Although DNA of bacterial and viral origin, as well as viral RNA, have been intensively studied as triggers of innate immune responses, the stimulatory properties of bacterial RNA and its role during infections have just begun to be deciphered. Bacterial RNA is a strong inducer of type I IFN and NF-κB–dependent cytokines, and it also can activate the Nlrp3 inflammasome. In this review, we focus on the receptors and signaling pathways involved in innate immune activation by bacterial RNA and analyze the physiological relevance of bacterial RNA recognition during infections. Furthermore, we present the concept that RNA modifications can impair RNA-dependent immune activation. RNA modifications differ between eukaryotes and prokaryotes; thus, they can serve to define the innate pattern that is recognized. In this regard, we discuss the role of ribose 2′-O-methylation as a potential immune-escape mechanism.
The innate immune system serves an important function in the early recognition of invading pathogens and can be activated within minutes after pathogen encounter. Activation of innate immune cells is achieved by the recognition of highly conserved microbial structures, termed pathogen-associated molecular patterns (PAMPs). Recognition of PAMPs is mediated by a limited set of germline-encoded pattern recognition receptors (PRRs) that are present on the surface or in the cytoplasm of any innate immune cell. Important PRRs include TLRs, nucleotide-binding oligomerization domain–like receptors (NLRs), retinoic acid inducible gene I (RIG-I)–like receptors (RLRs), and C-type lectin-binding domain receptors. It is recognized that microbial nucleic acids, DNA and RNA, constitute an important group of PAMPs that, depending on their subcellular localization, are sensed by either endosomal or cytoplasmic receptors. Cytoplasmic nucleic acid recognition recently attracted much interest and has been reviewed intensively (1–3).
Activation of endosomal TLR3 and TLR7/TLR8 is triggered by viral dsRNA and ssRNA, respectively, whereas TLR9 recognizes unmethylated CpG motifs present in prokaryotic DNA (4–9). In addition, TLR7 and human TLR8 can be activated by synthetic small molecules, including imidazoquinolines and nucleoside analogs (7, 10). In contrast to the human system, TLR8 was suggested to be nonfunctional in mice because it cannot be activated by ssRNA or small molecule agonists (7, 10). Unresponsiveness to these stimuli has been attributed to the absence of a 5-aa motif in the ectodomain of murine TLR8 that is required for ligand recognition (11). However, combined stimulation of murine PBMCs with a TLR8-specific imidazoquinoline in the presence of DNA oligonucleotides resulted in stimulation, arguing against general unresponsiveness of murine TLR8 (12). Among the cytosolic RLRs, MDA5 was identified to initiate antiviral signaling in response to long stretches of viral dsRNA, whereas RIG-I is a sensor of short dsRNA or ssRNA with 5′−triphosphate termini (13–16). However, RIG-I also was implicated in the recognition of RNA from Listeria monocytogenes (17, 18). Recently discovered cyclic GMP-AMP synthase and absent in melanoma 2 are cytosolic receptors that recognize microbial DNA (19–24). Protein kinase PKR, which is known to contribute to viral RNA recognition, also was shown to be activated by bacterial RNA (25), indicating that some of the signaling pathways for viral and bacterial RNA may converge. Although the relevance of viral RNA and bacterial DNA recognition for the initiation of innate immune responses is now well established, bacterial RNA is a less well-studied activator of innate immune responses. The current knowledge on bacterial RNA–induced immune activation is discussed in this review.
Species- and cell-type–specific sensing of bacterial RNA
One of the first studies addressing the potential role of bacterial RNA as an immunostimulatory PAMP was performed by Koski et al. (26), who demonstrated that transfection of total RNA derived from bacterial, but not eukaryotic, sources induced high levels of IL-12 secretion in human monocyte-derived dendritic cell (MDDC) precursors. Similarly, both human PBMCs and murine bone marrow–derived macrophages (BMDMs) responded to intracellular delivery of total bacterial RNA with TNF production (27–30). In addition to inducing classical NF-κB–dependent cytokines, bacterial RNA has been identified as a strong trigger of IFN-α in plasmacytoid dendritic cells (pDCs) and of IFN-β in myeloid dendritic cells (mDCs) in an IRF-dependent manner (28, 30–34). Depending on the model system, the involvement of IRF1, IRF3, IRF5, or IRF7 has been implicated (30, 31, 33, 35).
