⁵⁸ Still, their reported localization for lacZ transcripts in E. coli differs from the pattern reported by the Broude group, which is described above. The different patterns of localization might be due to the different methodologies that have been applied and, thus, localization of lacZ mRNA in E. coli remains an open question.

Figure 2.

Localization of RNA in bacterial cell. Shown is schematic presentation of transcripts (green chains) distribution patterns in bacterial cells reported thus far. (A) Localization near sites of transcription: transcripts remain close to their transcription site on the chromosome with limited dispersion in the cytoplasm. (B) Cytoplasmic localization: transcripts are distributed in the cytoplasm in a helix-like pattern. (C) Membrane localization: transcripts are distributed around the cell circumference. (D) Polar localization: transcripts localize near the cell poles. (E) Septal localization: transcripts localize to the forming septum during cell division.

The second model, presented by our group, suggests that mRNAs may localize to subcellular domains in bacterial cells, where their protein products function. Using imaging of live and fixed E. coli cells, as well as amplification of specific mRNAs from the cytosolic and membrane cellular fractions, we observed three different RNA localization patterns: non-homogenous distribution that follows a helical path in the cytoplasm, discrete distribution along the cell membrane, and foci near the poles (Fig. 2B–2D).⁵⁹ Significantly, the localization patterns of the transcripts that have been monitored correlated with the respective localization of their protein products. Moreover, localization of some transcripts was shown to correlate with the requirements for complex formation. Thus, bglG mRNA, which encodes a transcription factor, was shown to localize to the membrane when co-transcribed with its membrane sensor, with which BglG forms a pre-complex near the membrane. However, when expressed on its own, bglG mRNA localizes to the cell poles, where BglG associates with the general PTS protein EI.⁵⁹ Importantly, not all E. coli transcripts that we monitored were shown to co-localize with their protein products, as exemplified by the ptsI mRNA, which codes for the polar EI protein.⁴⁵

The two models described above are not necessarily contradicting and might be both operating in the cell or, as recently suggested by Campos M and Jacobs-Wagner,²³ in different organisms. Moreover, repositioning of genetic loci encoding membrane proteins closer to the membrane upon their expression, as shown by Goulian and coworkers, supports both models.⁶⁰ The spatial shift of the genetic loci towards the membrane upon induction may act as a mechanism to localize membrane proteins via the intermediary mRNA molecule. Of note, this mechanism applies only for mRNAs that encode membrane proteins, since Goulian and coworkers did not observe a similar shift in a genetic locus that encodes a cytoplasmic protein upon its induction.⁶⁰

Other examples for bacterial transcripts whose localization corresponds with localization of their protein products have been reported.⁶¹ We will mention two recent reports here. First, Gueiros-Filho and coworkers observed an additional pattern of mRNA localization. Using the MS2 system, the authors detected the transcripts of comE, the late competence operon in B. subtilis, near the septa in midcell and the poles (Fig. 2E); localization of the comE transcripts has been shown to depend on DivIVA and ComN.⁶² Second, Broude and coworkers have just reported the development of new method for RNA labeling in live cells, which is based on combination of the protein complementation approach mentioned above with a split aptamer approach, i.e., sequence-specific binding of two RNA probes that are complementary to two adjacent sites on the RNA target.⁶³ Using this approach, they revealed distinct localization pattern of the pstC mRNA in live E. coli cells, which has been verified by FISH. Significantly, the pstC transcripts do not co-localize with the bulk DNA in both live and fixed cells, supporting the model by which the RNAs move away from their transcription site. Taken together, the correlation between the localization of transcripts and their encoded proteins suggests that localized translation occurs in bacteria.

Finally, there are many non-coding RNAs in bacteria, with many of them being trans-acting sRNAs that anneal to their target mRNAs and stabilize or degrade them.¹⁹ Since the sRNAs and their target mRNAs are often transcribed from remote locations on the chromosome, there should be a mechanism that brings them together. One option is that either the sRNA or the mRNA or both relocate from their site of synthesis, may be via the help of the Hfq chaperone, to make the sRNA-mediated regulation possible. The option that the regions on the chromosome that encode the two transcripts, which need to pair, are spatially close is less likely, in light of the many targets of some sRNAs, which are often scattered throughout the genome.

