3.1. A Conserved Protein Fold between AGO, PIWI and Cas

The viral-derived RNase H-like protein superfamily is thought to have been a key driver of the evolution of complex life forms on the early earth, as it includes many essential enzymes in both eukaryotes and bacteria [88]. Members of the RNase H-like protein superfamily are mediators of many important processes including DNA replication, transposition, gene splicing, and the generation of adaptive immune responses [89]. The RNA interference (RNAi) pathways of eukaryotes and the CRISPR anti-viral defence of bacteria both rely on RNase H-like endonucleases. In RNAi, this is the Argonaute and PIWI-like proteins, and in CRISPR, the Cas proteins [89]. These enzymes all possess an RNase H fold, one of the most ancient and abundant protein folds known [90]. At a glance, these distinct pathways and mechanisms can initially appear to be completely unrelated. However, there are many shared features of these pathways.

3.2. CRISPR RNA and piRNA-Guided Immunity

One of the most striking similarities between eukaryotic and prokaryotic regulatory RNA pathways is the crRNA-guided CRISPR immunity of prokaryotes and piRNA-guided immunity in metazoans. The piRNAs and crRNAs of these pathways each represent a form of genome-encoded trans-generational adaptive immunity. In prokaryotes, CRISPR mediates defence against invasive nucleic acids from bacteriophage and plasmids [67]. In metazoans, piRNA-mediated immunity is the primary defence mechanism against transposable elements in the germline [91]. The majority of piRNAs are derived from piRNA clusters [92]. Similarly, crRNAs are also derived from distinct regions in the prokaryotic genome, the CRISPR loci [67]. In prokaryotes, short sequences of the invasive DNA are integrated into the CRISPR loci by the Cas endonuclease, separated by repeat sequences. In eukaryotes, both invasive RNA and DNA viruses can be integrated into piRNA clusters [93,94]. Both piRNAs and crRNAs are classes of regulatory RNAs that are complementary in sequence to the invasive nucleic acid, and that guide an endonuclease complex to the foreign target sequence [94]. In CRISPR, this is the Cas endonuclease that is guided by crRNAs, whereas piRNAs guide the PIWI endonuclease to the target sequence. In contrast to CRISPR, piRNAs can also direct histone modifications that mediate gene silencing [95]. These pathways are prime examples of the parallel functions and mechanisms of the distinct regulatory RNA pathways in eukaryotes and bacteria.

3.3. Shared Features of mRNA Decay in Eukaryotes and Bacteria

In eukaryotes, newly synthesised mRNAs are characterised by a 5′ 7-methylguanylate (m⁷G) cap and 3′ poly-adenylation [96]. General mRNA decay occurs due to decapping of the mRNAs and removal of the poly (A) tails in cytoplasmic Processing bodies (P-bodies) [97]. In bacteria, the degradosome is the key mediator of bulk RNA turnover, and newly synthesised bacterial mRNAs possess a 5′ triphosphate group that must be converted to a 5′ monophosphate group for efficient recognition by RNase E or RNase Y of the degradosome [98]. Interestingly, both miRNAs and siRNAs in eukaryotes are also characterised by 5′ monophosphate groups, and the enzyme responsible for this conversion in E. coli, RppH, is closely related to DCP2 of the decapping complex in eukaryotes [99,100,101]. The DCP2 decapping complex is also known to be involved in miRNA-mediated gene silencing in eukaryotes [102]. The 5′ monophosphate may enable the association of miRNAs and siRNAs with the nucleases of the bacterial degradosomes. Whether this shared evolutionary relationship between decapping enzymes could also enable miRNA-mediated regulation of bacterial gene expression through the degradosome, is another interesting prospect. A further example of this shared relationship between eukaryotic and bacterial RNA decay pathways is also demonstrated by Hfq of the trans-asRNA pathway, and its likeness to the eukaryotic Sm proteins.

3.4. Bacterial Hfq is Reminiscent of Eukaryotic Decapping Proteins

The function of Hfq is somewhat analogous to the RISC complex, as it facilitates interactions between trans-asRNAs and their target mRNAs [103]. Yet, it bears the most striking structural and functional similarities to the eukaryotic Sm and Sm-like proteins [104,105]. It contains the conserved Sm1 motif and, similar to the eukaryotic proteins, forms an oligomeric ring-shaped structure. Hfq facilitates the regulation of mRNA stability through interactions with trans-asRNAs, whereas in eukaryotes the Sm-like protein, Lsm1, is involved in mRNA decapping and degradation [106]. These conserved structural and functional components may enable trans-asRNAs of bacteria to interact with the mRNA decay machinery in eukaryotes. Alternatively, they may enable miRNAs that function through decapping enzymes to influence bacterial gene expression. Investigating the ability of eukaryotic regulatory RNAs to complex with Hfq or the bacterial degradosome, and of trans-asRNAs to interact with eukaryotic Sm-like proteins could reveal novel mechanisms of host-bacterial interactions.