Recognition of bacterial RNA is differentially regulated and cell-type specific. pDCs are professional producers of type I IFN and are equipped with a limited set of innate immune receptors, including TLR7 and TLR9. Experiments with FLT3L-generated murine pDCs demonstrated that IFN-α production upon bacterial RNA stimulation was abolished in TLR7-deficient pDCs (28, 34). Similar results were observed in human pDCs in which TLR7 expression was silenced by small interfering RNA (siRNA) (32). Thus, TLR7 in pDCs plays an important role in the detection of viral ssRNA, as well as in the sensing of bacterial RNA (Fig. 1). However, the role of TLR7 in the induction of type-I IFN in murine mDCs remains controversial. Mancuso et al. (30) reported that purified RNA derived from Streptococcus agalactiae (group B Streptococcus [GBS]) induced IFN-β in mouse mDCs in a TLR7-dependent manner. In contrast, Gratz et al. (33) proposed that DOTAP-packaged lysates from Streptococcus pyogenes induced IFN-β by an RNA-dependent, but TLR7-independent, mechanism. Similarly, production of the classical NF-κB–dependent cytokines TNF and IL-1β upon bacterial RNA stimulation did not require TLR7-mediated signaling in murine BMDMs and mDCs (27, 36), although these cells can be efficiently activated by TLR7 small molecule agonists, including imiquimod and R848. Potential redundancies in different nucleic acid–sensing receptors in recognition of bacterial RNA could be excluded by using cells with a combined deficiency for TLR3, TLR7, and TLR9 (36). TLR7-dependent production of type I and II IFNs also was described upon delivery of Borrelia burgdorferi RNA, yet an additional activation of NF-κB–dependent cytokines was observed (37).
Further analysis of the signaling pathways involved in sensing of bacterial RNA in murine BMDMs and dendritic cells (DCs) demonstrated a requirement for the common TLR adapter molecule MyD88, as well as for Unc93B1 (27, 36, 38, 39). Unc93B1 is a chaperone protein that is essential for the delivery of nucleic acid–sensing TLRs from the endoplasmic reticulum to the endolysosome (40). An H412R point mutation within Unc93B1 was described to disrupt signaling via TLR3, TLR7, and TLR9 and, therefore, was termed a “3d” mutation (40). Extending these findings, Brinkmann et al. (41) identified TLR13 as another receptor interacting with Unc93B1, rendering TLR13 a likely candidate receptor for bacterial RNA recognition. Indeed, TLR13 was subsequently identified as a sensor for bacterial 23S rRNA in murine macrophages and DCs using both TLR13 overexpression and siRNA-mediated gene silencing (38, 39). These findings were further strengthened by Li and Chen (38) using macrophages derived from TLR13-deficient mice generated in their laboratory. Remarkably, TLR13 is unique among innate immune receptors because it exclusively senses a specific nucleotide motif within bacterial 23S rRNA. This sequence was identified by Oldenburg et al. (39) as a highly conserved 5′-GACGGAAAGACC-3′ motif at the active site of the 23S rRNA. Similarly, Li et al. independently identified the 13-mer sequence 5′-ACGGAAAGACCCC-3′ as a minimal ligand for TLR13 (38). Of note, this sequence is also the binding site for macrolide, lincosamide, and streptogramin antibiotics in 23S rRNA. N6-methylation or N6-dimethylation within this sequence at a distinct adenosine residue (A2058, Escherichia coli numbering) by erm methylases (erythromycin ribosome methylase) or naturally occurring A2058G point mutations rendered bacteria resistant to macrolide, lincosamide, and streptogramin antibiotics (42, 43), as well as abolished activation of TLR13 (39) (Fig. 1). Thus, methylation-induced camouflage of 23S rRNA may serve as an immune-evasion mechanism in certain bacteria.
Importantly, TLR13 is not expressed in humans, and its human homolog receptor remains poorly defined. Early studies using HEK293 cells stably expressing human TLR3, TLR7, and TLR8 indicated that bacterial RNA can trigger IL-8 secretion by all three TLRs (29), and a potential contribution of TLR8 in transcriptional regulation of IFN-β production in human monocytes was suggested by Cervantes et al. (31). Yet, the current evidence for a role of TLR8 or other TLRs in bacterial RNA sensing in primary human immune cells is weak or, in the case of NF-κB–dependent cytokines, still lacking. Of note, a recent report identifying the molecular structure of TLR8 showed that this receptor recognizes degradation products of RNA, namely a uridine mononucleoside at one site and oligonucleotides like UG or UUG at a distinct second site (44). Thus, TLR8 might also sense bacterial RNA degradation products. The concept of recognition of RNA degradation products by TLR8 raises the hypothesis that phosphatases and/or nucleases of bacterial or host origin might play a role upstream of TLR8 activation (45), in analogy to the requirement for lysosomal endonuclease DNase II for the activation of TLR9 (46, 47).