Translation-Independent Localization of RNA in Bacteria and RNA-Localizing cis Elements

In eukaryotes, the transcription and translation machineries are separated by a membrane and, hence, the two processes are regarded as uncoupled. Various cis-acting RNA localizing elements (RNA zip-codes) were shown to play an essential role in localizing eukaryotic mRNAs, most of them mapping to untranslated regions within the transcripts.⁶⁴ The features that characterize the zip-codes are largely unknown, but are thought to include sequence and structure motifs.^65,66

In prokaryotic cells, in which the nucleoid and the cytoplasm are not separated by a membrane, transcripts were not considered to localize to regions outside the nucleoid, unless brought there by translation-related complexes, such as the signal recognition particle (SRP) complex.⁶⁷ This concept is now being reevaluated, raising questions regarding the mechanisms that underlie RNA localization in bacteria: Do bacterial transcripts localize independently of translation? Do bacterial transcripts contain RNA zip-codes, which determine their subcellular localization, or features that are embedded in the N’-terminus of their encoded proteins determine their localization? If RNA zip-codes exist, what is their nature? The organization of bacterial genomes in operons with relatively short untranslated regions⁶⁸ suggests that RNA zip-codes, if exist, might be included within coding sequences. Below we describe examples for translation-independent localization of RNA in bacteria, present evidence for the existence of bacterial RNA zip-codes, mainly within open reading frames, and review suggestions regarding their nature.

An example for a translation-independent localization of a bacterial RNA by a cis-acting RNA localizing element was reported by Anderson and Schneewind, who studied secretion of YopQ, a virulent protein in Yersinia enterocolitica. They have shown that the 3′ UTR of yopQ mRNA is dispensable for YopQ secretion, but the first 45 nucleotides of the open reading frame of yopQ contain an RNA element which is responsible for YopQ secretion.⁶⁹ Other Yop proteins, including YopE and YopN, were also suggested to use an RNA element for their secretion.^70,71 The important role that the mRNA sequence plays in Yop proteins secretion was demonstrated by the lack of effect of frameshift mutations within the putative element sequence, which altered the protein sequence, compared with the dramatic effect of silent mutations that altered the RNA structure, on secretion.^72,73

Using various methods to uncouple transcription and translation, our group has shown that several transcripts in E. coli localize to the cytoplasm, membrane or poles in a translation-independent manner.⁵⁹ These results imply that the RNA molecules contain RNA zip-codes. Our study has further limited the sequence for membrane localization of the bgl operon transcripts to the sequence encoding the first two trans-membrane helices of the BglF protein. Similarly, the sequence required for polar localization of bglG mRNA has been narrowed down to the sequence encoding the first 50 amino acids that compose the RNA-binding domain of the BglG protein.⁵⁹

Recently, Bibi has suggested that, contrary to the prevailing hypothesis that bacterial transcripts encoding integral membrane proteins are targeted to the membrane together with the translating ribosomes via the SRP and its receptor, the transcripts and the ribosomes are targeted to the membrane separately.⁷⁴ According to this model, transcripts may localize to the membrane independent of translation, whereas the ribosome are targeted to the membrane by the SRP receptor, rather than by the SRP itself. Using a computational approach, Prilusky and Bibi suggested that localization of transcripts encoding membrane proteins to the membrane is due to their uracil-richness.⁷⁵ Their suggestion relies on their finding that both hydrophobic and hydrophilic amino acids that are overrepresented in membrane-spanning protein domains show a strong bias for uracil-rich codons. Research on bacterial RNA zip-codes has only begun; apparently our understanding of their nature will increase as more mRNAs displaying specific localization patterns are discovered.

Factors that may relocate RNA in Bacteria

Following transcription, the RNA molecules may undergo translation (for mRNAs), exert their activity (for sRNAs) or be degraded. In eukaryotes, it is well established that mRNAs either stay in the nucleus or move out to other cellular locations in a process termed mRNA targeting; subsequently the mRNAs are translated, stored or degraded. Eukaryotic RNA is localized by mechanisms, such as active transport, diffusion and capture and vectorial export,⁷⁶ with the majority of the transcripts being targeted by means of cytoskeletal elements.⁷⁷ Which factors in bacteria are candidates for relocating RNA molecules? Below, we review the types of bacterial factors that may be involved in this process and compare them to their eukaryotic counterparts, where applicable.