5.1. Bacterial asRNAs Can Hijack Eukaryotic RNAi Pathways

Several studies have illustrated the presence of msRNAs in bacterial genomes using bioinformatic approaches over the last decade [125,126,127]. These bacterial-derived msRNAs can be delivered into host cells and incorporate into the eukaryotic host’s miRNA-processing machinery, thereby serving important functional roles for the bacterium [47,128]. For example, an msRNA derived from an RNA stem-loop in Mycobacterium marinum associates with eukaryotic RISC [128]. Furthermore, Salmonella can release RNA into the cytosol of infected IECs, and this RNA is processed by human AGO2, in a Dicer-independent fashion, into small functional msRNAs of 22 nucleotides in length. One of these miRNA-like molecules, Sal-1, serves as a critical virulence factor, facilitating the intracellular replication and survival of the bacteria [47]. These data demonstrate a role for bacterial-derived RNA molecules in the successful colonisation by human pathogens.

Similarly, it was observed that the periodontal pathogens Aggregatibacter actinomycetemcomitans, P. gingivalis, and Treponema denticola produce msRNAs in EVs that could enter eukaryotic fibroblast cells in vitro [129]. Exogenous transfection of some of the most abundantly expressed msRNAs by these bacteria led to a decrease in the anti-inflammatory cytokines IL-5 and IL-13 by Jurkat T cells [129]. These pathogens have been implicated in periodontitis, which is driven by the inflammatory response to the bacteria of the oral microbiota [130]. This work, therefore, highlights bacterial RNAs as a potential novel therapeutic target.

Additionally, trans-acting asRNAs OxyS and DsrA from E. coli, can impair C. elegans chemosensory behaviour and longevity through down-regulation of che-2 and the diacylglycerol lipase gene F42G9.6, respectively [35]. This is an RDE-4 dependent process, with RDE-4 being a key component of C. elegans RNAi, that processes dsRNA by complexing with Dicer and subsequently worm argonaute, RDE-1 [35,131]. In E. coli OxyS and DsrA function to co-ordinate the stress response through the regulation of the protein sigma factor RpoS in an Hfq-dependent manner [132]. Together, these studies show that a number of bacteria are known to produce RNAs that could mimic eukaryotic regulatory RNAs and feed into eukaryotic RNAi processing pathways. In addition, bacterial asRNAs can also have additional distinct functions in the eukaryotic host. Whether these relationships extend to host-microbiota interactions is an intriguing prospect that warrants further study. Investigating whether the production of RNA by members of the microbiota can modulate relationships with the host, and vice versa will be a key part of translating correlative microbiota studies into clinical benefits.

5.2. Proposal of Mechanisms for Regulatory RNAs in Inter-domain Communications

This review of the current literature leads us to propose two potential mechanisms by which regulatory RNAs could mediate bi-directional communications between eukaryotes and bacteria. (1) RNA molecules produced by one organism could integrate into one of the distinct regulatory RNA pathways present in the other. For example, the RNAs produced by Salmonella become functional upon processing by eukaryotic RNAi machinery, such as AGO2 and Dicer [47]. This is also true for E. coli, which exploits C. elegans RNAi machinery to regulate eukaryotic gene expression through its own bacterial trans-asRNAs [35] (2) RNAs are exported in EVs in conjunction with the molecules required for their functioning in the recipient (Figure 2). This is shown by Hfq which is exported in EVs by Yersinia pestis, and in an Acinetobacter baumannii Hfq mutant which displays reduced EV production [36,133]. Furthermore, eukaryotic Argonaute has also been detected in EVs [134]. These mechanisms could also occur in combination, with both host and recipient molecules important in inter-domain regulatory RNA-mediated communications.

Figure 2. RNA-mediated communications between the mammalian host and the bacteria of the gastrointestinal microbiota. (A) The bacteria of the mammalian intestinal microbiota can indirectly alter the host miRNA profile through the production of metabolites and activation of Toll-like Receptors (TLRs). It is known that bacteria can produce microRNA-size RNAs (msRNAs) in Extracellular Vesicles (EVs) that can integrate into host miRNA pathways [47]. Other RNAs such as trans-asRNAs could be exported in vesicles and could similarly integrate into host pathways. It is also possible that asRNAs could be exported along with their effectors in EVs (such as Hfq and RNase E) and could directly influence host gene expression. EV-independent export of asRNAs could be mediated by Hfq, as Hfq can create holes in the bacterial cell membrane [79]. (B) Plant exosome-like nanoparticles (ELNs) from diet and EVs from intestinal epithelial cells containing miRNAs can enter members of the microbiota and influence bacterial gene expression and growth [14,16]. These miRNAs could be exported with effectors such as Argonaute (AGO) to directly alter bacterial gene expression. Alternatively, eukaryotic miRNAs could become functional upon integration into existing bacterial asRNA regulatory pathways (Created with BioRender.com).