Inflammasome activation by bacterial RNA
The inflammasome is a multiprotein complex that is critical for the activation of procaspase-1 into the enzymatically active form that subsequently processes pro–IL-1β and pro–IL-18 into the active cytokine (reviewed in Refs. 48, 49). Activation of the Nlrp3 inflammasome, named after the NLR involved, requires two signals: a priming signal (signal 1) provided by an activated TLR, NLR, or cytokine receptor, followed by a second signal that can be delivered in the form of ATP, pore-forming bacterial toxins, or particulate matter. Thus, the Nlrp3 inflammasome can be activated by a plethora of microbial, as well as endogenous, danger molecules. Caspase-1 activation also can be achieved by intracellular delivery of total bacterial RNA derived from a variety of Gram-positive and Gram-negative bacteria (36, 50–52). Although RIG-I–dependent inflammasome activation by viral RNA requires sensing of 5′-triphosphate termini (53), triphosphate moieties, as they can be found in bacterial mRNA, were dispensable for Nlrp3-dependent caspase-1 cleavage in response to bacterial RNA in both murine and human immune cells (35, 36, 54). Interestingly, intracellularly delivered bacterial RNA seems to serve as both signal 1 and signal 2 for inflammasome activation (Fig. 2). The priming signal for inflammasome activation appears to be mediated in an Unc93B1-TLR13–dependent manner (36, 38), although the possibility of a redundant involvement of other TLRs or of cytosolic receptors cannot be ruled out because of the lack of experimental evidence. Bacterial RNA that gains access to the cytosolic compartment was suggested to function as signal 2, allowing assembly of the inflammasome by an ill-defined mechanism (35, 55). In contrast, it was suggested that bacterial RNA may interact directly with Nlrp3 (56). Alternatively, the RNA helicase DHX33 could serve as an additional sensor for cytosolic bacterial, as well as viral, RNA and facilitate Nlrp3 inflammasome activation by binding to both dsRNA via the DHX33 helicase C domain and to Nlrp3 via DHX33-DEAD-Box and Nlrp3-NACHT domain interactions (57, 58). Moreover, ubiquitination of DHX33 by TRIM33 was suggested to be required for cytosolic RNA-induced Nlrp3 inflammasome activation (59). How binding of bacterial RNA to Nlrp3 and/or DHX33 triggers inflammasome assembly remains to be determined. PKR, a dsRNA-activated protein kinase, recently was implicated as a common regulator of inflammasome activation by E. coli RNA and polyinosinic-polycytidylic acid, as well as by a variety of other stimuli, including ATP, monosodium urate crystals, cytosolic DNA, or infection with E. coli and Salmonella (60). However, the general role of PKR for caspase-1 cleavage could not be confirmed in subsequent studies by other groups (61–63). The reason for these discrepancies remains elusive, and further studies are needed to clarify the contribution of PKR to inflammasome activation.
The concept of inflammasome activation by cytosolic bacterial RNA raised the question of how microbial RNA gain access to this compartment. In the case of phagocytosed bacteria, it was speculated that an intrinsic phagosomal leakage may cause the release of microbial RNA into the cytosol, even in the absence of bacterial invasion, secretion systems, or pore-forming toxins (35, 55). Interestingly, Sander et al. (35) observed that live, but not heat-killed, E. coli was able to induce caspase-1 activation and that the absence or presence of RNA within the bacterium correlated with the ability to activate the inflammasome. Indeed, bacterial RNA was rapidly degraded in dead bacteria, and the addition of purified bacterial RNA to heat-killed E. coli restored inflammasome activation. Among different E. coli RNA species, mRNA was the most potent inducer of pyroptosis and IL-1β secretion, whereas rRNA and small RNAs had minor or undetectable effects in murine immune cells (35). Thus, the investigators proposed that the presence of prokaryotic mRNA signifies microbial viability to the host, promoting innate and adaptive immune responses. Therefore, microbial RNA was referred to as “viability associated PAMP” or “vita-PAMP.” In contrast to the murine system, a recent publication indicates that the Nlrp3 inflammasome can be triggered efficiently by all tested bacterial RNA species (including ribosomal, messenger, transfer, and small RNA) in human monocyte-derived macrophages and THP-1 cells (54).