Cytoskeletal proteins

Bacteria possess various homologs of actin, tubulin, and intermediate filaments, which are involved in a plethora of processes. The major ones are described below.

Tubulin homologs

In eukaryotes, microtubules play a key role during mRNA transport, e.g., kinesin and dynein, which are involved in oskar and bicoid mRNA localization in Drosophila melanogaster, respectively.^78-80 FtsZ, a highly conserved cell division protein in bacteria, is a counterpart of eukaryotic tubulin.^81,82 Although FtsZ bears low level of sequence similarity to tubulin, the two proteins share significant structural similarity. FtsZ forms a ring-like structure at the division site in midcell, prior to septum formation, which recruits cell division proteins and enzymes involved in peptidoglycan synthesis (Fig. 3A). Notably, the ring-like structure contains only part of the total FtsZ content in the cell; the rest of the FtsZ molecules, which are found in the cytoplasm in a hypothetical helix-like manner,⁸³ could play a role in RNA localization.

Figure 3.

Localization of cytoskeletal elements in E. coli cells. (A) FtsZ, a bacterial tubulin homolog that plays a major role in cell division, forms a ring-like structure, termed Z-ring, at the cell midpoint; the rest of the FtsZ molecules are distributed in a pattern that follows a helical path in the cytoplasm. (B) MreB, an actin homolog that maintains the rod shape of bacterial cells, is distributed in a pattern that follows a helical path along the cell circumference.

BtubA/B and TubZ are additional tubulin homologs. BtubA/B in Prosthecobacter dejongei shares a higher sequence similarity with eukaryotic tubulin than FtsZ, but its function is not known.⁸⁴ TubZ, which is encoded by a Bacillus subtilis plasmid and shares sequence similarity with both eukaryotic tubulin and FtsZ, assembles into polymers that are required for the stability of the plasmid.⁸⁵ It displays a straight, helical filamentous structure along the length of the cell.⁸⁶

Actin homologs

ASH1 mRNA in yeast is a well-studied example of mRNA which is translocated along actin filament via myosin motor protein Myo4p.⁸⁷ Homologs of actin are the most diverse family among bacterial cytoskeletal proteins. In fact, there are as many as 35 actin homologs identified by phylogenetic analysis in bacteria.⁸⁸ MreB and ParM were the first bacterial actin-homologs to be identified and studied, but others have also been studied over the years.

MreB, an essential protein found in most rod-shaped bacteria, is required for cell shape maintenance.⁸⁹ It has been suggested to form a helical structure in the cell, similar to actin cables, although recent data suggest that it organizes into discrete patches that follow a helical path, rather than a continuous smooth filament.^90-92 (Fig. 3B) Another actin homolog is ParM, which is encoded from a plasmid and is involved in this plasmid partitioning in the E. coli host. ParM exhibits an unstable dynamic polymerizing behavior, very similar to eukaryotic microtubules.⁹³ MamK, another actin-like protein, forms filamentous structure and is required for the arrangement of the magnetosome, an organelle in Magnetospirillum magneticum.⁹⁴

RNA chaperones

RNA is a very unstable molecule, with comparatively short half-life, particularly in bacteria, and it has a high tendency to form intermolecular structures that can affect its rate of translation positively or negatively. Consequently, RNA molecules interact with various factors that participate in determining their fate, such as RNA chaperones, which assist in RNA folding and in maintaining RNA levels by acting at the level of transcription or translation. Some RNA chaperones have a broad-spectrum of RNA targets, among them Hfq, which plays a central role in matching sRNAs with their mRNA targets,⁹⁵ and cold-shock proteins, comprising of the CspA family proteins - CspA, CspB, CspG and CspI, which are known to act at low temperatures,⁹⁶ but some RNA chaperones have specific target mRNAs, like ProQ⁹⁷ or plasmid-encoded FinO.⁹⁸

The subcellular localization of RNA chaperones has not been extensively studied thus far. Still, using cellular electron microscopy imaging, Hfq, the bacterial Lsm-homolog, was shown to significantly concentrate at the cell periphery, in addition to its presence in the nucleoid and in the cytoplasm.⁹⁹ The reservoir of membrane-proximal Hfq molecules might play an important role in the translational control of mRNAs encoding membrane proteins, but also in regulating the fate of sRNAs and mRNAs in general.