Significance of bacterial RNA recognition for innate immune responses against entire bacteria
Despite the increasing knowledge about bacterial RNA–mediated immune activation, little is known about the physiological relevance of RNA sensing for immune responses against entire bacteria. Because TLR13 was identified only recently as a sensor for bacterial RNA in mice, and because TLR13−/− mice are not yet generally available, most studies provide only indirect evidence for the significance of RNA sensing in antibacterial host defense. Two important points to consider are the release of RNA from bacteria and, thus, the mechanisms of how to get access to any innate immune receptor and second, a point that is often overlooked, the concentration necessary to trigger an immune response. From the studies using purified bacterial RNA, it seems obvious that bacterial RNA, in general, can activate innate immune cells, because Gram-positive and Gram-negative RNA preparations induced cytokine secretion (28, 30, 31, 36, 38, 39, 64). Yet the contribution of RNA to whole bacteria recognition has been less well analyzed.
The strongest evidence has yet been provided for streptococci, including S. pyogenes (group A streptococci) and S. agalactiae (GBS). In an attempt to identify the innate immune receptors involved in the sensing of S. pyogenes, two groups independently observed that production of NF-κB–dependent cytokines, including IL-6, TNF, and IL-12, was nearly abolished in murine MyD88-deficient mDCs and BMDMs, whereas single or combined deficiencies of TLR2, TLR4, or TLR9 did not attenuate cytokine production (65, 66). Similarly, IFN-β induction by S. pyogenes was fully and partially dependent on MyD88 in mDCs and BMDMs, respectively (33). These observations raised the hypothesis of either functional redundancy in TLR recognition or the existence of an unidentified MyD88-dependent receptor. Further analysis based on the transfection of RNase- or DNase-treated extracts of sonicated S. pyogenes demonstrated that bacterial RNA was critical for mounting IFN-β responses in murine mDCs, yet they were independent of TLR7. In contrast, S. pyogenes DNA was identified as an inducer of type I IFN in BMDMs (33). Together, these data raised the possibility that sensing of microbial RNA might be important for innate immune responses against S. pyogenes. Indeed, in a recent publication, Fieber et al. (67) provide evidence for a redundant involvement of TLR2 and TLR13 for TNF and IL-6 production by murine mDCs and BMDMs upon S. pyogenes infection in vitro. In contrast, TLR2 and Unc93B1 played a nonredundant role in vivo in a s.c. S. pyogenes infection model.
A role for bacterial RNA in innate immune activation also was demonstrated for S. agalactiae. Initial studies indicated that infection of murine macrophages and DCs with live S. agalactiae resulted in MyD88-dependent production of proinflammatory cytokines like TNF and IL-1β, but it did not require receptors classically involved in the sensing of Gram-positive bacteria (i.e., TLR2, TLR7, or TLR9) (68, 69). Dependency on bacterial internalization, phagosomal acidification, and Unc93B1 (27, 68) indicated a predominant role for the nucleic acid–sensing pathways in the recognition of S. agalactiae by an unidentified MyD88-dependent receptor. In light of the subsequent identification of TLR13 as a receptor for bacterial RNA in murine immune cells (39), Signorino et al. (64) investigated the immune responses induced by an erm methylase–overexpressing S. agalactiae strain in which TLR13 recognition is camouflaged. Remarkably, TLR13-mediated detection of 23S rRNA was required for in vitro cytokine induction in response to heat-killed bacteria, whereas no significant effect was observed when live bacteria were used as stimulus. However, in line with the observations of Deshmukh et al. (27), immune cell activation by live GBS was greatly diminished in Unc93B1 3d, but not in TLR7/8/9 triple-deficient, cells. Similarly, in an in vivo infection model, Unc93B1 mutation, but not erm-mediated masking of TLR13 recognition, increased lethality and bacterial burdens (64). Collectively, these data indicate that TLR13-dependent recognition of bacterial RNA plays a redundant role in the recognition of live S. agalactiae, likely due to functional compensation by other endosomal Unc93B1-dependent TLRs. However, it cannot be excluded that expression of erm methylases might be repressed during infection, resulting in residual TLR13 stimulation. Thus, it would be interesting to study the course of infection in TLR13-deficient mice.
Redundancy of bacterial RNA–mediated immune cell activation also was reported for Staphylococcus aureus. Heat-inactivated S. aureus strains carrying an erm-mediated methylation of 23S rRNA failed to induce IL-6 production in TLR2/3/4/7/9-deficient macrophages, whereas 23S rRNA methylation alone or a combined deficiency in TLR2/3/4/7/9 did not attenuate cytokine production (39). Accordingly, macrophages derived from mice with a combined defect in UNC93B1 and TLR2, but not those generated from UNC93B1 3d or TLR2−/− mice, were unresponsive to S. aureus challenge, indicating redundancy of TLR13- and TLR2-mediated effects.
The data obtained so far might indicate that bacterial RNA is an important trigger (dominant PAMP) within the genus Streptococcus and also contributes to immune recognition of staphylococci. Only a few Unc93B1-defective patients have been reported, so far with specific susceptibility only for HSV. However, infections with Streptococcus pneumoniae are observed in MyD88-defective children.
TLR8 also was implicated indirectly in bacterial susceptibility. Cervantes et al. (31, 70) showed that B. burgdorferi induced IFN-β in a TLR8-dependent manner and that whole bacteria recognition required TLR2 and TLR8 cooperativity. Along the same lines, B. burgdorferi extracts depleted enzymatically for nucleic acids showed dependency on TLR7 with respect to IFNs, yet they also induced NF-κB–dependent genes (37). Using murine DCs, Helicobacter pylori RNA induced proinflammatory cytokines (71), and phagocytosis of H. pylori activated TLR8 in human monocytes (72). Mycobacterium bovis bacillus Calmette-Guérin upregulated TLR8 in differentiated macrophages, and TLR8 gene variants were associated with susceptibility to pulmonary tuberculosis, arguing for a contribution of TLR8 signaling in mycobacteria infection (73).
Inhibition of bacterial RNA–mediated immune activation by RNA modifications
Because host and microbial nucleic acids are composed of the same four basic ribonucleotide building blocks, discrimination between self and nonself is a fundamental, but challenging, task. Selective recognition and discrimination between self and foreign RNA was suggested to rely on different components, including subcellular compartmentalization, sequence composition, secondary/tertiary structures, and specific nucleotide modifications (5, 74–76). In the case of RNA, >100 modifications at either the nucleobase or the ribose have been identified that are introduced posttranscriptionally. Importantly, the extent and kind of nucleotide modifications vary significantly depending on the RNA species and its evolutionary origin. In general, eukaryotic RNA is modified more abundantly than is prokaryotic RNA, with mammalian tRNA incorporating the greatest amount of modified nucleotides (up to 25% in eukaryotes). Although certain RNA modifications are shared between different kingdoms of life, a significant number of modifications are uniquely expressed in either bacteria or eukaryotes (77). Seminal work in the field of modification-dependent discrimination between self and nonself RNA was performed by the group of Drew Weissman. Analysis of the immunostimulatory capacity of different purified RNA subtypes revealed that immune activation inversely correlated with the number of modified nucleotides (i.e., highly modified eukaryotic RNA failed to induce TNF secretion in human MDDCs and E. coli tRNA, which is modified to a greater extent than other bacterial RNA species, induced lower levels of TNF compared with total E. coli RNA) (29). Remarkably, mitochondrial RNA was the only mammalian RNA species that potently induced innate immune activation. This observation is not surprising given that mitochondria are remnants of aerobic bacteria, according to the endosymbiont theory. As proof of principle for self/nonself discrimination by means of differential modification profiles, Karikó et al. (29) demonstrated that random incorporation of 5-methyluridine, 6-methyladenosine, 2-thiouridine, or pseudouridine into RNA in vitro transcripts attenuated their MDDC-stimulating activities. Similar inhibitory effects were observed upon 3′ polyadenylation, a classical feature of mammalian, but not prokaryotic, mRNA (26).