RNA helicases

In eukaryotes, RNA helicases are essential for RNA-related processes from its biogenesis to degradation. DEAD-box RNA helicases were shown to play prominent function in mRNA transport in eukaryotes.¹⁰⁰ Importantly, E. coli contains many DEAD-box helicases, which were shown to function in mRNA processing and decay, as well as in ribosome biogenesis.¹⁰¹ It remains to be determined whether they also play a role in affecting the localization of bacterial transcripts.

Polarity determinants

Every cell that divides to two fairly identical daughter cells needs to have a mechanism that marks the point of midcell. Hence, cells have systems that are dedicated to establishing cell polarity and determining the position of division. The Min system in E. coli is a classic example for such a system. It consists of three proteins- MinC, MinD and MinE, which perform these tasks by inhibiting the formation of the FtsZ ring elsewhere in the cell, except for in midcell. MinC directly interacts with the FtsZ ring to prevent its formation. The membrane-binding MinD and MinE proteins restrict MinC activity to the zones near the cell poles,¹⁰² thus creating a bipolar gradient of MinC.^103,104 The entire process is dynamic, with the three proteins oscillating from pole to pole (Fig. 4A). Implications of the Min system in various cellular processes have been suggested over the years, and its possible involvement in affecting RNA distribution is appealing.

Figure 4.

Localization of polarity-determining factors in E. coli cells. (A) Schematic presentation of pole-to-pole oscillation of the Min system. MinD anchors to the membrane at the poles and binds to the FtsZ–inhibiting protein MinC. MinE forms a ring near the cell pole and prevents the MinCD complex formation near the cell center. The complex extends toward the cell center as the cell grows and oscillates from one pole to another in a dynamic way. (B) The anionic phospholipid cardiolipin represents the non-random distribution of lipids in the bacterial membrane. Cardiolipin is enriched at the bacterial cell poles (top) and also at the septum in dividing cells (bottom).

Another category of polarity determinants, which may very well be involved in RNA distribution, either directly or by affecting localization of RNA-binding/targeting proteins, is that of phospholipids. The bacterial cell membrane is not homogenous and contains domains that are enriched in a specific phospholipid, mainly cardiolipin (CL) or phosphatidylethanolamine (PE).¹⁰⁵ Cardiolipin, localizes mainly to the poles of rod shape bacteria (Fig. 4B) and has been shown to affect localization of certain polar proteins.¹⁰⁶

Conclusions

Only in recent years has it become known that RNA localization, once considered a phenomenon restricted to eukaryotic cells, is also relevant for prokaryotes. Since research in the field of bacterial RNA localization has only commenced, the extent of the phenomenon in different bacterial species is not known yet. The discovery that subcellular distribution of various bacterial mRNAs correlates with the distribution of their protein products, but is independent of translation, implies that these RNAs are specifically localized due to information engraved within them. The resemblance between the distribution patterns observed thus far for bacterial proteins and RNAs¹⁰⁷ suggests that the processes that underlie their localization, although not coupled, are interrelated. Taken together, the evidence that accumulated up until now suggests that localized translation occurs in bacteria. Bearing in mind the properties of RNA molecules - bulky and most often bound by proteins/complexes - and of the bacterial cytoplasm,⁴¹ one may anticipate mechanisms for RNA transport to exist. Considering the conservation of factors used for RNA targeting, e.g., cytoskeletal proteins, from bacteria to higher cells, some mechanisms employed by eukaryotic cells for this task may operate also in bacteria. Since the field of RNA targeting in general is relatively young, the discovery that bacterial RNAs are specifically localized provides an opportunity to improve our understanding of this phenomenon by studying it in the relatively simple bacterial cells. All in all, investigating RNA localization would also provide insights on the complexity of bacterial cells, as well as on regulation of processes associated with mRNA metabolism, and will most probably lead to identification of novel players in this field.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank members of Orna Amster-Choder's lab for fruitful discussions.

Funding

This work was supported by the Israel Science Foundation founded by the Israel Academy of Sciences and Humanities.