Although most studies addressing the impact of RNA modifications on innate immune activation were based on artificial oligoribonucleotides with randomly incorporated modifications, two recent studies (32, 34) investigated the role of naturally occurring tRNA modifications in their physiological sequence context. They demonstrated that most bacterial tRNAs can trigger TLR7-dependent production of IFN-α in human pDCs. In contrast, a few bacterial tRNA species, including E. coli tRNATyr, as well as total tRNA preparations derived from E. coli Nissle 1917 and Thermus thermophiles, failed to do so. Both groups independently identified a single 2′-O-methylation of guanosine at position 18 (Gm18) as the structural determinant necessary and sufficient for TLR7 silencing. Notably, Gm18 rendered tRNAs nonstimulatory and also acted as a TLR7 antagonist (i.e., it dominantly suppressed immune activation by otherwise stimulatory tRNAs). Accordingly, tRNA preparations from an E. coli trmH mutant that lacks 2′-O-methyltransferase activity induced higher levels of IFN-α than did the respective wild-type sample (32, 34). Binding studies suggested that 2′-O-methylated RNA binds with a higher affinity to TLR7 than its unmodified counterpart, thus disrupting its interaction with stimulatory RNA (78, 79). The principal concept of immune silencing by 2′-O-methylation has been appreciated for a while, and 2′-O-methylated nucleotides are now commonly incorporated in siRNAs to avoid unwanted immunostimulation (reviewed in Refs. 80, 81). Yet, the aforementioned studies were the first reports addressing the role of RNA methylation in naturally occurring bacterial RNA. Further analysis of the sequence context required for TLR7 silencing by E. coli tRNATyr indicated a functional DmR dinucleotide motif, with D reflecting all bases except cytosine and R reflecting all purine bases (82). The constraints of this motif stayed intact when it was transposed from its natural location within the D-loop into the anticodon loop or T-loop. Subsequent studies indicated that Gm18 also exerted dominant negative effects on the production of NF-κB–dependent cytokines, including IL-6 and TNF, in human monocytes, likely by interfering with TLR8-mediated signaling (79). In light of the recently described TLR8 structure, it became clear that TLR8 itself recognizes uridine and RNA-degradation products. Thus, the inhibitory effects of 2′-O-methylation could also be due to inhibition of unidentified RNases that produce the necessary degradation products.
However, despite the striking phenotype in human cells, 2′-O-methylated RNA did not interfere with TLR13 activation by bacterial 23S rRNA in murine immune cells, indicating the existence of strictly species-specific inhibitory patterns (79). The potent immunosuppressive effect of 2′-O-methylation within bacterial RNA raised the hypothesis that this modification might serve as an immune-evasion mechanism for certain bacteria, in analogy to the recent discoveries regarding antiviral immunity. Indeed, 2′-O-methylation of the viral mRNA cap impaired activation of MDA5 and mediated evasion of host restriction by IFIT members in response to corona and flavivirus, respectively (83, 84). Yet, experimental validation for the role of bacterial tRNA sensing and its masking by 2′-O-methylation in the context of stimulation by whole bacteria is still pending. Database analysis for putative trmH family tRNA/rRNA methyltransferases indicates the presence of trmH homologs in different Gram-negative and Gram-positive bacteria. Thus, in future experiments it will be interesting to determine whether the methylation status of bacterial RNA correlates with virulence and infection outcome in response to relevant pathogens.
Bacterial RNA is a potent trigger of innate immune cell activation. In murine and human pDCs, bacterial RNA–induced secretion of IFN-α occurs in a TLR7-dependent manner, whereas production of NK-κB–dependent cytokines like IL-6 and TNF in murine macrophages and DCs is mediated by TLR13. TLR13 is unique among PRRs because it exclusively senses a highly conserved region within bacterial 23S rRNA. However, TLR13 is not expressed in humans, and its human homolog receptor remains poorly defined. Moreover, cytosolic bacterial RNA can activate the Nlrp3 inflammasome, a multiprotein complex mediating the cleavage of pro–IL-1β into the biologically active cytokine. Limited evidence suggests that sensing of bacterial RNA might be of physiological relevance for host defense against group A and group B streptococci, although potential redundancies with other nucleic acid–sensing TLRs cannot be excluded.
Discrimination between self and nonself RNA by innate immune cells is achieved by differential posttranscriptional modification profiles of eukaryotic and prokaryotic RNA. 2′-O-methylation of ribose residues occurs regularly in eukaryotic RNA and is a well established antagonist of RNA-induced immune activation. Incorporation of this modification into E. coli tRNA might exploit this principle and serve as a potential immune-escape mechanism. The future use of RNA receptor–deficient cells using knockout mice or molecular tools in human cells will allow deciphering of the physiological importance of RNA recognition during bacterial infections.
This work was supported by Grant DA592/5 from the German Research Foundation (to A.H.D.).
Abbreviations used in this article:
bone marrow–derived macrophage
group B Streptococcus
2′-O-methylation of guanosine at position 18 in E. coli tRNA
myeloid dendritic cell
monocyte-derived dendritic cell
nucleotide-binding oligomerization domain–like receptor
pathogen-associated molecular pattern
plasmacytoid dendritic cell
pattern recognition receptor
retinoic acid inducible gene I
small interfering RNA.
The authors have no financial conflicts of interest.